The myth of market failure : employment and the labor market in Mexico

/333S Public Disclosure Authorized MA4,9f& TmEMYTH OF MAET FAILURE Public Disclosure Authorized Employment and the Labor n M arkMaret Mexico Public Disclosure Authorized Public Disclosure Authorized Peter Gregory A World Bank Research Publication  The Myth of Market Failure A World Bank Research Publication  The Myth of Market Failure Employment and the Labor Market in Mexico Peter Gregory Published for The World Bank THE JOHNS HOPKINS UNIVERSITYPRESS Baltimore and Loncon  Copyright C) 1986 by The International Bank for Reconstruction and Development / The World Bank 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing May 1986 The Johns Hopkins University Press Baltimore, Maryland 21211, U.S.A. The World Bank does not accept responsibility for the viewsexpressed herein,whichare thoseof the author and shouldnot be attributed to the World Bankor to its affiliatedorganizations.The findings,interpreta- tions, and conclusionsare the resultof researchsupported bythe Bank; they do not necessarilyrepresent officialpolicy of the Bank. Libraryof Congress Data Cataloging-in-Publication Gregory, Peter. The myth of market failure. "Published for the World Bank." Bibliography:p. Includes index. 1. Labor supply-Mexico. 2. Wages-Mexico. 3. Migration, Internal-Mexico. 4. Mexico-Emigration and immigration. S. United States-Emigration and immigration. 1.World Bank. 11.Title. HD5731.A6G74 1986 331.12'0972 85-45927 ISBN 0-8018-3343-4 Contents Preface vii 1. Introduction The Employment Problem in Mexico 2 Earlier Studies of the Mexican Labor Market 4 Labor Market Models 7 Overview 10 Notes 12 2. Evolution of the Labor Force, 1940-80 14 Size of the Labor Force 18 Sectoral Distribution 25 Growth in Tertiary Sector Employment 34 Summary 48 Appendix: Technical Notes on Labor Force Measurements 49 Notes 62 3. The Underutilization of Labor: Unemployment and Underemployment 66 Open Unemployment 66 Hidden or Discouraged Unemployment 77 Underemployment 81 Disguised Unemployment 94 Summary 96 Notes 98 4. Agriculture and the Labor Market 101 Land Distribution 102 Rural Workers 108 The Demand for Labor and the Evolution of Employment Conditions 114 Employment and Underutilization 122 Evaluation of the Findings 130 Summary 138 Notes 139 v Vi CONTENTS S. Internal Migration and Regional Development 144 Flows 146 The Sizeand Directionof Migratiot. Research Findingson Migration 151 Migrationand Regional IncomeDifferentials 164 Summary 165 Notes 166 6. Migration to the United States 170 Measuresof Stocksand Flowsof Migrants 171 of the MigrantFlow 185 Characteristics CausalFactorsin Migration 191 The Impactof Migrationon the MexicanLaborMarket 204 Summary 210 Notes 211 7. Wages in Mexico, 1935 to the Present 217 Historyof MinimumWages 217 The Courseof RealWages,1940-55 219 WageData fromEconomic Censuses,1960-75 231 Wagessince1975 239 Effectsof the MinimumWage 252 Summary 260 Notes 262 8. Conclusion 266 Review of LaborMarketConditions 267 The Economic Crisis 269 RecentDevelopments 275 Short-Term Prospects 276 Notes 279 Bibliography 281 Index 293 Preface THIS BOOK is the outgrowth of research supported by the World Bank. The Bank has had a longstanding interest in the performance of labor markets in the developing countries. Over the past ten years, it has sponsored a seriesof inquiries into various aspectsof labor market opera- tion that has contributed greatly to our fund of knowledge and under- standing of labor market phenomena. I feel fortunate to have been a participant in these inquiries for they have afforded me an opportunity to conduct research in greater depth than would have beenpossiblewith the resources available to most academic schoLars.When I undertook this study of the Mexican labor market, I had no intention of writing a book-length manuscript. Rather, I expected to produce a report that would trace with broad strokes the changes that had occurred in the conditions of employment from about 1940 to 1980. My previous work on Mexico had partially prepared me for this task, and I expected that I would be able to complete it in expeditious fashion. Since I had been advised to base my study largely on published primary and secondary sources of information, I foresaw an effort constrained by a relative paucity of useful and reliable information. I had long been aware of the assessmentof employmentconditions that prevailed in Mexico. It was one that stood in contradiction to a record of steady and relatively rapid growth. Employmentconditions were held to be "bad" and even deteriorating. Virtually all of the literature dealing directly with employmentwas painted with the same brush. In view of the unanimity of Mexican and foreign observers,I was inclined to give some credence to this view, even though I detected elementsin the picture that were not entirelyconsistentwith it. Therefore,,I welcomedthe opportunity to examine the Mexican experience more closely, for Mexico seemedto me to pose a critical challenge to the hopes and aspirations of many developing countries. For, if Mexico, with its high and sustained rate of growth over a forty-year period, has been unable to realize an improve- ment in the general working conditions of most of its labor force, what hope could other countries with lesserrates of growth entertain of improv- ing the lot of their people? vii Viii PREFACE As I proceeded with my inquiry, I became increasinglydubious of the correctness of the common characterizations of current conditions in the labor market and of their evolution over time. Since my findings have proved to stand in such complete opposition to the establishedview, I felt that I could receivea hearing for my heretical notions only if I developed the case for them as carefully and as fully as I could. The result of that effort has grown into this book. I do not pretend that it addressesall of the questions that are of interest to students of labor markets and employment conditions. Limitations of both time and relevant information led me to pass over additional topics of interest. Rather, I have tried to focus on some of the elements that are critical to an evaluation of market perfor- mance, some of which are frequentlycited as basesfor conclusionsthat are sharply different from mine. Where I have found a traditional view want- ing, I have attempted to make explicit the inadequaciesof its conceptual or methodological underpinningsas I perceivethem and then to set forth the evidence and reasoning leading to a different conclusion. This work would have been much poorer but for the help and en- couragement of many individuals and institutions. Without the material support of the World Bank, this study could not have been undertaken, and I gratefully acknowledgethat support. Of equal importance has been the encouragement and critical comments of the several staff economists of the Bank who attended seminars on my research or who reviewed earlier drafts of my manuscript and offeredmany valuable suggestions.In particular, I am indebted to Mark Leisersonand GuyPfeffermannfor their careful readings and thoughtful comments. Although the Bank's support was critical to this undertaking, it could not have been accomplished without the cooperation and generousassistanceof many individualsand institutions in Mexico. To all of them I extend my heartfelt thanks. A special note of thanks goesto Micha Gisser,my colleagueat the University of New Mexico, who, during countlesshours of joggingwith me, servedas a proving ground for many of the ideas and conclusionsthat appear in this book. I also thank Alice Scott and Patricia Gonzalez for typing various drafts of the manuscript. Finally, I wish to acknowledge the patience, understanding, and encouragement given me by my wife, Doris, and our children, Beate and Tom, during the seeminglyendlessprocess of produc- ing this manuscript. 1. Introduction THE DECADE OF THE 1970s witnessed the emergenceof a widespread concern over employment conditions in developingcountries. This de- velopment represented a significant shift in emphasis from the concerns that heretofore had dominated discussionsof the problems of economic development. In the period shortly following World War lI, economic developmentwas viewedprimarily as a problem of capital accumulation and growth of output. Implicit in this emphasiswas the assumption that employment was simply a function of the ra':e of capital accumulation since capital-labor ratios were assumedto be fixed in most of the planning modelsof the times. Thus the emphasisin developmentplanning rested on maximizing the rate of growth in the capital stock, and little explicit consideration was givento the implicationsof growth for employment.If anything, it was assumed that the developingcountries of today would reproduce the patterns of labor force structure that the now-developed countries had manifestedduring successive stages of their earlierdevelop- ment. That an employment "problem" might be a cause of concern followed the observation that the structure of employmentin developingcountries was evolvingalong a path that differedgreatly from that followedduring the industrialization of the developedcountries.In particular, it was noted that, in developingcountries, while accelerateddevelopmentyielded early gains in employmentin the "productive" sectors of industry and construc- tion, these were soon surpassed by a rapid expansion of employmentin the tertiary (service)sector. Bycontrast, today's industrializednations did not experience rapid growth in tertiary sector employmentuntil a much later stage in their development.' This unexpected growth was widely inter- preted as unfavorable for the quality of employment.The tertiary sector came to be viewed as the sector of last resort, a sponge capable of absorbing, via jobs of very low productivity, all those unable to secure preferred employment in the "productive sectors." In view of the low absorptive capacity of the modern industrial sector and the rapidly in- creasing labor forces of most developing countries, an air of alarm re- 1 2 INTRODUCTION placed the complacencyof the previous decades. Numerous comparative and country studies appeared, cataloguing the form and extent of under- utilization of labor forces in developingcountries. Invariably, the studies concluded that unemploymentand underemploymentaffected a substan- tial proportion of labor forces in these countries, and that development strategies could no longer afford to ignore the implications for employ- ment of alternative policy paths. This concern over employmentwas also heightenedby an increasinginterest in the social consequencesof develop- ment and in the distribution of income. The Employment Problem in Mexico During the 1970s, Mexico beganto mirror these same concerns.Several studies of employmentconditions were spawned, all of which concluded that the labor force sufferedfrom serious underutilization. Indeed, it was frequently stated that, not only were employment conditions currently unfavorable, but that they had actuallydeteriorated over time. During the regime of President L6pez Portillo, finding a solution to the problem of employment was elevated to matter of high priority, and a national commission, the Comisi6n Consultiva del Empleo, was constituted to analyze the problem and propose policy measures for its amelioration. In 1979, the commission reaffirmed estimates that had been arrived at by other groups: an earlier national commission,the Grupo de Estudio sobre el Problema del Empleo, by the Regional EmploymentProgram for Latin America and the Caribbean of the International Labour Office (PREALC), and by others. They agreed that approximately half of the labor force was either unemployed or underemployedand that the full-time equivalent of this underutilization amounted to an open unemployment rate approx- imating 25 percent of the labor force. The integration of the rapidly growing labor force and the backlogged stock of underemployed labor was viewedto be a herculean task that calledfor the maintenance of a high rate of overall growth as well as a seriesof policy measures calculated to increase the employability of labor. Fortified by a rapidly swelling volume of revenues from petroleum exports and international credits, Mexico embarked on an expansionary course designed to stimulate growth and accelerate the rate of job crea- tion. It was estimated that a growth rate of 7.5 percent was required to absorb increases in the labor force and to reverse the "deterioration in employment conditions" that accompanied the recession of 1976-77.2 Beginningin 1978, the economy responded with annual growth rates of  EMPLOYMENT PROBLEM IN MEXICO 3 more than 8 percent that were to extend over four years. In view of the large volume of underutilized labor supposedly overhanging the labor market, one might reasonably have expected employers to be faced by ample supplies of labor. Yet, this did not prove to be the case. After only two years of growth at a rate of only about 2 percentage points above the long-run average growth rate of Mexico, unrnistakable signs of labor shortages began to appear in many sectors and regions.Shortageswere not limited only to skilled personnel-which would have been understand- able-but extended to unskilledlabor as well. "Help Wanted" signs could be seen posted at construction sites, and women began to appear in substantial numbers in construction crews. Although the legal minimum wage had customarilynot been exceeded,by 1981 the wage for unskilled labor in construction was reported to have risen to one and one-half times the legal minimum in severalcities around the cDuntry.The managers and owners of small and medium-sizeestablishmen ts in manufacturingwhom I interviewedreported a ratio of only 2 to 3 applicants to every advertised vacancy of entry-leveljobs, not all of whom were considered qualifiedfor employment. The commercial agricultural sector, heavily dependent on seasonal labor, similarly reported difficultiesin recruitingadequate labor forces. The rapidity with which a condition of surpluis labor had becomeone of shortage greatly surprised most Mexican observers,includinggovernment officials. How could the common perceptiornhave been so far off the mark? Was the fault to be found in the weaknessof the empirical data on which these appraisals had been based? Were the underlying concepts adopted to characterize conditionsin the labor market faulty? Or did the quality of analysisthat had been applied to the problem leavesomethingto be desired? These questions have more than a passing interest to economic policymakers. For example, had perceptions of the nature of the employ- ment problem been different, would the L6pez Portillo government have embarked on such an ambitious expansionary program that has proved, in retrospect, to have had disastrous consequencesfor Mexico? I have concluded that the misreadingof the nature of the employment problem derived not only from conceptual shortcomingsbut also from an incomplete or partial analysis,that is, a failure to include in the analysis more of the variablesthat are essentialto an accurate appreciationof labor market operation. To be sure, labor market information is relatively scarce, particularly in a consistent form over time; data are scattered and often of dubious quality. But had labor market analyses been more sys- tematic, had they been guided by a clear conceptualizationof the market relationships that must hold if an internally consistent construct is to 4 INTRODUCTION emerge, 1 believe that a very different perception of market conditions would have emerged. Earlier Studies of the Mexican Labor Market Throughout this book, I will referto the conclusionsof earlier studies of the Mexican labor market and will comment on the conceptual and empirical bases of those conclusionsin some detail. As a backdrop to my study, however, it would be useful to review briefly the nature of those appraisals as well as some of their salient features. By and large, Mexican studies of labor market conditions have focused on those prevailing at a moment in time, for example, as revealed by a population census. There have been intercensal comparisons made of measures such as labor force participation rates, unemployment, or the sectoral distribution of the labor force. Most of the analysis is static, however, and fails to address the determinants of change over time and thus does not explore fullythe significanceof observable changes. Conse- quently, one misses any sense of the dynamics of labor market operation and of the significantchangesthat have occurred in the welfareof the labor force over time. For example, it should come as no surprise to anyonethat the employmentconditions of a substantial proportion of the labor force are deplorable and that the income position of many households is pre- carious. Nevertheless, is the observation of such a state at a moment in time a sufficientbasis for concludingthat conditionsin labor markets are deteriorating and that unemploymentand underemploymentare increas- ing in gravity?' Ordinarily, evidenceof deterioration requires the compari- son of observations at two points in time at the least in order to establish a trend, yet one looks in vain for a careful empirical documentation of trends that are supportive of such a strong and pessimisticconclusion. The observations and their interpretation that underlie the pessimistic appraisal of Mexican labor market conditions before the economiccrisis of 1982 and the recession that followed echo those that had become commonplacein much of the literature of the 1960s and early 1970s on employment conditions in developing countries. Most worrisome to observers was the "rapid" expansion of tertiary sector employment, a growth that was viewed as premature and indicative of a failure of the development process to create a sufficient number of preferred jobs at acceptablyhigh levelsof productivity. In the absenceof significantbarriers to entry, the tertiary sector could absorb unlimited quantities of labor in low-paying, low-productivity jobs, which would often be of a self- employednature. Suchactivitywas usuallyinterpreted as seriouslyunder-  STUDIES OF THE MEXICAN LABOR MARKET S employing workers or as representing little mnorethan disguised unem- ployment. 4 In more recent treatments of this theme, references to the absorptive role of this sector commonlyuse th e word "informal" in place of "tertiary."5 Associated with the growth of tertiary sector employment was an accelerationof rural-urban migration. Rural poverty and limitedland and employmentopportunities were thought to be pushing the rural popula- tion toward the cities where, in view of the scarcityof job opportunities in the modern industrial sector, workers were forced to crowd into the tertiary or informal sector on unfavorable terms. 6 Migration came to be viewed as a process that simply transformed rural poverty into urban poverty. Hopes for increasingincome by migratingwere held to have been largely dashed. 7 A consequenceof the rapid growth of the urban popula- tion under such unfavorable conditionshas been the increasing "margin- alization" of the population, groups divorced from the mainstream of economic, cultural, and political life.' By and large, this appreciation of employment conditions in Mexico has never been subjected to rigorous testing, although the conclusions drawn havregiven rise to empirically testable propositions. For example, consider the conclusion that the growth of tertiary sector employment has been "excessive"; rather than being seen as a response to demand, this growth is believed to be a consequenceof an increasein the supply of labor. If, indeed,the surge has been supply determined, predictable consequences flowing from such growth should be observable. If the number of available workers is in- creasing more rapidly than is the quantity of labor demanded at the prevailingwage, then rates of earnings could be expectedto show declines over time. If wages in the tertiary sector were above the subsistencelevel before the onset of acceleratedgrowth in employment,these should have fallen toward the subsistence level. If wages had been at the subsis- tence level originally, then they could be expected to hover around the subsistence level, with declines below that level carrying with them in- creases in mortality rates, as a weakened and undernourished population succumbed to disease.Does evidenceof such a deterioration in remunera- tion levels exist? If not, on what basis can it be argued that the growth of tertiary employmenthas been "excessive"?UJnfortunately, any reference to empirical evidence of the secular course of wages is absent in these discussions. Similar questions can be posed about the rate of rural-urban migration. What is an appropriate test of whether or not migrationhas beenexcessive and has increasedpoverty?From the point of viewof the private welfareof migrants, migration becomesexcessiveif it is associated with a declinein income that is not expected to be offset by future increases. Stated more 6 INTRODUCTION precisely,migration might be deemed to be excessiveif the present valueof the expectedfuture stream of earningsand social servicesin the city (netof the costs of migration) is less than that of future earnings and social services in the countryside. Were this to describe accurately the fate of migrants, and migration were observed to continue at a high level, one would be led to the implausibleconclusion that migrants are irrational in opting for an inferior material standard of living. If it is difficult to conclude that migrants are irrational, then the premise that migration leads to impoverishment should be reexamined. Certainly,what happens to migrants' incomes is an empiricallyverifiablephenomenon as is their perception of the change in their material welfare. Indeed, in Mexico substantial literature exists on empirical studies of migration and its economic consequences for the migrants. What is surprising is that the findingsof such studies should be ignored in the various diagnosticstudies of the employment problem of Mexico. As indicated earlier, the conceptual and measurement aspectsof under- utilization of the labor force are open to question. Leavingthese aspects aside for the moment, let us grant that underutilization has amounted to the equivalent of a 25 percent open unemployment rate, as has been concluded by the studies cited earlier. A further question still remains, however, that is particularly relevant for policymaking. What is the sig- nificance of this unemployment rate equivalent for the operation of the labor market? Is it correct to assume that it represents a supply of labor effectivelyavailable to certain expanding economicsectors, and, if so, at what wage rate? Essentially,this question addresses the elasticity of sup- ply. Would employers in a rapidly expanding economy be faced with a perfectly elastic supply of labor as long as this "surplus" survived?While prediction is always difficult in economics,some helpful inferencesmight be derived from past experience. What has been the course of market- determined wages in the expandingemploymentsectors in the past? If real wages were found to have been rising coincident with annual rates of growth in the grossdomesticproduct on the order of 6 percent, is there any reason to believethat a more rapid rate of growth could be accommodated without an evenfaster rate of increasein wages?Indeed,if past increasesin employmentwere accompaniedby increasesin market-determinedwages, what operational meaningdoes a labor surplus of 2S percent of the labor force really have? By contrast, if expansion of employmentin the past has taken place at constant or evenfallingwages, this would support the view that the supply schedule of labor has been highly elastic and that an accelerated rate of growth could be accommodatedwithout serious wage inflation. Again, however, this is an empirical question that can be re- solved by reference to the past behavior of employment and wages.  LABDR MARKET MODELS 7 Unfortunately, none of the diagnoses of Mexico's employmentproblem includes such a component. Labor Market Models The analysis and interpretation of labor marketphenomena are likely to be most fruitful if they depart from a hypothesizedstatement of market relationships that can be empiricallytested. Essentially,what a student of the Mexican labor market would like to establish is a model that most closely describes how that market functions At least three models have beenoffered in the literature as descriptiveof labor markets in developing countries. These include the neoclassicalancldual labor market models; the latter is divided into those positing competitivemarket conditions in all sectors and those postulating the existenceof barriers to entry and of institutional interventionin at least one segmentof the labor market. Each model identifies the variables relevant to its operation and yields an observable outcome that flows from its underlying premises.Students of labor markets are thus provided with guidancein what to look for in their efforts to characterize any particular market. Let us consider these models and indicate how they might serve as useful guides to market analysis. The neoclassicalmodel provides the simplest descriptivestatement of market operation. Competitive conditions prevail in all segmentsof the labor market, and mobility of labor is unhindered and servesan equilibrat- ing function. In such a market, labor is fullyemployed, and open unem- ployment takes on the character of frictional unemployment. The exis- tence of wage differentials in different parts of the labor market for a particular class of labor would be expected to give rise to movement of labor from low- to higher-wageemployments,and, over time, differentials could be expected to shrink until they reflected only the monetary and nonmonetary costs of moving. An alternative scenario would be that, in order for an expanding sector to attract additional supplies of labor, a higher wage would have to be offered, whichwould open up a differential in the wages paid among sectors and stimulate mobility.In a dynamicand growing economy, equilibrium would be unlikely. Rather, the growth process would be characterized by successive movements away fom equilibrium as new focuses of growth emerged and by responses on the supply side that moved the market back toward a state of equilibrium.The empirical testing of such a model would focus on labor mobility and its correlates and on the course of wages in diFferentsegmentsof the labor market. Rapidly expandingsectorswould be expectedto face labor supply schedulesthat were upward sloping. 8 INTRODUCTION Another model of the labor market is offered by those who postulate the existence of a labor surplus in the rural sector that could be drawn upon for the expansion of the urban or industrial sectors.9 Such models posit the existence of a dual labor market. In the rural sector, the marginal product of labor is negligible or zero; labor is compensated at a rate equal to its average product. As long as the condition of surplus labor survives in the rural sector, the supply of unskilled labor to the expanding urban sector will be perfectly elastic at a wage equal to the rural average product of labor plus a differential that compensates movers for the economic and noneconomic costs of movement. Only when the surplus has been absorbed, that is, when the marginal product of rural labor equals that of urban labor, will wages begin to rise in both sectors as the process of urban 10 or industrial expansion continues. The relevance of this model to any particular labor market can also be tested by focusing on the course of productivity of low-skilled labor in the rural and urban sectors and on the course of wages over time in the expanding urban sector. If the economy is characterized by the existence of surplus labor, one would expect to find the expansion of low-skilled employment in the urban sector to have occurred at a stable level of wages over time. Urban open unemployment would be expected to be negligible or of the frictional variety. If, at some point in time, the labor surplus has finally been absorbed, subsequent expansion of the urban sector would be expected to become possible only at successively higher wages, with wages in the rural sector following. At such a point, the surplus labor model ceases to serve as the appropriate model of the labor market and is supplanted instead by the neoclassical. A third model, formulated by Michael Todaro, is a variant of the surplus labor model described aove.1 ' The distinguishing feature of this model is the division of the urban labor market into two segments. The modern or industrial segment is noncompetitive, with wages established by institutional intervention (legal minimum wages, collective bargaining, and so on) at levels far above the opportunity cost of labor. Barriers to entry limit access to these preferred employments. Although the probabili- ties of immediate absorption in the high-wage sector are very low, the relative attractiveness of employment conditions in this sector nevertheless spurs rural-urban migration. The result is the formation of an urban queue as migrants wait for opportunities to materialize in the sectors of preferred employment; successful entry is a function of the length of time spent in the queue. The rate of migration to the urban sector will be governed by the relation between the present value of the expected streams of future net income from urban and rural employment. In turn, the present value of the future income stream that can be derived from urban employment will be heavily influenced by the probabilities assigned to  LABOR MARKET MODELS 9 successful entry to a preferred employment in successivetime periods. Obviously, the shorter the urban queue, the more is that present value enhanced. Thus the equilibrating variable that regulatesthe flow of mi- grants is the length of the queue, for it is this tl at brings about the equality in the present values of the two streams of urban and rural future earnings. Those who are in place in the urban queue are not necessarilycompletely idle, although open urban unemploymentmay reasonably be expected to be high and durable. They may engagein activitieswith very low produc- tivity and at very low wages. Indeed,they may be willing to accept wages below their opportunity cost in the short run in the expectationthat these will be fullycompensatedby the high wagestFat will be forthcomingonce accession to the high-wage sector has been rzalized. The applicabilityof this model can likewisebe tested by referenceto the experience of migrants and the course of wage differentials within the urban market and between the informal, or zompetitive,segment of the urban labor market and the rural. Migrants to urban areas would be expected to experience great difficultiesin findinggainful employment.In addition, high rates of unemploymentwoulclbe found among unskilled workers in urban areas, and where wages are market determined they would appear to be depressedto levels approximatingor evenbelow those in rural occupations. The urban wage differzntialbetween the two seg- ments would be very wide and would show co tendencytoward narrow- ing until the rural surplus and the urban queue had been fully absorbed. The views of the Mexican labor market that prevailed before 1980 would appear to approximate most closely the conditions posed by this third model. Rural-urban migration was seen as "excessive" and as sur- passing the rate of job creation in the formal sector. Barriersto entry to the preferred sectors of employment forcedworkers to seek jobs in the infor- mal sector at market-determinedwagesthat were inferior and stagnating. In these characterizations, the labor supply schedule to the urban labor market seemsto have beenviewedas perfectlyelastic.Also consistentwith the third model is the belief that open unemploymentand underemploy- ment have been increasing over time. While one can frequently offer particular observations that seem to be consistent with a surplus labor model, it is not clear that the weight of the empiricalevidencewould lead to the conclusion that such a model represents the closest approximation of reality in Mexico. But whether it does or does not is empirically verifiable.In my opinion, it is not possible to test adequately the appro- priateness of any of the models discussed on the basis of information pertaining to a moment in time. Since neither the Mexican economy nor the models themselvesare static in nature, the performance of the labor market cannot be fully appreciated without an examination of its opera- 10 INTRODUCTION tion and responsivenessto dynamic elements in the economic environ- ment. Overview In undertaking this study of the Mexican labor market, I therefore chose to adopt as a frame of reference a time interval of forty years, 1940-80. Over that time span, significantchanges occurred in the struc- ture and productivity of the Mexican economy that were paralleled by changes in the occupational structure and sectoral distributionof the labor force. It is the latter set of changes that forms the central concern of this study. Not only are these developments scrutinized, but an attempt is made to identifythe forcesthat guidedthem and to evaluatethe effectthey have had on the material welfare of the labor force. At the conclusion of this exercise,the labor market model that most closely fits the Mexican experience should be discernible.These findingscould help to clarify the "employment problem" that Mexico faced in the years up through the 1970s and the relevance of this characterization for public policy. Although recent labor market conditions have changed radically from those prevailing before 1982, this report might be helpful in suggesting how labor market information could be organizedand analyzedfruitfully. The work is organized in the followingmanner. Chapter 2 charts with broad strokes the structural changes that have occurred in the Mexican economy over time as well as the changesin the sizeand distribution of the labor force. Particular attention is devotedto the growth of tertiary sector employment and its correlates. An initial effort is made to determine whether its growth has been "excessive." Chapter 3 examines the issue of the underutilization of the labor force and the conceptual and measurement problems associated with it. The question of whether the rate of underutilization of the labor force had been increasing before 1980 is also addressed. Then follows an overviewof the agricultural sector and its role as a source of labor for the rest of the economy. Essentialto an understanding of this role is a knowledge of agricultural productivity, the trends in the distribution of land among the rural population, and the links between rural and urban labor markets. These factors are explored in Chapter 4. As was indicated in the preceding brief review of labor market models, mobility of labor plays a pivotal role in each. The characterization of the flowof labor among differentsectors or markets is vital to comprehending how the labor market has functioned.Chapter 5 is thereforedevotedto an exploration of internal migration, largely rural-urban in character.  OVERVIEW 11 The following chapter extends the area of rnigration to include move- ment of labor across the border to the United States. Such migration has implications for the sizeof the labor force that must be accommodated in Mexico and may significantlyaffect the conditi,ons of employmentthere. The size of the migrant flow, however, has been a subject of intense debate, with various parties offering widely different estimates of the number of temporary migrants in the United States at a moment of time and of the number that "leak" into the U.S. labor force on a prolonged or permanent basis. In Chapter 6, 1 review and evaluate the available in- formation on this phenomenon. One of the key variablesin understandingthc character and functioning of labor markets is wages. Each of the models reviewed above posits a certain wage behavior for low-skilllabor that fcllows from the underlying assumptions about supply and demand conditions and the institutional characteristicsof labor markets. Chapter 7 brings togetherwage informa- tion from several sourcesto chart their course over time. In addition, the probable effect of institutional intervention on the course of wages is assessed. The final chapter provides an integrated summaryof the major findings of this study. It also compares the performance of the Mexican labor market with the outcomes predicted by the three labor market models outlined earlier in this introduction in order to determine which model offers the closest approximation of the Mexican experience. Some observations are offered regarding the relevanceof my findings for em- ployment policy. The chapter concludes wil:h a brief review of recent developmentsassociatedwith the economiccr:isis that erupted in 1982 and their implications for employmentand real wage levels. For the most part, the research on which this survey of the Mexican experienceis based has reliedon information culledfrom officialstatistical sources and from research reported by other scholars. Unfortunately, many of the basic statistical materials pose a seriouschallengeto anyone who wishes to use them. There are few data sources that provide con- sistency over time in the definitionof variablesor in coverage,and in the quality or reliability of the data. Even the population censuses, which mightbe expectedto fulfillthese conditionsmost closely,suffer from grave deficiencies.For some variables, information is availableonly for selected points in time. Thus, in order to fillin gaps,I havefrequently had to resort to scattered information, frequently of a less than rigorous quality, and culled from a variety of sources. The difficultiesposed by various data sourceswill be outlined in the text so that the reader will able to form an opinion on their significance.A more detailed description of data prob- lems encountered in the measurement and sectoral distribution of the 12 INTRODUCTION labor force has been relegatedto an appendix to Chapter 2. The product of this investigation can best be described as a mosaic, a panorama pieced together from a wide variety of sources.Just as a mosaic must be charac- terized by form and order if it is to convey successfullya recognizable image, so must the outcome of this study present an internally consistent account of labor market relations over time if it is to be judgedsuccessful. Although there are gaps remainingin the mosaic,by and large the findings of this report provide a reasonably consistent and convincingaccount of the evolution of employment conditionsin a substantial part of the Mex- ican labor market. Notes 1. Colin Clark, Conditionsof EconomicProgress,3d ed. (London: MacMillan1957), chap. 9. 2. Leopoldo SolisM., "Reflexionessobre el panaorama generalde la economiamexica- na," in El sistema econ6micomexicano (Mexico,D.F.: PremiaEditora, 1982),p. 348; see also JesfisReyesHeroles G. G., "Hacia una politica integralde ingresos,salarios, empleoe inflaci6n," in the same volume, pp. 250-76. 3. Comisi6n Consultiva de Empleo, Programanacional de empleo 1980-82: S(ntesis (Mexico,D.F.: Secretariadel Trabajo y Previsi6nSocial, 1979), p. 25. 4. Saul Trejo R., "El desempleoen Mexico: characteristicasgenerales," Comercio ex- terior,vol. 24, no. 7 (July 1974), p. 732. S. Thisis a broader conceptthat includesactivitiesin other sectorssimilarin characterto those formerlyascribedto only the tertiarysector. Informalsector jobs are characterizedby governedby legal employ- small-scaleproductive or serviceactivitiesthat are not effectively ment provisions. Handicraft production, many self-employedand unremunerated family workers, domestic servants, and petty commerce would make up the bulk of the urban informal sector. The rural informal sector comprises "small" or "traditional" farm units. Typically, the informal sector is viewed as having very limited capital resources and as suffering from low productivity and earnings. 6. Comisi6n Consultiva de Empleo, Programa nacional de empleo, pp.16-17,27; Grupo de Estudio del Problema del Empleo, "El problema ocupacional en Mexico: magnitud y recomendaciones," versi6npreliminar (Mexico, D.F., n.d.), pp. 143-51; Donald B. Keesing, "Employment and Lack of Employment in Mexico, 1900-70," in Quantitative Latin Amer- ican Studies: Methods and Findings, ed. James W. Wilkie and Kenneth Ruddle (Los Angeles: University of California, L.A., Latin American Center, 1977), pp. 3 - 2 1 . 7. Saul Trejo R., "Desempleo y subocupaci6n en Mexico," Comercio exterior, vol. 22, no. 5 (May 1972), p. 413. 8. Rodolfo Stavenhagen, "Marginalidad, participaci6n y estructura agraria en America Latina," Demograffa y economia, vol. 4, no. 3 (1970), pp. 267-91. 9. W. Arthur Lewis, "Development with Unlimited Supplies of Labour," Manchester School of Economicand Social Studies, vol. 22 (May 1954), pp. 139-91;John C. H. Fei and Gustav Ranis, Development of the Labor Surplus Economy: Theory and Policy (Home- wood, Ill.: Irwin, 1964). 10. Stated more precisely, this constancy in the supply price of labor to the urban sector  NOTES 13 depends on the constancyof the incomeper worker in agriculture.If the averageproduct of labor rises as mobilitydraws down the quantity of labor applied to rural production and if this increaseis not siphoned away by taxes or rent paym,nts to a land-owningclass,rural labor income will rise, thus raising the supply price of labor to the urban sector before equality is achievedin the marginal products of labor in the two sectors. 11. Michael P. Todaro, "A Model of Labor Migration and Urban Unemploymentin Less Developed Countries," American Economic Revieiv, vol. S9, no. 1 (March 1969), pp. 138-48. 2. Evolution of the Labor Force, 1940-80 THE PER IO D OF ACCELERATEDgrowth and modernizationin Mexico dates from about 1940. The groundwork had already been laid by the sweeping changes flowingfrom the precedingthirty yearsof revolutionary activity, which produced far-reachingsocial and institutional upheavals. The onset of World War 11brought with it a disruption of traditional sources of supply for industrial goods and stimulated the expansion of manufacturing production for internal consumption.The large-scaleland distribution program of the Cardenas regime during the precedingdecade led to a substantial increase in the land under cultivation and in agricul- tural output. Thus, the decade saw all of the principal sectors of the economy growing in tandem fashion.As is evidencedin table 2-1, the rates of growth of the three broad sectors showed only small differences. Following the end of the war and a return to more competitivecondi- tions in product markets, the governmentopted for a policy of fostering the further growth of domesticindustry behind protectivetariff walls and import controls. It also embarked on a large-scaleinvestmentprogram to provide infrastructure for an emerging modern commercialagricultural sector. Major irrigation and road-buildingprojects in the sparselypopu- lated north and northwestern provincesstimulated continued expansion of agricultural output for both domestic consumption and for export. With the exhaustion of "easy" import substitution, incentives were adopted for domesticproduction of consumerdurable goods and then for intermediate and capital goods.Throughout most of the past four decades, Mexico has succeededin maintaining a high and stable rate of growth, one that faltered significantlyonly during the mid-yearsof the past decade. Although the achievementsof the Mexican economy are surelyimpres- sive, there remain serious areas of concern. A rapidly growingpopulation of labor-force age poses a formidablechallengeto the economy's abilityto absorb it in productive employment.The pattern of development,particu- larly in agriculture, has been uneven. While a highlyproductive commer- cial agricultural sector has developed,the bulk of the rural population is to 14  EVOLUTION OF THE LABOR FORCE 1s be found in areas with scarce land and water resources,where increasesin productivity have been modest. Rural poverty remains a seriousproblem; some students of Mexico hold that the incom.-sof a sizablerural popula- tion have not shared in the generalincreasethat characterizesthe country as a whole, or have shared only to an insignificantextent. The relative distribution of income continuesto rank amongthe most unequal in Latin America and appears to be emergingas an acute social and political issue. The diverse character of agriculture is indicated by the large differences between the productivity of land and labor among the three classes of agricultural holdings. In 1970, the output per hectare on farm units employing "modern technology" was four times that of subsistencefarm units (thosein which only a smallpart of output was for market). The ratio of output per worker was even greater, on the order of 20:1.1 These differenceswithin the sector, moreover,have strong regionalassociations. In 1960, for example, the output per bectare was 20 times greater in Sonora than in Quintana Roo. In twelve other states, the productivity of land was no greater than one-fourth that of Sonora. 2 Large differencesin the earnings of agricultural wage earners provide a further indication of regional differences in employment opportunities and productivity. In 1970 the averagewage per worker was 14 times higher in Baja California Sur than in the more congestedcentral state o EHidalgo.3 These differences in employment conditions among regionsmay be presumedto provide at least a partial explanation for seasonal migrationsof agriculturalworkers as well as for the tendency for many agricultural households to supple- ment incomes from agriculture with earnings from nonagricultural em- ployment. The agricultural sector is not the only one to evincelarge differencesin productivity and earnings among productive units. The industrial sector also demonstrates such variance, although it is not as pronounced as that for agriculture. In 1975, value added and total remunerationsper worker in establishments employing more than 75) workers were 3.8 and 2.1 times greater, respectively,than in smallshops with 5 or feweremployees. 4 In the service sector, the differentialin the rate of earnings between small (5 or feweremployees)and large (more than 50 employees)establishments was more modest, on the order of 2.1:1 in 1975. In the same year, the differential in earnings between small and large establishmentsin com- merce was slightly narrower at 1.8:1.5 Equally impressiveas the differencesin productivity and earnings with- in each of the sectors enumerated here are the differencesamong them, especiallythose between the agricultural and nonagricultural sectors. In 1970, value added per worker in the latter sector was 5.2 times greater than in the former. The difference may have been even greater if (as I Table 2-1. Gross Domestic Product by Broad Sector, 1940-79 Gross domestic product Millions of 1970 pesos Annual rate of change (percent) Millions of 1950 pesos 1970 1940- 1950- 1960- 1970- Sector' 1940 1950 1960 1970 revisedb 1980 50 60 70 80 Primary Banco de Mexico n.a. 8,874 11,433 16,473 2 2.2 37 Unikel 5,221 9,242 13,917 17,607 52,123 75,704 5.9 4.2 2.4 3.4 Secondary Banco de Mexico n.a. 13,329 20,956 48,727 5.3 8.8 Unikel 6,789 12,466 24,603 52,203 145,070 296,046 6-3 7.0 7.8 7.4 Tertiary Banco de Mexico n.a. 22,183 39,404 76,278 250 474 481090 5.9 6.8 Unikel 10,931 19,352 35,696 82,431 ' - ' 6.3 8.7 6.7  Total Banco de Mexico n.a. 44,106 71,793 141,478 44452 41. 57 7.0 6.6 Unikel 22,940 41,060 74,215 152,251 ,271 84,855 6.0 6.1 7.5 Note: Throughout the tables, pesos are of course Mexican pesos. n.a. Not available. a. The sectors are defined as follows: primary-agriculture, cattle, forestry, and fishing; secondary-manufacturing, construction, mining, petroleum, and electricity; tertiary-transportation, communications, commerce and finance, government, and other services. b. The 1970 revised and 1980 GDP figures are derived from the revised national product series based on the input-output matrix of 1970 and on the 1970 weights of the various economic divisions. In addition, several changes have been made in the industrial classification of several activities. Consequently, the revised 1970 GDP and that for subsequent years are not comparable to those of earlier years. As a result of the methodological changes, the revised 1970 product emerges 6.1 percent greater than the earlier measure. T'he totals for 1970 revised and for 1980 are not equal to the sum of the sectoral values because of an adjustment involving the imputed value of banking services included in the net value added of the respective user sectors as well as of the tertiary sector. The adjustment amounted to - 5,395.5 and - 10,985.1 million pesos in 1970 and 1980, respectively. In the accounts for the other entries, the adjustment was made by including in their tertiary sector total the value of banking services net of the amounts included in the primary and secondary sectors. Sources: 1940-70-Luis Unikel, El desarrollo urbano de Mexico, 2d ed. (Mexico, D.F.: El Colegio de Mexico, 1978), tables VI-AS to VI-A8. 1950-70-Banco de Mexico. 1970 revised-1980-Secretaria de Programaci6n y Presupuesto, Sistema de cuentas nacionales deMexico, vol. 1, Resumen general (Mexico, D.F., 1981), and Sistema de cuentas nacionales de Mexico, 1978-80, vol. 1 (Mexico, D.F., 1982). 18 EVOLUTION OF THE LABOR FORCE believe) the agricultural labor force was underestimated by the 1970 census. It is not surprising,therefore, that most of the poverty in Mexico is to be found in the rural areas. On the basis of the 1975 income and expenditures survey,it has been estimated that approximately 52 percent of the poor familieswere in the agriculturalsector. Stated in other terms, 76 percent of all familiesin the sectorlivedin poverty. 6 The prevalenceand durability of large urban-rural income differentialsis likelyto haveserved as one important underlying factor explaining the large rural-urban migration as well as the tendency of rural residents to seek seasonal employmentin urban labor markets. LuisUnikelestimatesthat 4.5 million persons moved from rural to urban areas between 1950 and 1970.7 Thus, after four decades of rapid growth in Mexico, large income differencessurvive among households, givingrise to an income distribu- tion that must be ranked among the least egalitarian among semi- industrialized countries. Before 1980, most studies concluded that the relative distribution of income had shown little changeover the preceding twenty years or so. The share of the bottom decilesrevealedonly a modest increase.This gain, as wellas that of some of the intermediatelevels,came at the expense of the top decile. While the discussion above has highlighted the large differences in productivity and income that exist among different groups, it does not depict the substantial changes that have occurred within the labor force over time. Nor does it give any impressions of how the labor market has evolved in the course of development, or of how it has responded to, and facilitated, the changing structure of demand and output. In the sections that follow, I will trace and evaluate some of the major developments in the Mexican labor market and relate these to the general pattern of growth. Size of the Labor Force Analysesof the Mexican labor force are based on two principal sources of information. The decennial population censusesprovide the basis for charting historical changes over time. For the period since 1973, an additional source has appeared in the form of a quarterly labor force survey, the Encuesta continua sobre ocupaci6n (Ecso). 8 But only the three major metropolitan areas of Mexico City, Guadalajara, and Monterrey have been continuously surveyedsince its inception. As the decade pro- gressed, more regions of the country were added; by 1979, coveragehad been extended to the whole country. Unfortunately, only one report was issued that reflected the full national coverageand that pertained to the  SIZE OF T'HE LABOR FORCE 19 first quarter of 1979. Thereafter, the Secretatia de Programaci6n y Pre- supuesto reverted to abbreviated reports for the three major metropolitan areas. As of mid-1983, a new survey instrument with coverage limited to urban areas is being tested as a replacement :Eor ECSO. Population censuses are generally relied upon to provide benchmark measures of various population components, including the labor force. Particularly in the absence of continual and systematic sample surveys of households, the censuses assume a primary importance as sources of information about the size and characteristics of the labor force. In the case of Mexico, the 1950 census has come to be regarded as one of the more reliable censuses and has served as a benchmark for various popula- tion measures. The 1960 census is acknowledged to have been distorted by processing errors that yielded an inflated nmeasure of the labor force. Several efforts to correct the original published results have been made, and these are reviewed in some detail in the appendix at the end of this chapter. The data for 1960 that I have chosen to use and that are recorded in table 2-2 reflect the adjustments made by Altimir.9 The 1970 census has not been subjected to the close scrutiny accorded the 1960 census in view of its greater apparent "reasonableness." Nevertheless, it appears to me that the 1970 census significantly underenumerated the male labor force, and I have adjusted the data reported by this census accordingly. The 1980 census, by all general accounts, .s a distinct disappointment. In particular, changes in the definition of the labor force and deficiencies in the enumeration process have rendered the labor force measures of du- bious value and noncomparable with those of earlier censuses. Rather than adopt the implausible data of the 1980 census, I have chosen to estimate the labor force size in 1980 by applying to the census population of labor force age the labor force participation rates reported by the labor force survey (ECSO) for the first quarter of 1979. The details of these adjustment procedures employed are spelled out in the appendix to this chapter. As a consequence of the doubtful quality of the census results, there are limits placed on the confidence with which one can speak of the changes that have occurred in the size and distribution of the labor force. While the absolute values assigned to the various labor force measures are obviously subject to some error, I believe that they nonetheless reflect faithfully the broad sweep of changes that 'nave occurred over time. The size of the labor force is a function of the size of the population of labor force age (12 years and older in Mexico) and the participation rates of its components. In turn, the growth of the population of labor force age and the labor force follow with a lag the growth in the total population. Thus, as can be seen in table 2-2, the acceleration in growth of the population during the decades of the 1950s; and 1960s leads that of the 20 EVOLUTION OF THE LABOR FORCE Table 2-2. Population, Labor Force,and ParticipationRates by Sex, 195S-80 Labor Population force Inter- aged partic- censal Total 12 years Labor ipation rate of population and over force rate change Period (thousands) (thousands) (thousands) (percent) (percent) June 1950 Total 25,791.0 16,849.3 8,345.2 49.5 n.a. Male 12,696.9 8,166.3 7,207.6 88.2 n.a. Female 13,094.1 8,683.0 1,137.6 13.1 n.a. June 1960 Total 34,923.1 21,944.2 10,212.9 46.5 2.0 Male 17,415.3 10,796.7 8,496.2 78.7 1.7 Female 17,507.8 11,147.5 1,716.6 15.4 3.1 1969' Total 48,225.2 29,697.3 12,955.1 43.6 n.a. Male 24,065.6 14,625.6 10,488.8 71.7 n.a. Female 24,159.6 15,071.7 2,466.3 16.4 n.a. January 1970b Total 48,225.2 29,697.3 13,328.5 44.9 2.8 Male 24,065.6 14,625.6 10,674.2 73.0 2.4 Female 24,159.6 15,071.7 2,654.3 17.6 4.7 June 1980' Total 67,382.6 43,618.4 20,041.0 45.9 4.0 Male 33,295.3 21,411.8 15,266.6 71.3 3.5 Female 34,087.3 22,206.5 4,774.4 21.5 5.8 n.a. Not available. a. The 1970 census established the size of the labor force by reference to the entire year 1969. That is, anyone gainfully employed at any time in 1969 was included. This measure, therefore, does not correspond to the customary measure of the labor force at a moment in time. b. The 1970 census also established the size of the labor force during the week preceding the census. This census datum is adjusted for underenumeration of male labor force members in the estimated amount of 419,000. c. For 1980, the final version of the census reports a total population of only 66,846.8 thousand. Demographers in Mexico, however, believe this to be less accurate than the figure published in the preliminary census report. I have therefore relied on the latter source. Sources: 1950-Secretaria de Industria y Comercio, Direcci6n General de Estadistica, VII Censo general de poblaci6n, 1950 (Mexico, D.F., 1953); 1960-Oscar Altimir, "La medi- ci6n de la poblaci6n economicamente active de Mexico 1950-70," Demografia y economia, vol. 8, no. 1 (1974), pp. 50-83; 1969 and 1970-Secretaria de Industria y Comercio, Dirreci6n General de Estadistica, VIII Censo de poblaci6n, 1970 (Mexico, D.F., 1972); 1980-Secretaria de Programaci6n y Presupuesto, X Censo general de poblaci6n y vivienda, 1980, Resultados preliminares a nivel nacional y por entidad federativa (Mexico, D.F., n.d.).  SIZE OF TE{E LABOR FORCE 21 population of labor force age and the labor force. The annual rate of increase in population peaked at 3.4 percent during the latter decade and began a gradual declineto 3.2 percent during the early 1970s and to 2.5 percent by 1980.10 In contrast, the rate of increase in the population of labor force agepeaked during the decade of the i 970s and can be expected to decline during the present decade. Finally, the rate of growth of the labor force lagged behind both of these broader population measures between 1950 and 1970. During the 1970s, however, the labor force expanded at an annual rate estimated at 4 percent, comfortablyin excess of the 3.7 percent rate of increase in the population of labor force age. The modest growth in the labor force before 1970 was attributable in part to a decline in the aggregatelabor force participation rate, a decline that now appears to have run its course. Underlying this decline were changes in the participation rates of various age and sex subgroups, changes not unlike those that typically occur during the course of eco- nomic development. Participation rates of the male labor force declined continuously between 1950 and 1970; the declineswere heavily concen- trated in the very young and the very old age groups. In contrast, female participation rates increased in all age groups except for the over-65 group."1 However, these increases did not sufficeto offset completelythe declinein male participation rates, so that the labor force as a proportion of the population 12 years and older declined from 49.5 to 44.9 percent between 1950 and 1970. The past decade ha, seen a decelerationin the decline in male participation rates while rates for women continued to increase.As a result, the declinein the aggregateparticipation rate appears to have been stemmed and even reversed. In spite of the increasingincorporation of women in the labor force over the past three decades, the absolute rates still appear to be quite modest. The rates reported by the 1970 censuswere wellbelow those of any of the more advanced countries of Latin America. Tlheincreases that have been recorded by ECSO over the 1970-79 interval now yield rates that more closelyapproximate those of 1970 in such countries as Chile, the Domini- can Republic, Panama, Peru, and Venezuela." The failure of these rates to rise more rapidly has been interpreted, not as unwillingnessor lack of interest in employment among women, but as the absence of sufficient employmentopportunities for the potential l.bor force. To enable a fuller employment of the male labor force, women'; work aspirations have been sacrificed.Not only has discriminationbased on strong cultural traditions limited the number of employed women, but it has also militated against 1 3 Thus, the femaleadult the length of their participation in the labor force. population has been held to represent a large repository of hidden unem- ployment, and the appeal to "traditional values" is the means society 22 EVOLUTION OF THE LABOR FORCE employs to avert the surfacing of this potential availability of women in the form of overt unemployment.To the extent that this is true, it becomes difficult to establish empirically the actual size of the potential female labor force and to distinguish the hidden unemployed from those who would truly not be availablefor employmentevenif the "cultural values" did not discourage it. What we do seem to be observing in Mexico, however, is the weakening of whatever cultural impedimentsthere may have been to female labor force participation. The expansion of employ- ment opportunities has brought with it an increasein femaleparticipation rates. Furthermore, it would appear that a potential exists in Mexico for substantial continued growth in the size of the femalelabor force. Special tabulations of data drawn from the 1977 surveyof household incomesand expenditures indicate that higher labor force participation rates among women are associated with greater urbanization and higher levels of educational achievement.Since the trends in both of these correlates are positive, the secular increasethat we haveobservedin femaleparticipation rates is likely to extend into the future. The continuous decline in aggregate participation rates recorded be- tween 1950 and 1970, as well as the similar declinefor men, has frequently been interpreted as evidenceof a failureof the Mexican economyto create enough jobs to absorb the increasing population of labor force age. It is thus concludedthat disguisedor hidden unemploymentmust have beenon the increase. Bycontrast, a reasonable alternative interpretation of these dedines can be advanced. For instance, the declinein participation rates amongthe old, particularly men,may be attributable to increasinglongev- ity and to the extension of retirement benefits to a larger segment of the agingpopulation that has facilitatedan earliervoluntary withdrawal from the labor force. Among the very young, those aged between 12 and 19, small declinesin participation have been associatedwith large increasesin school attendance, whereas among those aged 20 to 24 both labor force participation and school attendance have risen. Among the population of labor force age, the secular increasein school attendance has been impres- sive. In 1940,the number of persons enrolledin schoolsabove the primary level was only 69,000. It increased to 155,000 in 1950. Thereafter it expanded more rapidly, reaching 1.9 million in 1970 and approximately 4.1 million during the 1977-78 school year.' 4 The proportion of the population 12 years and older that was enrolledin school increased from 6.5 to 12 percent between 1950 and 1970. The continuing trend in school enrollment is clearly in evidence in the findings of the labor force surveys of the past decade. By 1979, the proportion of the population of labor force agethat was enrolled in school had risen to 18.3 percent. Particularly notable has been the sharp increase  SIZE OF TH E LABOR FORCE 23 in the enrollment of females.Traditionally, a significantlysmallerpropor- tion of women of schoolage had beenin attendancethan had beenthe case among men. However,considerableprogress was made during the decade in closing the gap at all levels of schooling (seetable 2-3). An explanation of the notable expansion of school attendance can be offeredthat is based on demand and supply considerationsrather than on labor market failure. On the demand side, there is ample evidencethat the private returns to education must be substantial. A recent study based on the 1977 household survey of income and expenditures indicates that an additional year of schooling is associated with a 13.7 percent increase in income. In particular, significant income gains seem to follow from the successful completion of each schooling cycle--primary, secondary, or university. For example, completion of primary school yieldsalmost a 50 percent increase in income over noncompletion. At the upper secondary and universitylevels,completion is associatedvwith incomeincreasesof 37 and 79 percent, respectively."5Such differencesin earningsassociatedwith education would be expected to provide a povrerful incentive for young people to extend their stay in school. On the supply side, the Mexican government has made extraordinary efforts to expand educational opportunities. During the decade following 1970, the proportion of the gross domesticproduct devoted to education Table 2-3. Percentageof Labor Force Participationand School Enrollment Rates Among Young Men and Women, January 1970, and First Quarter 1979 Sum of labor force participation Labor force School and school participation enrollment enrollment Sex rates rates rates and age groups 1970 1979 1970 1979 1970 1979 Male 12-19 36.2 35.4 41.6 57.7 77.8 93.1 20-24 78.3 82.5 6.9 12.5 85.2 95.0 Female 12-19 16.6 15.4 32.0 52.2 48.6 67.6 20-24 25.0 33.4 2.5 6.8 27.5 40.2 Sources:Direcci6nGeneral de Estadistica,IX Censogeneraldepoblaci6n,1970:Resumen General(Mexico,D.F., 1972); Secretariade Programaci6n y Presupuesto,Encuestacontinua sobre ocupaci6n, 1st quarter 1979 (Mexico,D.F., 1980. 24 EVOLUTION OF THE LABOR FORCE almost doubled from 2.1 to 3.9 percent accordingto data provided by the education ministry. Between 1972 and 1980, the real per capita expendi- ture on education more than doubled. A notable expansion of secondary schools and universitiesmade it possible to accommodate an increasing proportion of the school-agepopulation. In 1980,almost 40 percent of the population of secondary school age was enrolled, and 15 percent of the group aged 20-24 was enrolledin institutionsof higher learning.In 1960, the corresponding proportions were 11 and 3 percent, respectively. 16 Furthermore, as a rapid expansion of vocationalschoolswas effected,the 1970s saw an increasing diversificationin the types of available educa- tional opportunities. From 281 vocational schools in 1970, the number increased to 1,573 by the 1979-80 academic year, and enrollment had increased fourfold.17 The significance of these increased enrollments is that they do not appear to have occurred at the expenseof labor force participation. To be sure, a slight decline is recorded in labor force participation rates for the youngestage group of both sexes, although this may be understatedif the 1970 census significantlyunderenumerated the labor force members of this group. If the declinewas greatest among the 12-14 year age group, as would seem likely,the participation rate of the 15-19 age group may have even increased. (Unfortunately,the labor force survey does not separate the 12-19 age group into subgroups as do the censuses.)Clearer are the trends among the young adults 20-24 years of age. While large increases in school enrollment have been recorded, labor participation rates have also increased notably.1 Thus, increases in school enrollment have not come at the expense of labor force participation. As enrollments have increased, so has the participation rate of the nonschool population. This can be seen if a comparison is made between the ratio of the labor force participation rate to the proportion of the population of labor force age not enrolled in school at different points in time as follows: 1970 1979 Labor forceparticipation rate = 0.449 0.510 0.455 = SSS i - Proportionof population 1 - 0.12 1 - 0.18 of laborforceageenrolled in school Between 1970 and 1979 the ratio increased from 0.51 to 0.555, which indicates that an increasein school enrollment was more than offset by an increase in the participation rate of the nonenrollers. In summary, it would appear that the secular intercensal declinein the aggregate labor force participation rate has run its course. Whereas the participation rates of the very young and the very old of both sexescan be  SECTORA.L DISTRIBUTION 25 expected to decline further, they are unlikelyto have as large an effect on the aggregaterate as they have had in the past. Furthermore, it is reason- able to expect that such declineswill be offset by continued increases in female participation in the labor force. Sectoral Distribution Ideally, the population censusesshould provide an accurate portrayal of the evolution of the labor force in the process of industrialization and urbanization by charting faithfullythe shiftsthat have occurredin employ- ment among sectors. Unfortunately, the Mexican censusesdo not provide an unambiguous basis for measuring the changes that have occurred. In part, this is the result of changing industrial classificationsof the labor force over time or changing practices in the classification of certain population groups such as the ejidatarios.19A substantial problem is posed by errors in processing the 1960 census that exaggeratedthe size of the rural labor force. Difficultiesin the 1970 census arose from an apparent underenumeration and from the very large number of individuals,almost 6 percent of the labor force, who could not be assignedto an economic sector of employment.A further complicationis the long referenceperiod adopted for defining most of the employment characteristicsof the labor force. While the 1970 census did establish the status of the labor force in the week preceding the census, an alternative measure of the labor force was also applied to include all those who had worked at any time during 1969-a procedure that served to inflate the size of the labor force. All census tabulations of employmentcharacteriitics are based on this latter population; thus unknown biases are introduced into the data and the comparability of the data with the other censusesis possiblyreduced. As indicated earlier, grave doubts about the accuracy of the labor force measures of the 1980 census limit the usefulnessof the data. Someof these issueshave beenaddressed by Oscar Altimirin the process of carefully reworking the labor force data appearing in the 1950, 1960, and 1970 censuses. 20 The details of the adjuittmentsmade appear in the appendix to this chapter, along with a discussionof some of the problems of interpretation posed by the census data. Altimir succeededin account- ing for most of the changes in the industrial classification of the labor force. A resort to rather broad sectoral or industrial groupings became necessarybecausethe data were not always availablein sufficientdetail to permit a consistentand detailedclassificationacross the three censuses.He also devised a way of classifyingaccording to sector a large proportion of those that had been left unassignedby the census. Although some prob- 26 EVOLUTION OF THE LABOR FORCE lemswith the data could not be resolved,Altimir's distributions are widely accepted as the most reliable for these years. I have adopted his estimates and present them in table 2-4. As can be appreciated from the distributions in table 2-4, economic development in Mexico since 1940 has brought with it very substantial changes in the industrial structure of the labor force. The decade of the 1940s saw a rapid growth in the labor force as a whole and in employment in all sectors. During the followingtwo decades, the industrial and con- struction sectors retained their role as "leading" sectors in the growth of employment, with rates well in excessof the growth of the labor force. During these three decades, the commercialsector showed decliningrates of growth in each successivedecade, while growth of the service sector oscillated within a narrow range. Accordingto Altimir's adjusted census data, the growth of the agricultural labor force over the 1950-69 period amounted to less than one half of 1 percent per year. These intercensal rates of change are likely to be misleading,however, since problems of comparability and enumeration have not been fully and successfullyre- solved in my opinion. The 1960 figure represents an attempt to adjust for the large errors incurred in the coding and processing of the census of that year. In addition, it reflects an attempt to improve the quality of the measurement of the agricultural labor force. In prior censuses, all eiidatarios were automaticallypresumedto be engagedin agriculture and were so classified for census purposes. Recognizingthat many ejidatarios did not work the land themselvesbut rather rented out their plots while they sought em- ployment in other sectors, the censusauthorities attempted to identifythe sector of principal activity in the course of the 1960 census.But the order of the questions probably resulted in only a limited success in properly identifyingthe principal sector of employment. 2" Alterations made in the 1970 census questionnaire, however, are quite likelyto have avoided the earlier difficulties.The importance of this change is that the agricultural labor force for 1950 and earlier is overestimatedcompared with that of subsequent censuses and reduces the observed rate of growth in the agriculturallabor force between 1950 and 1960. Alternatively,the greater accuracy in distinguishing the principal activity of eiidatarios in 1970 would tend to bias the size of the agricultural labor force downward compared with earlier years. An additional bias in the same direction stems from the underenumeration of the male labor force in 1970. Since the participation rates of prime-agemen appearedto be particularlylow in some of the predominantly rural states of Mexico, I am led to surmisethat most of those who escapedenumeration were actually engagedin agricul- ture. The fact that, in contrast to the June 1960 census, the 1970 census  SECTORAL DISTRIBUTION 27 was taken in January, a month of low agricultural activity, may also have contributed to this underenumeration. The lon!, reference period of a full year should have offset this possible source of bias, however. In any event, although the precise size of the agricultural labor force is arguable, there can be no doubt that it has grown at a much dower rate than the labor force as a whole. As for the evolution of the sectoral distriX,ution of the labor force between 1970 and 1980, 1 can offer only a tentative estimate. As was pointed out above, the 1970 census provided a measure of the labor force in the week prior to the census but no other information about its employ- ment characteristics. In the case of the 1980 census, the sector of employ- ment could not be specified for 30 percent Df the labor force. I have therefore resorted to the following procedure for distributing the labor forces in the two years. From the total labor force, as estimated in table 2-2, I first subtracted the number of unemployed. For 1970, this datum was given by the census; for 1980, it had to be estimated. For the first quarter of 1979, the national unemployment rate was reported to be 3.3 percent by the household labor force survey ECSO. Since the period from the second quarter of 1979 up to the census date of June 1980 saw a rapid expansion of employment (estimated by the Banco de Mexico to have been at an annual rate of approximately 6 percent), it is reasonable to suppose that the rate of unemployment had declined. I have arbitrarily assumed it to have reached 3 percent in June 1980. 1 then distributed the 1970 employed labor force over the various sectors in the same proportions as reported for the population that had been employed at any time during 1969. For 1980, 1 applied to the employed labor force the same pro- portional relationships among sectors that appear in ECSO'S survey results for the first quarter of 1979, the last such nationwide distribution 22 available. The results of this procedure are presented in table 2-5. It should be noted that these distributions are not comparable to those presented for the years prior to 1970 in table 2-5, because the classification of various activities was changed, and the 1979 data are not presented in sufficient detail to permit consistency with the pre-1970 period. The distributions yielded by these procedures identify construction and the tertiary sector as the leading generators of employment during the decade. Within the tertiary sector, commerce and government recorded the largest relative gains. Agricultural employ ment expanded only slightly. The precise rate remains in question, however, because of the suspected underenumeration of the agricultural labor fDrce by the 1970 census and the questionable accuracy of the ECSO survey's estimate of the size and distribution of the rural labor force. Within t'he extractive sector, employ- ment gains in petroleum extraction were partly offset by declines in mining Table 2-4. Sectoral Distribution of the Labor Force and Rates of Change, 1940-69 Averageannual rates Labor force (thousands) of change (percent) Sector 1940 1950 1960 1969 1940-50 1950-60 1960-69 1940-69 Agriculture, forestry, and fishing 3,832.4 4,864.9 5,048.3 5,292.7 2.4 0.4 0.5 1.1 Manufacturing, electricity, and extractive 709.7 1,237.5 1,760.3 2,829.1 5.7 3.6 5.4 4.9 Construction 115.9 263.8 414.2 609.8 8.6 4.6 4.4 5.9 Commerce and finance 413.7 732.6 1,083.4 1,397.0 5.9 4.0 2.9 4.3 Other services, 786.7 1,246.4 1,906.7 2,826.5 4.7 4.3 4.5 4.5 Total 5,858.5 8,345.2 10,212.9 12,955.1 3.6 2.0 2.7 2.8 a. Includes transportation, communications, government, and personal and other services. Sources: 1940-Luis Unikel, El desarrollo urbano de Mexico, 2d ed. (Mexico, D.F.: El Colegio de Mexico, 1978), table VI-A9; 1950-1969-Oscar Altimir, "La medici6n de la poblaci6n economicamente activa de Mexico, 1950-1970," Demografia y economia, vol. 8, no. 1 (1974), table 16. The observations drawn from the 1970 census refer to activity in 1969. Altimir's distribution of the labor force includes his redistribution of the unclassified workers over the several sectors.  SECTORAL DISTRIBUTION 29 activity and thus yielded only a small net increase. Intermediate rates of growth were evidenced by manufacturing and transport. The estimates presented here should be treated with caution. As noted earlier, there are no actual measures of employment for either year, and the bases from which I derived these estimatesare far from robust. At best, they provide only a rough indication of change in the distribution of employmentover the decade. The changes that have occurred in the sectoral distribution of the labor force over the entire forty-year interval are considerable.In 1940, agricul- ture predominated as a source of employment and accounted for almost two-thirds of the total. Currently it employs lzss than 30 percent of the labor force. In contrast, the tertiary sector has doubled in its relative importance, from 20 to more than 40 percent of total employment. Employment in manufacturing and construction activitiesalso doubled in relative importance and comprised more than 25 percent of the total by 1980. These structural changes in the labor force indicate parallel changes in the structure of output of the Mexican economy.The two sets of changes can be related to show trends in sectoral produc:tivityover time (tables2-6 and 2-7). The 1940-69 interval recorded broadly based increases in Table 2-5. Estimated Sectoral Distribution of the Labor Force, 1970 and 1980 Labor force Annual (thousanIs) rate of change Sector January 1970 June 1980 (percent) Agriculture,forestry, and fishing 5,362.7 5,645.3 0.5 Manufacturing, mining, and electricity 2,524.6 4,141.9 4.8 Construction 600.0 1,250.4 7.2 Commerce and finance 1,257.7 2,696.2 7.5 Services 3,082.6 5,704.9 6.0 Total employment 12,827.6 19,439.8 4.0 Unemployment 500.9 601.2 1.8 Total labor force 13,328.5 20,041.0 4.0 Note: Figures for the labor force are based on those n table 2-2. Sources:Secretariade Industria y Comercio,Direcci6nGeneral de Estadistica,IX Censo generalde poblaci6n, 1970 (Mexico,D.F., 1972);Secretaifade Programaci6ny Presupuesto, Encuesta continua sobre ocupaci6n,1st quarter 1979 (Mqexico, D.F., February 1980).  Table 2-6. Output, Employment, and Output per Worker in Mexico, 1940-69 (output in 1960 pesos) Averageannual rate of change (percent) Sector 1940 1950 1960 1969 1940-50 19so-60 1960-69 Primary, Output (millions) 8,543 15,442 23,970 32,912 6.1 4.5 3.6 Labor force (thousands) 3,832 4,867 5,048 5,293 2.4 0.4 0.5 Output per worker (pesos) 2,229 3,173 4,748 6,218 3.6 4.1 3.0 Secondaryb Output (millions) 12,447 23,467 43,933 94,362 6.5 6.5 8.9 Labor force (thousands) 826 1,490 2,175 3,439 6.1 3.9 5.2 Output per worker (pesos) 15,069 15,750 20,199 27,439 0.4 2.5 3.5 >, Tertiary' Output (millions) 27,663 48,061 84,127 153,469 5.7 5.8 6.9 Labor force (thousands) 1,200 1,988 2,990 4,223 5.2 4.2 3.9 Output per worker (pesos) 23,053 24,176 28,136 36,341 0.5 1.5 2.9 Total gross domestic productd Output (millions) 48,653 86,973 150,511 277,400 6.0 5.6 7.0 Labor force (thousands) 5,858 8,345 10,213 12,955 3.6 2.0 2.7 Output per worker (pesos) 8,305 10,422 14,737 21,413 2.3 3.5 4.2 a. The primary sector includes agriculture, livestock, forestry, and fishing. b. Includes mining, petroleum, manufacturing, construction, and electric power generation. c. Includes commerce, finance, transportation, communications, government, and other services. d. Since the totals have been adjusted for the value of interindustry financial transactions, they may be less than the sum of the sectoral products. Sources: Output data for 1940 and 1950-Luis Unikel, El desarrollo urbano de Mexico, 2d ed. (Mexico, D.F.: El Colegio de Mexico, 1978), tables VI-AS and VI-A6; for 1960-Banco de Mexico, Producto interno bruto y gasto, 1970-78 (Mexico, D.F., 1979), p. 61; for 1969-Banco de Mexico, Informe anual 1977 (Mexico, D.F., 1978), statistical appendix, table 3. Labor force data refer to the adjusted data of 0. Altimir, "La medici6n de la poblaci6n economicamente activa," Demografia y econom(a, vol. 8, no. 1 (1974), pp. 50-83.  DISTRIBUTION SECTORAiL 31 output and productivity (table 2-6). All the principal sectors participated in the developmentalprocess, although to varying degreesin each decade. The 1940s and 1950s featured impressive increases in total output and output per worker in the agricultural sector as the large investmentsmade in rural infrastructure began to yieldreturns. During the 1960s, this sector demonstrated less dynamism as the limits of cultivatable land were reached and the more obvious or easy investment opportunities in rural infrastructure (large-scaleirrigation projects, trunk roads, and the like) were exhausted. Gains in both output and output per worker were only modest. That these gains may not have been evenly distributed over the sector, however, is suggestedby the interregional differences(referred to above) in the productivity of agriculture. 23 During the 1940s, the secondary sector expanded its output by 89 Table 2-7. Output, Employment, and Output Per Worker by Sector, 1970 and 1980 (output in 1970 pesos) Annual rate of change Sector 1970 1980 (percent) Primary Output (millions) 54,123 75,704 3.4 Labor force (thousands) 5,377 5,503 0.2 Output per worker (pesos) 10,065 13,756 3.1 Secondary Output (millions) 145,070 296,046 7.4 Labor force (thousands) 3,133 5,244 5.3 Output per worker (pesos) 46,305 56,457 2.0 Tertiary Output (millions) 250,474 481,090 6.7 Labor force (thousands) 4,352 8,204 6.5 Output per worker (pesos) 57,555 58,641 0.2 Total gross domesticproduct Output (millions) 444,271 841,855 6.6 Labor force (thousands) 12,862 18,951 4.0 Output per worker (pesos) 35,541 44,422 2.5 Note: See notes to table 2-6. Sources:1970-Secretaria de Programaci6ny Presupucsto,Sistemade cuentasnacionales de Mexico, vol. 1, Resumen general(Mexico,D.F., 1981),table 2.2-2.5; 1980-Secretaria de Programaci6nPresupuesto,Sistema de cuentasnacionalesde Mexico, 1978-80, vol. 1 (Mexico,D.F., 1982). 32 EVOLUTION OF THE LABOR FORCE percent; this was accompanied by an almost equally large increase in employment.Sinceaccessto capital goods from abroad was limitedduring World War II, the growth of output was apparently achievedwith little or no deepening of capital. Also encouraging the expansion of the sector's employmentmay have been real minimumwagesthat were either stable or declining from their highly inflated levels of 1940, and apparently declin- ing real wages in the modern industrial segment. 2 4 The followingdecade produced an equally large increase in output as Mexico activelypursued policies of import substitution. In contrast with the 1940s, however, employmentgrowth lagged that of output. With the restoration of normal access to capital goods from abroad, capital deepening must have been a factor contributing to the sharp rise in output per worker of 28 percent over the decade. The rate of growth of the secondary sector's labor force declined in comparison with the previous decade but still managed to increase by 46 percent. The 1960s saw an acceleration in the rates of growth of sectoral output, employment, and productivity. The tertiary sector proved to be a full participant in, and contributor to, growth. Output increasesin each decade closelyapproximated the growth of GDP. This sector persistently reported the highest absolute levels of productivity in the economy, and the rate of increase in productivity accelerated in each decade. The recorded growth in productivitymight be expectedto follow from an increasingweightof highlyskilled employment within the sector, as well as from increments in physicalproductivity that occurred in some subsectors such as transport and communications. Obviously, any expansion in marginal activities that may have occurred within this sector was too small to offsetthe effectof these more favorable developments. In short, the broad sectoral data indicate a shift in employmentfrom the sector with the lowestproductivity, agriculture,to those of higherproduc- tivity. Moreover, this shift did not depress the rate of increase in produc- tivity to inconsequential levels in those sectors undergoing the most rapid growth in employment. In fact, the rate of increasein output per worker accelerated over the twenty-nine-yearperiod.' Furthermore, a significant narrowing occurred in the spread among the sectoral rates of increasein productivity over the three decades. By the 1960-69 period, the annual rates of increase had become clustered between 2.9 and 3.5 percent. Relative differences in output per worker between the sectors charac- terized by the highest and lowest productivity levels also narrowed, although the absolute differenceincreased markedly. In the decade of the 1940s, output per worker in the tertiary sector was more than ten times that in the primary. Although this ratio declined to 5.8 in the 1960s, the absolute difference increased from 20,824 to 30,123 pesos of 1960. The  SECTORAL DISTRIBUTION 33 same pattern of change is observable in the productivity differencesbe- tween the primary and secondary sectors. Changes over the most recent decade haye been estimated and are recorded in table 2-7. The output data employedfor this purpose are based on the input-output matrix of that year and on data from the 1975 economic censusesand thus are not comparable with measuresfor earlier years. While the output data can be regarded as reasonably reliable, the employment data must be considered more tentative for the reasons outlined above. The employment data are takcenfrom table 2-5 with an adjustment made to the 1980 data. Sincethe L970census measure of the employed labor force corresponded to January of that year I have adjusted the June 1980 measure of the labor force to its estimatedJanuary levelbut 26 It should be pointed out that there is maintained its relative distribution. not a complete correspondence between the ,ectoral assignment of pro- ductive activity by the national accounts and that of the labor force by the population census of 1970 and ECSO. Although this may result in some distortion in sectoral measures of output per worker, I consider it to be a lesserprobable sourceof error than that originating in my estimates of the absolute size of the labor force and its distribution over sectors.In light of these qualificationsregarding the quality of the labor force measures,the employment and productivity data must be viewed with caution. The resulting estimatesyield rates of growth in output per worker that are well belowthose recorded during the two precedingdecadesand bear a greater resemblance to the pattern recorded. for the 1940s. As in that earlier decade, agriculture emerges with the highest rate of increase in productivity, a rather unexpected result becauseof the relative stagnation characterizing the sector over much of the decade. This major increment may be unduly influencedby the very largeincrease in agricultural output in 1980, an atypical year blessedwith excellentweather and particularly favorable price incentives.The declinein the r ate of increaseof output per worker yielded for the secondary and tertiary sectors is not out of line with the productivity trends derived from the ecanomic censusesfor the first half of the 1970s. If my estimates for the decade as a whole prove to be reasonably correct, they would indicate that no significantrecovery in the rate of productivityincreaseoccurred duringihe second half of the decade. For the economy as a whole over the decade,:myestimatesyield an annual increaseof output per worker of 2.5 percent. Inview of the very rapid rate of growth of the total labor force by 4 percent per year, the achievementof even this rate of increase in productivity could be viewed as a positive development.Alsoencouragingwould be the apparent continuation of the past trend of decliningrelative differencesin output per worker among sectors. 34 EVOLUTION OF THE LABOR FORCE Growth in Tertiary Sector Employment Although the broad picture of the evolution of the labor force highlights the relative shift of labor from low- to high-productivitysectors, this has not relieved a longstanding concern about the implications of growth of tertiary sector employment. The conventional view of this sector tends to characterize it as an infinitely expandable sponge ready to absorb a burgeoning labor force that is incapable of finding satisfactory wage employment. Thus, workers are "forced" into a marginal employment and productive status in the subsectors of easy entry in commerce and services, often in the status of self-employedor unremunerated family worker. 27 Essentially, this view would seem to hold that employment growth in certain sectors has been largely determined by supply rather than a response to shifts in demand. An important inferencethat is drawn from this explanation of the growth of tertiary sector employmentis that the quality of employment has deteriorated over time. Unfortunately, the criteria employed for deciding that tertiary sector growth has been "excessive" are not explicitly set forth and verified empirically.Even a cursory look at the sectoral rates of growth, however, would not seem to indicate an "explosive" growth of marginal employ- ments. Until 1970, at least, both the commerce and service sectors, for example, expanded at rates below those of the "productive" sectors of industry and construction. Furthermore, much of the expansion of the service sector has been in technical and professional occupation catego- ries. Government employment expanded at approximately the same rate as total serviceemploymentand accounted forjust under one-fourth of the total in 1969. Finally,the growth of both servicesand commerceis likely to be closelyrelated to that of urban population; the growth in the latter would be expectedto shift the demand for labor in the former. 28 In the case of Mexico, the urban population, definedin the censusas those residing in communities with a population in excessof 2,500, grew at a rate of 4.8 percent per year between 1940 and 1970, a rate that exceededthe growth of tertiary employment. Yet one cannot judge whether the growth of the tertiary sector has been ".excessive"merely by looking at its absolute size. An examination of productivity and incomesgeneratedin the sector over time would be more instructive and would also provide a test of the widespread notion that employment expansion in this sector had been "supply-determined" and, therefore, "excessive." Consider the implications of the "labor surplus" and "overcrowding" characterizationsof growth of tertiary employment. If the wages of the unskilled and unorganized labor force had been at the subsistencelevel at the beginningof the period under observation, then we  GROWTH IN TERTIARY SECTOR EMPLOYMENT 35 should observe the expansion of employmentin the sector at a constant level of real wages, or, in the short run, at wages fallingbelowsubsistence. If wages began at a level above the subsistence level, then one should observe a decline in wages toward the subsistence level. If one could observe declines in real wages, this would provide support for the de- terioration hypothesis,at least within the sector.29 Alternatively,one could examine the distribution of employmentwithin the servicesector to seeif a large concentration of marginal employmentcould be detected. Let us first consider this latter course. Table 2-8 presents the distribution of employmentin the servicesector in detailed form according to the classificationemployedby the census in 1970. The various subsectors are grouped in ascendingorder accordingto the ease of entry, the latter being a function of labor skill or capital requirements. The largest singlecategory of serviceworkers are domestic servants, who account for just one-fourth of t:hetotal. Other subsectors employing relativelylarge proportions of low-skill or low-wageworkers are hotels, restaurants, and providers of cleaning and repair services. 30 Together with domestic servants, these categories account for 54 percent of service sector employment. It can be presumed that, if overcrowding were to characterize the service sector, this is the subsector in which one would expect to find it; all of the other subsectors would appear to pose various barriers to entry. Yet, within these low-skillsubsectors,there does not seem to be the overconcentration of workers in the categories of the unpaid family worker or the self-employed that is supposed to be the hallmark of "overcrowding." At 4.1 percent, unpaid familyworkers are no more prevalent here than in the nonagricultural sector of the economy as a whole, and only slightlymore prevalentthan within the servicesector. The proportion of the Group I work force that is self-employed, 19.3 percent, does diverge from that characterizinlgeither the service sector (15.9 percent) or the nonagricultural sector as a whole (16.6 percent), but not by a great deal. Indeed, other evidenceis;availablethat suggeststhe growth of serviceemploymenthas not been increasingly"excessive." One comprehensive study of employment found that, in 1970, for every nine urban inhabitants, there was one person employed in services-a rela- tionship that was similar to that held in 1950.31 Commerce is the other sector that is believed to harbor many of the marginally employed. On the surface it may seem to do so, for the proportion of unpaid familyworkers (8 percent)is twiceas high as that for the nonagricultural sector as a whole, as is the ratio of the self-employed (32 percent). However, employment in the commercesector lagged the growth of urban population, 4.3 to 4.8 percent a year between 1940 and 1969. Expansion of serviceemployment,4.5 percent a year, was also less 36 EVOLUTION OF THE LABOR FORCE than urban growth. Sinceserviceand commercialemploymentis predomi- nantly urban, these lags behind increasesin urban population would not seem to indicate patently "excessive" growth in the former. The persistentlyhigh and rising level of output per worker employed in the tertiary sector over time can be observed in tables 2-6 and 2-7. It suggeststhat this sector has not simplyservedthe passiverole of absorbing Table 2-8. Service Sector Employment by Employment Status, 1970 (number) Unpaid Total Self- family Servicesector employment employed workers Group I Hotels, restaurants, and bars 259,206 43,072 12,631 Cleaning and personal services 129,347 43,S89 5,156 Household domestics 541,063 87,049 21,705 Repair services 236,126 50,904 8,866 Subtotal 1,166,742 224,614 48,358 As percentage of Group I employment 19.3 4.1 Group II Radio, television, and communications 33,538 1,065 383 Entertainment and cultural 90,831 24,874 2,424 Business agents, rental services 78,313 16,473 1,986 Credit, insurance, finance 105,671 3,323 1,212 Subtotal 309,353 4S,735 6,005 As percentage of Group 11employment 14.8 1.9 Group III Education 319,574 16,656 10,313 Scientific research 7,073 303 82 Medical services (social) 168,252 6,635 2,839 Professional services 84,735 25,547 1,883 Subtotal 579,634 49,141 15,112 As percentage of Group lll employment 8.4 2.6 Group IV Miscellaneous services 80,291 19,299 2,590 Religious services 16,577 2,183 1,052 Inadequately specified 7,578 1,187 382 Subtotal 104,446 23,669 4,024 As percentage of Group IV employment 22.7 3.8 Sector total 2,158,175 343,159 73,504 As percentage of sector total employment 15.9 3.4 Source: Secretaria de Industria y Comercio, Direcci6n General de Estadlstica, Censo de poblaci6n, 1970: Resumen general (Mexico, D.F., 1972), table 44.  GROWTH IN TERTIARY SECTOR EMPLOYMENT 37 large numbers of workers unable to find remunerative employment in more productive activities. A clearer test of the relationship between supply and demand in these labor markets is provided by the course of earnings. This is carefullyand extensivelydocumentedin Chapter 7. The weight of the evidencereviewed reveals that real remunerations of low- skill sectors in servicesand commerce, as well as in manufacturing, have shown steady increasesat least since 1960. These increaseswould suggest that the growth of employmentin the two sectors has reflecteda response to greater shifts in demand relative to supply rather than the reverse. Domestic Service It may be instructive to examine more closelyone market that is con- spicuous in its importance to developingsocieties-the market for domes- tic servants. As was noted above, the 1970 census recorded more than a half million persons in this occupational category. It is a market with no barriers to entry and often serves as a port of entry to the urban labor market for women of rural origin. Since domestic service is frequently viewed as "an employment of last resort," i:his market should offer a sensitive indicator of the relative existenceof alternative employmentand wage opportunities. Focusing on this market may determine whether domestic employmentgrowth can be viewedprimarily as a function of an expanding supply of labor rather than as a response to shifting schedules of both supply and demand. While domestics formed the largest single category of employment within the service sector in 1970, their relaeiveimportance within the sector has been declining over time. From 1950 to 1969, the number so employedgrew at an annual rate of just under 3 percent, while the service sector, as defined and recorded in table 2-4, expanded at a rate of 4.4 percent. As a result, domestic employment as a proportion of total em- ployment in "other services" declined from 25 to 19 percent. This cate- gory of employment also lagged well behind the 5.6 percent annual increase in the population of larger urban zenters (which provide the principal market for such services)as well as behind the increase in the urban labor force. 32 It cannot be argued convincinglythat pressures of labor supply have led to a deterioration of employment conditions for domestics.As is documented in Chapter 7, the real cash wage of domestic servants in Mexico City is estimatedto have increasedon the order of 25 to 35 percent between 1963 and the early 1970s. While real wages dropped during the inflationary surge of the mid-1970s, they recovered and ad- vanced rapidly toward the end of the decade. The observed increase in real wages would suggestthat the growth in 38 EVOLUTION OF THE LABOR FORCE employmentin domesticservicereflecteda response to greater demand for such services.One mightreasonably expectthat changesin the demand for domestic services would be a function of changes in income and in the proportion of the population residing in urban centers, since domestic employment is very largely an urban phenomenon. Support for the pre- sumption of a relationship between income and the demand for domestic services is provided by the 1968 income and expendituresstudy, reported 33 in a study of the market for domestic services in Mexico City. Cross- sectional data contained therein revealed that the sizeof familyexpendi- tures on domestic serviceswas associatedwith the levelof familyincome. The income elasticity of demand calculated from these data proved to be particularly high for families in the middle income categories. The arc elasticities ranged from almost three to just over one. Thus the rapid increases in real incomesthat have occurred in Mexico over the past three decadescan be expectedto havebeen reflectedin an increasingdemandfor domestics. I have tried to quantify the relationship between changes in income, urbanization, and domesticemploymentby applyingthe followingproce- dure. On the basis of cross-sectional data for 1970 on the 32 states in Mexico, I have estimated a regressionof domesticservants per thousand population (DS) on per capita gross domestic product (GDP) and the proportion of the state's population residingin urban centersof 15,000 or more inhabitants (UP): DS = a + blGDP + b 2 UP. The employment date originate with the 1970 population census; income and urbanization data are from Unikel. The solution of the regression equation yields the following: DS = 2.03 + 0.00063GDP + 0.1308UP (0.0007) (0.0561) In this formulation the degree of urbanization proves significant at the 5 percent level while the income variable fails of significance.The presence of multicollinearity renders the standard error of the income term larger than the coefficient.(GDP and UP are significantlycorrelated with an r = 0.88.) The proportion of thevariance explainedbythesetwovariables, the adjusted R2 , is 0.595, about the same proportion that is explainedby either of the two variables alone. Substituting 1950 valuesfor income and urban population, the equation estimates the 1950 employmentof domestics as 7.75 per thousand population for a total of 205,088, well belowthe actual level of 310,165. However, this estimate is derived under the implicit assumption that the 1969 ratio of the price of domestic service to other  GROWTH IN TERTIARY SECTOR EMPLOYMENT 39 prices also held in 1950. If the price of domestic service, in fact, rose relative to other prices over this time interval, then the application of the 1969 coefficients should result in an underestimationof the 1950 employ- ment level. Unfortunately, I do not have data regardingthe changein relativeprices over the entire time interval, nor do I have a measureof the price elasticity of demand. There are wage data availablefor t:hedecade of the 1960s that record a sharp increase in the wages of domestic servantsin Mexico City relative to the general consumerprice index. ][fit is reasonable to assume that the same annual rate of increaseof 2.11 percent in the relativeprice of domestic serviceheld over the entire nineteen-yearperiod, 1950-69, and that the price elasticityof demand approximates unity, then the estimating equation can be amended. It can now include a term that reflects the relative price change over the entire period, P, times the assumed price elasticitytimes the 1970 employmentof domesticsper thousand popula- tion DS19 70 : DS19 50 = 2.03 + 0.00063GDP + 0.1308 TP + (- 1)(-P)(DS,970). The substitution of the values for the independent variablesof the equa- tion yields an estimated level of employment in 1950 of 348,139, or 12 percent greater than the actual level.Giventhe tenuousnessof the assump- tions made with respectto relative price changesand elasticityof demand, this estimate is not unreasonable. If the actual rate of change in relative prices was slower during the decade of the 1950s than it was during the 1960s, or if the price elasticityof demand were, in fact, less than unity, then the estimated levelof employmentfor 1950 would have more closely approximated the actual.3 Although this exercise may be lacking in precision, it does serve the useful purpose of supporting the view that ihe increase in domestic em- ployment has been associated with a shift in demand for such services rather than a response simply to the availability of a greater number of individuals seeking employment.Expressed in terms of a formal supply and demand apparatus, the observation of both an expansion of employ- ment and a rising real wage must have involved both a rightward shift in the demand for domestics and a movement upward along a demand schedule.Clearly,the shift in demand must have been greater than the shift in supply. In contrast, had the growth of employmentbeen supply-led,it would have involved a rightward shift in the supply of domestics that outran the shift in demand, with the unavoidableconsequenceof a decline in the real wage. There is every reason to believethat the past trends observable in this market will extend into the future, that the relative (if not the absolute) 40 EVOLUTION OF THE LABOR FORCE importance of domesticsin the labor force will decline,and that their wage will rise. In the past, this source of employmenthas provided a haven for women with disproportionately low levels of education. Almost 40 per- cent of the domesticssurveyedin Mexico City by a study in 1970 proved to be illiterate; another 46 percent reported less than a primary school education. 3 5 By contrast, 56 percent of the femalelabor force had achieved an education beyond the primary level.As educationallevelsof the female population continue to rise and approach those of males, it is reasonable to expect that the range of employmentopportunities open to women will expand, and their willingness to accept employment as domestics will wane. Furthermore, domesticservicehas represented an important point of entry into the metropolitan labor market for women migrating from rural or small urban communities. Over 60 percent of the domestics encountered by this same study originated in such communities, and 80 percent of these reported that their firstemploymentin the capital was as a domestic. As the relative size of the rural population continues to shrink, this source of domesticswill likewise loseimportance. Thus, the supplyof women to this market may be expectedto continue to declinein relation to total female labor supply even in the face of increasing labor force par- ticipation rates. Indeed, in an expanding economy, risingeducation levels among women that lead to greater participation and to higher wage employment opportunities will tend to reinforcethe influenceof increas- ing incomes and urbanization on the demand for domesticservices. De- mand schedules will thus shift further to the right and produce further gains in relative wages.This appears to have been occurringbetween 1979 and the onset of the economic crisis in mid-1982 (seeChapter 7). Food and Beverage Services In contrast to the domestic servicecategory of employment,the others includedin Group I appear to have grown at rates that exceededthe rate of growth of the total labor force or of urban population. For example, employment in the preparation and sale of food and beveragesmore than tripled over the 1950-69 interval, with an annual growth rate of 6.5 percent. This expansion and the substantial proportion of total employ- ment in the categories of the self-employedand unpaid family workers (about one-fourth of the total) may be misleading.It mightappear that this sector represents one of those frequently characterized as harboring the marginally employed, those unable to find employment at the "going" wage who are then reduced to self-employmentat levels of productivity and incomewell below those of wagelabor. However,a careful analysisof data reported in the service sector census of 1971 is not consistent with  GROWTH IN TERTIARY SECIOR EMPLOYMENT 41 such a pessimisticinterpretation of the observedgrowth in employment. The Censo de serviciospresentingdata for 1970 reports a coverageof 51.8 and 19.6 thousand establishmentsin the food preparation and beverage industries, respectively.Within both of these subsectors, establishments that employ no remunerated labor are in the majority, although they do not account for a majority of the covered employment. Someof the key data for such establishment; are presented in table 2-9. A sharp contrast can be observed in the measu:res of capital employed and net value added per person employed between those establishmentswith- out and with paid employees.With the exception of one of the eight subsectors that compose the food and beverage industry-restaurants- establishmentswith no paid employeesrather uniformlyreport net value added per worker as wellas gross salesper worker (not shownin the table) equal to about one-half those in establishmentswith paid help.Indeed,the net value added per worker in the former class of establishmentsin six of the eight four-digit classificationswas less than the wages paid in the establishmentswith paid employees.In the other two subsectors,net value added per worker in the former class of establ ishmentbarely exceededthe average wage in the latter. These comparisons would serve to confirmthe suspicionthat the indus- try harbors a substantial number of workers who are "marginally em- ployed." However, a comparison of the returns to labor in the two types of establishments suggests that these are not as different as may have been expected. For the establishmentswith paid employees,the census provides the sum of the wages and salaries paid as well as the number of paid employees,thus making it feasibleto calculatethe averageannual earnings of this class of workers. In the case of the establishmentswith unremuner- ated labor, it was necessaryto estimate the return to labor. This was done by imputing a return to the capital employed and subtractingthis from net value added with the balance thus derived assignedto labor. 3 6 As can be observed, the returns to labor of the self-eraployedand unremunerated workers, shown in parentheses,do not depari:very far from the wages paid to hired labor. To the extent that unremuneratedworkers are more likely than hired labor to be part-time workers, as the evidenceseemsto show, the actual differencesin average rates of reniuneration per time unit may be even narrower than those suggested by these data. 37 A further observation that does not correspond to a commonly held notion about the quality of employmentconcerns the level of earnings of labor in very small establishments.Table 2-9 presents data for all estab- lishments with no paid employeesfollowed by a similar set of data for all establishments with hired labor but limited to those with total employ- ment of one or two persons. It is commonly thought that very small 42 EVOLUTION OF THE LABOR FORCE Table 2-9. Employment, Productivity, and Earnings in Food and Beverage Services, 1970 (pesos) Net capital Net Average Average Employ- invest- value annual annual ment ment added wages total (num- per per and remuner- Subdivision ber) worker worker salaries ationsa Restaurants and cafes No paid employees 21,112 5,597 7,697 - 7,246 Paid employees 62,071 20,103 27,137 14,499 16,340 One or two workers 4,803 12,646 19,667 10,966 11,791 Food and taco stands No paid employees 14,594 5,795 9,226 - 8,768 Paid employees 13,410 7,239 17,545 8,566 9,549 One or two workers 3,271 7,758 21,091 8,018 8,831 Ice cream parlors and soda fountains No paid employees 10,725 3,989 6,944 - 6,606 Paid employees 3,781 11,480 17,132 7,492 8,337 One or two workers 1,412 11,732 19,071 6,458 7,134 Oyster bars No paid employees 2,382 5,490 11,167 - 10,807 Paid employees 2,568 8,304 21,269 12,272 13,446 One or two workers 558 8,608 23,701 12,535 13,403 Shops for preparation and sale of posole and tamales and similar establishments No paid employees 7,784 3,053 7,551 - 7,307 Paid employees 2,678 8,046 15,290 8,505 9,217 One or two workers 840 7,907 16,548 9,120 9,541 Taverns and bars No paid employees 6,986 9,617 12,500 - 7,328 Paid employees 14,769 16,566 25,662 11,699 13,191 One or two workers 5,689 19,107 28,667 12,056 13,249 -Not applicable. a. Figuresfor serviceestablishmentswith no paid employeesare imputed values. Source:Secretariade Industria y Comercio,Direcci6nGeneralde Estadistica,VI Censode servicios,1971 (Mexico,D.F., 1974), tables 8 and 16.  GROWTH IN TERTIARY SECTOR EMPLOYMENT 43 establishmentsform part of the "informal" sectorand are characterized by wage conditions and productivities significantlyinferior to those found in larger establishments. The census data, however, reveal no such differ- ences within each of the subsectors except for restaurants, a sector in which differencesin the quality of servicesancd labor employed are most likelyto span a very wide range. In all other sub sectors,net valueadded (as well as gross sales) per worker and remunerationsper employeein estab- lishments with no more than two persons employed either exceeded the average for all establishments with paid employeesor were not far re- moved from the average. The relationship of the average earnings to the legal minimum wage may be of interest . In 1970, the unweighted average of the urban minimum wage rates for all zones amounted to 9,092 pesos (Mex$) per year; the urban rates weighted by ihe labor force of each zone yielded a national average of Mex$9,851. 3 8 Only in the ice cream parlor and soda fountain sector did reported averageearnings depart significant- ly from the legal minimum. It is reasonable to expect that employment in the food and beverage serviceindustry would expand in the courseof economicdevelopmentand rising incomes. Data from household income and expenditures studies in Mexico provide strong support for a hypothesisthat the demand for such 39 Growth of employmentin the industry would servicesis income elastic. also be expected to accompanythe growth of the urban population. As in the caseof domesticserviceemployment,I tried to seewhether the growth in employmentcould be explainedby changes in incomeand urbanization. I proceeded to fit a linear regression of employment in this industry per thousand population (FB),on the grossdomesticproduct per capita (GDP), and the proportion of the population residing in urban centers with 15,000 population or more (UP) in 1970 across the 32 states. The form of the regression was as follows: FB = a + bl GDP+ b2 UP. Solving the equation for the coefficientsof the independent variables yielded FB = 0.434 + 0.0005GDP + 0.0547UP. (0.00025) (0.0199) Both coefficientsproved to be significant,at the 5 and 1 percent levels, respectively,and the adjusted R2 was 0.765. Substitutingthe 1950 values for GDP and UPand solving for the 1950 levelof FB, I obtained a value of 2.227 per thousand population. Multiplying the estimated employment per thousand by population (in thousands) for 1950 yielded a predicted 44 EVOLUTION OF THE LABOR FORCE employment for 1950 of 58,933. The actual level of employment in this industry was 60,145. Given the very close approximation of the predicted to the actual level of employment, the hypothesis that the growth of employment is largely a response to shifts in demand is again supported.40 Other Services Another subgroup of Group I in table 2-8, cleaning and personal services, unfortunately, does not lend itself to further decomposition, and its precise content cannot be determined in a way that would permit extensive comparisons with earlier population censuses. Nevertheless, some information about the quality of employment in this subsector of services can be gleaned from the census of services for 1970. Five kinds of establishments are reviewed in table 2-10. As in the case of the other subgroups of the service sector we have reviewed, there is a sharp differ- ence to be noted in the net value added per worker in establishments with and without paid employees. However, the earnings per worker reveal much narrower differences. As before, the earnings of paid employees in the smallest establishments, those with two or fewer workers, could not have compared unfavorably with those in larger establishments. In several cases, they actually exceeded the average wage payments made in all establishments with paid employees. Wage payments in all but one of the subgroups, shoeshine parlors, exceed the legal minimum wage. Shoeshine establishments yielded recorded earnings of paid employees which were, on average, only 77 percent of the weighted average legal minimum wage, although the estimated return to labor in establishments with no paid employeees was estimated to fall short of the minimum by only about 9 percent. Recorded payments to labor, however, may understate actual earnings of workers since establishment data do not include a considera- tion of any tips received in this category of establishment or any of the others that provide personal services directly to consumers. Finally, the remaining subsector from Group I of table 2-8 is examined in table 2-11. While one would normally expect workers employed in repair service activities to possess skills that would limit easy entry, it is also likely that the quality of such services would be characterized by wide variance and that only rudimentary skill would be evidenced by many practitioners, particularly in small establishments. The data in table 2-11 suggest that skill levels in this sector must be quite low, for productivity and wage levels appear to be low by comparison with the other subsectors of Group I. Annual wage payments below the weighted legal minimum wage appear in all but one of the strata of wage-paying establishments. This subsector resembles the others in Group I in the small differences in  GROWTH IN TERTIARY SECTOR EMPLOYMENT 45 the earnings of workers in the smallest establishmentsas compared with the averageof all paid employees.Yet the imputed returns to unremuner- ated labor are higher here relative to payments made to wage labor. If one could accept the data presented hereas representativeof generally low-skill serviceemployments,it would not appear that "overcrowding" exists or that productivities and earnings have been pushed down to very low levels. It could be argued, however, that the establishmentsthat are likelyto be enumerated in a census are unlikel) to be the repository of the Table2-10. Employment, Productivity, and Earnings in Personal and CleaningServices, 1970 (pesos) Net capital Net Average Average Employ- invest- value annual annual ment ment added wages total (num- per per and remuner- Subdivision ber) worker worker salaries ations' Barber shops No paid employees 12,091 4,794 8,689 - 8,305 Paid employees 8,653 5,039 16,274 9,598 10,420 One or two workers 4,141 5,479 15,236 9,041 9,671 Beauty shops No paid employees 5,364 6,809 9,992 - 9,447 Paid employees 7,898 9,436 18,748 10,417 11,678 One or two workers 3,211 9,661 18,525 10,377 11,497 Laundriesand dry cleaners No paid employees 5,104 6,419 10,440 - 9,926 Paid employees 13,349 33,003 27,211 11,326 12,975 One or two workers 2,706 48,181 44,310 12,465 14,359 Shoeshineparlors No paid employees 478 1,826 9,182 - 9,036 Paid employees 247 2,381 13,441 7,396 7,610 One or two workers 98 2,867 13,204 7,545 7,764 Car washes and lubrication shops No paid employees 259 20,888 13,185 - 11,514 Paid employees 4,243 29,254 27,435 9,981 11,575 One or two workers 591 51,589 35,096 12,677 14,908 - Not applicable. a. Figures for serviceestablishmentswith no paid eriployees are imputed values. Source:Secretariade Industria y Comercio,Direcci6nGeneralde Estadistica,VI Censode servicios, 1971 (Mexico,D.F., 1974), tables 8 and 16. 46 EVOLUTION OF THE LABOR FORCE "marginal" adherents of the sector. The latter may be expected to be self-employedindividualswith no establishmentor businesspremises.As "unregistered" enterprises, they could not be expected to find their way into the universe from which the census observations are drawn. Thus, those who are presumed to be the most disadvantaged workers in the sector simply escape detection. Although this may appear to be a persuasive observation, I would not wish to discount the relevance of the census data. Of particular signifi- cancewould be the data for the smallestestablishmentswith paid employ- ment, those with total (remunerated and unremunerated) employmentof two or fewerworkers. Suchtiny establishmentsare frequentlyjudged to be part of the "informal" sector and, therefore, free of the legal constraints Table 2-11. Employment, Productivity, and Earnings in Repair Services, 1970 (pesos) Net capital Net Average Average Employ- invest- value annual annual ment ment added wages total (num- per per and remuner- Subdivision ber) worker worker salaries ationsa Domestic electric appliance repair and installation No paid employees 6,615 4,707 11,069 - 10,692 Paid employees 5,766 17,016 22,314 9,480 11,072 One or two workers 1,948 7,347 17,138 8,349 9,349 Automobile repair No paid employees 11,774 11,698 9,586 - 8,650 Paid employees 36,236 21,583 26,183 10,238 11,969 One or two workers 7,013 14,235 20,715 9,039 10,258 Motorcycle and bicyclerepair No paid employees 2,658 5,156 10,109 - 9,697 Paid employees 920 13,328 17,716 8,076 8,732 One or two workers 565 14,050 19,326 8,078 8,645 Plumbingand soldering No paid employees 1,805 5,032 9,648 - 9,245 Paid employees 1,956 10,037 16,984 8,678 8,678 One or two workers 590 10,075 16,756 8,737 9,588 - Not applicable. a. Figures for serviceestablishmentswith no paid employeesare imputed values. Source:Secretariade Industriay Comercio,Direcci6nGeneral de Estadistica,VI Censode servicios, 1971 (Mexico, D.F., 1974), tables 8 and 16.  GROWTH IN TERTIARY SECTOR EMPLOYMENT 47 that govern employment conditions in large, modern employing units. Thus, one would expect remunerations in such establishmentsto reflect more closelythe opportunity cost of labor rather than the legal minimum wage. Indeed, we have observed that average remunerations frequently either barely approximated or actuallyfell belowthe legalminimumwage. These facts suggest that an important proportion of paid employees received less than the legal minimum and that the wage payments do respond to labor market conditions.4 1 Sincethe opportunity cost of work- ers employed in such small establishments is the earnings that would be forthcoming in a self-employedstatus, the remunerationspaid may not be a poor approximation of the former. On the contrary, if the imputed returns to labor in establishmentswith no paid labor can be accepted at face value, these returns are even more likely to approximate the oppor- tunity cost of such labor. They would thus be representativeof returns to self-employedlabor in similar activitiesin the informal sector. Commerce I mentioned earlier that, like the servic. sector, commerce is also frequently singled out as a sector that serves as an absorptive sponge for labor in "excess supply." Referencesto the ubiquitous street peddler are legion in the literature, as are those to tiny retail establishments with limited inventories and "underemployed" labor. Drawing on data for 1970 contained in the 1971 census of the commercesector, I conducted exercises similar to those reviewedabove for the servicesector. 4 2 Sincethe findingswere similar to those reported for the low-skill service sector, I will not presentthe findingsin detail. Sufficeii:to say that eventhe imputed earnings of workers in establishmentswith no paid employeesexceeded the legal minimum wage for 1970, as did thcse in the smallest establish- ments with paid employees but employingnlo more than a total of two persons. While the observation of remunerations at a moment in time-in this case, 1970-may lead to a conclusion that employmentconditions in the small establishmentstrata of the commerceand servicesectors were not as depressed as might have been expected, this night not be as important as the trend in conditions over time. If these sectors are viewed as flexible, accommodating, and expanding repositories for "excess" labor, then the true significanceof expanding employmentcan be gauged only if account can be taken of secular changes in earnings. This is undertaken in some detail in Chapter 7. The findingsreported there provide evidencethat the wage conditions of employmentimproved steadilyin all sizesof establish- ments in both of these sectors, at least over the interval 1960-75. Indeed, 48 EVOLUTION OF THE LABOR FORCE earnings in the smallest sized units appear to havegained relative to larger sized units. Therefore, I would view the evolution of the labor force and the structure of employment as having had favorable implicationsfor the lot of the Mexican worker. The shift of labor from the agricultural sector to others has occurred without causing a depression in earnings in the recorded parts of the "informal" labor market and has resultedin a steady improvement in wage conditions. A final observation that casts doubt on the view that labor market conditions have deteriorated over time is drawn from incomedistribution data for 1963, 1968, and 1977.43 If the economywere developingin such a manner that increases in productivity and wages in some sectors were beingpartially offset by declinesin others, then one would expectthis to be reflected in a sharp drop in the share of income accruing to the lowest income deciles. In fact, there has been great stability in that share. The share accruing to the bottom 40 percent of the households has varied within the narrow range of 10.5 to 10.9 percent." This implies that the incomes of these households have increased at approximately the same rate as has the average per capita income. Nevertheless,it is quite likely that some households, particularly in the agricultural subsistencesector, have had stagnant incomes and have not shared in the general increasein prosperity. The bottom decile, for example, has suffered a decline in its relative share of total income, from 1.69 to 1.08 percent. However, given the strong growth in per capita income, this still allows for an increasein real income for this decileof approximately 13 percent. For the most part, the lowest incomes in Mexico are to be found in remote stagnating agricultural regions of the country that often have large, partially assimi- lated indigenous populations-not in the expanding urban sectors. The other three deciles of the bottom four must have enjoyed substantial increases in real income, a phenomenon that would not ordinarily be consistent with a widespread depresssionof urban wages at the lower end of the wage structure. This conclusionwill be seento be consistentwith the earnings data reviewed in Chapter 7. Summary Concerns regarding the secular trend in the quality of employment in Mexico have been based on, among other things, observed declines in aggregaterates of labor force participation and a rapid growth in tertiary sector employment. Both have been interpreted as indicative of a failure of the Mexican process of economic development to create sufficient amounts of productive employment. Although these changes are empir- ically verifiable,they can be interpreted to be reflectionsof positiveforces  LABOR FORCE MEASUREMENTS 49 rather than indicators of a deterioration in the availability and quality of employment. The declines in participation rates of teenagers and young adults have gone hand in hand with increasesin school enrollments.Only if schoolattendance can be shown to be an unproductiveuse of time might one be led to hypothesizethat schools were me rerepositoriesof frustrated job seekers. Since there is clear evidence that investment in education carries with it generous rewards in the forrn of higher earnings, it is reasonable to hypothesizethat a decisionto substitute schoolingfor labor force participation represents a rational choice, especiallyamong young teenagers. Furthermore, the willingnessof farrilies to foregothe potential incomes of their children may reflecta continuing improvement in house- hold incomes which permits an increasingconsumptionof normal goods, including education. The decline in the participation rates of the elderly population parallels trends observable in more highly industrialized societies and may simply reflect the effects of the extension of social security pension coverageas well as risinghousehold incomes that permit the purchase of the elderly's leisure time. Nor does the expansion of tertiary sector eniploymentappear to have as ominous a character as has been generallysupposed. Until recently,it has expended at a rate below that of the growth of the urban population, the primary consumer of services.When the determinants of the demand for some of the low-skill services-the growth of the urban population and increases in per capita income-are taken into account, much of the growth in employment can be explained. Furthermore, the observation that the real remunerations of those engaged in establishmentsmost likely to be paying market-determinedwages have risen over time suggeststhat the demand for such serviceshas been increasingmore rapidly than their supply. In short, the concurrent increases in employment, productivity, and earnings in the major sectors of the economy would seem to point to a more optimistic appraisal of the course of employment conditions than that generally held by observers of the Mexican scene. Optimism at so early a point in this exposition derives frora a rather narrow empirical base; the chapters that followwill provide further evidencethat reinforces this appraisal. Appendix: Technical Notes on Labor Force Measurements In the course of Chapter 2, severalreferenceswere made to shortcom- ings in the available published labor force data. All of the population censusessince 1950 have beencharacterized byproblems that have greatly 50 EVOLUTION OF THE LABOR FORCE inhibited the charting of changes in the size and distribution of the labor force over time. This appendix examines the various procedures adopted by students of the Mexican labor force for adjusting the published data that refer to labor force participation rates and to the industrial distribu- tion of the labor force. Labor Force Participation Rates While the trend in rates of labor force participation is not in question, the precise amount of change is. Unfortunately, the population censuses since 1950 do not provide an unambiguous basisfor measuringchange. In part, this is attributable to variations in the concept of labor force par- ticipation that is applied from one census to another. One analysis of the evolution of the labor force holds that some of the decline in male par- ticipation rates observable in the 1970 census stems from changes in the criteria applied. 4 5 The 1960 census poses a particularly severe problem. Processingerrors resultedin gross distortions of participation rates, which necessitated a later revision of the published data. However, even the corrected data proved unconvincing.As a result, greater reliancehas come to be placedon the adjustments made by the Colegiode Mexico and on the work of Oscar Altimir.41 Yet, in light of the basic deficienciesof the census tabulations, even the adjusted data must be approached with care and not a little skepticism.Unfortunately,these two approaches to the adjustment of the census lead to different conclusionswith respect to the size of the labor force and to the internal structure of age-specific participation rates for the two sexes. Table 2-12 presents participation rates by age and sex derived from the population censusesand, for 1960, the adjusted data of the census, the Colegio, and Altimir. T1he difficultiesposed by the data for 1960 immediatelycome to light. Although the male participation rates seem plausible in the 30-60 age groups, they seem higher than one might expect, relativeto the 1950 rates, at the two extremes of the distribution. The female rates appear quite implausible,given the continuous increasesin participation rates for each age group from age 25 on. Altimir's adjustments result in lower participa- tion rates in all age groups for both sexes. However, they also yieldwider variations in the rates among the prime-age groups than one would normally expect to find. The Colegio's estimates for men fall somewhere between the census and Altimir estimates except for the two oldest age groupings; these show lower participation rates than Altimir's. Regarding the female rates, the Colegio's estimates lie above the other two for the young and below for the older women. In short, no one of the distributions  LABOR FORCE MEASUREMENTS Si is entirely satisfactory,that is, has a form that meets expectationsbased on international comparisons. Nevertheless, the literature on the Mexican labor force seemsto have adopted the Altimir e stimatesmore widelythan the others. I will do the same, although I retain some reservations. One obvious result of the different sets of participation rates is the different estimates of the size of the labor force for 1960. Whereas the census yielded a labor force of 11.253 million, the Colegio and Altimir adjustments yielded 10.631 million and 10.213 million, respectively. Obviously, the rate of intercensal changes in the labor force will vary depending on which of these numbers for 1960 is adopted. Less controversy surrounds the 1970 censu.sresults-perhaps because they have not been subjected to the same close scrutiny as that accorded the census for 1960. However, there are groands for believingthat this census underenumeratedthe labor force. Note, for example,the participa- tion rates for males in the prime age groups. These show significant declines from past levels and are lower than international comparisons would lead us to expect. In the case of women, the rates would suggest a declinein participation rates of most age groups in comparison with any of the three distributions for 1960. Again, in line with past trends in Mexico and with international experience, one might reasonably have expected female rates to rise. Indeed, an interesting paradox of the 1970 rates for females comes to light in the two measures that were made of the labor force. One was based on activity during 1969: anyone who had been employed at any time in 1969 was counted as a member of the labor force. The other measure was more conventionalartd referredto activity during a reference week prior to the census. One would normally expect the measure based on activity during 1969 to yield a larger labor force than that of a one-weekreferenceperiod duringthe firstmonth of 1970. Yet, in the case of women, the 1970 referenceweek yieldedthe larger labor force. Support for the hypothesis that the 1970 census underenumerated the labor force is provided by the quarterly labor force surveys which were initiated in 1973.47It is possible to compare tie participation rates of men and women in the Federal District as reported by the census with the survey's rates for 1973 (seetable 2-13). For rnen, the surveyparticipation rate amounted to 75.1 percent, expressed as an average of the four quarterly rates, as compared with the census rate of 71.7 percent for 1969 and 70.6 percent for January 1980. For women, the absolute and relative differencesare evengreater. Whereas the surveyyielded an averagerate of 35.0 percent, the two census measureswere reported as 29.7 percent and 28.8 percent. Although one would expect a labor force surveyto capture a larger and closer approximation of the actual labor force than a popula- Table 2-12. Labor Force ParticipationRates by Sex and Age Group, 1950-79 1960 Colegio Adjusted 1979 Age de by 1970 January- group 1950 Revised Mexico Altimir 1969 January March Male 2_14 } 25.8 32.2 22.4 15.3 12.8 15.5 35.4 15-19 79.0 77.6 68.2 61.6 49.9 52.2 20-24 93.3 94.7 91.8 86.8 79.6 78.3 82.5 25-29 97.3 96.7 94.7 92.5 90.6 88.2 95.9 30-34 98.5 97.1 95.9 94.8 93.2 89.6 . 35-39 98.8 97.4 96.6 96.0 94.3 90.2 96.7 40-44 98.7 97.3 96.6 96.9 93.9 89.8 45-49 98.6 97.3 96.3 97.2 93.9 89.6 93.4 50-54 98.3 96.6 95.4 94.4 92.3 88.0 J 55-59 98.1 96.3 94.0 96.4 90.6 86.2 85.1 60-64 97.8 96.0 91.8 95.4 86.1 81.5 65-69 97.1 94.4 88.8 94.6 81.1 76.7 1532 70+ 95.6 90.2 81.0 85.2 63.2 60.7 3 Female 10-11 5.9 - - - - 12-14 } 4.8 7.0 4.4 5.1 7.6 15.4 15-19 15.2 16.7 20.8 15.4 20.9 23.1 20-24 11.2 19.4 23.0 17.8 24.1 25.0 33.4 25-29 13.3 15.6 16.9 14.4 17.4 18.6 27.3 30-34 13.4 16.0 16.1 14.7 15.7 16.8  LABOR FORCE MEASUREMENTS 55 Table 2-14. Labor Force ParticipationRates of Men in Prime-Age Groups by Region, January 1970 and First Quarter 1979 Agegroup Region 25-34 3S-44 45-54 Federal District 1970 92.9 94.2 92.3 1979 96.7 97.9 94.4 North 1970 87.0 89.3 88.1 1979 93.3 94.3 87.3 Northeast 1970 90.3 91.7 89.9 1979 96.9 97.5 94.2 Northwest 1970 90.5 89.2 88.2 1979 95.2 97.1 93.8 North Central 1970 86.9 89.1 88.5 1979 90.0 97.1 97.8 Peninsular 1970 90.1 91.8 91.7 1979 98.1 98.5 96.3 Central Pacific 1970 85.4 86.2 85.3 1979 95.9 96.2 92.5 Central' 1970 87.4 88.7 87.6 1979 95.9 96.2 92.5 South Pacific 1970 85.0 87.3 87.0 1979 96.1 97.3 94.6 Central Gulf 1970 89.5 91.1 90.4 1979 96.9 99.1 99.0 All Mexico 1970 88.5 90.0 89.0 1979 95.9 96.7 93.4 Note: See table 3-2 for definitionsof the regions. a. Includesthose municipiosthat formpart of the MexicoCity metropolitanarea but that lie outside the Federal District. Sources:Secretariade Industriay Comercio,Direcci6nGeneralde Estadistica,IX Censode poblaci6n, 1970 (Mexico,D.F., 1972); and Secretaria de Programaci6n y Presupuesto, continua Encuesta vol. 7, no. 1 (1979),table2A. sobreocupaci6n, 56 EVOLUTION OF THE LABOR FORCE census amounts to approximately 419,000 males in the prime-age groups. Rather than a measured labor force participation rate of 89.1 percent for this group, as given by the census,the surveysyielda rate of 95.6 percent- much closer to the figure one would be led to expect from international comparisons. It is likely that the 1970 census also underestimated the size of the female labor force. One could reasonably haveexpected the secular trend suggested by the change from 1950 to 1960 to have continued to 1970.48 The femaleparticipation rate increased from 13.1 percent in 1950 to 15.2 percent in 1960 after the adjustments made by Altimir. Over the next decade, it increased by just over one percentage point to 16.4 (based on 1969 labor force experience).For the week prior to the census,however, a higher rate of 17.6 is reported. Nevertheless, even the latter rate may represent undercounting. The labor force surveys suggest that female participation may have been greater than that reported by the census.For example, the first survey, in the first quarter of 1973, reported a rate of 33.8 percent for the Federal District, more than a third greater than the census figure for January 1970. This difference, however, is not likely to be typical of the degree of underreporting that may have characterized the census as a whole. This can be inferred from the participation rates reported by the quarterly surveys for the first quarter of 1979 (seetable 2-15). There, the January 1970 census rate is presented, along with the rates for the more recent quarter, for each region.The recentnational participation rate can be seen to be 22 percent higher than the census rate. Of course, the entire differ- ence cannot be ascribed to underreporting in 1970; secular tendenciesmay well account for some of the interveningincrease.Over the decade of the 1970s, the largest increasesin participation rates appear to have occurred in those regions in which expanding urban centers exert a heavy weight. These areas include the Federal District, Central Pacific (which includes Guadalajara), Central (parts of metropolitan Mexico City), and North (Juarez). The apparent declinein participation in the Northwest region would seem to be misleading.This region includes the rapidly expanding border region of northern Baja California, an area in which light border industries have offered expanding employmentopportunities to women. The surveys report unusuallywide fluctuationsin the femaleparticipation rates for this region, possiblya consequenceof the sampledesign. The one for the first quarter of 1979 is among the lowest to be reported for recent quarters. For example, the rates for the first quarters of 1977 and 1978 were 17.9 and 18.7 percent, respectively.Therefore, skepticismabout the reported decline in female participation in that region would be war- ranted. While the data suggest that the 1970 census underenumerated the  LABOR FORCE MEASUREMENTS 57 female labor force, they do not provide a basis for estimating the extent of the underenumeration-that part of the difference between the 1979 and 1970 participations which reflects the secular increase in participation cannot be extracted. Thus, no attempt is made here to adjust the female labor force for 1970; the figures given are simply accepted. The participa- tion rates of men in the prime age groups of the labor force are less subject to significant secular trends. I have therefore adjusted the census labor force measures for 1970 (presented earlier in this chapter) by the 419,000 men whom I have estimated above as having been overlooked. The 1980 census results offer grounds for continued uncertainty about the accuracy of census measures of the labor force. The reported labor force appears to be larger than expected. For example, the aggregate labor force participation rate of 50.9 percent is considerably greater than the 1970 census rate of 44.9 percent and the household survey rate of 45.5 percent for the first quarter of 1979. The absolute size of the labor force is given as 22.066 million in June 1980, the month of the census, whereas the estimated size of the labor force during the first quarter of 1979 was 19.839 million. Even if one allows for a generous 6 percent growth over the ensuing year and a quarter, the labor force in June 1980 would still have measured only 21 million. A possible explanation of the larger-than-e:xpected measure may lie in the change in the definition of the labor force. In previous censuses and in the household labor force surveys unremunerated family workers were included in the labor force only if they had been occupied fifteen or more Table 2-15. Labor Force ParticipationRates of Women by Region, January 1970 and First Quarter 1979 January Fir;tquarter Region 1970 1979 Federal District 24.7 34.1 North 15.5 20.9 Northeast 17.0 19.0 Northwest 17.4 16.1 North Central 14.2 14.1 Peninsular 13.3 15.3 Central Pacific 16.5 22.7 Central, 15.6 26.7 South Pacific 14.5 18.2 Central Gulf 13.2 15.8 All Mexico 17.6 21.5 a. Includes those municipios that form part of the Mexico City metropolitan area but that lie outside the Federal District. Sources: Same as table 2-14. S8 EVOLUTION OF THE LABOR FORCE hours during the referenceweek. The 1980 censusremoved this qualifica- tion so that all suchworkers were includedregardlessof the hours worked. Such a definitional change could account for the large increase in the participation rates of both sexes. Unfortunately, it is not possible to determine whether this definitionalchangeis a sufficientexplanation. No information has yet been published on the numbers of hours worked during a referenceweek. In one table the labor force is classifiedaccording to employmentstatus, and one categoryis unremuneratedfamilyworkers. Although the proportion of the labor force of each sex in this category in 1980 is very similar to that of the 1970 census, a very large proportion (21.5 percent of the total labor force and 24.4 percent of the females)is not classified according to status. Even larger proportions of the labor force are not assignedto a sector of economicactivity (26.9 percent of the men and 36.9 percent of the women). Thus, I cannot observewhether there are unusually large concentrations of workers in sectors, such as agriculture and commerce, that often employlarge numbers of unremunerated family workers. Classificationof the Labor Force by Sector of Employment The second problem noted in Chapter 2 related to the difficultiesthat are encountered in relying on census data to trace the evolution of the sectoral distribution of employmentover time. Amongthe obstacles men- tioned were changes in the industrial classificationof individualactivities, the shifting treatment of ejidatarios,the overenumeration of the labor force by the 1960 census, and the large number of labor force members who could not be assignedto a sector of employment.The discussionthat follows is restricted to the last issue.Differenttreatment of the unclassified workers can yield sharply different perceptions regarding the secular changes in the quality of employment in Mexico. Let us examine firstthe sectoral distributions offeredby Earl L. McFar- land. Table 2-16 presents the distributionof the labor force as reported by the censusesof 1950, 1960, and 1970, with adjustments McFarland made in the 1970 census to account for changes in the classificationof subsec- tors. The 1960 census data are presented as reported, without any adjust- ment for the apparent overenumeration of that year. Included in the last column, however, is a revised sectoral distribution of the labor force for 1969. This adjustment reflectshis reallocation of a major portion of the unclassified group among most of the other sectors. The reallocation conforms to the sectoral distribution yielded by a 1969-70 sample survey of family income and expenditures. 4 9 Results of this survey led to the assignment of the bulk of the unclassified workers to the service and  LABOR FORCE MEASUREMENTS 59 Table 2-16. Economically Active Population of Mexico by Sector, 1950, 1960, and 1969 (thousands) Estimated Sector 1950 1960 1969 1969 Agriculture,livestock, forestry, and fishing 4,823.9 6,084.1 5,103.5 5,103.5 Mining and petroleum 97.1 141.1 180.2 178.7 Manufacturing 972.5 1,550.2 2,169.1 2,405.2 Construction 224.5 407.2 571.0 571.0 Electricity,gas, and water 25.0 41.3 53.3 93.4 Transportation and communication 210.6 356.1 368.8 456.9 Commerce 684.1 1,071.9 1,196.9 1,498.0 Services 600.6 1,048.9 2,158.2 1,792.7- 1,903.1 Government 278.8 470.2 406.6 763.3- 652.9 Insufficientlyspecified 355.0 81.6 747.5 92.3 Total 8,345.2 11,253.3 12,955.1 12,955.1 Source:Earl L. McFarland,Jr., "EmploymentGrowth in Services:Mexico 1950-1969," Ph.D. dissertation, Columbia University,1974, tables 1.2 and 1.4. government sectors (for these two sectors,he providesa range of possible numbers rather than a singlefigure).From these distributions,McFarland concluded that there has been an alarmingly rapid rate of increase in the labor force employed in the service sector and that it must represent an influx of persons unable to find satisfactory remunerated employment. Thus, the sector is seen as becoming a repository of labor employed in marginal activities at very low levels of productivity. A very different impressionof the evolution of the labor force by sector of employmentis gained from Altimir's approach. 50 His treatment involves the distribution of the unassignedlabor force for all three censuses.In the cases of the 1950 and 1960 censuses,he is able to reallocateto agriculture and mining a substantial number of workers by cross-referencewith the tables of the occupational groupings of these ;ame people. The balance is distributed over the sectors in the same proportion as the measured employment.The assumption is made that the failureto specifysector was randomly distributed with respectto sector.Altimir's adjustmentfor 1960 also includes a reduction in the size of the agricultural labor force to correct for the overenumeration of this group, an adjustment that McFar- land does not make. The problem for the 1970 census is considerablymore complex than for the others. First, the size of the unclassified group is 60 EVOLUTION OF THE LABOR FORCE much larger. Second, the occupational designation does not lend itself to reclassification since the bulk of those who could not be classified with respect to sector also defy classification by occupational group. Altimir attributes these failures to the long reference period to which the occupa- tional questions referred. The enumerators were expected to record the individuals according to their principal occupation during the whole of 1969 rather than as of a specific date (such as the week prior to the census). He concludes that the shifting of many individuals among sectors during the year complicated their assignment and yielded the large unclassifiable residual. This conclusion leads Altimir to conjecture that, in these cir- cumstances, it is reasonable to assume that the inability to specify a sector is dominated by random elements rather than by a systematic bias that fails to identify correctly persons who are employed in one or more particular sectors. Therefore, he proceeds to distribute the unclassified persons over the sectors in the same proportions as the measured specified labor force. It is the distribution so derived that is included in table 2-4. A comparison of the Altimir and McFarland distributions reveals sub- stantial differences. These differences affect the perception of both pro- ductivity growth by sector and the "healthfulness" of the employment increases associated with Mexico's model of development. Therefore, it is necessary to offer some judgments about the relative merits of the con- flicting estimates. I find it difficult to accept McFarland's assignment to the service sector of most of the unclassified workers of the 1970 census. His reliance on the sectoral distribution yielded by a sample survey of income and expendi- tures as the basis for adjusting the census seems to me to be misplaced. I have found that most Mexican surveys that have as their primary objective the collection of information relating to other subjects produce labor market information of dubious quality. Further reservations stem from an examination of the census and the character of the unclassified. Service sector employment is closely associated with urbanization. Indeed, the common characterization of the service sector as a "sponge" absorbing the ''excess" workers who cannot gain employment in preferred jobs is offered as typical of large urban areas. In addition, as might be expected, measured employment in Mexico's service sector is heavily urban. For example, although only 30.1 percent of the labor force is found in locali- ties with populations of over 50,000, fully 55.1 percent of measured 1 service sector employment is found there." If McFarland's conjecture were correct, it should follow that the unclassified workers should be heavily concentrated in the more urbanized areas. Yet, this does not appear to be the case. Whereas 30.1 percent of the labor force is to be found in localities with populations in excess of 50,000, only a slightly larger proportion of the unclassified-32.5 percent-is found there. At the other end of the  LABOR FORCE MEASUREMENTS 61 spectrum, 39.1 percent of the labor force is in localities with fewer than 2,500 inhabitants while 34.8 percent of the unclassified are reported there. The data for Mexico City also cast doubt on McFarland's thesis. If disguised service employment exists anywhere, one would suppose that Mexico City would harbor a disproportionate part of it. Yet, while the Federal District is reported to hold 17.2 percent of the labor force, it has only 11.7 percent of the unclassified workers. If the data are broken down by sex, a confused picture emerges. Among males, there is a slightly disproportionate share of the unclassified in larger urban areas: 27.2 percent of the male labor forze and 33.5 percent of the unclassified males are found there. Among females, however, the dispro- portionate share of the unclassified is in the rural areas: whereas 21.4 percent of the labor force is in localities with fewer than 2,500 inhabitants, fully 39.7 percent of the unclassified is there, Conversely, in the most urbanized localities, the two proportions are 46.0 and 30.3 percent, respectively. The heavier incidence of unclas:sified workers in less ur- banized areas would seen to offer a rationale for Altimir's treatment. To the extent that rural workers divide their time between agricultural and nonagricultural activities, the specification of a "principal" activity may have posed real difficulties for the enumerators and respondents. Two others aspects of the unclassified labor force may be noteworthy. Over half (56.2 percent) of the unclassified workers report that they were employed in a wage- or salary-earning capacity. One could reasonably expect that, if the unclassified workers were primarily those forced into marginal employment in the service sector, a disproportionately large number would be found in the categories of the self-employed or unpaid family workers. Yet, the proportions of the unclassified (by sector) work- ers in these two employment categories were only a little larger than those for the classified portion of the labor force. For example, 18.8 percent of the classified labor force is listed as self-employed as compared with 23.9 percent for the unclassified. The proportions for the unremunerated family 52 workers are 6.5 and 7.1 percent, respectively. Finally, there is a slightly higher incidence of nonclassification among the youngest and the oldest members of the labor force. These may be 53 rationalized as follows. To the extent that the youngest workers "shop" around for a desirable employment, they will tend to shift jobs over an interval of a year, subsequently making identification of the principal activity more difficult. Similarly, the oldest age groups may be active only intermittently and may show no permanent attachment to any single 54 sector of activity. These observations on the characteristics cf the unclassified workers may imply a less pessimistic appraisal of their employment situation than that offered by McFarland and may provide some justification for Alti- 62 EVOLUTION OF THE LABOR FORCE mir's procedure for distributingthem over the various sectors. It should be recognized, however, that Altimir's figures are also subject to an error of unknown dimension. Nevertheless,the rationale underlying his approach and the ensuing distribution seem sufficientlyconvincing to lead me to prefer his allocation to that of McFarland. Therefore, I have relied heavily on Altimir's work in the text. Notes 1. Enrique Hernindez Laos and Jorge C6rdova Chavez, "Estructura de la distribuci6n de ingreso en Mexico," Comercio exterior, vol. 29, no. 5 (May 1979), p. 506. 2. Luis Unikel, El desarrollo urbano de Mexico, 2d ed. (Mexico, D.F.: El Colegio de Mexico, 1978), table VI-AlS. 3. Luis G6mez Oliver, "Crisis agricola, crisis de los campesinos," Comercio exterior, vol. 28, no. 6 (June 1978), p. 719. The average wage was estimated by dividing total wage payments to permanent workers by the number of wage eamers in the census. 4. Secretaria de Programaci6n y Presupuesto, X Censo industrial 1976 (Mexico, D.F., 1979), table 8. 5. Secretaria de Programaci6n y Presupuesto, VIII Censo de comercio 1976 (Mexico, D.F., 1979), table 17; VII Censo de servicios 1976 (Mexico, D.F., 1979), table 17. 6. Poor families are defined as those households with incomes equal to less than one-half the estimated national mean. The proportions presented in the text should be taken only as rough approximations. The 1975 survey appears to have underestimated severely the in- comes of some groups, largely in the form of income in kind. The results of the 1977 survey, which are thought to be of better quality, are not available for employment of the head of household by sector. 7. Unikel, El desarrollo urbano de Mexico, p. 213. 8. The labor force survey report was originally entitled Encuesta continua de mano de obra. Beginning with the third quarter of 1978, the publication was renamed Encuesta continua sobre ocupaci6n and is commonly referred to by its acronym, ECSO. I refer to both the original and the subsequent publication by this acronym. Both were published by the Secretaria de Programaci6n y Presupuesto in Mexico, D.F. 9. Oscar Altimir, "La medici6n de la poblaci6n economicamente activa de Mexico 1950-70," Demografia y economia, vol. 8, no. 1 (1974), pp. 50-83. 10. Consejo Nacional de Poblaci6n, Secretarfa de Programaci6n y Presupuesto, Centro Latinoamericano de Demografia, Mexico: estimaciones y proyecciones de poblacidn 1950- 2000 (Mexico, D.F., September 1982), pp. 17-18. The population figures appearing in table 2-2 are taken from the official census publications. Revised estimates of total population appearing in this publication purport to adjust the census count for the underreporting of population, particularly of infants and children under 5 years of age. The revised estimates increase the total population by about 1.5 million in 1950 and 1970 and by about 2 million in 1960 and 1980. The intercensal rates of growth, however, are only slightly different from those yielded by the official censuses. 11. Detailed data on sex- and age-specific participation rates are presented in the appendix. 12. Organization of American States, Am&ica en cifras 1977 (Washington, D.C., 1979), vol.3, pp. 52-56. The low recorded participation rates for women probably understate their true participation in productive activity. Their rates are particularly low in the rural states of  NOTES 63 Mexico.It islikely that manyrural womenview their contributionto agriculturalproduction as part of their household duties and not as a separate activitythat would qualify them as members of the labor force. 13. Donald B. Keesing,"Employmentand Lackof Employment in Mexico,1900-70," in Quantitative Latin American Studies: Methods and Findings, ed. James W. Wilkie and Kenneth Ruddle(LosAngeles:University of California,L.A.,LatinAmericanCenter, 1977), pp. 3-21. 14. Secretaria de Industria y Comercio, Direcci6n Gmeral de Estadistica, VII Censo general de poblaci6n 6 de junio 1950, Resumen general (Mexico,D.F., 1953); IX Censo general de poblaci6n 1970, Resumen general (Mexico,D.F., 1972); Carlos Mufioz Izqui- erdo, "Educaci6n, estado y sociedad en Mexico (193(-76)," Revista de la educaci6n superior, no. 34 (1980), pp. 5-56. 15. Juan Diez-Canedoand Gabriel Vera, "La importariciade la escolaridaden la deter- minaci6n del ingreso en Mexico," in Distribucion del ingresoen Mexico, vol. 2, ed. Carlos BazdreschParada,JesusReyesHerolesG. G., and GabrielVera Ferrer(Mexico,D.F.: Banco de Mexico, November 1982), pp. 480, 502-08. 16. World Bank, World DevelopmentReport 1983 (NewYork: Oxford University Press, 1983), pp. 196-99. 17. Secretariade Educaci6nPiblica, Estad(sticas bdsicasdel sistemanacionalde educa- cion tecnol6gica1979-80 (preliminar)(Mexico,D.F., Aagust 1980), pp. 9-10. 18. In fact, the recorded data understate the proportion of the population enrolled in school.The labor force surveygivesprecedenceto the employed status over school enroll- ment of individualsin classifying the population of labor force age.Thus, thosewho are both employed and attending school are recorded only as employed. 19. Ejidatariosare membersof communities,or ejidos,that were granted land by the land reform processto be held in common.The ejidatario receivesrights to use the land but does not have title to it. 20. Altimir, "La medici6n de la poblaci6n economicamenteactiva." 21. In 1960, the question designed to determine the status of the respondent as an ejidatario or otherwise was asked first. The sector of principal activity was left to be determinedlater in the interview.However,manyof the censustakers, in spiteof instructions to the contrary, apparently accepted the status of eiida:ario as indicative of agricultural employmentand did not pursue the occupationalquestionsfurther.In 1970, the order of the questionswas reversed,so that the sector of employmentwasestablishedprior to determin- ing the status as ejidatario. 22. For purposes of this exercise, the respondents who did not specify a sector of employment,0.5 percent of the total, were distributed ovTer the severalsectors in the same proportions as measured employment. 23. The distribution of the gainsof agriculturalproductivitywillbe exploredin Chapter 4. 24. Stylianos Perrakis, "The Surplus Labor Model and Wage Behavior in Mexico," Industrial Relations, vol. 11, no. 1 (February 1972), pp. 80-95;Clark W. Reynolds,The Mexican Economy (New Haven, Conn.: YaleUniversityPress,1970), pp. 84-88; Timothy King,Mexico: Industrializationand Trade Policiessince I940 (London:Oxford University Press, 1971), pp. 26-27. 25. It is possible that the changes in output and productivity recorded for the 1960s understatethe actual rates of increase.The revisednational accounts,whichextend back to 1970, have resultedin an increaseof 6.1 percent in the 1970 GDP overearliermeasures.If the 1969 GNP has been understatedby a similarorder of magn tude,the rates of changein output and productivityappearingin the last columnof table 2-6 would prove to be conservative. 26. Employmentduring 1980 was estimated to have grown by 6.3 percent, or by 0.51 64 EVOLUTION OF THE LABOR FORCE percent a month compounded.Therefore, I reduced the June 1980 estimatedlabor force by 2.57 percent. Secretaria de Programaci6n y Presupuesto, Sistema de cuentas nacionales de Mexico, 1978-80, vol. 1 (Mexico, D.F., 1982), p. 58. 27. Earl L. McFarland, Jr., "Employment Growth in Services: Mexico 1950-69," Ph.D. dissertation, Columbia University, 1974, chap. 1; Saul Trejo R., "Desempleo ysubocupaci6n en Mexico," Comercio exterior, vol. 24, no. 7 (August 1974), pp. 411-16; Comisi6n Consultiva del Empleo, Programa nacional de empleo 1980182, Sintesis (Mexico, D.F.: Direcci6n del Empleo, November 1979), pp. 16-27. 28. Those who subscribe to the view that employment in the tertiary sector is supply determined emphasize the impact on labor supply of the rapid growth of the urban popula- tion. 29. A decline in wages within the tertiary sector would not provide sufficient grounds for concluding that average employment conditions had deteriorated if the growth of the tertiary sector occurred at the expense of an even lower wage agricultural sector. 30. While one might normally expect repair services to be heavily weighted by skilled labor, data to be reviewed below indicate the prevalence of wages that are too low to be consistent with high-skill levels. 31. Grupo de Estudio del Problema del Empleo, "El problema ocupacional en Mexico: magnitud y recomendaciones," versi6n preliminar (Mexico, D.F., n.d.), p. 162. 32. The urban population considered here is that in communities with 15,000 or more residents. A sharper decline in the relative importance of domestic employment can be noted for the Federal District. In 1930, it accounted for 16.3 percent of the labor force. Over the following three population censuses, this proportion declined to 14.3, 10.8, and 8.8 percent. "Anailisis del mercado de los servicios domesticos en Mexico," Cuadernos de empleo, no. 1 (Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, 1976), p. 64. 33. Ibid., p. 79. 34. In view of the multicollinearity present in the multiple regression equation, I also estimated the 1950 employment with a regression equation that omits the income term, but includes the relative price change term. The estimate of 1950 employment yielded a doser approximation of the actual level. A simple regression of employment per thousand popula- tion on the proportion of the population in urban centers took the following form: DS = 1.938 + 0.1761UP, with the coefficient of UP significant at the 1 percent level and an adjusted R2 of 0.597. This equation yields an estimate of only 181,219 domestic servants in 1950. However, the addition of the relative price change term raises the estimate to 324,468, only 4.6 percent greater than the actual level. 35. "Analisis del mercado de los servicios domesticos," p. 76. 36. The procedures followed in this exercise are described in some detail in Chapter 7. 37. Empirical evidence indicates that almost half of the unremunerated family workers in the large metropolitan urban labor markets of Mexico City, Guadalajara, and Monterrey work only part-time, that is, fewer than forty hours a week. In contrast, among hired workers, no more than 20 percent are so employed. Secretaria de Programaci6n y Presu- del empleoy el desempleoen puesto,Informaci6n basicasobrela estructuray caracteristicas lasareasmetropolitanasde lasciudadesde Mexico, Guadalajara y Monterrey(Mexico,D.F., 1979), table 32. 38. Comisi6n Nacional de los Salarios Minimos, Memoria de los trabajos 1972 y 1973 (Mexico, D.F., 1975), p. 358. 39. The 1977 study permitted the calculation of an income elasticity of demand on a cross-sectional basis over fourteen income groups. A regression equation in log form of semiannual expenditures on food and beverages consumed outside the home on semiannual income was run; it yielded an income elasticity of 1.163. Secretarfa de Presupuesto y  NOTES 65 Programad6n, Encuesta nacional de ingresos y gastos de los hogares 1977 (Mexico, D.F., n.d.), table P3.1. 40. The estimating procedure followed here omits a price variable since no data are available regarding the price movements of food and beverages served in commercial establishments. The sources of the data used in these calculations are the same as those reported for the estimating equations for domestic servic-. 41. In the absence of wage distributions by minimum wage zones, it is hazardous to make strong statements about the quantitative importance of wage payments below the legal minimum on the basis of the national average minimum and the national average wage. Many rates of wage payment that lie below the national avlerageminimum wage may either equal or exceed the minimum wage of their particular zone (and the reverse may hold for earnings rates that lie above the national average minimurn wage). However, if the national average wage were to fall "substantially" below the national average minimum wage, it would be likely to reflect payment of subminimum wages within zones, unless there is reason to believe that low-minimum wage zones are "overreprese nted" in the census enumeration. 42. Although the years in which the economic censuses are taken always end in 1 or 6, the data reported refer to the year preceding the census, that is, 0 or 5. 43. Hernandez and C6rdova, "Estructura de la distribuci6n de ingreso en Mexico." 44. All of the estimates of income distribution for Mexico are derived on the basis of assumptions that, by necessity, involve a high degree of arbitrariness. This arises from the large discrepancy, which has increased over time, between the estimates of national income based on the household income and expenditures studies and those derived from the national income accounts. Since the former serve as the basis of estimating the distribution of income, it is necessary to distribute this discrepancy over households in some manner. Different scholars have adopted their own method of distributing the discrepancy with rather sig- nificantly different results for the shares of particular de ciles. A thorough review of these studies is available in Joel Bergsman, Income Distribution and Poverty in Mexico, World Bank Staff Working Paper no. 395 (Washington, D.C., june 1980). 45. Brigida Garcla, "La participaci6n de la poblaci6n en la actividad econ6mica," De- mografia y economfa, vol. 9, no. 1 (1975), pp. 14-15. 46. Oscar Altimir, "La medici6n de la poblaci6n economicamente activa de Mexico 1950-70," Demografia y economia, vol. 8, no. 1 (1974), pp. 50-83; Colegio de Mexico, Dindmica de la poblaci6n de M&ico (Mexico, D.F., 1970). 47. Secretaria de Programaci6n y Presupuesto, Encuesta continua de mano de obra. 48. For a discussion of the difficulties involved in fully and accurately accounting for the labor force participation of women, see Mercedes Pedrero, La participaci6n femenina en la actividad econ6mica y su presupuesto de tiempo, Seri, Avances de Investigaci6n no. 3 (Mexico, D.F.: Centro Nacional de Informaci6n y Estadisticas del Trabajo, 1977). 49. Secretaria de Industria y Comercio, Direcci6n General de Muestreo, Ingresos y egresos de las familias en la republica mexicana, 1969-1970 (Mexico, D.F., n.d.). S0. Altimir, "La medici6n de la poblaci6n economicamente activa." 51. Secretaria de Industria y Comercio, Direcci6n General de Estadistica, IX Censo general de poblaci6n, 1970 (Mexico, D.F., 1972), table 37. 52. Ibid., table 40. 53. Ibid., table 35. 54. The census offers information for three age groups over 65 years of age, and it is for these that the incidence of nonclassification is greatest. In absolute terms, however, the numbers involved in these age groups are rather small. 3. The Underutilization of Labor: Unemployment and Underemployment As WAS PREVIOUSLY MENTIONED, two critical issues for students of Mexican economic development have been the degree to which the ex- isting labor resources have been utilized productively and the quality of the employment available to an expanding labor force. The empirical foundations upon which these concerns rest include the following ele- ments: (1) the observed decline in the labor force participation rates of males and the persistence of low female rates, trends that could mask high and increasing levels of hidden or discouraged unemployment; (2) the secular increase in rates of open unemployment reported by population censuses; and (3) the existence of underemployment as evidenced by the prevalence of less than full-time employment or of low earnings, or both, among a substantial proportion of the employed labor force. Chapter 2 dealt with the first of these elements. It was argued that the trends in male and female participation rates could be interpreted as positive responses to higher incomes and expanded educational opportunities rather than as reflections of deteriorating conditions of employment. Nevertheless, that analysis does not suffice to preclude the possibility that hidden unemploy- ment characterizes a significant proportion of the population of labor force age. Therefore, certain empirical data that permit an assessment of the prevalence of this phenomenon will be examined. This issue will be explored after a consideration of the most obvious form of underutiliza- tion of labor resources-open unemployment. The final form of underuti- lization-underemployment, in several of its presumed manifestations- will be discussed in the final sections of the chapter. Open Unemployment An analysis of secular trends in unemployment in Mexico faces severe obstacles. Reliable data are simply not available on a consistent basis prior 66  OPEN UNEMPLOYMENT 67 to 1973. The censuses since 1940 have presented measures of unemploy- ment, but the concept has varied from one census to another, as have the questions designed to explore the unemployed status of the labor force. Table 3-1 presents a summary of census measures of unemployment from 1940 to 1970, along with two noncensus measures for 1940 and 1979. The census reports yield an upward trend in measured unemployment, although even the highest rate of 3.8 percent reported for 1970 can hardly be considered a troublesome level. Normally. such a rate would not be viewed as departing far from the frictional level of unemployment. The problem of interpreting the historical data is illustrated by the appearance of two measures of unemploymeuit for 1940. The census of that year is the first to have included a clear reference to an unemployed status of labor force members. This category was limited to those who had been unemployed for at least a month-in this census, 1 percent of the labor force. An alternative measure of unemployment for that year was published in a series of annual unemployment rates for 1931-40 at- tributed by James Wilkie to the Direcci6n General de Estadistica.1 The origin of the rates is unknown, however, and some doubts have been expressed about the capacity of the agency t:o have obtained a reliable measure of unemployment. 2 If the "longer term" unemployment of over one month's duration was accurately recorded by the census, then the Table 3-1. Unemployment Rates in Mexico, 1940-79 (percent) Unemployment rate Year Total Male Female 1940a 1.0 1.C 0.7 1940a 3.1 n.a. n.a. 1950 1.3 1.3 1.2 1960b 1.8 1.S 1.3 1970 3.8 2.E 7.S 1979-lst quarter 3.3 2.IF 5.0 n.a. Not available. a. Censusincludesonly those unemployedfor at leas: a month at the time of the census. b. The measure of the labor force used here includes the adjustments to the census made by Oscar Altimir. Sources: Secretaria de Industria y Comercio, Direcci6n General de Estadistica, Censo general de poblaci6n, 1940, 1950, 1960, 1970; Secretal ia de Programaci6n y Presupuesto, Encuesta continua sobre ocupaci6n, 1st quarter 1979; James W. Wilkie, "New Hypotheses for Statistical Research in Recent Mexican History," La tin American Research Review, vol. 6, no. 2 (Summer 1971), table 4; Oscar Altimir, "La medici6n de la poblaci6n economi- camente activa de Mexico 1950-70," Demografia y economia, vol. 8, no. 1 (1974), pp. 50-83. 68 UNDERUTILIZATION OF LABOR Direcci6n's measure of 3.1 percent might very well represent a reasonable estimate of total open unemployment for the year. It would yield a 2.1 percent level of unemployment of a short-term, or frictional, character. If the Direcci6n's rate is thus accepted as a more likely and all-inclusive measure of unemployment for 1940, the onset of the upward trend in the census measures is moved forward to 1950. An interpretation of the census measures for 1950 and 1960 is compli- cated by the lack of precision in specifying the reference period and the conditions defining an unemployed status. It is quite possible that persons who were in fact unemployed may have been classified as not in the labor force. In addition, of course, the 1960 census is plagued by enumeration and processing errors (see the appendix to Chapter 2). The 1970 census is the first to adopt a precise definition of unemployment as inclusive of those who, in the week prior to the census, were without work and had actively sought employment. Perhaps the most striking change that appears in the census data is the emergence in 1970 of an abrupt increase in the female unemployment rate. In prior census years, unemployment among women appeared to be at insignificant levels and always below the male rate. Suddenly, in 1970, it emerges at a level 2.7 times that of males. Is this indicative of the failure of earlier censuses to properly capture the full incidence of female labor force participation and unemployment? Or does it represent a new phenomenon associated with the accumulated changes in labor market conditions and attitudes of women toward employment? In either case, it does not appear that this census measure was a temporary aberration. Throughout the decade of the 1970s, ECSo 3 has reported a consistent and clear tendency for female unemployment rates to lie significantly above those of males, not only in the aggregate (see table 3-1 for 1979), but also in virtually all the surveyed regions of the country. A consistent feature of the existing measures of unemployment is the relatively low incidence of long-term unemployment. We noted above that the 1940 census found that only 1 percent of the labor force had been unemployed for more than one month. In subsequent censuses, the unem- ployed were divided into those who had been so for twelve or fewer weeks and those who had been seeking work for thirteen or more weeks. In 1950, more than two-thirds of the unemployed were in the latter category, equivalent to about 1 percent of the labor force. In 1960 and 1970, however, this proportion shrank sharply to 16.3 and 13.4 percent of the unemployed, respectively; as a proportion of the total labor force, the long-term unemployed represented 0.3 and 0.5 percent in the two years, respectively. The 1970 census provided a more detailed breakdown of the duration of unemployment among those currently unemployed who had  OPEN UNEMPLOYMENT 69 held a job during 1969. Fully three-fourthsof these had been unemployed for four weeks or less, and an additional 15 percent had been unemployed between five and twelve weeks. The relative paucity of long-term unem- ployment suggeststhat an absolute scarcity ol:employmentopportunities has not been a chronic affliction of the Mexican economy. Although existing opportunities may frequently have offered wages that may not have appearedparticularly attractive, the absenceof unemploymentinsur- ance would be expected to render the prospect of continuing unemploy- ment an even less attractive alternative. Beginningin 1973, measuresof unemploymentbecame availableon a quarterly basis with the initiation of the household labor force survey (ECSO). While the first of these surveyswas lirnitedto only the three major metropolitan areas of Mexico City, Guadalajara,and Monterrey, the area covered was gradually extended until, by 1979, the whole country was included. The initial impression gained from the surveys is that the 1973 unemployment rates were significantlyhigher than those reported by the 1970 census. For example, the census reported the rate of unemployment in the FederalDistrict to be on the order of 5 percent in the week preceding the census in January 1970. Yet the first survey of the labor force under- taken by ECSO during the firstquarter of 1973 yieldeda rate of 7.2 percent, a levelaround which severalsubsequentsurveyshovered.It is possiblethat the unemployment rate had increased in the intervalsince the population census. However, I consider it more likely that most of the increase is attributable to the greater thoroughness with which labor force surveys determine the status of the population. It ha:, commonlybeen found that such specializedsurveysare more efficientin identifyinglabor force mem- bers and the unemployed than are censuses. In addition, an important difference in the definition of unemployment probably played a role in raising the survey rate above that of the census-the census classifiedthe population in accordance with its activity curing the week immediately preceding the census. In contrast, the sur,reys extended the reference period to two full months. Thus, anyone who was not working but had looked for employment at any time during this prolonged interval was counted as unemployed.I believethat these iwo factors sufficeto account for the major part of the differencebetween the census and survey mea- sures. Certainly,the subsequent surveysdo not evinceany seculartendencyfor unemployment to rise (see table 3-2). The economic recession that fol- lowed the devaluation of 1976 and the pursuit of increased economic stability by the new administration of PiresidentL6pez Portillo was reflected in a cyclical increase in recorded unemployment in the three major metropolitan regions. But, with the resumption of rapid growth in Table 3-2. Annual Average Unemployment Rates by Region, 1973-79 (percent) 1979 January- Region 1970 1973 1974 1975 1976 1977 1978 March 1. Federal District 7.1 7.6 7.4 7.3 6.7 8.1 7.1 6.3 2. Metropolitan Guadalajara n.a. 6.6 5.8 6.2 6.2 7.4 6.0 6.2 3. Metropolitan Monterrey n.a. 7.5 7.1 7.9 7.6 9.0 7.6 5.7 4. All large urban centers n.a. 5.6 5.5 4.8 3.5 5. Northeast 3.9 6.8a 5.7 4.5 3.9 6. Northwest 3.4 7.5' 5.7 3.7 2.5 7. North 4.0 7.2 4.3 3.7 3.8 8. North Central 2.9 2.6 1.8 1.9 2.1 9. Peninsular 2.0 2.2 1.7' 1.2 10. Central Pacific 3.3 4.3 3.4 3.2 11. Central 2.4 2.8b 2.9 2.3 12. Central Gulf 2.7 1.8' 1.5 13. South Pacific 1.5 0.9 All Mexico 3.8 3.3 n.a. Not available from 1970 census. Note: The regions listed comprise the following: (4) All42 municipios with a population of 100,000 or more in 1970; (5) Tamaulipas, Nuevo Le6n, and Coahuila; (6) Northern and Southern Baja California, Sonora, Sinaloa, and Nayarit; (7) Chihuahua and Durango; (8) Aguascalientes, San Luis Potosf, and Zacatecas; (9) Yucatan, Quintana Roo, Tabasco, and Campeche; (10) Colima, Jalisco, and Michoacan; (11) Queretaro, Guanajuato, Hidalgo, Morelos, Puebla, Tlaxacala, and Mexico; not included are those portions of the state of Mexico that form a part of the Metropolitan Mexico City area; (12) Veracruz; (13) Chiapas, Oaxaca, and Guerrero. a. October-December. b. July-December. c. January-September. Sources: 1970-Secretariade Industria y Comercio, Direcci6n General deEstadistica, IX Censogeneral depoblaci6n, 1970: Resumengeneral (Mexico, D.F., 1972), table 32. 1973-77-Direcci6n de Programaci6n y Presupuesto, Encuesta continua de mano de obra. 1978-79-Direcci6n de Programaci6n y Presupuesto, Encuesta continua sobre ocupaci6n.  OPE.N UNEMPLOYMENT 71 1978, the unemploymentrate declinedthroughout the country. Bythe first quarter of 1979, it had fallen to 3.3 percent, a rate beloweven that of the 1970 census. Clearly, the level of unemployment and degree of change since 1970 has varied by regions. The more highlyurbanized regions tend to report higher rates of unemployment than do the more rural, poorer regions. For example, the Federal District unemployment rate seems to hover at a level about twice the national average. The other two major metropolitan areas also report much higher ra tes in 1979 than the average. The average of the 42 municipalitieswith populations of 100,000 or more report rates intermediate to those of the larg,zmetropolitan areas and the less urbanized regions. Significantly, not a single region for which a comparison could be made reported a higher unemploymentrate in 1979 than in 1970. In view of the much more liberal definition of unemploy- ment adopted by the survey, the real improvement in employmentcondi- tions over the decade is likely to be understated by the decline in unem- ployment rates observed here. Incidentally, the entire reduction in the national rate is attributable to a significant(leclinein the unemployment rate among women from 7.5 to 5 percent (seetable 3-1).The rate for men remained constant at what must be close to the frictional rate. A decline in unemployment rates, of course, may appear to be less encouraging if it is accompanied by a declinein labor force participation rates. In other words, if the unemployed despair of findings jobs and withdraw from the labor force to join the ranks of the discouraged unemployed the measured rate of unemploymentwould decline, but so would the labor force participation rate. It is reassuring to note that in Mexico this has not been the case. Table 3-3 is a summary of changes in labor force participation and unemployment rates by region and sex between the census date January 1970 and the first quarter of 1979.Note that for this purpose, unemploymenthas beenmeasuredas a proportion of the population 12 years and older rather than as a proportion of the labor force in order to preserve a common denominator for the two measures. For the country as a whole, the labor force participation rate increased by 2.1 percent, while the unemployment rate declined by 0.1 percent. The increase in participation was markedly greater among women; neverthe- less, the unemployment rate registered a decline. Regionaldifferencesin the rates of changewould appear to be consider- able. Three regions, the North, Northwest. and North Central, appear with declinesin the aggregatelabor forceparticipation rates whichexceed the declines in the rate of unemployment, a relationship that could be interpreted as an unfavorable development.However, for the latter two regions, the 1979 first quarter observations ,eem to be sharply out of line with those for the immediatelypreceding quarters. For example, a com- 72 UNDERUTILIZATION OF LABOR Table 3-3. Summary of PercentagePoint Changes in Labor Force Participation (LFP)and Unemployment (U) Rates between January 1970 and First Quarter 1979 Total Male Female Region LFP U LFP U LFP U Federal District 3.8 0.8 -3.0 0.9 5.3 0.6 North -0.5 -0.2 -3.6 -0.3 4.4 -0.2 Northeast 0.4 0 0.1 0 2.0 0 Northwest -1.7 -0.9 - 0.5 - 0.9 -1.3 -1.0 North Central -2.4 -0.8 -3.0 -0.4 -0.1 -1.0 Peninsular 1.2 -0.8 2.4 - 0.7 2.1 -0.8 Central Pacific 4.4 0.1 4.9 0.3 6.2 0.2 Central 2.9 -0.1 1.2 0.1 5.9 -0.7 Central Gulf 5.2 -0.4 7.6 -0.3 2.6 -0.9 South Pacific 2.1 -0.6 0.9 -0.1 3.7 -0.6 All Mexico 2.1 -0.1 1.2 0 3.9 -0.2 Sources:Secretariade Industria y Comercio,Direcci6nGeneral de Estadistica,IX Censo general depoblaci6n, 1971:Resumen general (Mexico,D.F., 1972),table 32; Secretariade Programaci6ny Presupuesto,Encuestacontinua sobreocupaci6n,vol. 7, 1st quarter 1979, table 2. parison of the first quarter 1978 with the census measures would have yielded an increase in the participation rates of 1.3 and 1.7 percentage points, respectively,and 0.2 and - 1.1 percentagepoint changesin unem- ployment, changes that are more closely in line with those of the other regions. Thus, it may well be that the 1979 observations are the product of a sampling or processing aberration. In all the other regions, the labor force participation rate increased more rapidly than did the unemploy- ment rate, or increased while the latter declined.In either case,the findings would not be consistent with a hypothesisthat the decline in the unem- ployment rate merely disguises an increase in hidden or discouraged unemployment. Women appear to have experienceda very substantial improvement in labor market conditions. Participation rates were up in almost all regions, sharply in several. In spite of this, unemploymentrates either declined or showed little change. The exceptions were, again, the Northwest and North Central regions. As in the totals for these regions, femaleparticipa- tion rates showed declines in contrast to other recent sampling periods. The first quarter of 1978, for example,reported increasesin participation rates of 1.3 and 1.1points over 1970 and declinesin unemploymentof 0.3 and 1.1 points, respectively. In short, employment opportunities for  OPEN UNEMPLOYMENT 73 women during the decade appear to have expanded more rapidly than have their participation rates; this occasioned the decline in measured unemployment. In the aggregateamong men,employmentopportunities and labor force expanded at the same rate, netting out to no change in the measured unemployment rate. However, there were rather substantial differences among the various regions. The least favorable evolution in employment conditions would appear to have occurred in ihe FederalDistrict.There, a small decline in the participation rate was accompaniedby an increase in the unemployment rate. Otherwise, participation rates increased and un- employment rates either declinedor increased by less than the participa- tion rates. The Northwest and North Central regions again present a divergent pattern from other recent observations. Surveyresults for the first quarter of 1978 yielded an increase in participation of 3.1 and 2.5 percentage points and changes in unemploymentof 0.7 and - 1.1 over 1970 for these two regions. In brief, then, the data do not lend broad and consistent support to a hypothesis that open unemployment has been converted into a more disguisedform. If anything, employmentopportu- nities would appear to have improved. Indeed, given the more liberal definition of unemployment embodied in the 1979 survey data, the im- provement evident in these intertemporal comparisons was more likely to have been understated than overstated. The accelerated growth rates of 1978-81 would be expected to have brought about a further improvement in employment conditions. Unfor- tunately, the national coverageof ECSO was suspendedfollowingthe first quarter of 1979; labor force information for subsequentquarters is limited to only the three major metropolitan regions of the country. The expan- sion of employment in those areas was accommodated by increases in labor participation rates of both sexes and declinesin open unemploy- ment. Participation rates for men increased between 1 and 2 percentage points, those for women increased by a full 2 percentage points. Unem- ployment rates, definedhere in the conventionalmanner as a proportion of the labor force, fell by approximately one-third below the 1979 first- quarter level. If the national rate of unemploymentdeclined in the same proportion, a rate of about 2.1 percent at the end of 1981 and early 1982 would be implied. Unfortunately, the onset of an economic crisis in mid-1982 carried with it a rapid deterioration in employmentconditions. At the end of the year, open unemploymentc!iaracterizedabout 8 percent of the metropolitan labor force. As in most countries, the incidence of unemployment proves to be greater among young people of both sexesthan among the more mature adult population. Table 3-4 provides unemploymentrates by sex and age 74 UNDERUTILIZATION OF LABOR Table 3-4. Unemployment Rates by Sex and Age, January 1970 and First Quarter 1979 (percent) Men Women Age 1970 1979 1970 1979 12-19 5.1 6.5 7.8 10.9 20-24 3.5 4.7 6.5 6.4 25-34 2.2 2.2 7.3 3.4 35-44 2.0 1.4 7.8 2.2 45-54 2.1 1.2 7.6 1.8 55-64 2.2 1.2 7.6 1.0 65 and over 2.1 0.7 8.4 0.7 Total 2.8 2.8 7.5 5.0 Sources: Secretariade Industriay Comercio,Direcci6nGeneral de Estadistica,IX Censo generalde poblaci6n, 1970: Resumen general(Mexico,D.F., 1972), table 32; Secretariade Programaci6ny Presupuesto,Encuesta continuasobre ocupaci6n,vol. 7, 1st quarter 1979, table 2A. as reported by the 1970 census and by ECSO for the first quarter of 1979. Teenage unemployment rates are more than double the average in the more recent survey; they remain above average for the 20-24-year-old group and then taper off to very low levels. The rates yielded by the census reveal a smaller variance than that of the labor force surveys, although the pattern for men is generally similar to that for the more recent period. The census data for women, however, are quite curious. On the one hand, the unemployment rate seems to be invariant over the several age groups. On the other hand, the 1979 pattern exhibited by women in table 3-4 holds consistently both over time and across regions in the surveys. Unfortu- nately, there are no earlier census distributions of female unemployment by age group against which to compare these more recent distributions. The unemployment rates of the youngest labor force members tend to be quite volatile over time and show large variations over the several regions of the country. Teenage unemployment is highest in the large urban metropolitan areas. For example, in the ECSO survey for the first quarter of 1979, Mexico City reported unemployment rates for this youthful group of almost 18 percent. Guadalajara reported rates in the 12-15 percent range, while Monterrey's teenage unemployment rate was in the 15-17 percent range. These rates are in sharp contrast to those prevailing in the rest of the country. In virtually all the other areas, the male teenage rate lies below 10 percent; it dips as low as 2 percent in a few regions. Female rates in most regions lie above the male rates. The male  OPEN UNEMPLOYMENT 75 rates are lowest in those regions that are more rural and in which agri- cultural activity predominates. The labor force participation rates of the teenage males also tend to be higher in these regions. The relationship between the two rates is an inverse one, yielding a simple correlation coefficientof -0.6 that is significant at the 5 percent level. The female unemployment rates are also lower in the more rural regions. However, the labor force participation rates also tend to be lower; thus a positive correlation coefficientof 0.76, significant at the 1 percent level, results. The pattern of teenage unemploymentrates across regions is therefore similar to that for the total labor force. Unemploymentrates tend to be directly related to the degree of urbanization. A probable consequenceof this relationship is that, as the population continues to shift from rural to urban areas and the weight of the latter increa.sesin the national averages, the national unemploymentrate, total and teenage, will be higher than if constant weights were employed. Thus, the secular trend in unemploy- ment may be obscuredunless the data are decomposedby regionor by the degree of urbanization. Some further useful insights into the characteristicsof the unemployed can be gained from a set of special tabulations prepared by the Secretaria de Programaci6n y Presupuesto, based on thc ECSO surveyfor the second quarter of 1978.4 These tabulations were limited to the three major metropolitan areas of the country. While the coverageis thus limited, it does include the regions with the highest rates of unemployment.These data confirm my earlier observation that unemployment is dispro- portionately heavy among the young. Of the unemployed with previous work experience, between 50 and 60 percent are between the ages of 15 and 24. In the absence of more detailedinforrnation about this group, it is difficult to assess the significanceof this unemployment.In more highly developed countries like the United States, higher-than-average unem- ployment rates among young people are often attributed to their search for "desirable" jobs-jobs to which they are willing to make a permanent commitment. When these workers know little about the characteristicsof the available jobs, the search becomes a process of trial and error or movement among jobs. It is quite possible that at least part of this higher incidence of unemployment among Mexican youth is of a similar na- ture. Among those unemployed with no previous work experience, the young again predominate, as might be expeci:ed. Betweentwo-thirds and three-fourths are teenagers; the bulk of the remainder is aged between 20 and 24. It has frequently been observed that the unemployed in developing countries tend to have attained higher levels of education than the em- ployed or the labor force as a whole. In part, this has been explained in 76 UNDERUTILIZATION OF LABOR terms of the higher income levels of families with more highly educated members, a condition that permits the luxury of unemployment and a more leisurely search for the "ideal" job. The survey data for Mexico, however, depart from such a pattern: there appear to be no consistent and significantdifferencesin the educational levelsof the unemployed and the employed. Information about the length of time that elapsedsincethe last previous 5 Unfortunately, this employment is also provided by this special report. time interval is not synonymouswith the duration of unemployment;thus, we have no information more recent than that provided by the 1970 census. As a rough indicator of duration, it may be noted that the majority of the unemployed had been separated from their previous employment within the past six months. The proportions ranged from 55 percent of the unemployedwith previouswork experiencein Monterrey,to 62 percent in Mexico City, and 75 percent in Guadalajara. The presence of numerous individualswho had not held a job for more than two years (or over five), especially among women, suggests that many of these were simply re- entering the labor force after a period of withdrawal. Indeed,among those reporting the longestinterval since the last employment,the predominant reason offered for separation from that employmentwas "personal."' The interval of time sincethe previous employmentseemsto be directly related to the level of skill of the respondent as indicated by the occupational classificationof his or her last employment. 7 For everyone who reported their previous status as an agricultural laborer, the interval between jobs was less than six months; this was alsothe case for about three-fourths of the nonagricultural laborers. At the other extreme, only 42 percent of the professional and technical occupational group in Mexico City and 25 percent of that in Monterrey reported so short an interval. Within the managerial occupational group, 60 and 50 percent did so in these two regions, respectively.It is difficult to believethat prolonged absence of employmentin the last two occupationalgroups could be ascribed to other than voluntary reasons, especially since a substantial number of these individuals had not been employed for over two years. Among the unemployed with previous work experience,the predomi- nant reasons for separation from their last jobs were either personal or attributable to the end of seasonal employment.The former accounted for almost half of the responses, the latter for about one-third. The more detailed information for Mexico City indicates that personal reasons are offered by women twice as often as by men, while men offer seasonal employmentas a reason for termination twiceas often as do women. The data suggestthat close to one-half of the separations were voluntary in all  HIDDEN OR DISCOURAGED UNEMPLOYMENT 77 three regions.From this observation-and the one made previously about the short duration of unemployment-it would seem that workers do not view the labor market as offering very limited employmentalternatives. If workers assigneda low probability to the chances of findinganother job, one would not expect them to abandon so readily a job they already had. Hidden or Discouraged Unemployment Early in this chapter, the possibility was mentioned that labor force participation and open unemployment rates may not provide accurate measures of the quantity of available labor resources.This may hold true especiallyamong women. To the extent that individualsof labor force age believe that employment opportunities are nonexistent, they will neither search for work nor report themselvesas being unemployed.Some may have actually searched for work but given up after an unsuccessfuleffort. These individuals constitute the pool of the hidden or discouraged unem- ployed. It is presumed that they are availablefor employmentand would accept it, were it offered. The labor force surveys have made an attempt to determinethe impor- tance of this category of individuals. All people of labor force age who were neither employed nor actively searching for work were asked, "Would you be willing to accept immediatelyeither a paid part-time or full-time job?" Those who responded in the affirmativewere classifiedas hidden unemployed. As an indication of the size of the group that re- sponded in the affirmative, the first quarterly survey for 1979 yields a figure of almost 950,000, or 2.2 percent of the population of labor force age. If this number were added to both the labor force and the open unemployed, an "unemployment rate" of 7.8 percent would result, a rate more than twice the measured national open unemploymentrate of 3.3 percent. Table 3-5 is concerned with the incidenceof this hidden unem- ployment by sex and region. It would appear to be more prevalent among women than among men: the female national rate was about 25 percent greater than the male. In only three of the thirteen regions does the male rate exceed the female.Moreover, there are substantial differencesin the reported rates among regions. These range from 0.5 to 3.1 percent of the population of labor force age. Differencesamong regionsdo not appear to be related to any obvious regional characteri'stics.The hidden unemploy- ment rate might be expected to vary directly with the open rate of unem- ployment since the prospects of findingjobs might be perceivedas poorer, the higher the visibleunemployment.Yetthis does not seemto be the case. 78 UNDERUTILIZATION OF LABOR The correlation across regions between the rate of open unemployment and the proportion of the population classifiedas hidden unemployed is positive at 0.24 but proves to be statistically insignificant. A closer examination of the data for metropolitan Mexico City from 1975 through the middleof 1979 revealsthat the hidden unemployedhave remained within the range of 1.7 to 2.5 percent of the population. How- ever, fifteen of the eighteen quarterly observations fall within the much narrower range of 1.9-2.2 percent. Rates among males are concentrated within 1.7-1.9 percent of the population. Those for women show a somewhat greater variance, although the bulk of the latter are in the 2.2-2.6 percent range. As in the case of the regional cross section, there seemsto be no consistent relationship between changesin the open unem- ployment rate and the hidden rate over time. Neither the cyclicalincrease in unemployment in 1977 nor its subsequent sharp decline into 1979 seemsto have been accompanied by paralled movementsin hidden unem- ployment. Yet the economicboom that endured through the first half of Table 3-5. Incidence of Measured "Hidden" Unemployment by Sex and Region, First Quarter 1979 Hidden unemployment Unemploy- (percentageof) ment rate population of (percentage labor force age of labor Region force) Men Women Total North 3.8 4.1 1.0 3.1 Northeast 3.9 1.6 2.0 1.8 Northwest 2.5 0.9 2.6 1.7 North Central 2.1 3.5 1.6 2.6 Peninsular 1.2 0.3 0.9 0.5 Central Pacific 3.2 1.8 3.1 2.5 Central 2.1 2.0 3.5 2.8 Central Gulf 1.5 0.7 1.0 0.8 South Pacific 0.9 2.3 2.0 2.2 Metropolitan Mexico City 6.1 2.1 2.5 2.3 Metropolitan Guadalajara 6.2 1.9 2.4 2.2 Metropolitan Monterrey 5.7 1.3 1.7 1.5 All other large urban areas 3.5 1.4 1.5 1.4 All Mexico 3.3 1.9 2.4 2.2 Source:Secretariade Programaci6ny Presupuesto,Encuestacontinua sobre ocupaci6n, vol. 7, 1st quarter 1979, table 2.  HIDDEN OR DISCOURAGEI) UNEMPLOYMENT 79 1982 did appear to have reduced the measured rates slightly in Mexico City and Guadalajara and moderately in Monterrey. The lack of a consistent responsiveness in tb is measure to differences or changes in open unemployment raises some question about what in fact is being measured. One would expect that changes in the objective condi- tions of the labor market would lead individuals to reevaluate the prob- ability of finding employment, with predictable consequences for the rate of hidden unemployment. Of course, a behavioral change would depend on a widespread perception of a change in Iabor market conditions. If actual changes are not perceived or are too small to be taken seriously, then the expected behavioral response migh; not occur. The issue then revolves around the question of what constitutes a significant or percepti- ble change. For example, between the first quarters of 1977 and 1979, the open unemployment rate for the Mexico City metropolitan region de- clined from 8.3 to 6.1 percent. The rates for both sexes fell by about the same proportion, one-fourth. Yet the hidden unemployment rate moved in the opposite direction from 1.9 to 2.3 percent of the population of labor force age, an increase of over 20 percent. One possible explanation for the lack of conformity to the "expected" relationship between open and hidden unemployment may be that the latter is an imperfect measure of the phenomenon it purports to represent. It may be argued, for example, that the responses to such a hypothetical question about availability for employment are likely to be misleading, that many more persons are likely to answer in the affirmative than are actually available. Indeed, this would seem to be the case in Mexico, just as it has been shown to be commonly true of other countries.' The special tabulations based on ECSO'S second quarterly survey of 1978 illustrate that, of those who initially indicated that they were available for employ- ment, only a small minority actually proved 1:o be so.9 In the three metro- politan areas of Mexico City, Guadalajara, aid Monterrey, only 4.1, 6.7, and 7.4 percent, respectively, of the populations were actually available for immediate employment. If one were to add these to both the labor force and to the ranks of the open unemployed, their impact would be minimal. The unemployment rate would be increased on the order of 1 percentage point. Clearly, the vast majority of those classified as hidden unemployed were only conditionally available. The special report does not identify the conditions placed on the respondents' availability nor the reasons for not having actively sought employment. However, it does report the current activity of the respondents so classified. Among men, the overwhelming majority, over three-fourths, were students, a status that effectively pre- cludes an active labor force status except on a part-time basis. Among 80 UNDERUTILIZATION OF LABOR women, involvement in household duties was by far the most frequent current activity. It was cited by over 75 percent of those classified as hidden unemployed; school attendance ranked second. Together, these groups accounted for virtually all of the respondents. It is noteworthy that between one-third and one-half of the respondents had had no previous work experience,and an additional 20-25 percent had not beenemployed for over five years. Over 40 percent were teenagers. These characteristics suggest that those classified as hidden unemployed had only a weak attachment to the labor market. Thus, it is unlikelythat the measured rate of hidden unemployment can be considered an accurate indicator of the number of discouraged workers currently available for employment. It would seem that if the concept of hidden or discouragedunemploy- ment were to have any operational meaning, the true status of those professing their availabilityfor employmentwould need to be established in a more precise manner. Such persons should also be asked why they have not activelysought employment. Only if they respond that they did not do so out of a conviction that no jobs were availableshould they be eligiblefor consideration as hidden unemployed.Another condition that would have to be satisfiedif individualswere to qualify would be that their reservation price (the minimum pay for which they are willing to work) not be above the current market price for the class of servicesthey could offer. Only then could an individual be considered to be involuntarily excludedfrom employment.The practical difficultiesof relating a reserva- tion price to the appropriate wage in the market, however, are over- whelming; consequently, it is rarely, if ever, attempted. The resulting ambiguity in measures of hidden unemploymentthus impedes any assess- ment of its true significance. Nevertheless, it is also clear that the appearance of job opportunities may induce individuals who had not previously contemplated employ- ment to enter the labor force. Given time, some of those who are not currently available for employment might be able to make arrangements that would permit them to seek and accept employment.This would be particularly true of women engaged in household activities(for example, effectivesubstitutes could be found in the form of hired domestichelp or a relative who would release a woman for employmentpurposes). It would seem that the large expansion in output and employmentduring 1979-81 was effected by an increase in active participation rates, particularly in those for women. It is tempting to point to this response as evidencethat hidden unemployment must have been present in significant amounts. However, it does not seem to me that this necessarilyfollows. As noted above, the observed increases in participation rates did not necessarily  UrNDEREMPLOYMFNT 81 come about by the incorporation into the labor force of people who had previously been classified as hidden unemployed. The proportion of the population of labor force age classifiedas such declinedonly slightly,by considerably less than the increasein the participation rate. Furthermore, to the extent that the expansion of employmentopportunities materialized at levels of remuneration higher than those previously available, a supply response would be expected from those whose reservation price had exceededthe previously prevailing and availablewage.The conclusionto be drawn from these observations is that the procedures currently avail- able and used for measuringthe sizeof the potentially availablelabor force are quite inadequate to the task. Data professing to offer such measures should be viewed with great care, if not with skepticism. Underemployment Since open unemploymentin Mexico appears to be low in comparison with that reported for many other developingcountries,concern with the underutilization of labor resourceshas centeredmuch more heavilyon the presumed high level of underemployment. Unfortunately, most discus- sions of underemploymentof labor in Mexico reflectthe same problemsof concept and measurement that prevail generally. No single criterion for the definition of underemployment is generally accepted as an ideal, nor do the available data conform to the requirements of most of the abstract definitions. In spite of the different definitionsadopted by various labor of underemploymentin market studies,a striking feature of their estimiates Mexico during the 1970s is their concentrationwithin a relativelynarrow range. Estimates of the proportion of the labor force that is underem- ployed range from about 45 to 52 percent. It would be premature, how- ever, to conclude that the convergence of the estimates confirms the correctness of these measures. Rather, their similaritymay be due to the inclusion of some common elements or to their derivation from related distributions. A common component of most measures of underemploymentis that portion of the labor force that is employed on a part-time basis during some reference period, for example, a week, a year, or both. Once a normal work week or work year has been definedin terms of "normal" hours or days, it is no difficult task to idertify those who have worked fewer hours or days than "normal." The difiference between a full-time or "normal" work period and the time actuallyworked provides a measure of underemployment.This may be expressedeither as a proportion of the 82 UNDERUTILIZATION OF LABOR labor force or in terms of an equivalent rate of open unemployment."o This is frequently referred to as "visible" underemployment since it is so readily amenable to measurement. If a measure of underemployment is to have any operational value (for example, for determining the employment objectives of a development program), it should relate to the volume of labor services that are poten- tially available to the labor market. In other words, the conversion of all short-time work into unemployment equivalents implicitly assumes that all part-time employment is involuntary. If, on the other hand, some part of such employment is voluntary, and the workers so employed prefer not to work longer hours or more days, the measured underemployment will exaggerate the seriousness of the employment problem. Until recently, only scattered data have been available from census reports about the time intensity of employment of the labor force. Unfor- tunately, different bases of measurement were adopted by the various sources. Therefore, no consistent basis exists for making intertemporal comparisons of the degree of utilization of the labor force. The 1950 population census reported that 7.5 percent of the labor force had worked four or fewer days during the week preceding the census. It appears, however, that this proportion includes the 0.4 percent of the labor force that had been unemployed for more than thirteen weeks as well." The 1970 census offered a different basis for determining underutilization, the number of months in 1969 during which each labor force member worked. A substantial proportion of the labor force, 12.6 percent, is recorded as having worked fewer than six months; another 6.5 percent reported six to nine months of employment. The distribution of men and women over these classes was not very dissimilar, although the incidence of shorter-term employment was some- what greater among women.' 2 As might be expected, the seasonality of employment in the agricultural sector gave rise to a greater incidence of employment of fewer than ten months. Fully 56.6 percent of all men working fewer than ten months had held their major employment in that sector. As a proportion of the men employed principally in agriculture, this group accounted for 23.1 percent.'3 After agriculture, the manufactur- ing sector gave rise to the largest volume of male employment for fewer than ten months. Here, however, the incidence was lower than that in construction, which ranged from 14.8 to 29.2 percent. Amongwomen, the incidence of short-year employment was greatest in the agricultural, ser- vice, and manufacturing sectors. About 21 percent of the labor force in each of these sectors reported employment of nine months or less. Since the censuses make no attempt to distinguish voluntary part-time employment from involuntary, these data are likely to lead to an over-  UNDEREMPLOYMENT 83 estimation of the true extent of underemployment attributable to part- time work. Another shortcomingof the 1970 censusdata as a measure of underemployment is that they are not related to the number of months duringwhich respondents participated in the Labor force. New entrants to the labor force appear throughout each year who will not have been availablefor employmentduring the entireyear. There are also otherswho retire from the labor force during the year. [n the 1970 census, both of these types of groups will have been registeredas having been part of the labor force in 1969 and their duration of employmentduly recorded. If the labor force is growing at about 3 percent per year,the sum of retirees and new entrants could amount to as much as 5 percent of the labor force. Thus, factoring in these entrants and retirees could reduce the volume of "measured underemployment" by approximately one-fourth. The elim- ination of workers who may havebeen voluniarily employedfor only part of a year would further reduce the volume of those who were truly underemployed. An alternative approach to the measurement of underemployment is pursued by the quarterly labor force surveys.The employedlabor force is classified according to the number of hours worked during the week preceding the survey. While the normal wotk week in Mexico generally consists of forty-four hours, for purposes cf ESCO, those employed for fewerthan forty hours are considered as having worked less than full time and are further divided according to whether they are voluntarily or involuntarily so employed. The survey publication for the first quarter of 1979 provides a summary table for the national labor force." 4 Fully 15.7 percent of the employedlabor force worked fewerthan forty hours during the reference week. Women were three times as likely as men to have worked a short week (32.6 to 10.3 percent). It is surprising that the agricultural sector did not report the highest incidence of part-time em- ployment, in spite of the fact that the first quarter of the year is the trough of the agriculturalcycle; only 9.9 percent of sector employmentproved to be part-time. The significanceof this finding,however, is open to question in view of the survey sample's underrepresentation of the rural labor force. The highest incidence of part-time work was found among professional and technical occupational groups, 39.5 percent of which worked fewer than forty hours. These surveys underline the problems that occur from classifyingall people who work a short week as underemployed.Fully two-thirds were found to be voluntarily so employed. Voluntary part-time employment was more prevalent among the women working a short work week than among men, but not by a very largemargin: 73 to 62 percent. Even among the men in agriculture, 63 percent of those working fewerthan forty hours 84 UNDERUTILIZATION OF LABOR were recorded as doing so voluntarily. Taken as a proportion of the employed labor force, workers who were involuntarily employed for fewer than forty hours amounted to only 5.1 percent. Obviously, if this were to be reduced to full-timeunemploymentequivalents,the unutilized proportion of the labor force would amount to only a fraction of this percentage. Disaggregation by region shows some variation in the prevalence of part-time employment.The highest incidenceoccurs in the metropolitan areas of the capital and Guadalajara, on the order of 20 percent of the employed labor force. Yet these regions also report the next to the lowest incidenceof involuntary short-weekwork: around 29 percent. Most of the other observations of part-time workers as a proportion of the labor force cluster closely around the national average of almost 16 percent. The Northwest and South Pacificregions presented the largest departures on the low side, with 10.3 and 11.9 percent, respectively,of their employed labor forces working part-time. While the latter also reports the least amount of involuntary underemployment-19 percent of those working short hours-the Northwest records one of the highestincidencesat 43.5 percent. Note that these two regions provide a sharp contrast in their characteristics.The Northwest, which combines a highly productive agri- cultural sector with important urban concentrations, enjoys one of the highestlevelsof per capita incomesin the country; the latter is largelyrural and very poor. Indeed,it is difficult to observe any systematicrelationship between the recorded levelsof part-time employmentand regional charac- teristics. The more detailed tabulations based on the second quarterly survey of 1979 provide one further basis for disaggregating the part-time workers in the three major metropolitan regions." 5 The employed labor force is clas- sified according to employment status and hours worked. As is usually assumed, a short week is more characteristicof a self-employedstatus and of unremunerated familyworkers than of others. In the Mexico City and Monterrey metropolitan regions, the incidenceof part-time employment among the self-employedwas about 1.5 times the average for the labor force as a whole; in Guadalajara, it was only 1.17 times greater. Among unremunerated family workers, approximately half in each region was employedpart-time. Unfortunately,ECSo does not use employmentstatus to distinguishfurther between those who were voluntarily working short work weeks and those who were involuntarily so employed. The self- employed, however, are also more likely to work significantlybeyond a normal week of forty to forty-eighthours. Whereas the proportion of the employed labor force working over forty-eight hours ranged fell between 26 and 31 percent in the three metropolitan areas, the proportion among  UNDEREMPLOYMENT 85 the self-employedranged between 39 and 46 percent. Among the unre- munerated family workers, the incidence ofE"overemployment" was slightly greater than average in Mexico City, but well below the average for all employed persons in the other two large urban areas. Most of the measures of underemploymentinclude workers who earn an income below some arbitrary level.A receni:estimate for Mexico of 47 percent underemployment adopted the legal minimumwage of each zone as the dividingline between the fully employed and the underemployed." 6 A 1976 estimate by the RegionalEmploymentProgramfor Latin America and the Caribbean (PREALC) adopted the average income of the modal interval in the household income distribution as the socially determined minimum income level. It is this applicationo Ean income criterion to the determination of the status of workers that resulltsin the large proportions of the labor force deemedto be underemployed.The rationale underlying the inclusion of all low-incomeworkers within the ranks of the underem- ployed derives from the presumption that their low productivity is evi- dence that they are not employed in a way that makes full use of their productive capacities. Included among these workers are those whose earnings are low because of part-time employment and those whose productivity and incomes are very low in spite of full-time employment. One can presume that it is this last category of individualsthat makes up the bulk of the group considered to be underemployed. Although an income criterion is widely used for identifyingthe under- employed, the usefulnessof this characterization depends on the use to which it is put. Generally, it is used to derive an estimate of the number of new jobs that must be created in order to ab,orb fully the labor force at "satisfactory" levels of productivity. The underemployed are first con- verted into unemployment equivalents,which then are used to define the number of new jobs required to eliminate underemployment.Adding this number to the expected growth of the labor force and to the number of open unemployed provides a measure of the total expansion required to achieve full employment.The Grupo de Estudio del Problema del Empleo followed this procedure in 1974. They estimated that just keeping unem- ployment and underemploymentat 1969 levelswould require an increase of 6.8 million new jobs between 1970 and 1980. For full employmentto be achieved, jobs would have to be created to address the needs of an additional 5.8 million individuals who were underemployed and who 17 representedthe equivalentof 3 millionunemployed. The estimated inter- censal increase in employment of about 6.6 million might lead one to conclude that employment conditions could not have improved signifi- cantly and might evenhave deteriorated slightly.Yet, in light of the course of wages during the 1970s for various groups reviewedin Chapter 7 and 86 UNDERUTILIZATION OF LABOR the tightness of the labor market during 1980-82 visible evidence of deterioration is not at hand. The use of the earnings criterion as the basis for an estimate of the new job requirements for the economy is very likely to lead to a substantial overestimate of those requirements. In the first place, part of those low earnings are the result of a voluntary decision to limit the amount of time devoted to market employment.Recallthat the bulk of part-time employ- ment was found above to be voluntary. These workers are probably individualswho are not availablefor "fuller" employment.Furthermore, an unspecified proportion of the low-income earners is already fully employed in a time dimension and may not consider itself available for other jobs. Nor can it be taken for granted that those whose current earnings are low have the skills or education necessaryto qualifyfor new, high-productivity employments. Most of the recipients of low earnings have corresponding low levelsof formal education, or they are likelyto be either very young or beyond an age at which an adaptation to new employment opportunities is likely to be possible. These observations receiveconsiderablesubstantiation from a profile of the urban labor force employed at low earnings drawn from the data in a special supplement to the labor force survey ECSO for the last calendar quarter of 1976.'"The supplementwas designedto investigatethe charac- ter of "informal" sector employment. This was defined as employment with earnings no greater than 10percent above the legalminimumwage of the included urban areas (the three major metropolitan areas plus a grouping of 43 municipalities with populations of 100,000 or more in 1970). In addition to low earnings, all those grouped in the informal sector had jobs that lacked at least two of the following attributes: access to prepaid medical services;fringebenefitssuch as paid vacations, retirement benefits,or life insurance; employmentstability definedby referenceto the contractual relationship of the worker to his place of employment; and unionization of the work force. All unremunerated family workers were alsoincluded. Sincelow or zero earnings servedas the common denomina- tor of this population, it qualifiesfor inclusion in the category of the urban underemployed as this group is generally defined." 9 The findingsof the ECSO surveypoint to some of the difficultiesinherent in any attempt to characterize the "employment problem" by referenceto the levelof earnings of labor force members at a moment in time. A review of some of the distinguishingdemographicfeaturesof the recipientsof low earnings follows. An attempt will be made to identify some of the condi- tions in the workers' relationship to their employment that seem to be associated with those earnings.One possibleinterpretation of the data will  UNDEREMPLOYMENT 87 be offered as will a comment on their significar.ce for the phenomenon of underemployment. The ECSO supplement's rough breakdown of the labor force by sex reveals a higher incidence of low earnings among women than among men.Whereas 45.5 percent of all employedwornenreported low earnings, 35 percent of the men did so. The probability of being counted among the low-earnings recipientsis alsogreater the fewerhours worked, as might be expected. One-third of the low-earnings recilpientsworked fewer than forty hours; only one-fifth of those with higher earnings did so. Among part-time workers, roughly half of both sexes reported low earnings. For men working more than forty hours, this proportion falls to 32 percent; among women, it declines modestly to 42.5 percent. 20 Unfortunately, the survey does not distinguish between voluntary and involuntarypart-time employment.It may be presumed, however,that the distribution of these two conditions is similar to that regularlyreported by ECSO. As was mentioned above, two-thirds of part-time workers were voluntarily so employed. If earnings had been converted to an hourly basis, the incidence of low wages would surely have been reduced sig- nificantly. For example, the survey reports that almost one-tenth of the low earners were drawn from the occupational ranks of professionals, managerial or supervisory personnel, and clerical workers, all of whom should normally expect rates of earnings well above the legal minimum wage." As noted earlier, however, these are occupational groups among which the incidenceof part-time employment is highest, a factor that may account for their inclusion in the informal sector. Surely, one would not usually expect to find these occupational groups contributing to a general problem of underemployment. Among the demographic characteristicsthat distinguishthe low earners from the rest of the labor force are their relative youth and low level of formal schooling or technical training. The incidence of low earnings is clearly a declining function of age. Ninety-three percent of those in the labor force under 15 years of age qualify as informally employed,as do 66 percent of those 15 to 19 years old and 35 percent of those 20 to 24 years old. Thereafter, the incidence declines, only to turn up again among workers 55 years and older. If viewed in a differentlight, fullyhalf of the informally employed are either under 25 or over 55 years of age. These same age groups form 38 percent of the total urban labor force. 22 With respect to the educational attainments of the low-earningspopulation, the following associations emerge: the highest iricidenceof low earnings, 66 percent, is to be found among those with no formal education, followed by a 55 percent incidence among those with incomplete primary school 88 UNDERUTILIZATION OF LABOR education. Expressed in other terms: although workers with an incom- plete primary school education formed 38 percent of the total urban employed labor force, they made up 57 percent of the informal labor 23 force. Related to these findings is another: the younger the age of entry into the labor force, the greater the probability that the individual will realize low earnings. Whereas almost half of those entering before age 16 receivedlow earnings at the time of the survey, only one-quarter of those beginning work between ages 18 and 29 did so. An apparent anomaly appears in the statistic showing that over half of those entering the labor force after age 30 received low earnings. Almost all of these individuals were women, but nothing can be inferred about their personal or job characteristics. Unfortunately, the findingsof the ECSO study are presented in the form of simple bivariate distributions of the population. No multivariate analy- sis was undertaken that might have enhanced our understanding of the phenomenon of low earnings, and this limits our ability to speak with authority on its significance.Nevertheless, the information does at least provide a basis for a tentative interpretation of this phenomenon that stands in sharp contrast to the explanation based on market failure. The relative youth and poor education of the low earners suggest that they possess only modest endowments of human capital. To the extent that these workers are employed in occupations utilizing general skills-a likelyevent-the low earnings may simplyreflectthe incidenceof the costs of training. Human capital theory holds that the costs of acquiringgeneral skills will be borne by the individual. They will be financed either by out-of-pocket payments for institutionally provided training or by the acceptance of a wage during the training period that reflects the low marginal product of an inexperiencedworker. 2 4 Particularly among those with very little or no formal education, the concept of "general skills" may be extended to include the development of attitudes and personal discipline that are prerequisites to the finding and holding of a well- remunerated employment. What the data suggestis that, with the passageof time and an increasein work experience, workers do improve their earnings. The cross-section data for males indicate that the chances of holding a job that pays at least 110 percent of the minimum wage vary with age. They are only one in twenty for teenagers between the ages of 12 and 14; three in ten for those aged 15 through 19; two out of three for workers 20 through 24 years; and three out of four for those in the prime labor force age groups.' Further support for this human capital interpretation of low earnings is supplied by the findingthat wages are an increasingfunction of tenure in the current employment.Whereas S6 percent of workers holding their current jobs for  UNDEREMPLOYMENT 89 less than one year reported low earnings, this proportion falls to one-third of those with two years' tenure, one-fifth of those with four to ten years', and 14 percent of those with longer tenure. 2" Formal education or technical training ar pears to serve as a partial substitute for experience,although to what extent this may be so cannot be determined from this survey without more detailed cross-tabulations or multivariate analysis.The data suggestthat low earnings are by no means a permanent condition for the bulk of the male wage earners in the labor market. Rather, there appears to be considerablemobility up the earnings ladder in the urban labor market. In short, for most male entrants to the labor market, low earnings appear to be only a temporary phenomenon and hardly indicative of a chronic condition over a working life. If viewed in this light, the measures of underemployrnent based on an earnings criterion appear to include a substantial number of individuals in the process of acquiring general skills and bearing the costs of training. The earnings situation for women exhibits features similar to those for men but in a less definitive form. The earnings data for females are muddied by the greater incidence of part-time employment and the ab- sence of tables relating earnings to hours worked. This problem was present among males as well but to a lesser extent. Whereas only 20 percent of men reported fewerthan forty hours of work, over 36 percent of the females did so. Nevertheless, the incidence of low earnings among women drops very sharply as age increases,up to the age of 30; this closely resemblesthe findingsfor men reported above.After age 30, the number of low earners increases to around half of the remaining age groups. Since these are the ages during which most women are likely to bear substantial domestic responsibilities,the increased incidenceof low earnings may be heavily influenced by the frequency of part-time employment. Further- more, the female data are heavily weighted by the preponderance of women with very short tenures in their current employment.In fact, if the incidence of low earnings among women is related to tenure, their situa- tion matches very closelythat for men in this tespect. Thus, tenure may be the principal factor accounting for the larger concentration of women in the low-earnings category. In view of these observations, the use of an earnings criterion for definingunderemployment would seem to lead to an exaggerationof the extent of an employmentproblem and to a mnisspecification of its nature. The fact that low earnings prevail within a segmentof the labor force does little by itself to advance our understanding either of the workers' rela- tionship to the labor market or of the appropriate policies required to ameliorate employment conditions. Indeed, the consistent use of income as a criterion in definingunderemploymentwould permanently consigna 90 UNDERUTILIZATION OF LABOR large portion of the employed labor force to that status-regardless of the country's stage of economicdevelopmentor its current levelof income. In the absence of complete wage equality, even the most highly developed country will have a substantial body of workers employed at earnings below some modal or other arbitrary level.Alternatively,the incidenceof low earnings might be considered as indicative of the extent of poverty. Even for this purpose, however, earnings of individualsare likelyto be an imperfect guide to the incidenceof poverty; poverty is more appropriately defined on the basisof household incomes.From the demographiccharac- teristics of this sample of low earners-most of whom are youths and women-one may surmise that a signficant proportion of these were secondary wage earners rather than the sole sourcesof household support. A brief review of the various estimates of underemploymentadvanced for Mexico illustrates the difficultiesconfronting attempts to measure the phenomenon. In 1976 PREALC published an estimate of Mexican under- employmentrates that prevailedin 1970.27 Differentmethods were used to estimate rural and urban underemployment.In the caseof the agricultural sector, the extent of underemploymentwas calculated by subtracting from the man-days of labor available to the sector the number of man-days required to produce the sector's output. The latter requirement was de- rived from estimates of the production functions for different crops and different size farm units. The labor input coefficientsof these modelsserve to define the labor requirements of the sector. The available supply of labor was taken to include all male adults. It was further assumedthat "the work of women and children is available according to labor demand."" 8 This method yields an estimate of the agricultural labor force that is underemployedequal to 46 percent of the total, a proportion that includes rural open unemployment as well. One of the difficultiesassociated with the estimation of labor require- ments in this way, however, is that it may not capture all of the productive activitiesundertaken by household members.Labor time devoted to home production or to activities ancillary to agricultural production (such as marketing) is quite likely to escape detection and inclusion in these esti- mates. Furthermore, what appears to be a surplus of labor to the agricul- tural sector may not prove to be a surplusto the economyas a whole. Since seasonal movementsof labor from the agriculturalto the nonagricultural sector are a common phenomenon in Mexico, it cannot be assumedthat the agricultural labor "surplus" is idle when not directly employed in agricultural production. Finally, some of the labor force that is available for peak season employmentin agriculture may not be availablefor work at other times. The reported resistance of small farmers to technological  UNDEREMPLOYMENT 91 changes that would require increased labor may be a further indication that farm familiesdo not viewthemselvesas repositoriesof surpluslabor. 29 There is some precedent for the beliefthat the application of technical coefficients to an estimate of labor requirements is likely to lead to an overestimation of the amount of nonproductive time of agriculturalwork- ers that might be available for application elsewherein the economy. A pair of studies of the agriculturalsector of Guatemala illustratesthe point. Based on alternative measures of labor supply and on the technological labor input requirements of the sector, one study concluded that, about 1970, the proportion of availablelabor actuallyutilizedranged from 48 to 70 percent.3 0 A reasonable implication that might be drawn from these estimates is that a substantial body of manpower resourcesis availablefor other productive purposes. However, another study undertaken in the middle 1960s by the Comite Interamericano de Desarrollo Agricolafound that, once all the various tasks and obligations of typical minifundistas were taken into account, including market and home production as well as community services, only 13 percent of their time was unoccupied. An unexpected findingwas that the highland minifundistaswho would ordi- narily be consideredunderemployed actuallyhired outsideworkers for up to 35 percent of the labor required on their plots.31 It should not be surprisingto findin Mexico, as well, a much smaller surplusof labor time than that yielded by production models of the agriculturalsector. Indeed, the evidencepresented in Chapter 4, which deals in greater detail with that sector, would appear to confirm this expectation. In the PREALC estimate of underemployment in Mexico, a different methodology was applied to the urban sector. The sourceof the estimates was the household survey of income and expenditures for 1968.Underem- ployment was definedexclusivelyby reference to income.Sincethe income data referred to household rather than individual worker incomes, an adjustment had to be made according to the presumed number of em- ployed members per household. This study chose to assume that the number of labor force members per househcld was inverselyrelated to household income. 3 2 Unfortunately, the basis;for this choice is not dis- cussed. A contrary assumption would seem more reasonable,namely, that a larger number of labor force participants is more likelyto yield a higher level of household income rather than a lower one. This was clearly revealed to be the case by a household income and expenditure study of urban households in Costa Rica. 3 3 The number of employedpersons per household increased progressively,from 0.87 7:n households with monthly incomes of under 500 Costa Rican colones tD about two workers at an income level of 2,500 Costa Rican colones. At higher income levels, the 92 UNDERUTILIZATION OF LABOR number of workers stabilized at about two. To the extent that household incomesin Mexico are likewisedirectlyrelated to the number of employed members, PREALC's assumption would have the effectof understating the earnings of workers and overstating the extent of underemployment.A further bias in the same direction is the result of the inclusion of an unknown number of household memberswhose low incomes were not the product of earnings, but of low nonlabor incomes such as retirement pensions. As indicated above, the incomelevelby which underemploymentwas to be estimated was the average income of the modal interval in the urban household income distribution. One-third of the workers in the modal income group and all of those in the still lower income groups were then deemed to be underemployed.To be sure, the choiceof an income level is bound to be quite arbitrary; there is no obvious choice. In this case, the selection of the modal group is an accident of the choice of the limits of each income class. Altering these limits could generate a whole family of modal incomeintervals. Choosingthe appropriate interval from such a set of intervals would pose a severe challenge. The application of this procedure to the urban sector yielded a figure of 27.9 percent of the labor force as underemployed. In terms of open unemployment equivalents, this amounted to an unemployment rate of 11.8 percent. For the rural sector, the open unemploymentequivalentwas 35 percent; the weighted average for the Mexican labor force as a whole amounted to 25.4 percent. 34 Tlhevagaries of measures for underutilization of labor are illustrated by PREALC'S later estimate for 1970, the same year of the estimate given immediately above. The criteria employed in this later study are quite different from those applied earlier, and the extent of underutilization of the labor force assumes different proportions. 3 5 The openly employed form an obvious component of the underutilized labor force. Underem- ployment is defined differently for the rural and urban sectors. 3 6 Rural underemployment includes all workers who are self-employedin tradi- tional agriculture, as well as all unpaid family members (adjustment is made for an imputed additional participation of women and minors).The urban informal sector, the repository of the urban underemployed, in- cludes the self-employedand unremunerated family members, but not people in professionaland technicaloccupational groupswho report these same two employment statuses. Nor are salaried workers with low earn- ings included in this characterizationof the underemployed.The portions of the labor force thus deemed to be underemployedare 53 percent of the rural and 35 percent of the urban, figures significantlygreater than those given in the earlier study. However, the method used to convert the  UNDEREMPLOYMENT 93 underemployed into open unemployment equivalents yielded a measure below the 25.4 percent of the earlier study. The open unemployment equivalent of the underemployed is given as 15.3 percent of the labor force. When the rate of measured open unemployment is added, a total open unemployment equivalentof 19.1 percent results. This study charts the changes in the degree of underutilization from 1950 to 1980 and concludes that it has declined from 23.7 to 17.0 percent over that period for the labor force as a whole. 37 However, the average conceals two divergent trends. A sharp decline in rural underemployment has been partially offset by an increase in urban. The Grupo de Estudio del Problema del Erffpleoconcluded that in the early 1970s underemployment characterized about 44.8 percent of the national labor force. 3" Regional variations were considerableand ranged from 33.4 percent in central Mexico to 61.5 percent in the southern peninsula. Rural rates were higher than those for urban. The underem- ployed were defined to include all part-time workers plus those earning less than the legal minimum wage. Since no distinction could be made between those workers who were voluntarily working part-time and those who were not, the measure is upward biased. Using the legal minimum wage to distinguish between the underemployed and fully employed is again arbitrary. The application of this criterion guaranteesthat the degree of underemployment will fluctuate with the current relationship of the minimumwage to the market price of unskilledlabor. For example,in the late 1930s,when the legal minimumwage wa:, far above the market price of labor, the measured underemployment would have encompassed an overwhelmingproportion of the labor force. That proportion would have fallen dramatically over the followingdecade as the real value of the legal minimumwage declinedand the market price of unskilledlabor surpassed the legal minimum. Therefore, the utility of su,cha criterion is in question. It could yield widely different perceptions of the employment problem even though the objective conditions of the labor force might have changed very little. Dr. Saul Trejo's estimate of underemployment in 1972 was derived from the proportion of the labor force that was employed at a capital- labor ratio below the average for the economy as a whole. This method yielded an estimate of 52 percent of the labor force. 39 No attempt was made to convert this proportion into its open unemployment equivalent. Since the amount of capital with which labor is combined tends to be directly related to the productivity and incomeof labor, it is not surprising that this method should yieldan underemploymentratio that departs only modestly from those based on an income criterion.However, this measure can only account for the quantity of physical capital combinedwith labor 94 UNDERUTILIZATION OF LABOR and ignores the human capital endowment of workers. It may be this omission that accounts for the larger rate of underemploymentyielded. Would underemploymentbe eliminated if all workers were provided with an amount of physical capital equal to the average for the economy? Clearly not, if the criterion were applied consistently. In fact, this would only increase the proportion of the labor force that would have to be considered underemployed-the remedy would have had the effect of raising the average capital-labor ratio, and a larger number of workers would now lie below the average. The supplementto the ECSO survey citedabove yieldedan underemploy- ment figure of 38.2 percent of the labor force (for urban centers with populations of 100,000 or more). 4 0 This proves to be significantlylarger than PREALC'S estimate of 27.9 percent for the urban labor force and may simply reflect ECSO'S choice of a higher income standard. No basis is provided for the conversion of this percentage into open unemployment equivalents. The object of this discussionhas not beento belittleprevious attempts to characterize or quantify the employment problem. Rather, it has been to illustrate the difficulties of developing a meaningful measure that is not subject to variations unrelated to actual conditions in the labor market. For example, the use of an earnings criterion that is tied to a certain concept of average income could lead to a paradoxical conclusion-that employment conditions were deteriorating and underemployment in- creasing, even though real wages of all workers were actually increasing steadily.This would be the case if a rising generalwage levelwere accom- panied by widening skill or income differentials. Even if the relative distribution remained constant as wages rose generally,no improvement in the condition of underemploymentwould be apparent. Surely, a proce- dure that ignores absolute changes in the well-being of workers leaves something to be desired. This method, if applied consistently, would preclude the attainment of full employment. How will we know when underemployment has been eliminated? Disguised Unemployment One of the components of underemploymentthat merits separate atten- tion is disguised unemployment.This refers to a phenomenon that might be categorizedas "make-work" employment.It is presumed to exist in a labor market that offers so few salaried employment opportunities that individuals are forced to create their own jobs. These are usually of a casual nature and presumably at very low levels of productivity and  DISGUISED UNEMPLOYMENT 95 remuneration. Itinerant vendors, street entertainers, bootblacks, parking guards, and an assortment of other offerers of goods or services are frequently assumed to constitute this category. Unremunerated family workers are also sometimes included here. An essential element in the characterization of the disguised unemployed is the assumption that the activityperformed is not the product of a voltntary choicebut rather one of last resort. Relativelylittle empirical evidenceis availablefor testing the validity of the assumptions regarding this component of the labor force. Therefore, it is of interest to review the findings of a recent survey of self-employed casual workers in several areas of metropolitan Mexico City.4 ' The study undertaken by the Department of SocialStudiesof Banamex, one of the leading banks in the country, was directed toward "those persons with the following characteristics: self-employed,unattached in any formal way to a firm or any other ide:3tifiableestablishment, and whose earnings from work are unstable" (translation mine). 4 2 Two sepa- rate surveyswere undertaken: one in January 1982 when economicactiv- ity and employmentwere at very high levels,the other in Septemberof the same year during an economic crisischaracterized by a sharp contraction of the economy and rapidly rising unemployment. One hundred and fifty individuals were interviewedin the first survey and an additional hundred in the second. Those intervieweddid not constitute a scientifically random sample of the designatedpopulation. Rather they were selected at random on the streets in various sections of the city by the interviewers.'Whilethe nonrandom nature of the samplebecomes evidentin the sharp contrasts in personal characteristicsof the two sampled groups, much less variation is evidenced in the responses to most of the questions posed. The intervieweesexhibited a certain degreeof ambivalencetoward their current labor market status. On the one hand, a majority indicated that they would like to hold paid employment for its greater security, frinige benefits, and training opportunities. On the other hand, these workers professed an overwhelming preference for self-employmentbecause the work hours were flexible and they were free of supervision. 4 3 Indeed, virtually all of those interviewedliked their current occupations." Nor can it be said that those surveyed labored for long hours at low rates of remuneration. In both samples, the median hours worked were between seven and eight per day; 40 percent of the workers in each sample averred that they were not available for full-time employment. 4 5 An examination of the reported earnings reveals that the remunerations can hardly be consideredmarginal. Only 17 percent of each of the two samplesindicated monthly earnings of less than Mex$5,000 at a time when the legal mini- mum wage was Mex$6,720. In view of the large number of part-time 96 UNDERUTILIZATION OF LABOR workers, however, this rate of earnings probably excecdedthe minimum wage in most cases. The median rate of earnings for the January sample approximated Mex$8,800. By September it had risen to Mex$10,000, while the legal minimum had been increased to Mex$8,730 pesos. Four- teen and 21 percent of the two samplesreported monthlyearnings of more than Mex$15,000. The average daily earnings of virtually every occupa- tional group exceeded the legal minimumwage. Only the "fire-eaters," a small group of very young teenage street entertainers, reported earnings well belowthe legal minimum.Casual observationsuggests,however, that they work only a few hours a day during the heaviest commuter traffic. Even bootblacks and those dependent on gratuities-for example, auto watchers and organ grinders-reported daily earnings in excess of the legal minimum .46An interesting finding was a lack of a close correspon- dence between educational and earnings levels in this particular labor market. Nor can one characterize the sampled individuals as coming from impoverished households. The incidence of additional workers in the households of the sampled workers was quite high. In the households of the 150 workers sampled in January, there were eighty-two other em- ployed persons; in the 100 households of the September sample, there were eighty. The earnings distributionof employedhouseholdmembers in the January survey resembledcloselythat for the interviewedworkers. The corresponding distribution from the later surveygenerallyrevealedhigher earnings among household members than among the interviewed.How- ever, the interviewed group was unable to specifythe earnings of approx- imately one-fourth of the employed household members. 47 In short, the results of the Banamex survey do not lend support to the classificationof casual workers as truly marginally employed or as dis- guisedunemployed. On the contrary, they appear to have a strong attach- ment to their present employment, either because their earnings exceed their opportunity cost or because they value highly the freedom and flexibilitythat self-employmentallowsthem. Although these findingsmay be challenged on the grounds that they are not the product of a scientific sampling process and thus may be suspect, there is a sufficient degree of consistency in the respondents' replies to warrant questioning the more conventional view of the casual self-employedworker. This is clearly a segment of the labor force deservingof greater empirical investigation. 48 Summary My findings for Mexico indicate an absence of a serious problem of open unemployment. Unemployment rates for the mature adult labor  SUMMARY 97 force have approximated frictional levels, and no secular trend in the average rate is discernable. There appears to be no real shortage of employment opportunities. Periods of unernployment tend to be very short, and workers seemto have sufficientconfidencein a high probability of reemployment that they are not loath to quit jobs voluntarily. Underemployment has been held to be the greater problem as evidenced by the incidence of less-than-full-timeemployment and low earnings. Other studies have set the degree of underutilization of the labor force at an open unemployment equivalent rate of 25 percent. I have offered reasons for believingthat this greatly exagge:rates the degree of underuti- lization and also fails to provide a reliable measure of the potential supply of labor that is readily available to new sources of employment. Within the urban labor force, too little attention is givento the possible reasons for low earnings and to the personal characteristics of those receiving them. A proper consideration of these factors might lead to a very different appreciation of the significanceof the observed low earn- ings. Given the youth of many urban low earners, the state of "underem- ployment" may reflecta transitory phenomenon-a stage in the working livesof young people during which they acquire generalskillsand bear the cost of doing so. The prevalence of individualswith very low educational levels speaks of human capital deficiencies.Ihese may be overcome with age and increased work experience in many cases but, in others, may permanently disqualify workers from significant improvements in em- ployment status. Underutilization among the rural population is usually estimated with arbitrary criteria that cannot fullyaccount for the wholeof the productive activity of the agricultural labor force. Whileone may be certain that low earnings are more prevalent in rural than in urban areas, it does not follow that low earners are either immediatelyavailableor qualifiedfor employ- ment in newly created jobs at high levels ol0productivity and earnings. Chapter 4 deals exclusivelywith the agriculturallabor force and includesa discussionof the reliabilityof conventionalmeasuresof underemployment as applied to this sector. The limited usefulnessof current measures of labor force underutiliza- tion for predicting the effective availability of labor was clearly demon- strated by Mexico's experience during the recent economic boom. As noted in Chapter 1, the acceleration of the growth process that got under way in 1978 led to widespread labor shortages by 1980. Clearly,the large labor surplus that was thought to exist failed to materialize. Rather than providing a reliable measure of the labor available for employment,the commonly applied earnings criteria for measuringunderemploymentmay be more appropriate for the identification of the population in poverty. The shortcomings in using individual earnings for this purpose, however, 98 UNDERUTILIZATION OF LABOR have already been indicated. Little can be gained from reclassifyingthe phenomenon of low earnings or poverty as underemployment.It contrib- utes nothing to an understanding either of an individual's relationship to the labor market or of the appropriate policies for the alleviation of poverty. Notes 1. See James W. Wilkie, "New Hypotheses for Statistical Research in Recent History," Latin American Research Review, vol. 6, no. 2 (Summer 1971), table 4. 2. Donald B. Keesing, "Employment and Lack of Employment in Mexico, 1900-70," in Quantitative Latin American Studies: Methods and Findings, ed. James W. Wilkie and Kenneth Ruddle (Los Angeles: University of California, L.A., Latin American Center, 1977), p.5. Keesing observes that the trend in the unemployment series for the 1930s does follow the recorded changes in output of the economy. However, he was unable to venture a judgment regarding the accuracy of the recorded levels of unemployment. 3. ECSO is the acronym for both Encuesta continua de mano de obra and Encuesta continua sobre ocupaci6n; see Chapter 2, note 8. 4. Informaci6n basica sobre la estructura y caracteristicas del empleo y el desempleo en las areas metropolitanas de las ciudades de Mexico, Guadalajara y Monterrey (Mexico, D.F., n.d.), table 46. Cited hereafter as Informaci6n basica. 3. Ibid., table 58. 6. Ibid., table 49. 7. Ibid., table 50. 8. Regional Employment Program for Latin America and the Caribbean (PREALC), Em- ployment in Latin America (New York: Praeger, 1978). The authors conclude that, "as for hidden unemployment, most recent investigations tend to show that its quantitative sig- nificance is relatively slight" (p. 1). 9. Informaci6n basica, table 59. 10. On the basis of the work week, this conversion is made by first summing up over the employed labor force the number of hours by which workers fall short of the normal work week. That sum is then divided by the length of the normal work week. The dividend would represent the equivalent of some number of fully employed persons. 11. Secretaria de Industria y Comercio, Direcci6n General de Estadistica, VII Censo general de poblaci6n, 1950 (Mexico, D.F., 1952), p. 240. 12. IX Censo general de poblaci6n, 1970, p. 723. 13. While respondents were classified by the sector of their principal employment, the duration of employment was intended to apply to all jobs held, irrespective of sector. 14. Ecso, vol. 7, 1st quarter (1979), table 10. 15. Informacion basica, table 10. 16. Comisi6n Consultiva del Empleo, Programa nacional de empleo 1980182, vol. 1 (Mexico, D.F., 1979), p. 42. 17. Grupo de Estudio del Problema del Empleo, "El problema ocupacional en Mexico: magnitud y recomendaciones," versi6n preliminar (Mexico, D.F., n.d.), p. 41. 18. Secretaria de Programaci6n y Presupuesto (spp) and Secretaria del Trabajo y Previsi6n Social (STPS), La ocupaci6n informal en areas urbanas, 1976 (Mexico, D.F., 1979). 19. The adequacy of the attributes selected by the survey authors for defining the informal  NOTES 99 sector or, for that matter, the usefulnessof the conceptof an informal labor market is not a matter of concern here.The relevanceof the study for my purposes rests on the inclusionof that segmentof the labor force employedat low rates of earnings. 20. sppand STPS, La ocupaci6n informal en areasurbanas,p. 191. 21. Ibid., p. 170. 22. Ibid., pp. 126-31. 23. Ibid., pp. 137-45. 24. Gary S. Becker, Human Capital (New York: Columbia University Press, 1964), chap. 2. 25. srp and STPS, La ocupaci6n informal en areas urbanas,p. 127. 26. Ibid., p. 233. 27. PREALC, Employment in Latin America,chap. 1. 28. Ibid., app. A. 29. Teresa Rend6n, "Utilizaci6nde mano de obra en Ia agriculturamexicana,1940-73," Demografiay economia,vol. 10, no. 3 (1976),p. 353. -lowever, this resistancemay simply be indicativeof the perceivedmarginalproduct of that additional labor input applied to the family plot. If the marginal product is "small," it ms.y not be viewed as compensating adequately for the disutility of the extra effort required. 30. Erik Thorbeckeand EverardusJ.Stoutjesdijk,Employmentand Output:a Methodol- ogy Applied to Peru and Guatemala (Paris: Organisation for Economic Co-operationand Development,1971), pp. 104-10. 31. Comite Interamericano de Desarrollo Agricola,Tenencia de la tierra y desarrollo socioecon6micodel sector agricola:Guatemala(Washington,D.C.: Pan AmericanUnion, 1965), pp. 100-02 and 159-61. The term minifundistacan be translated looselyas "small subsistencefarmer." 32. PREALC, Employment in Latin America,p. 20. 33. Peter Gregory, "Legal Minimum Wages as an Instrument of Social Policyin Less DevelopedCountries, with Special Referenceto Costa Rica," in The Economicsof Legal Minimum Wages, ed. Simon Rottenberg(Washington,I).C.: AmericanEnterpriseInstitute, 1981), pp. 398-99. 34. PREALC, Employment in Latin America,p. 25. 35. PREALC, Dindmica del subempleo en American Latina, Estudios e Informes de la CEPAL, no. 10 (Santiago:United Nations/International Labour Office, 1981), pp. 29-30, 81-101. 36. For purposesof this study,the so-calledurban sector actuallyencompassesthe entire nonagricultural sector. 37. PREALC, Dindmica del subempleo,p. 26. 38. Grupo de Estudio, El problema ocupacionalen Mexico, p. 37. 39. Saul Trejo R., "Desempleoy subocupaci6nen Mexico," Comercioexterior,vol. 24, no. 7 (August 1974), pp. 411-16. 40. spp and STPA, La ocupaci6n informal, p. 126 41. Departamento de Estudios Sociales,Banamex,Encuestade subempleados(Mexico, D.F., May 1983). 42. Ibid., p. 1. 43. Ibid., pp. 30, 31. 44. Ibid., p. 40. 45. Ibid., pp. 44, 49. 46. Ibid., pp. 18, 52. 47. Ibid., pp. 54-56. 48. The onlyother referenceI havefound to the economicstatus of ambulatorypeddlersis 100 UNDERUTILIZATION OF LABOR drawn from a survey of households in Mexico City as part of a study of migration by social scientists from the Colegio de Mexico. That study recorded the occupational status and earnings of the labor force members of the sampled households. Two percent of the labor force was found to be employed in this capacity. Average earnings in 1969-70 of this group were reported to be Mex$1,113, or 31 percent greater than the legal minimum wage (48 percent for men and 7 percent for women). Since no information about work hours is provided, however, earnings per time unit cannot be determined. The only other datum provided that characterizes this group is the average level of schooling completed: a low 3.2 years. Humberto Munoz Garcia, Orlandina de Oliveira, and Claudio Stern, "Migraci6n y marginalidad ocupacional en la Ciudad de Mexico," El perfil de Mexico en 1980, vol. 3, 6th ed. (Mexico, D.F.: Siglo Veintiuno, 1979), pp. 354-57, 345. 4. Agriculture and the Labor Market WHEN DEVELOPING COUNTRIES are at an early stage of moderniza- tion, the agricultural sector is the dominant area of employment. As the country develops, labor shifts from agriculture into the expanding sec- ondary and tertiary sectors. During this process, closer links are forged between the agricultural and nonagricultural labor markets. In 1940, when Mexico was at the beginning of the period of accelerated growth and modernization of its economy, about two-th irds of the labor force were still employed in the primary sector of the economy. A vast pool of labor services was thus potentially available to the other sectors as they ex- panded. However, the manner in which labor resources shift and the rate at which they do so is determined not only by the rate at which new employment opportunities appear in the eccnomy but also by the eco- nomic and institutional conditions prevailir.g in the countryside. This chapter reviews the evolution of the agricultural sector over recent decades and the implications this has had for the supply of labor to the rest of the economy. Also addressed is the degree to which available labor resources are involuntarily underutilized. The antecedents of the revolution that reformed the sector earlier in this century date back to the latter half of the nineteenth century. A major reorganization of the agricultural sector began during the 1850s with the expropriation of extensive church holdings and their redistribution in large units to private landholders. This was f.ollowed by a series of "re- forms." In 1885, the Porfirian administration began seizing communal lands and private holdings with uncertain ti:les and giving them, along with federal lands, to land development companies and favored indi- viduals. Justifications for this reform were that agricultural output was stagnant and that the indigenous and mestizD rural population was too stupid and lazy to increase output. These events occurred at a time when the government was emphasizing the rapid d&velopment of the domestic economy and an expansion of exports.' The effect of this reform was to deprive 90 percent of the villages in the central plateau of all of their 101 102 AGRICULTURE AND THE LABOR MARKET common lands. In the country as a whole, 90 percent of all rural families were left landless, while 50 percent of the rural population became inte- grated into the hacienda system. 2 One result of this upheaval was the creation of a large force of wage labor. Stimulatedby expandingexport opportunities and favorable prices, output expanded and, between 1885 and 1898, the real daily minimum market-determined wage for agricultural laborers rose by about 37 per- cent. The increasewas abetted by the expansion of urban production and employment.At the turn of the century,the terms of trade for agricultural exports had begun to decline, mechanization of industrial production proceeded to displaceartisans, and urban labor began to drift back to the rural areas. By 1911, the real daily wage in agriculture had fallen back to the level of 1885, or by 27 percent from the peak reached around 1897. The combination of a decline in the demand for labor and falling wages gave rise to increasing pressures for the return of the communal and private lands that had been distributed to the large commercial agricul- tural enterprises. The mounting demands erupted in a violent revolution, which was again to change the face of the agricultural sector. Land Distribution For about ten years, anarchy prevailed in the countryside. Forcible seizures of land and armed strugglesfor the control of the land took their toll in lives and lost output.3 Beginning in the early 1920s, new land reforms were directed by government decree. The state, however, did not possess effective centralized powers of enforcement, and the titling of transfers proceeded very slowly. In the face of uncertain tenure, neither investment nor productivity responded. Not until the administration of Lazaro Cardenas (1934 10) did the reform process become thoroughly institutionalized, orderly, and enforceable.Bythe end of Cardenas's term, some 25.5 millionhectares had been redistributed, 70 percent of this total during his administration alone. At the end of 1969, the total land trans- ferred amounted to 75.6 millionhectares. Although large amounts of land were redistributed during the 1960s during the terms of L6pezMateos and Diaz Ordaz, most of this was pastureland or of low productivity. For example, of the 18.1 millionhectares distributed between 1965 and 1969, only 8.7 percent was arable cropland. 4 Approximately half of the redistributed land was assignedto ejidos in an effort to recreate the communal system of landholding that charac- terized precolonial Mexico. Under the land reform law, the land is en- trusted to a village or community, which then assignsusage rights to its  LAN4D DISTRIBUTION 103 members on an egalitarian basis. Frequently,h aciendaswere expropriated and simply reconstituted as ejidos. The origiral intention was to provide each familywith a ten-hectare plot, but this goal was not achieved in the majority of cases. During the 1920s and early 1930s, the average plot consisted of two to three hectares, five to seven during the rest of the 1930s, and seven to eight hectares through the mid-1940s. 5 While most of the ejidos assignedusage rights to individual rnembers,some were consti- tuted as cooperative agricultural enterprises, especiallyin the northwest- ern region, with the land cultivated as a single unit under a formal managerial system. Farm output failed to increase as quickly as expected, however. It was therefore necessaryto reconsiderthe desirabilityof continuing a pattern of reform that would lead to the fragmentation of farmland and create disincentives for investment among the remaining larger units still in private hands. 6 Greater domestic and foreign demand for cash crops resulted in changes governing the distribution of lands in areas in which large-scaleirrigation projects had been undertaken. Land was distributed in the form of private holdings rather than as communal property to ejidos, and the size of the units was considerably larger than the typical plot in the latter. From 1925 through the 1940s, the combination of large-scale irrigation projects and heavy investmentin rural infrastructure created very favorable conditions for large private farmers, particularly in the arid northwestern and coastal regions. lTechnologically progressive and highly motivated to save and invest, tie new agricultural entre- preneurs spearheaded a sharp increase in the sector's production, which appeared to "solve" the perceivedproblem of leficient output.7 However, a disproportionately small share of government investmentin rural infra- structure and credit has been directed towarcl small private farmers and ejidatarios. As a result, large disparities in productivityand incomes were created between the smaller beneficiariesof land reform and the larger private commercial farmers. While the land reform satisfiedthe immediatedemands for land of many rural workers, it could not do so in units of sufficientsize to assure an adequate level of living. Particularly in the crowded regions of central Mexico, where distributions to ejidospredominated, the individualfamily plots were especiallylimited in size.As of 1960,10 percent of the ejidatar- ios had no more than one hectare at their d;.sposalfor cultivation; 44 percent had no more than four hectares. Only 16 percent held plots with over ten hectaresof cropland, and these held 45 percent of all the cropland assigned to ejidos. Since approximately a thirclof the cultivatable land is left fallow in any crop year, the income-producing area is still further reduced. In the private sector, there was an even greater preponderance of 104 AGRICULTURE AND THE LABOR MARKET small plots. Minifundistas with fewer than five hectares of cropland accounted for 77 percent of the 1.2 millionprivate farmers. Amongthem, they held less than 11 percent of the land in private hands. At the other end of the distribution were 3.1 percent of the private farmers with fifty hectares or more, accounting for 63 percent of the cropland.! A consequenceof so many smallplots is a yieldper farm unit (in terms of the value of output) that is insufficientto provide even a subsistencelevel of income. Fully half of the plots have been classified as "sub-subsistence" units on the basis of 1960 output levels. The annual value of crops produced on these units was estimated to average Mex$500 annually, or US$40 at the 1960 exchange rate. The addition of the value of livestock and forest products would increasethe figureby no more than 50 percent. These small plots accounted for only 4 percent of the total agricultural output of Mexico. Furthermore, over the decade 1950-60, no change occurred in the value of output per farm unit in this class.9 Typically,these small plots cannot fullyemploya household's availablelabor resources.It has been estimated that during the mid-1960s they required only between 75 and 150 man-days per year for direct cultivation.'" Another one-third of the units fallsinto the categoryof subfamilyfarms, units that employ fewer than two man-years of labor. In 1960, they accounted for 17 percent of total farm output. Between 1950 and 1960, the average value of harvested crops per farm unit increased in real terms by 24 percent and reached a level of Mex$3,101, or US$240,in the latter year. If the value of livestock and forestry production is added, total value of output per unit would be increased by an estimated 50 percent. It is considered unlikely that the average income per unit exceeded Mex$5,000, or US$400,per year. These plots could barely employfullya single individual for a year; annual labor inputs were estimatedto average 250 to 350 days." In view of this, it is reasonable to expect that a large number of families in this and the sub-subsistencecategorieswould have to supplementfarm income with earningsfrom other sources.' 2 For many regions dependent on rain-fedagriculture,the uncertainty surroundingthe amount of rainfall-and subsequentlythe sizeof harvests-has lent even greater urgency to the pursuit of supplementarysources of income." 3 The remaining 16 percent of the farm units may be consideredto have achieved levels of output adequate for the maintenance of a family. They accounted for 79 percent of Mexico's agriculturaloutput in 1960 and for most of the growth in the sector's output between 1950 and 1960. Indeed, fully 45 percent of the increasein total output was achievedby the largest farm stratum, which employed more than twelve man-years of labor or produced output valued at more than 100,000 pesos in 1960. These units represented only 0.5 percent of all farm units but accounted for one-third  LAND DISTRIBUTION 105 of total sector output. The next stratum, that of units employingfour to twelve man-years of labor or producing output valued at Mex$25,000- 100,000 a year, accounted for 3 percent of all farm units-22 percent of sector output in 1960, and 35 percent of the total increasein agricultural output over the decade. 14 In short, the picture of the agricultural sectorthat emerges from these data is one of a majority of subsistenceunits with stagnant output levels on the one hand, ancl,on the other, a minority of commercial farms that account for the bullk of total output and for the increase in output over time. On average, the largest farms produce an output approximately 513 times greater than that produced on the smallest plots. Data from the 1970 agricultural census confirmthe reduced incomes obtainable from small landholdings. Private farms of less than fivehect- ares yielded an average net income of only Mex$918, or about US$75. Those over five hectaresyielded net incomestwenty timesgreater. Among ejidatarios, annual net incomes from ho]dings of all sizes averaged Mex$4,426, or about US$350, an income substantially unchanged from that reported for 1960. Within the ejido sector, however, the same large differences among different-size holdings can be noted as among the private sector units. Plots up to one hectare in sizeyielded net incomes of only Mex$875, or US$70, while those with more than twenty hectares yielded almost thirty times more. 15 P. Lamartine Yates's interpretation of the agricultural censuses,however, suggestsa less static picture of overall productivity levels than that gained from observation of only the tiniest plots. He holds that, between 1950 and 1970, for example, output per hectare on private farms of over fivehectares increased by 147 percent, while increasing by 73 percent on smaller units. For ejidos of all sizes, output per hectare increased by 113 percent. These figures suggestthat output, even on relatively small plots, must have recorded significant 6 increases." As indicated above, disparitiesin the sizeof cultivatedplots lead to very large differencesin incomes from the land. Differences in productivityper hectare, however, are much narrower but apparently have been widening over time. For example, although the outpu.tper hectare of private hold- ings of five hectares or more was only 6 percent greater than that on smaller private plots in 1940, the margin increasedto 40 percent in 1960. In 1940, output per hectare on ejido plots actually exceededthat on the larger private farms; in 1960,the latter reported output levelsper land unit 25 percent greater than that for ejidos.17 These data imply that somewhat narrower differencesexist in productivity of land than might have been expected from more aggregated data available on a regional basis. The distribution of land and irrigation over the rural population is very 106 AGRICULTURE AND THE LABOR MARKET uneven. Moreover, these disparities have strong regional associations. The Northwest and Northeast, which enjoy greater access to irrigation facili- ties and in which larger farm units predominate, report significantly higher levels of output per hectare and per worker than do the Peninsular and the heavily populated Central regions. In 1970, the value of output per hectare in the Northwest and Northeast was Mex$3,435 and Mex$2,002, re- spectively, but only Mex$1,397 and Mex$1,436, respectively, in the Peninsular and Central sections. The differences in output per worker were even greater: Mex$16,603, Mex$10,723, Mex$4,201, and Mex$2,778, respectively. In the Central region, the paucity of agricultural resources other than labor is particularly striking. The region contains a third of the national agricultural labor force but only 11 percent of the cultivated land. Less than 20 percent of the land is irrigated. The average amount of cultivated land per member of the agricultural labor force is 8 reported to be no more than 1.3 hectares." Indeed, fully 70 percent of the smallest private plots are located in four of the states comprising this region (Hidalgo, Mexico, Tlaxcala, and Puebla), plus Oaxaca in the South. Given the small size of so many plots and the limited incomes it is possible to wrest from them, it should come as no surprise to observe a reduction in their numbers as their owners have either sold out or aban- doned their holdings. Between 1940 and 1970, there has actually been a decline in the number of private smallholdings and in the total land area 9 they occupy." Since ejidatarios do not possess title to their holdings, they cannot sell them. Nor can they retain the right to use this land if they do not cultivate it. Nevertheless, these farmers have found ways of avoiding active cultiva- tion of their land so that they are free to seek off-farm employment. The practice of sharecropping is becoming more widespread."0 A certain amount of renting of ejido land is also believed to be taking place; however, since the practice is illegal, it is impossible to obtain estimates of its incidence. In retrospect, the initial impact of the land reform was to provide a floor under the incomes of the recipients of land. In the process, it created a more permanent tie to the land for the majority of recipients who received land through ejidos. Indeed, the prospect of receiving land induced some farm workers who had migrated to urban areas to return to their villages so that they might exercise their right to land. Yet the land reform fell far short of its objective, for it failed to provide its beneficiaries with enough land and other resources from which to achieve incomes adequate for subsistence, much less keep those earnings rising over time in tandem with wages in other sectors of the economy. Furthermore, the increasing pres- sures of population on the available land supply have meant a narrowing  LAND DISTRIBUTION 107 margin for satisfyingthe demands for land. One of the ironies of the land reform has been that, in spite of its being one of the most sweeping of its kind in the world, so many agricultural workers remain landless. Indeed, in 1960, the number of such workers exceededby an estimated 15 percent that which existed in 1940. The 1963 household survey of income and expenditures of the Banco de Mexico repcrted that half of all heads of households in the agricultural labor force were landless. According to the population censuses, agricultural wage laborers formed the fastest grow- ing component of the sectoral labor force and, by 1970, accounted for just half of the total. An increaseof almost 16 percent in their number occurred between 1960 and 1970 even as the number of self-employedproducers was declining. Over this same interval, the number of unremunerated family members showed a marginal increase. Further distribution to landless people is obviously constrained by the limited availability of suitable land. The possibilities have already been exhausted for making vast new arid areas cultivatablethrough large-scale irrigation projects. However, some scope remains for increasing arable lands in rain-fed areas with small-scaleirrigation projects that tap local supplies of underground and surface water. Colonization projects in the south are proceeding, but they require large expenditures on reclamation projects to make the land arable and will result in distributionsto the local population rather than to the workers of the congested Central region. Some lands may become available as individual farm holdings are dis- covered to exceed legal size limits. The Sistema Alimentario Mexicano, introduced in 1980, provided for the expropriation and distribution to ejidos of those cattle ranches deemed suitable for crop production by the Ministry of Agriculture. 2" As yet, the extent to whichthose provisions have yielded a significant amount of land for redistribution cannot be deter- mined. The other remaining source of arable land is the larger private holdings that survived the land reform process in the older agricultural regions, or those larger holdings that were created in the northern regions of the country. Since these units have been the most dynamic sources of growth of agricultural output over the past thirty years, a decision to expropriate and redistribute them would be a difficult one for the Mexican govern- ment. Should it satisfythe continuing demands of the landless for land, or maintain the high and growing levels of agricultural output? As recent Mexican governmentshave espoused the goal of greater self-sufficiency in food supply, the risks inherent in breaking up efficient production units may appear too great to be assumed. In short, the prospects of converting much of the landless labor force into an independentpeasantry appear to be quite remote: 108 AGRICULTURE AND THE LABOR MARKET Rural Workers There is a large number of landless householdsand of landownerswith plots too small to employ fullytheir households' availablelabor resources. This implies a potential supply of labor for employmentwithin the sector on those larger units that are net buyers of labor servicesas wellas on those smaller units whose households are net sellers of services. In fact, the sector has long had an active wage market. In the older agricultural regions the large haciendas-survivors of the land reform process-pro- vide employment for workers drawn from the surrounding countryside. Large farm units created by the reform, located mostly in the North and Northwest, depend on wage labor for the bulk of their labor requirements. The diversity of climatic conditions and cropping patterns in different parts of the country givesriseto differentplanting and harvestingseasons. This provides employment opportunities that do not conflict with the requirements of the smallholders' own plots. Migrant Labor The prevalence of migrant farm labor within Mexico has long been established. In 1940, it was estimated that 200,000 farm workers madeup a migrant labor force availablefor the harvestof various crops. In Chiapas alone, some 35,000 indigenous peasants would descend from the moun- tains to harvest coffee on farms situated along the Guatemalan border. 22 Peasants from Puebla and other states of central Mexico had well- established contacts with sugar-growing areas along the Gulf coast; migrations during the cane-cutting season were an annual phenomenon. 23 The tobacco growers of the state of Nayarit rely on migrant workers from within the state as well as from neighboring states for the harvesting and processing of their crop. 2 4 Ejidatarios and their families annually desert entire villages in northeastern San Luis Potosi for the sugar and cotton harvest in Tamaulipas. 2 5 The large commercial farms of the Northwest also depend heavily on migrant labor for the harvest of crops requiring large labor inputs, such as cotton and vegetables. Since the population density of the region is relativelylow, active recruitment of farm migrants is undertaken by roving recruiters in the largelyrural states of the Central region. The wages paid for such labor are effectivelygovernedby the high legal minimum wages prevailing in that area. Although employers are responsible for the housing of migrants, the facilitiesprovided are fre- quently very primitive and lack adequate sanitary conditions. 2 1  RURAL WORKERS 109 A high incidence of short-term migratiori of rural workers has been recorded by numerous studies of rural communitiesby social scientists. For example, Balan, Browning, and Jelin report that approximately 59 percent of the ejidatarios of a poor ejido near Cedral, Nuevo Le6n, had been short-term migrants.2 7 Foster found that half of the men in the village of Toxi had been to the United States,many of them ten or more times. 28 Even more remote villages,such as Tilatongo in Oaxaca, are characterized by internal migration of a temporary or permanent kind. About one-third of the male population had migrated either temporarily or on a more extended basis, while another third had beccme permanently established 2 9 DeWalt found both temporary and permanent migration in other places. to be a common phenomenon in his study of the ejido of Puerto de las Piedras in the state of Mexico, especiallyamong the landless. Even those with land were participants in the process. 30 An alternative source of farm employmentwas provided by the bracero program of the United States during World War 11and the two succeeding decades. During this interval, some 5 million Mexicans were contracted for employment in the United States. Concornitantwith the formal con- tracting of farm workers was a substantial flow of illegal migrants to the United States, many of whom also found employmentin agriculture but without the contractual protection provided by the formal bracero pro- gram. The end of the bracero program, of cou:rse, did not put an end to this movement of workers but rather converted a larger proportion of it to illegal migration. This phenomenon will be discussedin greater detail in Chapter 6. Off-Farm Employment While much of the migration involved movement between jobs within the agricultural sector, the prevalence of employment opportunities in nonagricultural activitiesshould not be overlooked. At least over the past forty years or so, agricultural workers have sought and found temporary employment outside the sector. During the 1940s for example, the wide- spread construction of feeder roads in rural areas provided off-season employment opportunities. With the completion of these projects, work- ers increasinglyturned to urban areas for employment.The construction industry in the larger urban centershas long pi-ovidedan important source of seasonal employment. In Mexico City, the large central market district providesanother source of temporary employmentfor stevedoresor mate- rials handlers. Street vending of newspapers, fruits, vegetables,candy, and so forth is also held to offer employment opportunities to temporary migrants. Women seeking temporary employmententer domestic service, 110 AGRICULTURE AND THE LABOR MARKET street vending, or work as helpers or dishwashers in food service estab- lishments.3 1 With the exception of construction jobs, these occupational opportuni- ties are relatively unaffected by cyclical variations in economic activity. Apparently, short-term migrants seeking temporary employment have relativelylittle difficultyin findingcasual jobs that pose no entry require- ments or long-term commitment. For example, one study of agricultural communities in the valley of Toluca in the state of Mexico reported minimal time spent in search of off-farm employment. While 4,423,600 -dayswere so employed, only 113,300 were spent in job searches-a ratio of 39 days worked to each day of search. 32 To be sure, the earnings yielded by such employmentsare typically low, no higher than the legal minimum wage and frequently below it. Nevertheless, given the low opportunity cost of the labor so occupied, even these wages must appear attractive. While these descriptive accounts indicate that substantial numbers of rural workers are involvedin labor market activity,they do not conveyan appreciation of the full importance to farm households of off-farm em- ployment. Fortunately, such an appreciation can be gained from studies undertaken at the household level over the past twenty years. Viewed as a whole, they provide a picture of the farm household that departs substan- tially from the conventional one of subsistencefarmers and their families eking out an existence on undersized plots and facing few options in the disposition of the labor resources available to their households. What emerges is a much more complex set of alternativesfor such households. The mix of opportunities rangesfrom own-accountemployment (on one's own land), to wage labor within the agricultural sector, wage labor in nonagricultural productive activities,and nonagricultural production in a self-employedcapacity-not only for the head of the household but also for other members of the unit. Most of the household studies to be found, however, limit their attention to workers who have accessto land either as ejidatarios or as private owners. Very little information is availableon the sources of income and employment of landless farm households. A study of 100 ejidos dispersed through six distinct regions in the state of San Luis Potosi during the mid-1960s highlights the importance of employment off the family plot for the majority of the sampled popula- tion. Giventhe small sizeof the averagelandholding, two to fivehectares, the average number of work days devoted to its cultivation generally averaged less than fifty a year. Only where irrigation was available and double cropping prevaileddid the familyplot requiremore than a hundred work days. Most households provided wage labor for various activities outside the ejido; the proportion of total income derived from such  RURAL WORKERS 1ll sourcesgenerallyexceededby a considerablemargin that generatedon the ejidatario's own plot. The highest proportion of outside income was reported by the ejidos located in an arid zone close to the capital city of the state. There, the ejidatariosreported that 87 percent of their incomeoriginated from wages earned outside the ejidos. Of the remaining 13 percent, most income reportedly came from irrigated plots. Althoughthe largestsinglesource of off-farm employment was wage labor within the agricultural sector, the majority of workers were employed in a variety of rural and urban nonagricultural activities ranging from construction and mining to com- merce and the gathering of wood for sale. Cinly9 percent of the sampled ejidatarios declaredthat they were either not interestedin outsideemploy- ment or were unable to find work. 33 In three other districts, similarlysmallproportions indicated an absence of off-farm employment. On average, earnings from off-plot activities constituted more than half of household income, ranging from 56 to 70 percent. Since these districts were farther from urban centers, however, nonagricultural work accounted for less than half of off-farm income. Daily wages varied widely, even within agr:.culture.In general, the daily rate for agriculturalday labor on small holdingsaveragedless than half the applicable legal minimumwage. Bycontrast.,large farms were reported as offering wages two to three times those paid by smallholders. 4 Earnings rates in nonagricultural activities were considerably greater than those available from most agricultural employments. The monumental study of the agricultural sectorby SergioReyesOsorio and his associates,undertaken during 1966--68,confirmsthe importance to farm households of employmentoff their own lands. Amongthe private producers, 6 percent also held jobs as wage laborers in agriculture in addition to cultivating their own land. Another 38 percent had held employment in nonagricultural activities during the reference year. Among ejidatarios, the proportion was 23 percent in each type of outside 35 activity. The nonagricultural employmentswere distributedover severalsectors. The principal sector of off-farm employmentwas in commerce, in which 35 percent of the jobs were to be found. At 15 percent, a distant second sector was formed by the sum of unskilled jobs held in manufacturing, construction, mining,and transport activities.Another 13 percent were in services.Included in this listing of off-farmernploymentwere those involv- ing the rental and operation of farm machinery; these accounted for 13 percent of the total. Another 13 percent of the sampled workers held skilled jobs without reference to sector; handicraft production occupied 112 AGRICULTURE AND THE LABOR MARKET less than 10 percent. However,there were sharp differences in the distribu- tion of employments among sectors in the sampled areas. Proximity to urban areas appears to have been associatedwith a greater concentration in commerce and in industrial-type jobs (the second category above). More isolated areas reported a greatear incidenceof handicraft produc- tion and employment in skilled jobs. 3" As might be expected, the incidenceof employmentoff the family farm was greater among those with subsistence-size or smallerplots. Indeed,for these groups, the great bulk of individualand familyincome derivedfrom sources other than the familyplot. But even among those producers with family-sizeunits, outside employmentoccupiedsignificantproportions of total days worked and contributed as much as a third of familyincome. 37 Although the incidence of outside employment was greater among peasants with rain-fed lands, those holding irrigated plots, which yield higher returns and might be expected to offer more days of employment, were not far behind in the frequencywith whichthey sought off-farmjobs. A study of El Bajfo,a fertile but denselypopulated region in the state of Guanajuato, offers further confirmation of the importance of off-farm employment. 3 8 The data have been ordered according to the status of the producer, ejidatario, or private farmer; by size of plot; and by rain-fed or irrigated agriculture. In general, farmers in rain-fed areas or with small plots spend fewer days per year working on their own plots and are thus availablefor more off-farm employment.For example, among producers with infrasubsistence plots in rain-fed areas, the proportion of days worked off-farm to total days worked was 39 percent for ejidatariosand 62 percent for private farmers. Other small-scale farmers spent from 15 to 25 percent of their working days at off-farm labor. While these jobs included day labor in agriculture, they were far less numerous than those devoted to nonagricultural activity. Among ejidatarios, days spent in nonagricultural activity were more than three times as numerous, and among private cultivators almost seven times as numerous, as the days spent as farm laborers. Sincewages are higherin nonagriculturaljobs, the importance of earnings from this work relative to farm day labor is even greater. In most strata, earnings from hiring out within the farm sector bordered on the insignificant. The importance of wage employment as a supplement to self- employment for cultivators and their families has also been extensively documented in the six regional studies undertaken during the decade of the 1970s by A. Rene Barbosa-Ramirezand the Centro de Investigaciones Agrarias.39 The prevalence of small plots, too small to provide either full-time employment or an adequate income, implies the existence of  RURAL WORKERS 113 strong incentives to enter the wage employment market. Indeed, more days were occupied in wage labor than in self-employment in most of the regions studied. For example, the region of Valsequilloin the state of Puebla is one in which small farms, averagingonly 2.5 hectares,predomi- nate. Almost 80 percent of the land was irrigated, and "modern" tech- niques of production, as evidencedby the nature of inputs, prevailed.Yet outside employment occupied the majority of days worked by both the cultivator and his familymembers. Only 30 and 13 percent, respectively, of his and his family's work days were spent on the familyplot. 4 0 Even in the much poorer and more isolated region of Mixteca Baja in the state of Oaxaca, 60 percent of the days worked by the familyunit were off-farm. 4" A region in the Toluca valley of the state oi: Mexico revealed a strong inverserelationship between the sizeof the farnilyplot and the proportion of days worked in off-farm employment. Cultivators with less than one hectare of land devotedalmost seventeentimesmore daysto off-farmthan to own-account employment. It is not u:itil the average plot size approaches approximately eight hectares that days of on-farm employ- ment exceed those for off-farm. 42 For the region as a whole, 84 percent of all days worked were off-farm. A recent study, undertaken in the Mixteca rlegion of Oaxaca, illustrates the wide range of productive alternativesavailableto the rural population. In this very poor and relatively isolated region, the importance of off-plot activity is striking. Although the study does not account for the distribu- tion of days worked, it does provide information on the sourcesof income. For example, in the caseof producers on rain-fed lands, almost 64 percent of their incomes originated in off-plot employment,while an additional 4 percent was derived from artisan production. For the region as a whole, the value of artisan production amounted to 32 percent of all household production, exceeding the 19 percent originating in the value of crop output on rain-fed lands. Furthermore, between a third and a half of the labor force migrates to other regions of the country and to the United States in search of temporary employment. The importance of these activitiescan be gauged from the size of the remittances that flow from both domesticand foreigndestinations of the migrants: more than Mex$2 billion in a recent but unspecifiedyear (probably 1982 or 1983), an amount wellin excessof the valueof all agriculturaland livestockproduc- 43 tion of the region. It is thus amply clear that, in virtually all of the regions that have been studied, regions with very different quantitative and qualitative endow- ments of land, off-farm employmenthas accoanted for a substantial, if not dominant, proportion of the total for both cultivators and their families. 114 AGRICULTURE AND THE LABOR MARKET The Demand for Labor and the Evolution of Employment Conditions As was discussed in Chapter 3, there is a widely held view in Mexico that the countryside shelters a large mass of underemployed labor. Further- more, it is believed that employment conditions for the agricultural labor force have been deteriorating over time. The lack of growth in output for much of the small-farm sector is held to be a contributing factor. But more important for the recent deterioration has been an agricultural labor force that is alleged to be growing in the face of stagnating, or possibly even declining, labor input requirements for the sector as a whole. One recently published study of the sector by the Centro de Estudios en Planeaci6n Agropecuaria (CESPA), an agency within the agricultural minis- try, points to a steady reduction in labor inputs over time as agricultural technology and cropping patterns have changed. It offers the following estimates of the rates of change in the total number of days of direct labor that are required by cropping and livestock activities: Crop production Livestock Total 1946-48 to 1964-66 5.0 5.2 5.1 1964-66 to 1976-78 -2.9 2.8 -1.3 Until the mid-1960s, the rapid expansion of cultivated lands carried with it a great increase in labor requirements. Since then, the rate of expansion in cultivated hectarage has declined substantially and has provided an inadequate offset to the employment effects of technological and cropping changes. According to this view, shrinking labor requirements in the context of a growing rural population have implied a deterioration in job opportunities for the actual and potential agricultural labor force. This decline has motivated the large movement of population from rural areas to the cities, where workers are destined to suffer conditions of underem- ployment in an overcrowded tertiary sector.' A less pessimistic appraisal of the course of the demand for labor is offered by Teresa Rend6n over an interval that overlaps the more recent years of the CESPA study. Over the 1960-73 interval, Rend6n calculates the total labor requirements of the annual crop sector to have increased by about 15 percent, and by almost 20 percent for the agricultural sector as a whole.45 Her contrasting findings are apparently due to the use of technical coefficients that differ from those applied by the CESPA researchers. In addition, her interval includes several years of the early 1960s, during which the sector enjoyed a considerable expansion in the area under  DEMAND FOR LABOR 115 cultivation and growth in output; the beginning of the recent CESPA interval corresponds to a period of significantlyslower growth in both. Changes in Technology and Cropping Patterns As indicated above, secular changes in labor input requirements are generally ascribed to changes in agricultural technology and in cropping patterns. By all indications,the degreeof mecha.nization has been increas- ing. For example, the use of tractors has shown a significantexpansion over the past two decades (seetable 4-1). Until 1970, the irrigateddistricts, with only 15 percent of the cultivated land, employed over half of the tractors. This, however, is not surprising in view of the fact that the irrigated districts contain farm units that are considerablylarger than the average in the rain-fed-areas. What is striking, however, is the rapid extension of tractors in the rest of the agriculturalsector since 1970. (I will return to a discussionof the possiblesignificanceof this increaseat a later point in this chapter.) Broader measures of mechanizationprovide further evidenceof chang- ing technological coefficients.The value of machinery used by each agri- cultural worker increased from an average of Mex$142,600 in 1940 to Mex$380,500 and Mex$709,900 in 1950 and 1960, respectively,all in 1960 pesos." The rate of increasewas much greater on the large private farms, which are also the principal employers Dfwage labor. The increase on these large units was more than sevenfold over this interval, from Mex$216,700 to Mex$1,506,000 in 1940 and 1960. Over the same interval, little mechanization was reported among private smallholders, while in the ejido sector use of machinery more than tripled. However, these data may understate significantlythe use of mechanical traction Table 4-1. Number of Tractors by Type of Agricultural Area Irrigated Rain-fed Year Total districts areas 1963 63,184 34,s53 28,631 1970 89,720 48,506 41,214 1975 119,318 48,174 71,144 1976 134,346 51,342 83,004 1977 148,412 50,020 98,392 1978 158,247 54,442 103,805 Source: CESPA, El desarrolloagropecuario de Mexico, vol. 6, El empleo de manodeobra en las actividades productivas agropecuarias (Mexico, D.F., 1982), p. 76. 116 AGRICULTURE AND THE LABOR MARKET among smallholders, particularly for certain tasks such as plowing. Mechanized services are available for hire, and various regional studies reveal that these are widelyutilizedby smallfarmers. 4 7 Finally,the propor- tion of land subject to mechanization has increased sharply from 3.9 percent in 1950 to 8.2 and 21.5 percent, respectively,in 1960 and 1970.48 Within the irrigated sector, a 25 percent increaseoccurred in the mecha- nized hectarage during the 1960-70 decade. This boosted the proportion of land employingmechanicaltraction to over 50 percent for the firsttime; another 39 percent of irrigated land was reported to be partially mecha- nized. The trend was carried forward during the 1970s. For the crop year 1972-73, the mechanized part of the irrigated districts encompassed 57 percent of the total area, while the proportion partially mechanized de- clinedslightlyto 35 percent. For the sector as a whole, the rapid extension of mechanizationbetween 1960 and 1978 can be seen in the growth of the depreciatedstock of farm machineryat an annual rate of 7.1 percent. 4 9 The labor displacement effectsof mechanizationhave been estimated by Ren- d6n to have been on the order of an annual reduction of 0.3 percent in sectoral labor input requirements between 1960 and 1973.50 The CESPA study, however, calculates a much greater displacementeffect:3.2 percent a year from 1964-66 to 1976-78.51An alternative measure of displace- ment holds that each tractor servesto replacefrom ten to twelveworkers. 52 Changes in cropping patterns have also tended to reduce labor require- ments on each unit of cultivated land. Table 4-2 illustrates the changes in cropping patterns accordingto the degree of labor intensity of production and the associated changesin labor requirements.During the twenty years 1946-66, when the harvested area expanded rapidly in associationwith the land redistribution process, most of the growth occurred in crops classifiedas semi-intensivein their use of labor. The crops accounting for most of this increase were corn and beans, traditional subsistencecrops, and sugarcane. Significantrelative increases also occurred in hectarage and labor devoted to labor-intensive crops, although their absolute im- portance is overshadowed by the semi-intensivecrops. A similar change occurred in the area and labor devoted to extensive crops. Thus, during this period of dynamic expansion of the sector, all three classesof produc- tion grew in step with each other. The more recent 1964-78 period provided a sharp contrast. The area on which semi-intensivecrops were grown declined.This was largely attrib- utable to a reduction in the planting of subsistencecrops, beans and corn. The increase in hectares used for extensive crops more than offset the declinein the semi-intensivecropping area, as the production of safflower, sorghum, barley, and soybeansexpanded. There was a moderate increase  D]EMAND FOR LABOR 117 Table 4-2. Harvested Area and Labor Inpvt Requirements by Degree of Labor Intensity of Land Use, 1946-48 to 1976-78 Hectares in cropsa Labor input required (thousands) of workdays) (millions Crop Inten- Semi-in- Exten- Inten- Semi-in- Exten- year sive tensive sive Total sive tensive sive Total 1946-48 752 4,631 820 6,664b 62 159 12 264- 1964-66 1,701 10,954 1,779 14,434 145 459 34 638 1976-78 2,171 9,704 3,275 15,150 169 264 33 466 rateof change Annualpercentage 1946-48to 1964-66 4.6 4.8 4.4 4.6 4.8 6.1 6.0 5.0 1964-66 to 1976-78 2.1 -1.0 5.2 0.4 1.3 -4.5 -0.2 -2.6 a. Extensive crops require 15 or fewerwork days lor eachcultivated hectare,semi- intensivecrops 16-40 days,and intensivecrops41 or nioredays. b. The total includes441,000unclassifiedhectares. c. The total includes days. 31 millionunclassified Sources: CESPA,EldesarrolloagropecuariodeMexico, "ol. 6, Elempleodemanodeporo productivas en las actividades agropecuarias(Mexico,D.F.,1982),pp. 65-66. in land devoted to labor-intensive crops, but there was a shift within this sector away from cotton toward alfalfa and some vegetable crops. The employment effects of these shifts are held to be considerable. The 42 percent decline in labor requirements for the semi-intensive sector far outweighed the expansion of the intensive, and requirements in the exten- sive sector remained virtually unchanged despite the large increase in hectarage. Thus, the annual net decline in the sector's labor requirements is estimated at 2.6 percent. Within the annual crop subsector, cropping changes are estimated to have reduced labor r equirements at an annual rate of 0.5 percent, not much different from the 0.6 percent annual rate offered by Rend6n.53 Table 4-3 is a summary of the effects that various developments have had on the labor requirements of the sector as calculated by CESPA and Rend6n. Rend6n's estimates refer to an annua.. crop sector composed of thirty-one commodities; CESPA'S figures refer to groups of eighteen and twenty-one crops, respectively, in the two intervals cited. These selected crops, however, account for the overwhelming proportion of the hectar- age dedicated to annual crops, 97.9 and 93.3 percent in the two time 118 AGRICULTURE AND THE LABOR MARKET Table 4-3. Factors Accounting for Changes in Labor Requirements for the Principal Agricultural Crops (annual percentage change) CESPA' Rend6nb 1946-48 1964-66 Factor to 1964-66 to 1967-78 1960-73 Area harvested 4.7 0.0 0.6 Mechanization 0.8 - 3.2 - 0.3 Cropping changes 0.1 -0.5 -0.6 Increased yields n.a. n.a. 0.5 Total 5.6 - 3.7 0.2 n.a. Not available. a. Includes the twenty-one most important annual crops. b. Refers to the annual or "short-cycle" crop subsector. Sources: CESPA, El empleo de mano de obra, p. 62; Rend6n, "Utilizacion de mano de obra," p. 369. intervals, 1946-48 to 1964-66 and 1964-66 to 1976-78, respectively. Note that, in Rend6n's calculations, the negative impact of mechanization and cropping changes is offset by increases in labor requirements associ- ated with increasing yields per hectare. In addition, Rend6n observes that increases in the area cultivated also added to the labor requirements of the sector. The CESPA calculations, however, show a zero contribution from this source during the more recent interval even though more land was reported harvested; this expansion amounted to 3.5 million and 0.965 million hectares between 1960-70 and 1970-77, respectively.54Another indicator of the declining labor requirements of the sector is the number of days employed on each cultivated hectare. Rend6n estimates that, in the subsector comprising the most important annual crops, labor inputs per hectare declined by 5 percent between 1960 and 1973.55 For the agricul- tural sector as a whole, CESPA offers a much sharper decline of 36.7 percent between 1964-66 and 1976-78.5" Both of these analyses concur that an increase in the production of perennial crops, largely high-value fruits, has had a favorable impact on employment. Between 1960 and 1973, the hectarage devoted to perennials grew by about one-half, and labor requirements by 40 percent. By and large, the labor input for each hectare cultivated is large-an average of 112 days in 1973.57Finally, more labor has been required in the rapidly expanding livestock sector. Given the low labor intensity of the sector, 58 however, its contribution to total employment has remained modest.  DEMAND FOR LABOR 119 Measuresof Labor Requirements The CESPA study offers an estimate of the total direct and indirect labor needsof the agricultural and livestocksector fcr 1977. The direct require- ments amounted to 692.4 milliondays, of which almost 70 percent were occupied in agricultural crop production. Diirectrequirements, in turn, represented almost 85 percent of total labor requirements. The indirect labor requirements refer to those tasks anciltary to cropping activities, such as administration, maintenance, and repair of machinery. Labor needs for this type of work were estimated at: 103.1 million days, or 14 percent of the total. Finally,23.3 milliondays were occupiedin construc- tion and other investment activities. The sum total of days required was 818.8 million." 9 Of course, given the seasonalityof agricultural produc- tion, the demand for workers is not evenlydistributed over the year. Peak months of employment are associated with planting and harvesting sea- sons: June through August for planting and cultivation,and October and November for harvesting. The months of lowestactivity are January and February. The peak month of July in 1977 re(quired 84.4 milliondays, 69 percent more than the trough month of February. 60 On the basis of field studies undertaken in 1977 by two other govern- ment agencies, the CESPA study reports the duration of employment, in months, of the agriculturallabor force (table4-4). Ascan be observed, the two sets of findingsshow considerable divergencein the distribution of individuals by duration of agricultural employment.The CESPA authors indicate their belief that the FEDA/BANRURAL results represent a more faithful portrayal of rural reality, although the reasons for this conclusion are not spelled out. In both data sets, approximately 0.5 millionmembers of the estimated agricultural labor force remain unaccounted for. These workers are presumed to representthose who stand in the most precarious position in the labor market and are the mo:stunderemployed.Although more than half of the labor force appears to be employed at least eight months of the year,the number of days workMd in each month is estimated to have averaged only fifteen. The implication is that even this relatively "permanent" portion of the agricultural labor force is not "fully em- ployed." Beyond identifyingthe above-mentionedhalf million workers as marginal, however, the study makes no aternpt to quantify the degree of underutilization of the labor force. It is presumed to be self-evidentthat the available labor resources are seriouslyunderemployed and will con- tinue to be so in the near to intermediate f ature. The regional case studies by Rene Barbosa-Ramirezduring the middle of the 1970s did make an effort to measure the volume of unutilizedlabor resources in fiveof the six regions studied. Thesecrude estimatessuggested 120 AGRICULTURE AND THE LABOR MARKET Table 4-4. Agricultural Labor Force by Number of Months Worked in 1977 Reported by Reported by PROCAP FEDA/BANRURAL Months worked Number Percent Number Percent 1-3 192,660 2.9 797,297 12.2 3-6 488,785 7.5 985,398 15.0 6-8 33,387 0.5 620,886 9.5 8 2,874,700 43.8 n.a. n.a. 8 or more n.a. in.a. 3,726,730 56.8 12 2,340,746 35.7 n.a. n.a. Unaccounted 626,722 9.6 426,689 6.5 Total 6,557,000 100.0 6,557,000 100.0 n.a. Not available. Source: CESPA, El empleo de mano de obra, p. 158. that the proportion of the "available" labor time that was actually used ranged between 32.8 and 57.2 percent; only one of the five surveyed regions recorded a proportion greater than 50 percent.6" The study ac- knowledged that the time not occupied in production on or off the family plot could not be equated with the conventional concept of unemploy- ment, since the actual availability of time for employment was not univer- sally determined. Nevertheless, it was held in most of the regional accounts that these crude estimates sufficed to indicate the seriousness of the employment problem. (I will return to a discussion of some interesting aspects of this study below.) A set of measures for labor utilization provided by the CESPA study suggests that rural underemployment has increased in recent years as a result of declining labor input requirements and growth of the agricultural labor force. These measures take the form of an estimate of the average number of days of annual employment for each agricultural worker at three points in time and is derived from the corresponding technical coefficients of labor inputs. According to these estimates, the average number of days worked increased sharply from 77 to 168 during two decades of dynamic growth, 1946-48 to 1964-66. Subsequently, the trend was reversed; by 1976-78, the days worked had declined to only 120. This number is probably understated, however, because the size of the agricultural labor force was substantially overestimated.6 2 A more realistic labor force size would yield an average work year of 132 days. However, these estimates are limited to days of direct labor. They omit  DEMAND FOR LABOR 121 days spent in indirect or investment activi:ies that account for about twenty-three additional working days. Another widely cited source would place the deterioration of employ- ment opportunities for landless wage earners even earlier than does the CESPAstudy. ReyesOsorio concluded that the number of days worked by wage laborers declined from 190 in 1950 to only 100 in 1960.63This estimate was based on agricultural census r eports of total wages paid divided by the averageagriculturallegal minimumwage.The total number of utilized days thus obtained was then divided by the number of agricul- tural laborers to derive the average number of days worked by each. There are plausible reasons for believing that this procedure yields a declinethat is exaggerated.The legal minimurnwage is not a reliable guide to the wages actuallypaid at all points in time. Particularlywhen it departs substantially from the market price of labor, t'helegalminimum is likelyto be widely ignored by employers, and especiallyby small farmers. 6 4 With respect to the two years employed in Reyes Osorio's estimate, the legal minimum wage in 1950 reflected a decade-longdeclinein real terms and probably was close to the market wage. In contrast, the real legal mini- mum was increased substantially during the 1950s; it reached a level 40 percent above the 1950 minimumby 1960.65 To the extent that the market wage lagged behind the legal minimum, the procedure used by Reyes Osorio would understate the days of employment in 1960. Indeed, a preliminary version of this same study explicitly acknowledges a diver- gence between the legal minimum and the actual prevailing wage and suggests that use of the former results in an underestimation of days worked of from 10 to 20 percent. 66 Although one might quibble over details of the various statistical mea- sures enumerated here, they all seem to poinmto a growing gap between days of labor required by the agricultural sector and the available labor supply. It is virtually unanimously concluded that these trends reflect a deterioration of employmentconditions for the rural labor force and that the sector faces an increasing volume of surplus labor. The CESPAreport summarizes its finding as follows:67 [As a result of technological changes that have been enumerated] an important part of the total mandays available to the sector were not employed.... Agricultural workers were afforded limited opportuni- ties for year-round employment.This signifiesthat there have accumu- lated very high levels of visible underemploymentand unemployment among those persons who found it impossibleto be employedover the entire productive process. The scarcity of employment opportunities was aggravated by the agricultural crisis faced by the country [that is, 122 AGRICULTURE AND THE LABOR MARKET the relative stagnation of the sector since the mid-1960s; translation mine]. One of the most important government agricultural planning docu- ments of recent years also views the principal goal of the agricultural sector as that of increasingthe demand for labor in order to alleviate the underutilization of the existing labor force." During the past ten years, agriculture has lost some of its dynamism and a number of problems have arisen. The chief symptoms of this decline include the need to import large quantities of grains during the last two years, a fall in exports, an increase in rural unemployment, and the existenceof unsatisfactory rural livingconditions.... A review of em- ployment in the sector reveals that: (1) during the 1950s employment opportunities increased, generallyspeaking, as rapidly as the working- age population, so that there was no increasein eitherunemploymentor underemployment; (2) during the 1960s availablejobs increased more slowly than the working-age population; and (3) at the present time about 40 percent of the potential work force is not being used effec- tively. A number of studies have yielded the same conclusion with regard to the rate of agricultural underemployment.... The challenge at the present time is how to increasemanpower absorption and at the same time narrow the income gap between [the large commercialfarm- ers and poorest segments of the agricultural labor force]. Employment and Underutilization In view of the overwhelming consensus among Mexican observers in their assessment of the gravity of the employment problem within the agricultural sector, it may appear foolhardy for me to offer one that runs counter to the prevailingview. Yet I cannot help but feel that the analyses of the employment conditions of the rural population on which the somber conclusionsare based have been too narrowly focused and have tended to overlook considerations that might lead to a less distressingset of conclusions.Two major weaknessesof these analysescan be identified. One derives from estimates of the supply of labor available to the sector. The other is the general omission from consideration of that part of the rural supply of labor that is diverted to nonagricultural labor markets. This diversionreduces the number of workers who are idle and available for employment within agriculture.  EMPLOYMENT AND UNDERUTILIZATION 123 Estimates of the Agricultural Labor Force The measurement of the agricultural labor force poses formidable dif- ficulties.A large part of the difficultystems from the seasonal nature of production. As noted above, the number of persons employed and the number of days expended vary sharply over the year. One can presume that those individuals employed during the trough month of the year represent a sort of core, or permanent, part of the agriculturallabor force. But what about the remainder-what is the nature of their attachment to the agricultural labor force? Consider the CESPA estimates of monthly employment. During the trough months, an estimated 3.7 million workers are employed and may be assumedto work throughout most of the yea.r.In table 4-4, this number corresponds to those laborers employed at least eight months of the year according to the FEDA/BANRURAL source. Supplementing this core are workers employedfor shorter periods. In the months of peak employment, their numbers rise by 2.4 million and yield a tol:alemployedlabor force of approximately 6.1 million. 69 This supplementarylabor force thus repre- sents about 40 percent of the total in the peak months. If the seasonally employed persons in agriculture are considered to be available for year- round employmentwithin-and only within-the sector, then the conclu- sion that has been reached by various observe rs (that 40 percent of the agricultural labor force is underemployed) would clearly follow. How- ever, can the assumption that the peak labor force is availablefor employ- ment throughout the year be consideredvalid? No representativedata are offered that permit us to reach definitive conrclusions, but I consider it unlikely that this assumption is a realistic one. The difficultiesinvolvedin estimatingthe size of the agricultural labor force in terms of the work days available are illustrated by the series of regional studies undertaken by Ren6 Barbosa-Ramirez.In the first of the regions reviewed,the valleyof Toluca in the state of Mexico, the available days were defined as 252 days in the year times the number of family members between the ages of 15 and 55 without regard to sex. 70 In the second, the Mixteca Baja in Oaxaca, the age group was extended to include all those between 12 and 55 while the work year remained fixed at 252 days. 7 ' In the third region,Valsequillo in Puebla, this same age group was taken, but the work year was extendedto 3 00 days. 7 2 The study of Los Tuxtlas, Veracruz, adopted still another criterion for calculatingthe avail- able labor supply; it included all those who, during the nine-month reference period, had worked for at least ten days with or without remuneration. 7 3 The work year again was definedas 300 days. Finally,in 124 AGRICULTURE AND THE LABOR MARKET the study of Las Huastecas, San Luis Potosi, the labor force was defined simply as consisting of those who had worked at least thirty days during the reference period. 74 Obviously, the absence of a universally accepted definition of the labor force is bound to giverise to different and noncom- parable measures of unemployment or underemployment. It is interesting to note the evolution of perceptionsof Ramirez-Barbosa with respect to the size of the rural labor surplus. By the second phase of the study, which coincidedwith the Valsequillostudy, he began to harbor some doubts about the legitimacy of the measurement procedures fol- lowed, particularly after the publication of a pilot study done for a national household survey of incomes and expenditures.That study found that "a surprisingly low proportion of the idle population was willing to work" (translation mine). 7 5 The author then resolvedto undertake a much more careful investigation of the availability for employment of the population of labor force age. The change in procedure was not im- plemented fully until the final regional study was undertaken. While the findings are clearly limited to the region of Las Huastecas, they serve to raise strong cautionary signals about the usefulness of the traditional methods of estimating rural labor surpluses. As indicated above, the availablelabor force was originallyestimated in an arbitrary manner by assuming that everyone of labor force age was available for employment for a full year. How far this assumption falls short of reality was illustrated in Las Huastecas when various subgroups of this population were subjected to more careful examination. For exam- ple, the population of labor force age that had not been activelyemployed for at least thirty days during the referenceperiod included 9,258 indi- viduals. Of these, 92.8 percent stated they were not available for employ- ment. Of the remainder, 587 individualsexpressed an interest in full-time employmentand 78 in only occasionalemployment.Insteadof 2.8 million work days available fom this population, no more than 190,000 were actually available. 76 In the case of individualswho are employed,there is a common proce- dure for estimating the days available for additional employment. The normal work year is definedas consisting of a number of days (300, for example); from that is subtracted the number of days required for the cultivation of the family plot. Sometimesthe days of employmentoff the plot may also be deducted. For either case,the remainder is definedas time availablefor employment.In Las Huastecas, the head of the family,as well as the other family members who were employed on the familyplot, were asked to enumerate the number of work days lost from productive activi- ties on the plot during the year. In the stratum comprisingnuclear families, the average familyhead reported an average of sixty-sevensuch days. Of  EMPLOYMENT AND UNDERUTILIZATION 125 these, twenty-six days were occupied in ritualiEtic or civic obligations and on personal errands; thirteen days were lost because of illness, eighteen to inclement weather, and only ten to a lack of work. All other employed family members reported an average of twenty-four days for each house- hold not spent in productive activity on the family plot, of which five were attributed to a lack of work to be done. While the number of days lost by the producer and the members of extended family and other forms of household organization did vary somewhat, the distribution of days lost 77 for various reasons was quite similar. About one-third of those individuals who worked on the family plot also reported off-plot employment. These workers were asked why they sought outside employment as well as why they did not seek more such employment. The most frequently offered response to the first question was a desire for more money. Over half of the family heads and 40 percent of family members so responded. A related reason of "necessity" was given by an additional 12 to 15 percent of the respondents. Lack of work on the family plot or insufficiency of land together accounted for most of the remaining answers. Of greater interest, however, are the reasons cited for not seeking more outside employment. Seventy-eight percent of the household heads and 64 percent of other employed household members cited the time-consuming cultivation of their plots as limiting their avail- ability for additional employment. Only 14 percent of the family heads and 17.5 percent of the members found the absence of employment opportunities to be a limiting factor.78 Estimates of Unemployment The study concluded by estimating the volume of available and unem- ployed labor time. The total days available numnbered4.939 million. After deducting the number of days worked and those voluntarily dedicated to nonwork activities, 1.120 million days of involuntary unemployment were left. This total was composed of three separate components. The first-459,900-was derived from the "registered" or reported days lost from on-plot employment mentioned above. These days of "involuntary unemployment" included days lost to illness or inclement weather and amounted to 71 percent of the total. Only 29 percent of this measure of involuntary unemployment consisted of days when no work was available on the plot. One may question whether days lost to illness or inclement weather represent a measure of open unemployment in the sense that workers are then available for employment in the labor market. The second major component of estimated "unemployment" came from family workers (other than the head of the household) who had 126 AGRICULTURE AND THE LABOR MARKET blocks of unaccounted time. Those who had been establishedas "work- ers" were assumedto have 300 days of availabletime for employmentjust as did the family heads. But household members could not account fully for the disposition of time between employment and registered days of voluntary and "involuntary" unemployment (as defined above); there remained 60 days per household for which an accounting could not be made. The authors of the study did not establish the actual availability for employment of these days but adopted the normative assumption that the behavior of these individuals should match that of household heads, for whom all time was successfullyallocated. The total unemployed days yielded by this procedure amounted to 584,522. This is obviously an outside or maximum figure, for it may be questioned whether all of the unaccounted days were actually available for employment. Finally, the remaining 73,400 days of available time were ascribed to the 587 individuals who were not currently working but who indicated they were available for immediate employment. 79 The sum of the "unem- ployed days" over the several household subgroups amounted to 23.8 percent of the estimated total available work days. This figure can be compared with a 63.2 percent unemploymentrate that would result from the crude procedure followedin the earlier regionalstudies. (Accordingto that method, the available labor resources were defined in terms of the product of all the household members of labor force age times 300 days.) Even this lower estimate overstates the quantity of surplus labor avail- able for employment outside the household's own farm unit. As noted above, work days lost becauseof illness are not available for employment anywhere. Neither are those lost to inclementweather since such days are not predictable events or of sufficientduration to permit or justify a job search. Over half of the total days of "open unemployment" have been so designated because they were not worked by household members and could not otherwise be explained. Of course, there is no way of determin- ing the extent to which such idlenesswas voluntary or involuntary. The fact that only a modest minority reported that it did not seek more off-plot employment because of an absence of job opportunities suggests that at least some of this idleness was intentional. But, even if it was not volun- tary, whether it was time that could have been made availablefor employ- ment elsewheredepends on the frequencyand duration of the idle periods. Unless idleness occurs in blocks of time uninterrupted by on-plot work obligations, the days not worked may not effectivelybe "marketable." Another factor is that the costs of job search may loom too high relative to the earnings that can be anticipated from a brief period of outside employ- ment. This would seem to be the case particularly when available job opportunities lie beyond a commuting radius of home.  EMPLOYMENT AND UNDERUTILIZATION 127 A more meaningful estimate of the rate of open unemployment can be achieved with a few adjustments. For example, one procedure would alter the size of the pool of "available" work days by excluding those days lost to voluntary causes, illness, and inclement weather. Unemployed days would then include those days idle from lack of work on the family plot plus the available days of workers not currently employed but desirous of employment. To these days one might add some proportion (such as one-half) of the days of family workers for which no accounting can be offered. Calculated in this way, the unemployment rate is reduced to only 11.5 percent. While this may either understate or overstate somewhat the actual rate, depending on how one wishes to tre at the unaccounted time of family workers, the absolute value of this ratio would still prove to be far lower than that generally assumed to hold in the countryside. It would also come closer to approximating the volume of labor services likely to be available in the labor market. A more recent survey of employment and underemployment in agricul- ture was undertaken on a state-by-state basis by the Banco Nacional de Credito Rural (BANRURAL) during 1980-82.8o The study attempted to account for all days worked as well as for those not worked by all those employed on the sampled plots at the time of the survey. The study adopted 300 days worked as a measure of full employment. Multiplying the number of interviewed workers by 300 thus yielded the total number of work days assumed to be available for employment. Of this total 20.5 percent were not worked, but almost half of all days not worked were voluntarily forgone. Although 10.4 percent of all the available days were involuntarily idled, not all were available for alternative employment. Almost a fourth of the days lost from work involuntarily were attributable to illness or inclement weather. Only those id]e days associated with the seasonality of agricultural production might be considered available. In the eighteen states for which the research reporl:s were available, such days represented only 8.1 percent of the days in a fully employed year. Hired Labor Other observations support the view that households harbor limited amounts of surplus labor that is readily available for employment. For example, one of the features of peasant agriculture that is strikingly documented by the Barbosa-Ramirez study is the prevalence of hired labor on even the smallest of plots in most parts of the country. Given the size of rural households and the number of family members of productive ages, one would have expected that the labor requirements of family plots could and would be satisfied by the household unit itself. Yet, this does not seem 128 AGRICULTURE AND THE LABOR MARKET to be the case. For example, in the region of Valsequillo,Puebla,workers were hired for more than half the days expended on both the private and ejido plots. On average, the cultivator provided 27 percent of the total days expended, other family members another 17 percent, while hired labor accounted for 56 percent of the total."' As noted earlier, only a small proportion of the total days worked by the cultivator and other family members was devoted to the family plot. Farm households clearly pre- ferred to hire themselves out in rural or urban labor markets and to substitute hired labor for their own. Apparently, there was still no short- age of household labor time available for cultivating the plot. The study estimated that no more than 30 percent of the total days potentially available from household members of labor force age was actually en- gaged productively. In the region of Mixteca Baja, Oaxaca, one-fourth of all the labor was hired, in spite of estimates that only 18 percent of the time of available household workers was employedon familyplots, and lessthan half of the available days was occupied productively either on or off the family holding. In this case, too, a majority of employed days was off the family 82 plot. In the Toluca valley, half of the labor requirements were met by hired labor, while family households allocated over 80 percent of their productive time to off-farm employment. According to the study, the number of family members in the sampled households who were idle (other than heads of households) amounted to twicethe volume of actual- ly employedlabor." 3 In viewof this potentially availablelabor, the decision to hire seemsincongruousunless those "idle" labor resourcesdo not really consider themselvesavailable for employment of any kind. Severalpossible explanations have been advanced for the prevalenceof so much hired labor in spite of the presumably large surpluses of family labor. In some regions, the growing season is so short that labor inputs must be concentrated in large amounts within brief intervals. At such times, these requirements exceedthe supplyof familylabor. This explana- tion would be more convincing were it not for the large quantities of household labor that are potentially available and probably could be pressed into service for short periods of time. Another explanation offered for areas such as Mixteca Baja with large indigenous population concentrations points to the breakdown of tradi- tional practices of labor exchange within the community. Given the sur- vival of traditional production techniques, the size and timing of labor requirements have remained unaltered. The increasing monetization of economicactivity,however,renders an exchangeof free servicesunaccept- able. Wage labor has come to replace bartered labor even though no change has occurred in the patterns of family and nonfamily inputs of  EMPLOYMENT AND UNDERUTILIZATION 129 labor." Although some part of the hired labo:r may be attributed to this change, it is not likely to account for a large part of the hiring. If hired labor reflected merely a monetization of a traditional exchange of labor services,then one would expect the landholders to report a figure for the time they worked as farm day laborers that was comparableto the number of days they used hired labor. Yet very few days are so reported. Indeed, several strata of farm households in the ReyesOsorio study reported no days at all in farm wage employmentfor their members eventhough they were employers of wage laborers. Alternatively,the prevalence of a large amount of hired labor in some heavilypopulated regions has been attributed to paymentsmade to family 85 members, primarily adult sons who had little or no land of their own. While this may provide a partial explanation, it is unlikely to explain a significant part of the hired input. For example, on the basis of his six regional studies, Barbosa-Ramirezconcluded that family labor is unre- munerated almost in its entirety." Finally, some interviewed observers attributed the prevalence of so much wage labor to a concern of the rural communities for the economic status of the landless. In order to avoid extreme deprivation, the landlessare hired as a way of effectingan income redistribution within the community. Thus, even the poor are willing to share with those who are poorer still. In addition, there is status attached to being an employer of labor, so that this form of charity is viewed as yielding some psychic rewards. Perhaps more persuasive than these explanations is one that views the hiring of labor as a rational decision arrived at by a comparison of the cost of hiring labor with the opportunity cost of the time of landholders and their families. The field studies of rural Mexican communities report a very sensitive perception of the opportunity cost of time and household responsesleading to an allocation of labor that most efficiently realizes the 87 To the extent that the rural labor force is desired level of familyincome. not homogeneous,individualswillhave differentopportunity costs.Many landholders may find it more profitable to accept employment off-farm and hire lower-wage laborers to cultivate their plot. A not unreasonable assumption is that, as a class, landholders are likely to possess greater endowments of human capital and labor market connections then the landless rural labor force. Thus, the existenceof an activemarket for wage labor suggests that labor is a scarce resource even in poor rural communi- ties. This conclusion would be consistent wiih studies of some resident fieldresearcherswho have observedrelatively little involuntary underem- ployment. It also jibes with the experience of the Plan Puebla program which encounteredresistance among subsistencefarmers to technological changes that would increase labor requirements. 88 130 AGRICULTURE AND THE LABOR MARKET To be sure, most of the literature concentrateson households cultivating their own land and ignores the plight of landless laborers. If one were to accept the observation of ReyesOsorio, cited earlier in this chapter, that the number of days of annual employment had declined to only 100 by 1960, this would indeed be indicativeof a precarious situation. However, those estimates were limited to days of agricultural employment and ignored all others. Another study suggests that the number of days of annual employmentincreases substantiallyif account is taken of all work sources. For example, in the mid-1960s, landless laborers reported total number of days worked annually ranging from 177 in Laguna, to 226 in several zones of Michoacan, and 238 in Tlaxcala. 8 9 Although this would still leave a significant number of days potentially availablefor employ- ment, there is no way of knowing what proportion of these would actually have been made available. Evaluation of the Findings If one takes an objective view of the economic conditions of the agri- cultural population at a moment in time, there is much reason for concern. A more encouraging assessmentmight emerge,however, if one places the changes observable in the agricultural sector in a secularperspective. It is generally accepted that these changes add up to a gradual process of modernization of Mexican agriculture. I would add to this that they are also indicative of a greater degree of integration of agricultural and nonagricultural labor markets. What I find interesting about the assess- ments made of these developmentsis that they are held to be intensifying the "employment problem" of the rural sector. The increasein mechaniza- tion is seen as causing a reduction in labor utilization. Analytically,this is treated as an autonomously determined independent variable that ex- pands the ranks of the openly unemployed. One could legitimately ask, however, whether the line of causation doesindeed run from technological change to unemployment.In the case of mechanization,for example, one might instead hypothesize that the recent acceleration is a response to changes in the availabilityof labor and to a changingprice ratio of labor to machinery. This hypothesishas not been posed and so remainsuntested as far as I know. Yet there is a recognition that relative factor prices can determine how much labor and machinery will be employed, even though no effort has been made to explore the actual contribution of changing prices to past changes in the factor proportions employed. 90 Some authors attribute the extension of mechanization to government encouragement and to the subsidization of capital purchases by low-interest loans and tariff-free  EVALUATION OF THE FINDINGS 131 imports. 9 ' Although such support may have played a role, to be decisiveit would have to be shown that the price of machinery was fallingcontinu- ally relative to that of labor as the pace of mechanizationaccelerated. Unfortunately, the prices of farm machinery and of farm wage labor are not available over time; therefore, relative price trends cannot be estab- lished. And, of course, since all observers of i:heagricultural sector have harbored the conviction that it is endowed with a large surplus of labor, the possibility that an absolute or relative scarcity of labor could have motivated the increasein mechanization has never even been considered. Yet this possibility should not be so lightly discarded. Similarly,cropping changesin the direction of less labor-intensivecrops are viewed as autonomous developmentsthal reduce labor requirements and contribute to the underutilization of agricultural labor. There are straws in the wind, however, that suggestthat the cropping changes may reflect the rational response of output to changing demand for different food commodities as well as to the opportunity cost of labor to the agricultural sector. One of the striking findings of a recent study of this sector is the importance of livestock and poultry as a source of income to small farmers, those with fewer than four hectares of land. 9 2 In twelve of the twenty-fivestates surveyed, the income from these sources exceededthat from crop sales,usually by a substantial margin; the addition to crop sales of the value of home consumption would not changethis relationship. In several of the other states, livestock and poultry sales were significant in size even though they were outweighed by the value of crop production. Earlier in this chapter, it was noted that surveys of the sector during the 1960s valued livestockand poultry output at no more than 50 percent of the value of crops. If that was an accurate estimate of small farm output, then it would appear that even such plots, which one might expect to be devoted to subsistence production and cult)vated in a labor-intensive fashion, are increasinglybeing devoted to animal and poultry production. Hence an increased demand for forage cropswould follow, and an expan- sion of production of such crops and of livestock could be expected to displacethat of the traditional subsistencecrops of the small farm sector. Still remaining, however, is the question of why these changes have occurred, especiallyat the level of the small fa:rmunit. Unfortunately, it is impossible to offer definitiveexplanations. But it is possible to advance some plausiblepropositions for which some support can be found in many of the empirical studies of the sector that have been cited. The following is one scenario that might reasonably be expected to unfold in an economy that has been undergoing a high and steady rate of economic growth and that has been characte:rized by large disparities in 132 AGRICULTURE AND THE LABOR MARKET earnings between agriculture and nonagriculturalemployment.As will be demonstrated more clearly in Chapter 7, urban wages, even in the low- skill labor market, appear to have beenrising steadilysince 1940 and have thus increased the opportunity cost of agricultural labor. Those rural residents with the demographic characteristics required for successful employment in the nonagricultural sector will have strong incentives to seek employmentthere. They will work first on a seasonal basis and later, once the probability of success is perceivedto have exceeded a certain level,perhaps on a permanent basis. To the extent that the value of output of the vast majority of small plots had stagnated at very low levelsbetween 1950 and 1970 (as indicated by some of the studies cited earlier in this chapter), the gap between farm earnings and potential nonfarm earnings would have increased and thus created a greater incentive to shift from agricultural to other employment. Such a shift might then precipitate changes in the organization of agricultural production. Plotsyieldingvery low returns could simply be abandoned or consolidated with other plots into larger, more economic units either through sale or rental. During a transition period, cultivators could devote increasingproportions of their time to nonagricultural work while continuing to cultivate their land by using wage labor or that of family members. Alternatively, the labor requirements of production on smallholdingscould be reduced by changes in cropping patterns toward less labor-intensiveproducts; forage crops and livestock or poultry production would so qualify. Such a shift in output would also be motivated by the high income elasticityof demand for meat products in the face of rising average family incomes in the country as a whole. Intersectoral mobility of labor forestalls an increase in the size of the agricultural labor force even as the harvested area continues to expand. Indeed,the effectivesupply of labor to the sector may actuallybe declining if a numerically stable labor force allocated a shrinking proportion of its available working days to agricultural production in favor of nonfarm employment.In such a context, a rising land-labor ratio also serves as a stimulus to the adoption of technological and cropping changes that are labor-saving.The use of farm machinery would be encouragedas would a shift to less labor-intensive crops. Increases in output per unit of labor employed would follow. In view of the rising opportunity cost of labor to the agricultural sector and its increasing relative scarcity, it would be reasonable to expect the price of agricultural labor to rise. An added stimulus to this process would thus be provided, especiallyif this price rises relative to that of farm machinery services. Gradually, the flow of labor out of the sector would be expected to lead to an intensificationof the land consolidation process. This would continue until units were  EVALUATION OF THE FINDINGS 133 formed that could yield incomes competitivewith those availablein other sectors. Clearly,this processof adjustmentis likelyto be a gradual one, but one could argue that it has already beenunder way for some three decades and that it accelerated during the decade of the 1970s. Consider the data presented in table 4-5. Up until 1970, there was a steady expansion in the area harvested and in the number of plots, expan- sion that outran the growth of the agriculturallabor force. This resultedin a gradual increasein the land-labor ratio until, in 1970, it stood 67 percent above the 1940 level. One of the striking features of the table is that, in spite of the increased number of agricultural plots, the number of self- employed farmers has been declining since 1950. This has been particu- larly sharp among private farmers since the absolute number of such plots declined visiblyby almost 30 percent between 1950 and 1970. However, the table masks the probability that, in spite of a notable increase in the number of ejido plots (804,000, or 58 percent), the number of self- employed ejidatarioshas also declined,perhaps significantly. By 1970, the number of plots outnumbered the number of s Alf-employed farmers by 1.1 million. Underlying these observable changes may be any or all of the hypo- thetical adjustments that were described above as possible concomitants of dynamic economic growth. For example.,small private farmers are reported to be abandoning the cultivation of :heir holdings completelyin favor of nonagricultural employment. If they sell their holdings to other surviving farmers, the land and potential incomes of the latter should increase. If they retain their land, they may continue to cultivate it with hired labor and family members. This would be profitable as long as the owner's wage in off-farm employmentexceedsthe wage of rural day labor and the reservationprice of familylabor. Ejidatarios, of course, cannot sell their land and may forfeit their usufruct if the land is not cultivated.They Table 4-5. Number of Harvested Hectares, Cultivated Plots, and Agricultural Workers, 1940-70 (thousands) Plots Labor force Area harvested Self- Labor- Family Year (hectares) Private Ejidos Total employed ers workers Total 1940 6,637.15 1,211 1,223 2,434 1,726 1,913 192 3,831 1950 8,528.00 1,366 1,378 2,744 2,905 1,567 340 4,812 1960 11,297.00 1,346 1,532 2,869 2,340 2,204 504 5,048 1970 14,840.00 997 2,182 3,179 2,076 2,499 528 5,103 Source:CESPA, El empleo de mano de obra, tables 14 and 15. 134 AGRICULTURE AND THE LABOR MARKET therefore have less incentive to abandon the land, and one would not expect to find the number of ejido plots declining.The plots may continue to be cultivated with hired and familylabor, however, even as the ejida- tario himself shifts to off-farm employment. Therefore, some part of the increased discrepancybetween the number of agricultural plots and that of self-employedfarmers may reflect an increase in the average number of plots cultivated by each individual farmer. Part may also be ascribed to the substitution of hired and family labor on lands that are retained even though the head of household is no longer activelyengaged in exploitingthem. Unfortunately,the agricultural censuses do not permit us to delineate the contribution of each of these factors-plots are recorded individually even though severalmay belong to or be cultivated by a single individual. Thus, any changes that may be occurring in the pattern of land ownership or usage are obscured. The presumption that farm households have been heavily involved in off-farm employment has, of course, been substantiated by numerous studies, some of which have already been cited above. There is no way of establishing empirically the secular trend in the proportion of available work time devoted to off-farm employment.One can estimate the rate at which entire rural households haveabandoned the countrysidein favor of urban residence (internal migration is the subject of Chapter 5). We can also observe the considerable importance attributable to sources of in- come other than own-plot production for the current farm population, especiallyhouseholds with plots of under four hectares. In several states, the averageincome receivedfrom off-farm employmentand familyremit- tances exceedsthat from the sale of farm produce. 93 On the one hand, this should facilitate a shift from subsistenceto market-oriented production. On the other, it may reflect a state of transition for rural households as their status shifts from agricultural to nonagricultural producers. It is reasonable to expectthat this process will continueinto the future if the overall growth rate of the Mexican economy can be restored to levels approximating the past secular rate. The relativeattractivenessof nonagri- cultural wages should continue to stimulate mobility out of agriculture. Viewedfrom the opposite perspective,in order for agriculturalproduction to be worthwhile, it will have to yield an increased return to labor and at least keeppace with returns in alternative occupations. This is likelyto be possible only as the agricultural sector undergoes an increasingcommer- cialization, with a progressiveconsolidation of smallholdingsinto larger units capable of providing an income competitive with that obtainable from without the sector.  EVALUATION OF THE FINDINGS 135 This process seems to be well under way. I have noted the increasing land-labor ratio with its likely consequences for labor productivity. I can also point to earnings data reported by the 1970 and 1980 population censuses; they suggest that earnings in agr:culture have increased sig- nificantly relative to those in other sectors. I have compared the changes in the median monthly earnings derived from agricultural employment with those originating in the nonagricultural sec':ors from 1970 to 1980 by state. Although some questions may be raised about the quality of the reported earnings data, these do seem to evince a large and reassuring 94 measure of internal consistency. As can be seen from table 4-6, median agricultural earnings in twenty- six of the thirty-two states increased considerably more rapidly than those originating in the rest of the economy. In four other states (Colima, Quintana Roo, Sinaloa, and Sonora), three of which were relatively high- earnings states, income in the two sectors aclvanced at roughly the same rate. In only two, Nayarit and Baja California Sur, which boasted rela- tively high agricultural earnings in 1970, did agricultural income lag nonagricultural over the decade. Equally significant is the observation that the largest relative increases in reported agricultural earnings occurred in states such as Durango, Queretaro, Zacatecas, and San Luis Potosi, in which the absolute levels of farm earnings were lowest in 1970. In other words, a process of narrowing clifferentials in income from agricultural and nonagricultural employmernt appears to have occurred within states during the decade as well as a narrowing between rich and poor states. The latter tendency is confirmed by the strong negative correlation between the income level in 1970 and the percentage change in the median income between 1970 and 1980. The correlation coefficients for the agricultural and nonagricultural sectors are -0.51 and -0.63, respectively, both significant at the 1 percent level. Furthermore, the degree of dispersion, among states, as measured by the coefficient of variation also declined, from 49.9 to 41.1 percent in agriculture and from 24.5 to 18.9 percent in the nonagricultural sctors. The intrasectoral and intersectoral narrowing that appears to have occurred would be expected to occur only as labor supply conditions generally become "tighter." Certainly the observable narrowing trend would not be expected to occur in a rural environment characterized by "surplus labor" readily available for employment. To the extent that the relatively rapid increases in incomes of labor force members reflect increasing prices of labor services, these might represent contributory explanations for the accelerated pace of mechanization and cropping changes. Table 4-6. Median Monthly Earnings by State (pesos) Total Agricultural Nonagricultural Percent Percent Percent State 1970 1980 change 1970 1980 change 1970 1980 change Aguascalientes 492.8 3,422.4 594.46 359.3 2,612.8 627.27 670.0 3,734.3 457.39 Baja California N. 1,332.8 5,917.7 344.00 929.6 4,816.7 418.13 1,414.3 6,104.6 331.62 Baja California S. 899.0 5,957.9 562.69 692.7 4,129.0 496.09 1,050.6 6,508.2 519.47 Campeche 451.8 3,608.4 698.74 340.5 2,560.5 651.91 659.3 4,376.0 563.72 Coahuila 685.1 4,348.1 534.63 377.9 2,323.3 514.71 858.7 4,769.8 455.48 Colima 581.6 4,269.7 634.18 531.2 3,383.6 537.01 768.4 4,878.8 534.92 Chiapas 247.0 2,671.6 981.50 215.1 1,919.0 792.17 506.5 3,601.3 611.00 Chihuahua 719.1 4,532.1 530.27 402.9 3,233.2 702.53 896.5 4,788.8 434.17 Durango 426.8 3,522.5 725.36 292.3 2,366.0 709.30 703.6 4,193.6 496.05 Federal District 1,027.8 5,843.1 468.47 598.9 5,637.4 841.23 1,053.4 5,599.3 431.53 Guanajuato 405.5 3,258.8 703.60 259.5 2,185.1 742.00 682.2 3,726.3 446.21 Guerrero 270.9 3,160.0 1,066.36 171.7 1,592.1 827.42 733.9 3,917.0 433.73 Hidalgo 272.2 2,395.4 780.06 194.1 1,213.8 525.18 675.3 3,641.0 439.13 Jalisco 614.7 4,427.8 620.34 422.6 2,829.5 569.51 803.2 4,771.7 494.10  Mexico 421.5 4,124.9 878.54 265.7 1,981.2 645.72 753.3 4,741.7 529.48 Michoacan 381.1 3,127.6 720.64 291.7 2,446.4 738.57 606.9 3,744.0 516.91 Morelos 492.1 3,623.9 636.48 382.3 2,386.8 524.40 790.3 4,256.4 438.61 Nayarit 554.8 3,398.1 512.46 531.7 2,960.3 456.81 740.7 4,213.7 468.86 Nuevo Le6n 883.3 5,354.9 506.22 391.8 2,818.9 619.41 970.6 5,636.5 480.69 Oaxaca 187.6 2,291.9 1,121.69 159.8 1,261.7 689.53 426.7 3,300.8 673.55 Puebla 287.5 2,693.1 836.65 198.6 1,325.3 567.16 687.9 3,920.1 469.88 Queretaro 391.2 3,618.6 825.03 230.2 2,083.2 804.90 689.0 4,330.2 528.47 Quintana Roo 507.5 4,507.4 788.23 332.8 1,979.8 494.82 893.1 5,322.9 495.98 San Luis Potosi 307.0 2,771.0 802.58 213.1 1,524.4 615.33 637.7 3,702.3 480.58 Sinaloa 742.8 4.496.5 505.35 615.4 3,533.3 474.12 904.2 5,210.1 467.23 Sonora 883.5 4,946.0 459.84 753.6 4,142.9 449.78 998.6 5,535.1 454.28 Tabasco 406.9 3,879.1 853.36 308.6 2,428.7 686.95 804.3 5,398.8 571.23 Tamaulipas 720.7 4,499.1 524.25 414.1 2,783.1 572.00 904.4 4,865.7 438.01 Tlaxcala 367.9 2,925.8 695.29 263.4 1,739.3 560.43 630.9 3,520.7 458.04 Veracruz 424.0 3,157.7 644.69 292.6 1,971.2 573.67 833.9 4,480.4 437.26 Yucatan 272.0 2,609.3 859.26 189.7 1,129.2 495.21 602.0 3,528.5 486.13 Zacatecas 312.4 2,785.2 791.43 233.5 2,147.7 819.79 553.2 3,353.4 506.13 Seret-r-ade yC -ources-de Fstadistica. IX Censo de poblaci6n. 1970 (Mexico. D.F.. 1972): and Secretaria de Direcri6n Gener1l f-rril Programaci6n y Presupuesto, X Censo de poblaci6n y vivienda, 1980 (Mexico, D.F., 1983, 1984), 32 vols. 138 AGRICULTURE AND THE LABOR MARKET Summary Althoughit has beenpopularto describe the agriculturalsectoras the repositoryof considerable open unemployment and underemployment, theevidence seemsto suggest that thisperception correct.It isnot entirely is true that earningsderivable from agriculture havebeenlow.This has beenlargelythe resultof a disproportionately greatnumberof verysmall farmunits,manyof whichlackadequate supplies of cooperating factorsof production, especially a reliablesourceof water. The sector has been undergoing a gradualprocessof change, however, and an appreciation of the fullsignificance of that changecannotbe gainedif thesectoris viewed in isolationfromtherestof theeconomy. Thefarmsectorhouseholds have long beenintegratedinto regionaland nationallabor markets.Income from off-farmsourceshas beensignificant in mostregionsfor whichwe have information,eventhose seemingly most isolatedfrom largeurban areas.The processof modernization of the sectoris evidentin the rising land-laborratio andthe increasing mechanization ofproduction. It is also reflectedby the shift in croppingpatternsawayfrom traditionalstaples and labor-intensive cropsto oilseed, grain,or foragecropsand livestock and poultryproduction.Ratherthanviewing theseas changes contribut- ing to an aggravation of the ruralemployment problem,one shouldtreat themas possibleresponses to a tightening rurallabormarketin whichthe cost of laboris risingand in whichabsolute scarcitiesmaybe occurring at criticaltimesin the croppingcycle. Clearly, the traditional methods of estimating the degree of underem- ployment of the agricultural population need to be reconsidered. The assumption that all adults in farm households or that all those working in agriculture for any length of time during the year are interested in, and available for, full-time employment represents an untenable simplification of the determinants of labor force participation. A careful investigation of the supply of labor effectively available for employment is likely to con- clude that farm households make as full use of that supply as they deem desirable and that involuntary unemployment or underemployment char- acterizes only a relatively small proportion of the rural population. Fur- thermore, the practice of measuring underemployment by deducting from the presumed size of the agricultural labor force the number of work days required for recorded levels of output overlooks the importance of em- ployment outside the agricultural sector. By treating the agricultural sector in isolation, one can uncover a huge surplus of labor. However, what may appear to be a surplus to the agricultural sector may not be surplus to the economy as a whole. To the  NOTES 139 extent that agricultural households divide their time between intra- and extrasector employment, the latter cannot be ignored. Nor is off-farm work to be viewed as an act of desperation on the part of impoverished farm households. All the evidencepoints to 2. much more careful consid- eration by households of the returns to labor from differentemployments. In addition, household labor resources are allocated in a way that is most likely to minimize the resource costs of gaining the income target estab- lished for the household. When the opportunity cost of the household head or other members of the unit exceeds the returns to own-account employment, these workers seek off-farm employmentand substitute less costly hired or other family labor. As links to the nonagricultural labor market become firmly established, it is reasonable to expect that the process of mobility from a farm to a permanent nonfarm status will be facilitated. In short, a case can be made that Mexico is undergoinga processsimilar to that observedduring the processof modern izationand industrialization of the now more highly developedeconomies.Labor resources, respond- ing to differences in potential earnings, gradually are redirected from agriculture to employments characterized by higher productivity. As the rural labor force shrinks, changes in the structure of agriculturalproduc- tion occur, the productivity of rural labor increases, and the returns to farm labor also increase. The maintenance of sustained growth of the national economy may then be expectedto lead to a narrowing in some of the extreme intersectoral differentials in earnings. Notes 1. ClarkW. Reynolds,The MexicanEconomy (NewHaven,Conn.: YaleUniversity Press, 1970), p. 136. 2. World Bank data. 3. The cost in human life of the revolution must havebeen veryhigh; the censusof 1921 reported a declinein the population from 15.2 millionin 1910 to 14.8 million.During the preceding decade, the population had increasedby 1.4 million,or 11 percent. 4. U.S. Agencyfor InternationalDevelopment(USAID), SpringReview of Land Reform, June 1970, 2d ed., vol. 7 (Washington,D.C., 1970), pp. 19-20; Sergio Reyes Osorio, Rodolfo Stavenhagen,Salomon Eckstein,and Jaun Ball steros, Estructuraagrariay desa- rollo agricolaen Mexico (Mexico,D.F.: Fondo de Cultura Econ6mica,1974), pp. 43-52. S. Reynolds,MexicanEconomy, p. 141. Reynolds' figuresapparendy refer to the size of arable plots assignedto each family.If the total area of tie ejidois dividedby the number of recipients of land, the averagearea for each member is considerablylarger-the total area includes pasture or other common lands as well as that dieemed unsuitablefor purposes of cultivation. Measures of total area are available in JamesW. Wilkie, The Mexican Revolu- tion: FederalExpendituresand SocialChangesince 1910 (Berkeley: Universityof California Press, 1968). 140 AGRICULTURE AND THE LABOR MARKET 6. Hacendados were permitted to retain for their own use 100 hectares of irrigated land or 300 hectares of rain-fed land. In addition, under provisions of nineteenth-century laws governing "colonization," it was possible for large landowners to divide their estates into smaller units. These were thus ineligible for expropriation and could be distributed among family members. The several units could then continue to be cultivated as a single entity. This practice was outlawed by legislation in 1962. USAID, Spring Review, pp. 20-22. 7. Reynolds, Mexican Economy, pp. 142-44; uSAID, Spring Review, pp. 29-30. 8. Reyes Osorio and others, Estructura agraria, pp. 56-58. 9. Ibid., pp.197-98. It should be noted, however, that these output figures may understate the production accruing to cultivators of minifundia. The census records each plot separately, while some small farmers may cultivate more than one unit. It is not unusual, for example, to find ejidatarios who also own private plots. It is of interest to note that over 60 percent of even these small units marketed at least part of their output. 10. Ibid., p. 349. 11. Ibid. 12. The value terms employed here are gross value of output, not net income. For the smallest category of farm units, there are probably few purchased inputs, and the amortiza- tion of capital equipment is likely to be minimal. Thus, the difference between gross and net incomes is probably minor. The importance of these factors and of hired labor, however, may be expected to increase with farm size; therefore, the difference between gross and net income for larger units would also increase. 13. An account by Aleman illustrates the uncertainty posed by climatic conditions in rain-fed agriculture. A study of sixteen ejidos located in the state of San Luis Potosi found that 37 percent of the ejidatarios lose their entire crop as a general rule. In 1965, some ejidatarios were reported to have gone five consecutive years without a harvest, in spite of which they persisted in annual plantings in hopes that rainfall would prove adequate. Farmers who successfully harvested a crop reported very low yields. Eloisa Aleman, Investigaci6n socioeconomica directa delosejidos deSan LuisPotosif(Mexico, D.F.: Instituto Mexicano de Investigaciones Econ6micas, 1966), p. 68. 14. Reyes Osorio and others, Estructura agraria, pp. 199-201. 15. P. Lamartine Yates, Mexico's Agricultural Dilemma (Tucson: University of Arizona Press, 1981), p. 162. 16. Ibid., p. 134. 17. Reyes Osorio and others, Estructura agraria, p. 214. 18. World Bank, "The Economy of Mexico," table 7.4 and 7.5. 19. Yates, Mexico's Agricultural Dilemma, pp. 156-59; Teresa Rend6n, "Utilizaci6n de mano de obra en la agricultura Mexicana, 1940-1973," Demografia y economia, vol. 10, no. 3 (1976), p. 366. 20. Kaja Finkler, "From Sharecroppers to Entrepreneurs: Peasant Household Production Strategies under the Ejido System of Mexico," Economic Development and Cultural Change, vol. 27 (October 1978), pp. 103-20. 21. Yates, Mexico's Agricultural Dilemma, p. 159. 22. Enrique Contreras S., Estratificaci6n y movilidad social en la ciudad de Mexico (Mexico, D.F.: Universidad Nacional Autonoma de Mexico, 1978), p. 375. 23. Rend6n, "Utilizaci6n de mano de obra," pp. 367-68. 24. Gabriel Maldonado Lee, La muier asalariada en el sector agricola (Mexico, D.F.: Centro Nacional de Informaci6n y Estadisticas del Trabajo, 1977), pp. 69, 74. 25. AlemAn, Investigaci6n socioeconomica directa, p. 112. 26. Yates, Mexico's Agricultural Dilemma, p. 163; Roger D. Norton and Leopoldo Solis M., eds., The Book of CHAC: Programming Studies for Mexican Agriculture (Baltimore,  NOTES 141 Md.: Johns Hopkins Universtiy Press, 1983), pp. 346-47; Centro de Investigaciones para el Desarrollo Rural (CIDER), Mercado de trabajo del algod6n (Mexico, D.F., n.d.). 27. Jorge Balan, Harley L. Browning, and Elizabeth Jelin, Men in a Developing Society (Austin: University of Texas Press, 1973), p. 79. 28. George M. Foster, Tzintzuntzan: Mexican Peasants in a Changing World, rev. ed. (New York: Elsevier, 1979), pp. 275-76. 29. Douglas S. Butterworth, Tilitongo: Comunidad mixteca en transici6n (Mexico, D.F.: Instituto Nacional Indigenista, 1975), pp. 173-93. 30. Billie R. DeWalt, Modernization in a Mexican Ejido (London: Cambridge University Press, 1978), p. 221. 31. Lourdes Arizpe, Migraci6n, etnicismo y cambio ec2n6mico (Mexico, D.F.: El Colegio de Mexico, 1978), pp. 230-31. Workers from rural communities often migrate together, taking with them one or two women to keep house for them. Sometimes villages specialize in a certain occupation: one village might be a source of mas ons, another of welders, and so on. 32. A. Rene Barbosa-Ramirez, Empleo, desempleo y subempleo en el sector agrope- cuario, vol. 1 (Mexico, D.F.: Centro de Investigaci6nes Agrarias, 1976), p. 159. 33. Aleman, Investigaci6n socioeconomica directa, pp. 69-71. 34. The existence of a substantial differential between wages reportedly paid day laborers by smallholders and large farm units is puzzling. One would expect the local rural labor market to be highly competitive and for wage differentials among various employing units to be narrow. It is possible, of course, to hypothesize abo It factors that could give rise to a differential within such a market. These would include quality differences among workers, longer commuting distances to large farms, a difference in the timing of peak periods of hiring by small and large farmers, and so on. It may be that the l arger farms pay wages equal to the legal minimum even though the effective supply price of labor lies below that level. However, since rural employers in most parts of Mexico at that time were reputed to ignore the legal requirements of employment, why should they not have ieen doing so here? Unfortunately, Aleman offers us no clue. 35. Reyes Osorio and others, Estructura agraria, p. 353. 36. Ibid., p. 401. 37. Ibid., pp. 399, 1084-85. 38. A. Rene Barbosa-Ramirez, El Bajio (Mexico, D.F: Centro de Investigaciones Agrar- ias, 1973), chap. 3. 39. A. ReneBarbosa-Ramirez, Empleo, desempleo y subempleo en elsectoragropecuario, 3 vols. (Mexico, D.F.: Centro de Investigaciones Agrarias, 1976, 1977, and 1979). 40. Ibid., vol. 2, p. 135. 41. Ibid., vol. 1, p. 229. 42. Ibid., pp. 132-33. 43. Gobierno Constitucional de los Estados Unidos Mexicanos and Gobierno Constitu- cional del Estado de Oaxaca, Programa de desarrollo rv'ral integral de las Mixtecas Oaxa- queiias Alta y Baja 1984-1988 (Mexico, D.F.: Secretaria de Programaci6n y Presupuesto, 1984), pp. 45-59. 44. Secretaria de Agricultura y Recursos Hidraulicos. Centro de Estudios en Planeaci6n Agropecuaria (CESPA), El desarrollo agropecuario de Mexico, vol. 6, El empleo de mano de obra en las actividades productivas agropecuarias (Mexizo, D.F., 1982), pp. 53, 61, 84-85, 186. 45. Rend6n, "Utilizaci6n de mano de obra," pp. 363, 372-73. 46. Cynthia Hewitt de Alcantara, Modernizing Mexican Agriculture: Socio-economic Implications of Technological Change 1940-70, Report no. 76.5 (Geneva: United Nations Institute for Social Development, 1976), pp. 64-72. 142 AGRICULTURE AND THE LABOR MARKET 47. Barbosa-Ramirez, Empleo, desempleo y subempleo, vol. 1, p. 138. 48. CESPA, El empleo de mano de obra, p. 49. 49. CESPA, El desarrollo agropecuario de Mexico, vol. 9, Inversion predial y formaci6n de capital (Mexico, D.F., 1982), p. 28. 50. Rend6n, "Utilizaci6n de mano de obra," pp. 371-72. 51. CESPA, El empleo de mano de obra, p. 62. 52. Ministry of the Presidency, Govemment of Mexico, "A Program for Mexican Agricul- ture," in The Book of CHAC, ed. Norton and Solis, p. 191. 53. Rend6n, "Utilizaci6n de mano de obra," pp. 372-73. 54. CESPA, El empleo de mano de obra, p. 52. 55. Rend6n, "Utilizaci6n de mano de obra," pp. 369-71. 56. CESPA, El empleo de mano de obra, p. 78. 57. Rend6n, "Utilizaci6n de mano de obra," p. 374. 58. CESPA, El empleo de mano de obra, pp. 103-06. 59. Ibid., p. 113. 60. Ibid., p. 135. 61. Barbosa-Ramirez, Empleo, desempleo y subempleo, vol. 1, pp. 173, 239; vol. 2, p. 136; vol. 3, pp. 120, 261. 62. CESPA, El empleo de mano de obra, p. 83. CESPA advances different estimates for the size of the labor force without venturing an explanation for the differences. In deriving the estimate of days worked in 1977, the authors computed a labor force size of 5.754 million. Earlier in the same study (p. 52), a figure of 6.557 million was used. Both numbers would appear to represent substantial overestimates of the actual agricultural labor force. For example, the latter number would imply an annual rate of growth of 3.6 percent since 1970, far above the approximately 0.1 percent rate of the preceding decade. In table 2-S I estimated its growth at an annual rate of 0.5 percent. The unadjusted census data for 1970 and 1980 indicate an annual growth rate of 1 percent. 63. Reyes Osorio and others, La estructura agraria, pp. 343-44. 64. Teresa Rend6n cites a study undertaken by the Colegio de Mexico in thirteen rural localities in 1975. It found that in eleven of these communities the median agricultural wage was below the legal minimum. In six, the median was at least 40 percent below the legal. Rend6n, "Utilizaci6n de mano de obra," p. 36Sn. 65. The reader is referred to Chapter 7 for a discussion of the course of minimum wages since 1940. 66. Centro de Investigaciones Agrarias, Estructura agraria y desarrollo agricola de Mex- ico, vol. 1 (Mexico, D.F., 1970), p. 593. 67. CESPA, El empleo de mano de obra, p. 183. 68. Ministry of the Presidency, "A Program for Mexican Agriculture," pp. 162-63, 171-72. 69. CESPA, El empleo de mano de obra, pp. 157-62. 70. Barbosa-Ramirez, Empleo, desempleo y subempleo, vol. 1, p. 168. 71. Ibid., p. 222. 72. Ibid., vol. 2, pp. 127-28. 73. Ibid., vol. 3, p. 114. 74. Ibid., p. 208. 75. Ibid., vol. 2, p. 128. 76. Ibid., vol. 3, p. 252. 77. Ibid., p. 249. 78. Ibid., p. 254. 79. Ibid., pp. 259-61.  NOTES 143 80. BancoNacional de Credito Rural (BANRURAL), Fideicomiso para Estudiosy Planesde Desarrollo Agropecuario (FEDA), Empleo, subempleo y desempleo en el sector rural,por subregioneseconomicas,31 vols. (Mexico,D.F., 1981-83). Althougha researchreport was publishedfor each state studied,only eighteenwere availableto me. The data cited,therefore, refer only to these states. 81. Barbosa-Ramirez,Empleo, desempleoy subempAeo, vol. 2, p. 123. 82. Ibid., vol. 1, pp. 222-39. 83. Ibid., pp. 130-33. 84. Ibid.,pp.226-27. Another exampleof the change:n the traditionalallocation of labor servicesin rural communitiescan be found in the passingof the practice of tequio.This was the provision of free labor by local residents for various social or public works, such as maintenance and construction of local roads, schools, public buildings, and churches. Volunteer labor has been largely replaced by wage labor. Ibid., pp. 230-31. 85. Reyes Osorio, La estructuraagraria,p. 426. 86. Barbosa-Ramirez,Empleo, desempleoy subemp'eo, vol. 3, p. 317. 87. Ibid., p. 112. Seealso, Arturo Warman,... Y venimos a contradecir(Mexico,D.F.: Ediciones de la Casa Chata, 1976), as cited in ibid., p. 112. 88. MichaelBelshaw,A VillageEconomy (New York:ColumbiaUniversityPress,1967), pp. 135-54; Rend6n, "Utilizaci6n de mano de obra," p).353n. 89. Rodolfo Stavenhagen,"Los jornalerosagricolas,' Revista de Mexico agrario,vol. 1, no. 1 (1967), p. 165. 90. Luz Maria Bassocoand RogerD. Norton, "A Quantitative Framework for Agricul- tural Policies,"in Norton and Solis,The Book of CHAC, chap.5; Ministryof the Presidency, "A Program for Mexican Agriculture," p. 191. 91. Hewitt de Alcantara,ModernizingMexicanAgriculture,pp. 64-72; Yates,Mexico's Agricultural Dilemma, p. 129. 92. Banco Nacional de Credito Rural, S.A., Fideicomiso para Estudios y Planes de Desarrollo Agropecuario, "Principales indicadores del estudio del empleo, subempleoy desempleo rural" (1983), table 9-A. 93. Ibid., table 9-A. 94. As is the case with severalof the tables reporting labor force-related data, the 1980 censusreports a considerablenumber of individuals(abo at 16percent of the total) for whom no earnings data were available or who could not be classifiedaccording to a sector of employmenteven though earningswere reported. In calculatingthe sectoral median earn- ings, I have limited myselfto only those for whom earnings and sector of employmentare reported. Earnings recipientsare dividedinto two categories:those assignedto agricultural employmentsand all other. The totals for each state, however,includethe earningsof those who were not assigneda sector of employment. 5. Internal Migration and Regional Development THE P RO C ES S OF economic development involves both an increase in the quantity of productive factors and a reorganization in the manner in which existing resources are employed. Expansion of these factors is usually dominated by growth in capital, both physical and human. The reorganization of production involves concomitant change in the structure of demand for labor that will have several dimensions: sectoral, occupa- tional, and geographic. Successful development will therefore require a labor force that is willing and able to adapt to these changes. In a free labor market, the changing structure of demand for labor will be reflected in changing wage relationships that are designed to encourage flows of workers to expanding areas and occupations. The responsiveness of the labor force (as well as of potential members of the labor force) to these inducements to move or adapt will be governed by a number of economic, cultural, and institutional factors. The greater the responsive- ness of workers to the shifting structure of demand, the smoother will be the process of growth and the more favorable the prospects for increasing the degree of equality in the distribution of income over time. Mexican workers would seem to be very willing to migrate when the incentives are present. Seasonal migration of labor within the agricultural sector, as well as between the rural and urban sectors, has long been a feature of the Mexican labor market. Short-term migration across the northern border to the United States has a history extending back into the nineteenth century when the building of the transcontinental railroads through the Southwest attracted large numbers of Mexican laborers. That migration has continued into the present century, in the form of both formal recruitment programs, for example, the bracero program of the 1940s and 1950s, and informal, illegal border crossings. Within Mexico, the rapid economic development that has characterized the country since 1940 has seen large numbers of people willing to make permanent moves in response to new and improved opportunities of employment. The accelerated development of an improved transportation 144  INTERNAL MIGRATION 145 system and the expansion of the means of communication facilitatedthe mobility of labor and provided for the disseminationof information that was vital to its efficientredeployment.' The institutional changes flowing from the revolution also removed any barriers to mobilitythat might have been posed by the traditional organization of agriculture. Unlike those in other poor lands, agricultural workers were not tied to the land or to a particular landowner by caste or bonds of servitude.Thus, they were free to exercisechoicewith respectto their location and, to a lesserextent, to a sector of employment. In this chapter, the sizeand direction of internal migratory flowsin the years since 1940 will be examined. The correlates of these flows will be reviewed, as these data have been defined by econometric studies of migration. In addition, the characteristicsof migrants will be described, along with their economic status prior to and following their moves. Attention will be given to the process by vvhichthe migrant becomes integrated into the labor market at the destinationpoint. Migration would be expectedto representa factor shaping regionaldifferencesin productiv- ity and incomes.Therefore, in the finalsection of this chapter, there will be a discussionof changes in regional measuresof income that have emerged over four decades of economic expansion arid migration. The literature is dominated by two broad views of the underlying motivation for and consequences of migration in Latin America. One, which is descriptive and is based on more or less casual empiricism,holds that migration is motivated by rural poverty, which "pushes" the popula- tion out of an economicallyhopelessenvironment into an urban environ- ment that is as bad or worse. In this scenario, employmentopportunities adequate to the absorption of the migrant flow are held to be lacking. Survival becomes dependent on the meager returns from casual self- employment or very low-wage jobs in an urban "informal" sector, par- ticularly in the servicesector. The literature speaks of migration as leading to the "marginalization" of populations or tc the "tertiarization" of the urban economy. Migration is said to contribute neither to increased efficiencyin the use of resources nor to an improvement in the lot of the migrants.Essentially,this represents a rather negativeevaluationof migra- tion and is a view that is frequently expressed in referencesto the migra- tion phenomenon in Mexico. The alternative approach treats migration as a product of the economi- cally rational decisionsof individualswho are respondingto the perceived costs and benefitsof their moves.The analytical framework for the study of migration derives from modern human capital theory. Individualsseek to maximize the present value of their net income streams by evaluating their employmentprospects in alternative settings. In this view,migration 146 INTERNAL MIGRATION represents a response to differences in wage levels and employment oppor- tunities, and it contributes both to increased economic efficiency and to improvement in the economic well-being of the migrants. In the case of Mexico, this evaluation is supported by the findings of anthropologists who have studied peasant communities or followed migrant populations in urban areas over time. It is also given credence by analyses of aggregated data drawn from population censuses or special sample surveys. Once the available evidence has been reviewed, validity of these two clashing views can be determined. The Size and Direction of Migration Flows The primary source of information regarding migration is the various population censuses. Unfortunately, the information that is contained in the censuses before 1970 provides no more than a partial indicator of population movement. Prior to that year, the only pertinent data were a count of persons residing in a state other than that of their birth. As such, this measure represented a stock of persons who at some time in their lives had changed their state of residence at least once and were currently residing in a state other than that in which they had been born. Not included, therefore, were those who, at the time of the census, were living in the state of their birth but who had resided elsewhere and had returned. This measure is to be distinguished from an accounting of migration that is a flow of population per unit of time. Furthermore, the procedure takes no account of intrastate migration. As a measure of migration, changes over time in the number of persons residing in a state other than that of their birth are far from ideal. First, they do not reflect the gross flows of populations, only the net flows. Thus, the movement of one non-native person in each direction between Jalisco and Queretaro would not be detectable as two cases of migration since they would leave unaffected the number of non-natives in each state. In addition, this measure does not provide a way to detail the multiple moves of any individual. At best, it may provide only a rough indication of the relative importance of population movements for the course of population change in the states over time. Yet, the data probably do serve to identify faithfully those states that have been significant losers and gainers of population through migration. The 1970 census marked a significant improvement in the way informa- tion relating to migration was presented. The traditional measure of non-native residents in each state appeared. Moreover, all those who reported ever having made an interstate move were classified according to  SIZE AND DIRECTION OF FLOWS 147 the length of residence in the state of current r esidence.Not only did this measure include migrants returning to the state of their birth but it also permitted the isolation of the migrants who changed their state of resi- dence within certain time intervalsprior to the census.Beforea discussion of the 1970 census, however, a review of the data based on non-native residents of the states is in order. 2 For each census year between 1940 and 1970, table 5-1 shows the non-native population as a percentage of the total population of each state and the relativeimportance of the non-nativepopulation in the intercensal growth. For the country as a whole, the data suggestan increase in the frequency of migration coinciding with the onset of rapid growth in the 1940s. The 1940 census recorded 10.6 percent of the population as residing in a state other than that of their birth. The proportion rose over the next two censuses and leveled off at just under 15 percent of the population. As might be expected, there was considerable variation among the states in the proportion of the population consisting of non- natives as well as in the rates of change in that subgroup over time. States with low income levels per capita tended to r eport small proportions of non-natives. Among these would be the southern states of Chiapas, Oax- aca, Guerrero, Tabasco, and Yucatan, as well as the depressed states in the Central and North Central region: San Luis Potosi, Michoacan, Hidalgo, Puebla, Queretaro, Tlaxcala, and Zacatecas. At the other extreme are the high-incomestates, whichhave constituted the development poles of the Mexican economy. In absolute terms, the entity with the largest concentration of non-natives is the FederalDistrict. The three censuses from 1940 through 1960 recorded over 40 percent of the population as being non-native.Between1960 and 1970, however,the natural rate of increasein the District became a more important contribu- tor to population growth, and the proportion of non-natives fell to one- 3 However, the third of the total. continued growth in the concentration of economic activity in the Mexico City metropolitan area resulted in spill- over effects.When the limits of the absorptive capacityof the District were reached, further activity and population growth engulfed the state of Mexico, which surrounds the District on three sides.This also occurred to a lesser extent in the state of Morelos, which abuts the District on the south. In 1940, only 3.5 percent of the state of Mexico's population was non-native. By 1970, this proportion had risen to over one-fourth. During the 1960s, this state was the principal destination of people emigrating from the District, accounting for one-third of rhesemigrants. The increase in non-natives in the state of Mexico accounted for 37 and 40 percent of the total increase in population during 1950-60 and 1960-70, respec- tively. 148 INTERNAL MIGRATION Table 5-1. Non-Native Population as a Proportion of Total Population and as a Source of Population Growth, by State, 1940-70 Percentage of Percentage of non-native intercensal growth' residents to total accounted for by change in population non-native population State 1940 1950 1960 1970 1940-S0 1950-60 1960-70 Aguascalientes 18.3 19.7 17.8 17.1 28.4 11.4 15.1 Baja California N. 48.2 S9.9 59.3 39.6 66.1 58.8 10.5 Baja California S. 4.5 4.4 14.2 16.9 4.3 43.0 21.6 Campeche 6.9 9.3 12.7 16.8 15.5 21.9 25.1 Coahuila 21.6 19.7 17.2 12.3 13.9 6.9 -51.8 Colima 22.8 24.2 27.2 27.7 27.5 33.5 28.7 Chiapas 2.0 2.3 4.1 2.3 3.4 9.4 - 97.6 Chihuahua 11.3 12.1 16.8 10.2 14.2 27.3 -63.7 Durango 11.1 10.9 10.3 6.6 10.4 7.3 - 86.8 Federal District 46.7 45.4 40.2 33.0 43.6 31.4 15.6 Guanajuato 3.7 4.2 6.2 5.6 5.9 12.7 3.5 Guerrero 2.0 3.1 3.6 3.3 7.4 5.3 2.4 Hidalgo 4.4 4.2 6.4 4.2 1.5 19.7 -110.5 Jalisco 4.5 5.3 7.9 10.2 8.5 14.4 16.8 Mexico 3.5 5.0 13.4 26.9 12.4 36.6 40.2 Michoacan 3.7 3.9 4.2 3.7 4.5 5.2 2.0 Morelos 20.7 23.8 26.3 26.6 29.9 32.5 27.1 Nayarit 15.4 15.8 16.1 15.1 17.4 16.6 12.5 Nuevo Le6n 14.7 18.4 23.6 23.6 28.4 35.0 23.6 Oaxaca 1.6 3.1 3.2 2.7 10.9 3.6 - 4.2 Puebla 4.4 5.7 5.9 5.9 10.5 7.0 5.9 Queretaro 5.4 5.5 5.8 8.3 6.3 7.0 14.9 Quintana Roo 26.1 24.8 39.2 43.3 22.0 56.0 48.7 San Luis Potosi 7.1 7.5 7.0 6.5 8.9 4.9 4.2 Sinaloa 6.4 6.6 9.1 11.7 7.6 16.8 16.9 Sonora 10.1 12.3 18.2 14.8 17.7 29.2 6.4 Tabasco 3.3 3.0 4.0 6.3 1.9 6.9 10.6 Tamaulipas 23.9 30.9 28.5 21.8 43.3 22.7 6.2 Tlaxcala 5.2 6.0 6.2 6.1 8.8 7.2 5.4 Veracruz 8.4 8.1 10.2 8.6 6.9 16.4 4.7 Yucatan 1.8 1.5 2.1 2.0 0.4 5.1 1.6 Zacatecas 6.6 5.7 4.8 4.6 0.3 0.9 3.4 Mexico 10.6 12.8 14.9 14.5 n.a. n.a. n.a. n.a. Not available. a. A negative sign indicates a decline in the non-native population. A negative percentage greater than - 100 indicates a decline in non-natives greater than the growth in population. Source: Secretarfa de Industria y Comercio, Direcci6n General de Estadistica, Censo de Poblaci6n, 1940, 1950, 1960, and 1970.  SIZE AND DIRICTION OF FLOWS 149 The other focal points of growing concentrations of non-native resi- dents are to be found in the northern border states, which are also relativelyhigh-incomestates. Most notable among these has beenthe state of Baja California. Between 1940 and 1970, the population of the state expanded elevenfold,an increase attainable only with a large infusion of migrants. As can be seen from table 5-1, the expansion of the non-native population accounted for approximately 60 pcrcent of the increasein total population during the firsttwo intercensalperiods. Over those same years, Tamaulipas, Sonora, and Nuevo Le6n likewise recorded significant growth associated with large increases in the non-native population. A clearer picture of the distribution of the migrant population both by state of origin and state of residence is provided by table 5-2, which is based on the 1970 census.The informationpresented is of two kinds. One is that based on the state of residencein 1970 and state of birth. The ratio of the number of non-natives residing in a given state to the number of persons originating in that state but residing in others provides a rough notion of relative gains and losses of population from migration. For example, the ratio of 7.5 for Baja California Norte implies a gain of 7.5 non-native residents for every person born in that state who was found residing in other states. Conversely,the ratio of 0.12 for Zacatecas implies that for every increase of one non-native in its population, the state recorded a loss to other states of more than eightpersons. The second basis of recording gains and lossesis a broader one; it refers to the destination and state of immediate previous residence of all those reporting any change in the state of residence.This measure thereforeincludesthose who have moved back to the state of their birth from another state. Although the size of the ratio changes for some states, the general orders of magni- tude are very similar. The states in table 5-2 are ranked in accordance with their per capita income in 1970. The direct relationship between the migrant ratios and income is readily apparent. The correlation between income and the broader measure of migratior.yieldsa coefficientof 0.41, significant at the 2 percent level. As mentioned above, the use of the changingproportions of non-native residents as an indicator of geographic mobility has severe limitations. Some cluesto the limitations of this measure canbe gleaned from the 1970 census inclusion of movers by length of residencein their current state. In the country as a whole, some 7,190 thousand persons, or 14.9 percent of the total population, were reported to have changed their state of resi- dence at any time in the past. Over half of these,or 55 percent,had made at least one interstate move during the precedingten years. As a proportion of the total population, this group represented about 8.1 percent. More revealing,however, is the extent of movement in the singleyear preceding 150 INTERNAL MIGRATION Table 5-2. States as Sources and Destination of Interstate Migrants, 1970 State in rank Ratio of Ratio of order by in-migrants in-migrants per capita to native to all GDP in 1970 out-migrants out-migrants Federal District 3.96 2.95 Nuevo Le6n 3.13 2.92 Baja California N. 7.50 5.15 Sonora 1.94 1.72 Coahuila 0.60 0.60 Tamaulipas 2.11 1.83 Baja California S. 1.53 1.32 Sinaloa 1.02 1.01 Chihuahua 1.44 1.39 Campeche 1.50 1.45 Mexico 2.32 2.23 Colima 1.96 1.69 Veracruz 0.98 0.98 Jalisco 0.64 0.71 Tabasco 0.70 0.69 Yucatan 0.16 0.18 Morelos 2.53 2.01 Aguascalientes 0.69 0.69 Quintana Roo 5.97 5.36 Durango 0.26 0.28 Nayarit 1.03 1.01 Queretaro 0.32 0.35 Guanajuato 0.23 0.26 San Luis Potosi 0.25 0.27 Puebla 0.37 0.39 Guerrero 0.23 0.24 Chiapas 0.39 0.41 Michoacan 0.14 0.16 Hidalgo 0.15 0.17 Zacatecas 0.12 0.13 Tlaxcala 0.23 0.26 Oaxaca 0.16 0.18 Source: Secretaria de Industria y Comercio, Direcci6n General de Estadistica, IX Censo de poblaci6n, 1970. Per capita GDP from Luis Unikel, El desarrollo urbano de Mexico, 2d ed. (Mexico, D.F.: El Colegio de Mexico, 1976), appendix table VI-A3.  RESEARCH FINDINGS 151 the census. Fully 723,000, or 10 percent of all movers,reported a change in their state of residencewithin the past year; tFey represented1.5percent of the total population. If 1969 was a typical year with respect to the frequency of mobility, this would imply a much higher rate of mobility than is suggested by any of the other measures reviewed thus far. What may also be implied is that, once an initial mcve from the home state is made, an individual is more prone to make a further interstate move. This act would not normally be detected as a move by the data as it is currently garnered by the census. Another indication of the size of net population flows may be gleaned from the census-basedmeasures of growth of the urban and rural popula- tions. Such flows would, of course, include iritra- as well as interstate movements. In common with most other developing countries, Mexico has experienced a significantshift of population from the countryside to towns and cities. If one adopts the census definitionof urban centers as those with populations of 2,500 or more, then the proportion of the population that is urban increased from 35 percent in 1940 to almost 59 percent in 1970. The annual rate of growth of the urban population over the thirty-year period was 4.8 percent. In contrast, that for the rural population was only 1.5percent. Estimatesof the net migration flow from the rural to urban areas amounted to some 3.755 million for the decade 1960-70.4 This was equivalentto 35.4 percent of the intercensalgrowth of the urban population. Although the rural-urban flowwas a rather general phenomenon, not all urban areas were recipientsof the flow. Urban areas in several states have grown at a rate below thc natural rate, indicating a net out-migration. As might be expected, these tend to be located in the lower-income states. For the decade 1960-70, they included Aguasca- lientes, Coahuila, Durango, Hidalgo, San Luis Potosi, Yucatan, and Zacatecas.' Severalof these states had suffered net out-migration of their urban populations during the preceding decacleas well. Unfortunately, the final 1980 census results did not become available for the country as a whole in time for me to report the contribution that migration during the 1970s may have made to the distribution of popula- tion in 1980. Research Findings on Migration There is an extensiveliterature on internal migrationbased on empirical research. This literature is divisible into two broad categories that are based on the character of the methodologyemployedand the data sources. One of these groupings uses data drawn from the population censuses. 152 INTERNAL MIGRATION The treatment of the data ranges from pure description to the analysis of the correlates of migration by the application of econometric techniques. The other class of research is based on sample surveys of migrants in two metropolitan areas (Mexico City and Monterrey) and on anthropological studies of rural communities and urban migrants. These studies are rich in detail and permit one to gain considerable insight into the motivations underlying migration. They also shed light on the process of migration and the integration of workers into the labor market at their destination point. Most of the other empirical studies adopt an approach that is based explicitly on a human capital framework or have collected information that can be analyzed and interpreted within that structure. The econometric studies typically use multiple regression analysis. This method relates variations in the rates of migration among states to a number of independent variables that might be expected to influence the costs and prospective gains of migration. Among the variables are distance between origin and destination points, income or wage levels at each point, unemployment rates, size of total or urban population, and a variety of other ecological or social variables.6 By and large, the results conform to expectations generated by human capital theory.7 Distance proves to be highly significant and is inversely related to migration rates. Longer distances may have a deterrent effect, not only because of the higher costs associated with a move, but also because information may be less readily available. Destinations characterized by high wages or average incomes have a more powerful drawing effect than do poorer regions. Clearly, the probability of improving one's economic position is greater if the destination is a high-wage rather than a low-wage area. Unemployment rates tend to be lower in the destination and higher in the points of origin; this finding suggests that there is movement from areas with more restricted employment opportunities to areas with less. The only unexpected result is the positive sign of the income or wage variable at the point of origin. Human capital theory would lead one to expect a negative sign, that is, that lower incomes would be associated with higher rates of emigration. However, several other empirical studies have reported a similar direct relationship. The rationalization that has been offered for it holds that, in developing countries, the poorest groups are too poor to migrate-they are unable to muster the resources to defray the costs of migration. In addition, the poorest areas are likely to be the most deficient in human capital, especially education, so that prospects of improvement in employment and income may be perceived as being very dim. In short, these three variables emerge as the most significant factors in explaining the variance in migration rates among alternative pairs of points. 8  RE EARCH FINDINGS 153 As indicated above, the sample surveys studied the migrant populations of Mexico City and Monterrey. These cites would appear to be natural choices for students of migration since both have been among the principal foci of migrants. Mexico City has long held a dominant position in the country, not only because it serves as the seat of the national government, but also because of its role as the major financial and service center for the rest of the country. In addition, the rapid inclustrial development of the metropolitan area over the past four decades has provided expanded employment opportunities, not only in the industrial sector, but also in the others that serve it. An appreciation of the impact that migration has had on the population of the metropolitan area can be gained from the 1970 census datum of the proportion of non-natives in the total: 33.4 percent, equivalent to 2.8 million people. Monterrey has emerged as a prominent metropolitan center only in more recent times. In 1930, the census reported a population of only 134,202. Within twenty years, the population of Monterrey had virtually tripled, and by 1970 had reached 1.1 million, Its growth has been char- acterized by the rise of a modern heavy industrial sector that currently represents the largest source of employment; it provides jobs for 37 percent of the employed labor force.9 The fact that the rapid population growth of Monterrey is more recent explains the large relative importance of migrants in the population. In 1965, only 29 percent of the population was native born. An additional 14 percent had been born there, but had emigrated at some time in the past and subsequently returned. If the latter are not counted among the migrants, the non-native population still accounted for 57 percent of the total. Because men are believed to have a greater independent voice than women in the decision to migrate, most studies of migration limit their samples to male migrants. The Mexican studies are no exception. All of them concern themselves primarily with male migrants. The characteriza- tion of the migration that follows should be understood to apply only to men unless otherwise explicitly noted."0 This preoccupation with males would be understandable if they dominated td e migrant flow. In the case of Mexico City, however, female migration has exceeded male by a large, though secularly declining, margin. During the 1930s, female migrants outnumbered male by 40 percent and by 20 percent in the 1950s. " During the 1960s, the margin declined further to 17.2 percent.'2 Characteristics of Migrants The sample surveys and other empirical studies confirm the importance of distance as a factor that channels migration flows. Two-thirds to 154 INTERNAL MIGRATION three-fourths of the migrant flow originated in neighboring states. In the case of migration to Monterrey, this proportion includes migrants origi- 3 Between80 and 90 percent nating within the state of Nuevo Le6n itself." of the migrants originated within a radius of 600 kilometers of the metro- politan areas.' 4 Rural areas have long been a sourceof population that has contributed to urban growth. It is only more recently, however, that migrants with rural origins have constituted the majority in the flowto Mexico City. For example,of those arrivingprior to 1955,only 40 percent had originated in communities with populations under 5,000 whereas the cohort arriving between 1965 and 1970 was 53 percent rural." 5 The rural component of the flow to Monterrey has been marginally greater than that to Mexico City, and it, too, has shown a slight increase over time.' 6 Most of the migrants to Monterrey with rural origins, almost 60 percent, movedto the city in a single step with no intermediate moves; the balance was almost evenlydividedbetween those who had left their place of birth less than and more than ten years prior to arriving in Monterrey." 7 One interesting observation made by Contreras is the inverse rela- tionship between the distance traveled to Mexico City and the relative importance of rural residents in the flow." Migrants from neighboring areas within 400 kilometers were 70 percent rural, and the percentage rural declines to 43 and 28 as the distance increases to between 400 and 600 kilometers and to more than 600 kilometers, respectively. (In this case,rural communitiesare definedas those with populations of less than 10,000.) One may hypothesize that the greater prevalence of urbanites among the long-distancemigrants reflectstheir higher levelsof education and the greater access they had to information that would have reduced the risks of migration. Given the large income differencesbetween urban and rural familiesthat exist in Mexico, it is also likely that the interurban migrants were better able than rural residentsto financethe longer moves. Only Anson reports that a significantproportion-41 percent-of the migrants had considered alternative urban destinations (to Monterrey), while Balan, Browning,and Jelinfound that only 11 percent did so." Yet it cannot be said that moves are made blindly, without considerable in- formation. Migrants to Monterrey were found to have been influencedby a prior acquaintance with the city. Nearly two-thirds had visited the city for some purpose prior to moving there, and this proportion shows a steady increase among the more recent cohorts of migrants. Furthermore, 84 percent of the migrants had friends or relatives already living in Monterrey before their move. Adult migrants to Mexico City reported the presence of friends and relatives almost as frequently as did their Monter- rey counterparts, and over half had made prior visitsto the city. 20 As might  RESEARCHFINDINGS 155 be expected,the significance of relativesand friends already at the destina- tion point was not limited to their value as a sourceof information. They also servedas an important sourceof assistanc. upon arrival. Over half of the adult migrants receivedfree room and board. This form of assistance probably far outweighed any help received from families at the point of origin. Most migrants receivedno aid at all from the latter. What aid was 21 received was limited to some cash or a one-way bus ticket. It has generallybeen held that migrationis a selective process; migrants are not chosen at random from the population at the point of origin. Indeed,one would expect that the migrants would be drawn from among those who are best equipped to succeedand who have the greatest prob- ability of recouping their investment in migrating. Thus, migrants would be expected to be young and to be drawn frora among those with higher levels of education. Mexican migrants conform to this characterization. Actually, a very substantial number of urban non-natives, about one- third, leave their place of origin prior to reacting age 16, presumably as part of a family move in most cases.The bulk of the remainder, who are more likelyto move independentlyof their families,also fall into the ranks of the young. Only 14 percent of the migrants to Mexico Citywere over 30 years of age at the time of arrival. In Monter-ey, fully two-thirds of the migrants were no older than 21.22 Rural migrarLtstend to be older than the interurban, a factor that may pose an obstacle to prospects for upward occupational mobility. It is more difficult to generalize with complete confidenceabout the educational levelsof migrants because of differencesin the way in which data are presented. The data for Mexico City suggest that, among those whose education can be presumed to have been completedprior to their move, the number of years of schooling completed has been rising over time. 23 Whilethe educational attainments of the migrants would appear to compare favorably with those of the general population of Mexico (3.5 years of primary schooling on average in 1975), they remain well below the average level of the native population of Mexico City. For example, male migrants aged21 to 40 who arrived in the Mexico City metropolitan area between 1965 and 1970 reported an averagelevelof schoolingof five completedyears. Since,in 1970, only 16.8 percent of the population over 15 years old had completedsix years of primary education, the migrants were, on the average,clearly better educated than the population at large. In comparison with the average educational levels of native males, however, the migrants' achievementswere more modest. Native males aged 21-30 had completed eight years while those 31-40 had completed 7.6 years. Furthermore, it would appear that the difference between the educational levels of migrants and natives has been increasingover time. 156 INTERNAL MIGRATION The oldest group of migrants, who arrived prior to 1935, reported only 1.3 years of schooling less than the natives. This increased to a two-year difference for young adult males arriving between 1935 and 1955 and further to three years for the most recent young arrivals included in the study. Of course, men over 40 among the recent arrivals to the area reported much lower levels of education than did the young arrivals. The gap between their level of schooling and that of their native peer groups was even greater than that noted for the younger migrants. But the older recent migrants were relatively few in number, and it may be presumed that at least some of them did not migrate for economic reasons. A more adequate notion of the selectivityof migrants requires a com- parison of the educational levelsof migrants with those of the population at the point of origin rather than at the destination. The availableevidence seemsto confirmthe expectation that the migrants are those with charac- teristics most likely to achieve success.The anthropologicalstudies of the village of Tzintzuntzan in Michoacan found that the migrants to urban centers came from the minority of households with above-averagestan- dards of living, education, and occupational skills. Half of the men who had achieved a sixth grade education in the interval 1942-79 had 24 migrated. A more formal comparison of educational attainments of migrants and nonmigrants is provided by Balan, Browning, and Jelin for migrants to Monterrey.' The basis of comparisonis the proportion of men among the migrants with six or more years of schooling and the proportion in the zones in whichthey originated. The zones are classified into fivecategories according to the level of socioeconomicdevelopment. The proportion of migrants who completed primary school is significantlygreater than that in the population in their zones of origin. However, the difference has narrowed over time. For example, among those migrants who arrived before 1941, the proportion with completed primary education was ten times greater than that in the zone they had left for the two lowest of the five socioeconomic categories. The proportion was six times greater in another category and a little more than twice as great in a fourth. Among those arriving in the decade of the 1950s, the ratios had narrowed signifi- cantly from a high of three for one zone and to two or less for the remaining four zones.This suggeststhat the degree of selectivityhas been declining even as the average level of schooling of migrants has been increasing. Two reasons may be offered for this apparent decline in selectivity. First, there has been a significant narrowing of differencesin average educational attainment among regions. The progressive extension of primary educational opportunities to the less urbanized areas of the  RESEARCH FINDINGS 157 country has had a larger relative impacton educational achievementthere than it has had in the largest urban centers. These centers have enjoyed greater accessto education, particularly through the primary level, for a longer period of time. The increasingly mass nature of migration has tended to dilute the distinctiveness of the migrant. In particular, the increase in the number of migrants with rural origins, whose educational attainments are more modest than those of urban natives, would tend to lower the average educational level of migrants relative to the region of origin." It should not be surprising to find an increasingflow of rural migrants to the larger urban areas. Findings by Kemper and others suggest that this increase was a consequence, in part at least, of the termination of the bracero program. Many of the migrantsfrom Tzintzuntzan and the Maza- huan region, for example, were men who, as surplus labor in the village, had been siphoned off as temporary migrants 1othe UnitedStates.Follow- ing the end of the program in 1964, the rate of migration to Mexico City increased sharply. These moves, however, proved to be different-neither temporary nor seasonal. Only 6.6 percent of these emigrants returned to 27 the village. Other fundamental changes have contributed to this increase. Of the greatest importance has beenthe extensionof educationalopportunities to rural areas; this has enhanced the competitivenessof the rural population in urban labor markets. Furthermore, education, once acquired, appears to make little differencein the income levels attainable in rural areas. The economicreturns to education are significantlygreater in urban employ- ments. In 1968, an additional year of schoolingwas associated with an increase in monthly earnings among male nonagricultural workers that was 3.2 times greater than that accruing to workers in agriculture, Mex$283 to Mex$89. By 1977, the relativedifferencein the returns from an additional year of schoolinghad increasedto 4.5, the absolute returns being Mex$854 and Mex$189 in nonagriculture and agriculture, 28 Given these very substantial differences in returns from respectively. education, the incentivesto migrate from rural to urban areas must appear very powerful to those with some schooling. Indeed,the most important reasons given for migrating to the metropolitan areas were economic; 60 to 65 percent of the respondents offered employment or income-related reasons for moving. Of course, these reasons for moving could also be interpreted in a somewhat differentlight. Moves could be male, not so much in response to the attractiveness of opportunities at the destination point ("pull" factors), as to the absolutely desperate economicconditionsprevailing in the place of origin.("push" factors). However. the push factors would not 158 INTERNAL MIGRATION seem to be present in an extreme form among the sampledmigrants. Most workers were employed before migrating and expressed a high degree of job satisfaction in their last job. In the Mexico City sample, only 4.2 percent of the migrants had been unemployed and activelysearching for work before migrating. Another 0.5 percent were classedas discouraged unemployed; about 13 percent reported that their jobs had recently terminated. 2 9 The proportion of men in the Monterrey studies whosework had ended apparently was higher, but the significanceof this is not clear. The period between the end of employmentat the point of origin and the departure for Monterrey was reported by Anson to be very short, 10.3 days on average. Migrants to Mexico City also reported only a short interruption between the end of employment and departure. Fully 43 percent left the very next day, another 31 percent left within a month; 5 percent were being transferred by their employersand did not experience any interruption in their work. Nevertheless, one should not discount completely the significanceof poverty and restricted employment opportunities in the rural areas of origin as a compellingforce for migration. The declinein mortality rates resulted in an increasing rate of population growth in areas with limited land resources. Although many of the plots assignedto landless laborers during the land reform movements of the 1930s and 1940s had been adequate to support the rural population at an improved standard of living, pressures on the land began to become apparent during the 1950s. By 1960, the ratio of cultivated land to labor in some of the more congested and poorer regions of Mexico, including the central and south- ern zones, appears to have fallen to belowtwo hectares per member of the agricultural labor force. With shrinkingland resources availablefor redis- tribution, the diminishingprospects of acquiringone's own holding might be expected to induce the more adventuresome among the rural popula- tion to seek their fortunes elsewhere. The rural origins and modest educational attainment of most migrants is reflectedin their occupational experiencebeforemigration. Agricultural employment was reported as the last premigration job for from 40 to 55 percent of the migrants; this proportion shows a secular increase among successivecohorts. Approximately one-fourth of the migrants had held low-skilledjobs in construction or artisan or factory production. In short, from two-thirds to three-fourths of the migrants were unskilled manual workers prior to migrating. The balance were employedin commerceand services;some of these may also have been in low-skilledjobs. Those with prior experience in skilled, technical, or professionalemploymentappear to have numbered no more than 15 percent of the migrants. 30 Only limited information is available about the employment status of  RE5EARCH FINDINGS 159 migrants prior to migrating, and that refers largely to the status of those whose jobs had been in agriculture. By and large, it is the wage laborer and the unpaid family worker who migrate, which, in view of the youth of the migrants, is not surprising. Ejidatarios or aparceros (share croppers) appear as a small minority in the flow, on the order of 11 percent in the Mexico City sample.3 ' The degree of stability in employment was reported to be low for Anson's Monterrey sample; only 31 percent held an empleo 32 de planta (permanent job). In contrast, the survey of migrants to Mexico City found that three-fifths of the respondens held permanent remuner- 33 ated jobs before migration. Integration in Urban Labor Market The process of integration into the metropolitan labor market appears to have been a relatively painless one. Although only a small proportion, approximately one-quarter, either were certain of a job or had the promise of a job prior to migration, the delay in finding employment was short. 34 Over 60 percent found a job within two week; of arrival in Mexico City. Anson's migrants averaged 11.5 days to find their first job in Monterrey.3" The importance of friends and relatives in the destination point becomes apparent again, for between one-half and three-quarters of the new arriv- als found their first jobs with their help. Most migrants arrive with few definite job preferences. In view of the prevalence of agricultural workers with nontransferable skills, this should not be surprising. Seventeen per- cent of the migrants to Mexico City expressed a preference for employ- ment in manufacturing, construction, or unskilled service occupations; another 15 percent wished to continue exercising their artisan skills. Only 8 percent aspired to nonprofessional white-collar occupations.3 6 Migrants' first jobs are heavily concentrated in the unskilled manual categories in manufacturing and construction and, to a lesser extent, in services. However, an important distinction can be drawn between those who had and had not held jobs before migrating. Those who took their first job ever in Mexico City reported an occupational structure that was more favorable than that of natives in their first job. Particularly notable were the larger proportion entering nonmanu al jobs, including the highly skilled occupations, and the lower proportion entering unskilled manual work. Conversely, those who reported prior work experience fared sig- 37 nificantly worse than natives in their first employment in the city. The work histories of the migrants offer litt le support for the commonly expressed view that migration is the cause of che "tertiarization" of cities. The proportion of the migrants entering the service sector in low-skilled 160 INTERNAL MIGRATION jobs shows a progressive decline among the cohorts arriving in each decade since 1930. Of the migrants arriving during 1930-39, almost half found their first employment in the service sector. By the decade of the 1960s, this proportion had fallen to 27 percent. 3 8 In fact, nonmigrant unskilled entrants to the labor force were to be found in servicesin greater proportions than were migrants during that decade, and a growing pro- portion of migrants flowedinto the industrial sector. During the 1960s, 56 percent of the unskilled entrants found jobs there, a larger proportion than that among nonmigrants. Nor should it be assumedthat the manufactur- ing activity in which the migrants became employed was of an artisan or "informal" nature; fully70 percent of those enteringmanufacturingdid so in the capital goods sector. Construction activityhas also shown a gradual declinein importance as a port of entry. Whereas about one-fourth of the unskilled male migrants entered that sector during the 1930-49 interval, only 17 percent did so in the more recent decade.The fact that none of the migrants to Mexico City reported employmentas itinerant vendorsor in a self-employedstatus for their first jobs is significant.39 Butterworth also found it "noteworthy that none of the migrants [to Mexico City of Mixtec origins from Oaxaca] is an ambulatory salesman, 'peddler' or vendor of any kind." 40 Migration to the metropolitan areas resulted in an increasein employ- ment stability. Anson found that, on the one hand, 36 percent held a permanent job, or empleo de planta, in their first year, but, by the end of the interval of his study, 52 percent enjoyedthat status. On the other hand, he found that in each year between 29 and 51 percent of his sample had sufferedat least four weeks of unemployment.Whether that represents an improvement or deterioration as compared with the premigrationemploy- ment cannot be determinedin the absenceof data about the latter.4'To the extent that premigration jobs were in agriculture, some seasonal unem- ployment might have been expected. The authors of the Mexican City study define employment instability in terms of the frequency of job changes rather than unemployment.In that case,rather than representing a negative quality of employment,changesof jobs proved to be a medium of occupational mobility and improvement in job status. 42 Migrants to Mexico City were more heavily concentrated at the lower end of the earnings distribution than were native male members of the labor force. The proportion of male migrants aged 21 to 60 earning less than the legal minimum wage was 28.1 percent, as compared with 15.8 percent for the natives.However, the averageperformanceof the migrants was heavily weighted by the number of rural migrants with agricultural work histories, who tended to fare lesswell; 32.6 percent of these workers earned less than the minimum, in part becausethey are less educated and  RESEARCH FINDINGS 161 older as a group upon arrival in Mexico City. Migrants whose first employment was in a nonagricultural sector fared no worse than natives with respect to earnings below the legal minimums. Earnings were also a function of the length of residence in Mexico City. The most recent arrivals, workers with fewer than ten years of residence, were most likely to be earning less than the minimum; 38.6 percent did so. Among those with ten to nineteen years of residence, 28.2 percent earned below minimum wages; 26.4 percent of those with more than 20 years in the city did so.43Since Mexico City had among the highest legal minimum wages in the country, exceeded only by some of the zones bordering the United States, this finding does not preclude the possibility that unskilled migrant workers in the metropolitan area realized earnings greater than those of the vast majority of unskilled workers in the rest of the country. The anthropologists' studies of migrants also concur that migration enhanced their economic welfare. Kemper found most Tzintzuntzenios earning the minimum in Mexico City in 1969-1970, or the equivalent of US$2.54 per day. By contrast, few Tzintzuritzan villagers earned more than a dollar a day in the traditional occupations in farming, pottery making, and shopkeeping." Butterworth's Mixtecan community in Mex- ico City reported earnings ranging from Mex$400 to Mex$1,000 a month in 1960, which averaged Mex$550, or about 33 percent more than the minimum."5 Nine years later, in 1969, only 7 percent of the workers employed in agriculture in the state of Oaxaca earned as much as Mex$500 a month. In manufacturing jobs, 30 percent did so, as did 46 minorities of employees in construction, commerce, and services. Arizpe reported a growing differential rate of earnings between those migrants from the Mazahua region in the western parr of the state of Mexico who had become established in Mexico City and agricultural laborers in the Mazahuan villages. The weekly wage of an albanil in construction in the capital averaged Mex$90 in 1955, had doubled by 1965, and had reached Mex$224 by 1971. By contrast, the daily agriculture wage rose from Mex$42 a week in 1955 to Mex$60 in 1965, and to Mex$90 in 1971. The inflation of the mid-1970s apparently reduced the relative advantages of urban employment, however, and Arizpe reports a trickle of return mi- grants in the face of disillusionment with the high cost of living in Mexico 47 City. In spite of this reservation, all of the studies conclude that, in most cases, migration to the metropolitan areas resultecd in an immediate improve- 4 ment in the employment status of migrants. " However, the full implica- tion of migration for the employment position of workers depends not only on the immediate gains but also on the opportunities for subsequent improvement. In this respect, the evidence is again encouraging. Balan, 162 INTERNAL MIGRATION Browning, and Jelin analyzed the experience of men entering the Monter- rey labor market in an unskilled category and found that, within a ten-year interval, 54 percent had moved out of the unskilled category. Even the remaining 46 percent showed a high frequency of occupational change within the unskilled ranks. Furthermore, the younger groups showed a higher incidence of upward mobility than did the older, 59 to 48 percent, respectively."9 In Mexico City, migrants entering the city's labor force in an unskilled manual job reported a high frequency of upward movement, 25 percent a one-step move and 44 percent a two-step move within a seven-step occu- pational structure. This experience compared favorably with that of na- tives whose first employment was in an unskilled job. Their proportions of one- and two-step movements were 25 and 50 percent, respectively. Furthermore, migrants entering skilled manual or nonmanual occupations reported higher frequencies of upward mobility than did the natives." Among manual workers, industrial employment offered a higher prob- ability of advancement than did any other. This was probably attributable to the practice in modern industry of advancing workers through an occupational hierarchy according to seniority. Workers with premigration jobs in agriculture tended to experience somewhat lower rates of upward mobility than did others. This may reflect the fact the the former tended to migrate at a later age and to have less education than those with nonagri- cultural backgrounds. Although workers with agricultural backgrounds may thus appear to be at a disadvantage relative to those with other work experience, compared with those who remain in agriculture, where the opportunities for upward mobility are extremely limited, they have done well. All of the surveyed populations view themselves as having made sig- nificant improvements in their status and living standards as a result of migration. Overwhelmingly, they pronounced themselves satisfied with the economic results of their move. Anson is the only researcher to have gathered detailed information about earnings before and after migration, and these show very substantial improvements."1 Whereas 43 percent of the migrants earned less than Mex$10,400 in the year prior to migration, only 5 percent did so in the first year following the move. (The income figures are all in terms of constant 1975 Mexican pesos.) At the other end of the earnings structure, 31 percent earned more than Mex$20,800 before migrating, while 51 percent exceeded that amount in their first year in Monterrey. By the end of the five-year interval under study, fully 80 percent reported earnings in excess of Mex$20,800. On the average, family incomes for the first and last postmigration years exceeded the last premigration income by 30 and 64 percent, respectively.  RESEARCH FINDINGS 163 Anson then proceeds further to calculate the returns to migration after adjusting for differencesin the cost of livingbetween Monterrey and the migrants' places of origin and for changes in their opportunity cost. 52 In the first year, a median internal rate of return nf 13.5 percent was realized. On the assumption that income increased at a constant rate over the interval studied, the rate of return increases to 37.8 percent by the year prior to the study.5 3 To be sure, some respondents, about 10 percent of the sample, reported either smallpositiveor negativereturns in their first year in Monterrey. In virtually all cases,however, these were migrants who had previously enjoyed "high" incomes, so that they were not denied the essentialsof livingby their move. Compensatingfor this lack of immediate improvement was the more stable or permanent character of their employ- ment. Furthermore, by the end of the period, half of those who initially had experiencednegativereturns had begunto realizepositivenet returns. A final point of interest is the very low ccst of migration reported by Anson's sample. The total cost for rural migrants amounted to Mex$471 (US$38),and Mex$899 (US$72)for interurban migrants. This difference in cost can largely be ascribed to the differencein forgone earnings during the move-the cash outlays differed by only Mex$76. The low cost of migration reflects the low price of bus transportation in Mexico and the important subsidy in the form of free room and board which most mi- 4 grants receive upon arrival in the city.5 The limited information available about female migration indicates many similaritiesto that of males.5 5 The age structure is very similar as is the proportion originating in rural areas. As among men, there is an inverse relationship between distance and the number of migrants who come from states outside the Federal District. The greater the distance moved, however, the greater is the incidenceof urban antecedentsamong the migrants. Women are much less likelythan men, 24 versus57 percent, according to one sample, to have had any work experience prior to migration; they also have had fewer years of ,chooling than men. Among those with prior work experience, the incidence of nonmanual employ- ment appears to be higher than among male migrants or amongthe female labor force as a whole. But this group does not offset the lack of work experience and low levelsof formal education of the bulk of the migrants. Female migrants in the labor force are heavily concentratedin low-skilled jobs, particularlyin services.One of the most common ports of entry to the urban labor market is domestic service. Given the ready availability of work and the assuranceof room and board, it reduces the risks and costs of migration considerably. The researchersof El Colegiode Mexico provide some earningsdata for femalemigrants that are differentiatedby length of time sincemigration to 164 INTERNAL MIGRATION Mexico City. By and large, women's earnings were lower than those of men at all points in time, but especially in the first employment in Mexico City. Whereas 64 percent of recent female migrants earned less than the legal minimum wage in their first employment, only 27 percent of males did so. Earnings of both sexes improved with longer residence, however. Indeed, females exhibited a relatively higher incidence of improvement if earnings above the minimum wage were taken as the criterion. Among long-established migrants, the proportion still earning less than the mini- mum wage fell by 46 percent to 34.7 percent of the women, and by 38.4 percent to 16.7 percent among men. Migrant females did relatively worse than men when compared with natives of their own sex. Whereas the proportion of long-time resident male migrants receiving less than the legal minimum wage was only slightly greater than that among natives, the proportion among migrant women was almost twice that of natives. Thus, in spite of the considerable absolute improvement in the economic status of migrant women, they appear to be more disadvantaged than native women. Because these data do not control for differences in age or educational levels, however, they may reflect very different demographic characteristics of the two popula- 56 tions, rather than simply their status as migrants or natives. Migration and Regional Income Differentials To the extent that migration flows respond to differences in income among regions, it is to be expected that regional disparities in incomes would tend to decline over time. As the population moves to regions with high average incomes, the rate of growth in earnings might be expected to decline, while the exodus from low-income regions might be expected to raise the rate of increase there. Typically, however, the process of narrow- ing may be expected to be spread over several decades. Indeed, early in periods of accelerating growth, regional income differentials are more likely to increase; sustained narrowing occurs only after the development process has begun to approach maturity. From 1900 to 1940, the metropolitan region comprising Mexico City and the surrounding area of the state of Mexico led all others in the rate of increase in both population and per capita income. Two other high- income areas, the Northwest and Gulf regions, experienced higher rates of growth in both per capita income and population than did the remaining regions. Conversely, the poorer regions suffered slower rates of growth in both income and population.5 7 Income disparities among the regions thus increased during the early part of the century. Between 1940 and 1970, the  SUMMARY 165 income position of the capital and the state of Mexico declinedrelative to all but one region, but the high-income border regions in the North and Northwest gained relative to all the others. The course of incomedispersion can best be summarizedby referenceto the coefficient of variation (V), taking Unikel's measures of per capita gross domesticproduct of the individual states as observations.In 1900, V had a value of 52.5 percent. A sharp increasein dispersionoccurred in the interval to 1940; V in that year was 80.5 percent. Over the next decade, V declined to 66.6 percent and remained at that approximate levelthrough 1970. Thus. while some decline in dispersion seems to have occurred between 1940 and 1950, resource mobility was not sufficient to bring about a continued narrowing of regional income differentials.Indeed, the poorer states in the southern and central regions continued to lose ground relativeto the averagefor the country as a whole through 1970. During the ensuing decade, however, returns from the 1980 census suggest that a reversal in this wideningtendency may have set in. (Seetable 4-7 and the accompanying discussionfor evidenceof such a reversal.) Whereas Un- ikel's data are derived from the national accounts, the population census reports monthly incomes earned by labor force members. These income data yield measuresof dispersionthat are cons;iderably smallerthan those derived from the national accounts. In 1970, V stood at 49.4 percent. By 1980, it had declined to 27.1 percent. In view of the reservations I have expressed regarding the reliability of the 1980 census, it would seem prudent to await the emergence of additional evidencebefore accepting the finding of narrowing income differentialsat face value. Summary The evidencereviewedhere offerslittle support for the view that migra- tion has been "excessive" or immiserizing.Ir terms of private costs and benefits, it would seem that migration has proceeded in directions pre- dicted by theory and has resulted in substantial net gains to the migrants. processwherebyirdividuals who already have Migration is a self-selective the best chances of succeedingmove to a more challengingenvironment. In the Mexican context of steady and rapid economic growth, urban employment opportunities have emergedat wage levelsthat are comfort- ably above rural alternatives and have survived in spite of the large migratory flow. The expansivenature of urbarnlabor marketsis illustrated by the promptness with which most migrants were absorbed on advan- tageous terms. I attach considerablesignificanceto the persistenceof large urban-rural differentialsand to the favorablecharacter of the employment 166 INTERNAL MIGRATION conditions of migrants prior to migration. They suggest strongly that migration representeda response to perceivedopportunities for economic improvement, rather than simply an act of desperation to escape impossi- ble economic conditions at home. Nor are economicgains limited to the short run following migration. Apparently, upward mobility was com- monly realized among migrants. Within a few years, their socioeconomic position (at least, for males) was indistinguishablefrom that of urban natives with similar demographic characteristics.Regionaldifferentialsin income may be expectedto narrow as a result of migration, but only after a lag. In Mexico,prior to 1970, there was little such change,but preliminary indications suggestthat a significantnarrowing may have finallyoccurred during the past decade. Notes 1. For example, during the decade of the 1940s there was an almost fourfold increase in the number of kilometers of paved road from 768 in 1939 to 2,623 in 1950. The number of radios increased tenfold between 1940 and 1960, from 325,000 to 3,200,000. Rising literacy levels also contributed to improved information flows. 2. Throughout this book, references to Mexico's states should be understood to include the Federal District, which is not a state, and Baja California Sur, which until recently had the status of a territory. 3. The proportion of non-natives in the population of the metropolitan Mexico City area in 1970 was approximately the same, 32 percent. 4. Gustavo Cabrera Acevedo, "Poblaci6n, migraci6n y fuerza de trabajo," in Mercados regionales de trabajo (Mexico, D.F.: Instituto Nacional de Estudios del Trabajo, 1976), pp. 283-84. Unikel estimates the flow to have included 4.5 million people between 1950 and 1970, but he defines urban centers as those with 15,000 or more population (Luis Unikel, El desarrollo urbano de Mexico, 2d ed. [Mexico, D.F.: El Colegio de Mexico, 1978], p. 213). 5. Unikel, El desarrollo urbano, appendix table VI-A19. 6. Michael J. Greenwood and Jerry R. Ladman, "An Economic Analysis of Migration in Mexico," Annals of Regional Science, vol. 12, no. 2 (1978), pp. 16-3 1; Jonathan King, "Interstate Migration in Mexico," Economic Development and Cultural Change, vol. 27, no. 1 (1978), pp. 83-102; Joseph P. Stoltman and John M. Ball, "Migration and the Local Economic Factor in Rural Mexico," Human Organization, vol. 30, no. 1 (1971); Luis Unikel, C. Ruiz Chiapetto, and 0. Lazcano, "Factores de rechazo en la migraci6n rural en Mexico, 1950-1960," Demografia y economia, vol. 7, no. 1 (1973), pp. 24-57; Unikel, El desarrollo urbano. 7. Of the studies cited, that of Greenwood and Ladman offers the clearest results, and their regressions have the greatest explanatory power of all. The others suffer weaknesses that impair their effectiveness. King includes a large number of variables, many of which are closely interrelated; as a result, his regressions are plagued with problems of multicollinear- ity, which pose serious problems for the interpretation of his findings. Unikel's approach appears to have several shortcomings. Some of his equations are misspecified; for example, he includes measures of both inputs and outputs of a production function as independent variables in the same equation. In addition, some of his variables might be expected to be  NOTES 167 interrelated; there is thus the possibility of multicollinearity.He groups states of origin by referenceto a particular characteristicand runs separate regressions for each group, a method that reduces the variancewithin the sampleof the indeptndent variablespresumablyassoci- ated with migration rates. Finally,in his analysisof destinations,his groupings result in a smallnumber of observationsrelativeto the number of independentvariables,thus reducing the degrees of freedom. 8. In the Greenwood and Ladman study ("An EconcmicAnalysisof Migration, p. 21), these three variables, plus population size, accounted for 70 percent of the variance. 9. Secretarfade Programaci6ny Presupuesto,Encuest2continuasobreocupaci6n(ECSo), vol. 7, 1st quarter (1979), p. 238. 10. An exception is represented by Enrique Contreras S., Estratificaciony movilidad socialen la ciudad de Mexico (Mexico,D.F.: UniversidadNacional Autonoma de Mexico, 1978), p. 89. His sampleof migrantsconsistedof heads ot families;as a resultit includedonly a small number of female migrants,which can hardly be consideredrepresentative. 11. Gustavo Cabrera A., "Poblaci6n, migraci6ny fuerza de trabajo." 12. Secretaria de Industria y Comercio, Direcci6n Seneral de Estadistica, IX Censo general de poblaci6n, 1970, Resumen general (Mexico,D.F., 1972), table 14. 13. Ricardo Anson,"La relaci6nentre la politica de migraci6ninterna y la realidad de las consecuenciaspersonalesde la migraci6ninterna: el caso deMonterrey" (Monterrey,Nuevo Leon: Instituto Tecnologico y de Estudios Superiores de Monterrey, n.d.; processed), p. 14; Jorge Balan, Harley L. Browning,and ElizabethJelin, Men in a DevelopingSociety (Austin:University of TexasPress,1973), p.64; Mercede& PedreroN., "Corrientesmigrator- ias internas de Mexico (1950-60)," in Conferenciaregionallatinoamericana de poblacidn, Actas I (Mexico,D.F.: El Colegiode Mexico,1972),pp. .547-52; Contreras,Estratificaci6n y movilidad, p. 68. 14. Pedrero reports ("Corrientes migratorias internas," pp. 547-52) that 81 percent of the migrants to the Northwest region, another area of dynamic growth in recent years, originated in neighboring zones. 15. Humberto Muiioz, Orlandina de Oliveira,and Claudio Stern,Migracionesinternasa la ciudad de Mexico y su impacto, Temasde la Ciudad no. 8 (Mexico,D.F.: Delegaci6ndel D.D.F. en Venustiano Carranza, 1978), p. 35. 16. Balan, Browning,and Jelin, Men in a Developing Society,p. 64. 17. One-step migrationto Mexico Citywas found to b typicalof more recentmigrantsby other studiesas well.Whereassomeof theearliermigrantswould first moveto regionalurban centers, later migrants overwhelminglyleft for Mexico City directly. George M. Foster, Tzintzuntzan: MexicanPeasantsin a ChangingWorld, rev.ed. (New York:Elsevier,1979), p. 273; Douglas S. Butterworth, "A Study of the Urbanization Process among Mixtec Migrants from Tilatongo in Mexico City," in Peasants in Cities, ed. William Manglin (Boston,Mass.: Houghton-Mifflin,1970), p. 101. 18. Contreras, Estratificaciony movilidad,pp. 70-7i. 19. Anson,"La relaci6nentre la politicade migraci6n,"p.14; Balan,Browning,andJelin, Men in a Developing Society, p. 154. 20. This informationabout the migrants to Mexico City was extracted from unpublished tables (hereafterreferred to as Tabs.), to which I was kindly givenaccessby Orlandina de Oliveira at the Colegio de Mexico. The presence of relat ves and friends in Mexico City is factor guidingmigrationby LarissaA,Lomitz,Networks and Marginal- cited as a significant ity: Life in a Mexican Shantytown (New York: Academ.cPress, 1977). 21. Balan, Browning, and Jelin, Men in a Developing Society, p. 160; Tabs. 22. Ibid., p. 150; Contreras, Estratificaci6ny movilidad, pp. 82-83. Cabrera reports ("Poblaci6n,migraci6ny fuerza de trabajo") that migrantsin Mexico generallyare young.In 168 INTERNAL MIGRATION each of three intercensaldecades since 1930, about half of the male migrants were between the agesof 10 and 24 years and an additional 14percent between25 and 29. The proportion of women in the 10-24 age bracket consistently exceededthat of men although by a narrow margin. Among both sexes, the proportion arriving in the youngest age groups showed a small upward trend. 23. Mufioz, de Oliveira, and Stern, Migracionesinternas,p. 47. 24. Foster, Tzintzuntzan, p. 273; Robert V. Kemper,"Tzintzuntzenos in Mexico City: The AnthropologistAmong RecentMigrants," in Anthropologists in Cities, ed. GeorgeM. Foster and Robert V. Kemper (Boston,Mass.: Little,Brown, 1974), p. 83. 25. Balan, Browning,and Jelin, Men in a DevelopingSociety, pp. 144-47. 26. The difference in educational attainments between urban and rural migrants is suggestedby Anson's finding("La relaci6nentrela politica demigracion," p. 14)that, among his sample of adult ma]emigrants arriving in Monterreybetween1969 and 1974,the former had completed seven and the latter three years of school. 27. Kemper, "Tzintzuntzenios in Mexico City," p. 83; Lourdes Arizpe,Migraci6n,etni- cismo y cambio economico (Mexico,D.F.: El Colegio de Mexico, 1978), pp. 78-79. 28. These returnswere estimatedfrom a multipleregressionequationof monthlyearnings on education, age, occupational status (that is, employer,employee,self-employed,and so on), and size of communityof residence.The data were drawn from the 1968 and 1977 household surveys of incomesand expendituresand were processed for this project by the Bancode Mexico,S.A.The absenceof largedifferences in incomeassociatedwithdifferences in educational levelsamong men employedin agriculturewas also demonstratedby unpub- lisheddata drawn from the 1970 population censusby theGrupo de Estudio delProblemadel Empleo. 29. Tabs. 30. Balan, Browning,and Jelin, Men in a Developing Society, pp. 154-55; Mufioz,de Oliveira,and Stern,Migracionesinternas,pp. 43-45; Anson,"La relaci6nentre la politicade migraci6n interna," p. 14. 31. Tabs. 32. Anson, "La relaci6n entre la politica de migraci6n interna," p. 21. 33. Tabs. 34. Ibid. 35. Anson, "La relaci6n entre la politica de migraci6n interna," p. 12. 36. Tabs. 37. Humberto Munioz and Orlandina de Oliveira,"Migraci6ny movilidadocupacional," in Migraci6ny desigualdad socialen la ciudad de Mexico,ed. Humberto Munioz,Orlandina de Oliveira, and Claudio Stern (Mexico,D.F.: El Colegio de Mexico,1977), p. 94. 38. Munioz,de Oliviera, and Stern, Migracionesinternas,p. 43. Migrants with prior employment in agriculture have consistentlyevinced a higher incidenceof servicesector employment as a port of entry than those with nonagricultural or no previous work experience.See also Humberto Murloz and Orlandina de Oliveiros,"Migraci6n, oportuni- dades de empleo y diferencias de ingreso en la ciudad de Mexico," Revista mexicana de sociologia,vol. 38, no. 1 (1976), pp. 55-56. 39. BrigidaGarcia, Humberto Mufioz,and Orlandina de Oliveira,"Migracionesinternas y grupos populares urbanos: Ciudad de Mexico (1950-70)," in Poblaci6ny desarrolloen America Latina, ed. Victor L. Urquidi and Jose B. Morelos (Mexico,D.F.: El Colegio de Mexico,1979),p. 122.The groupof migrantsdescribedhere is limitedto thosewho had been employed prior to their arrival in Mexico City. It thus excludes migrants who arrived as children and later entered the labor market. 40. Butterworth, "A Study of the UrbanizationProcess," p. 106.  NOTES 169 41. Anson, "La relaci6n entre la politica de migraci6n," pp. 21-22. 42. Garcia, Munsoz, and de Oliveira,"Migracionesinrernasy grupospopularesurbanos," p. 123. 43. Humberto Mufioz,Orlandina de Oliveira,and ClaudioStern, "Migraci6ny margina- lidad ocupacional," in Migraci6ny desigualdad,pp. 8)-83. 44. Kemper, "Tzintzuntzefiosin Mexico City," p. 87. 45. Butterworth, "A Study of the UrbanizationProcess," pp. 106-107. 46. Censo generalde poblacion, 1970, table 48. 47. Arizpe, Migraci6n,etnicismo y cambio econ6m,co,pp. 85-86. 48. A caveat must be entered regarding the interpretation of this finding.Migrants who failto improvetheir economicstatus in the city may simplyreturn to their placesof originand thus escape detection. 49. Balan, Browning, and Jelin, Men in a Developing Society, pp. 132-33. Another dimensionof socioeconomic mobilityassociatedwith m.grationis the intergenerational one. Migrants improvenot only their premigration position but also their socioeconomic status relative to their fathers. See, for example,Instituto Nacional de Estudios del Trabajo, La incorporaci6nal trabajo obrero en un medio de indust:rializaci6n reciente,FolletoTecnico no. 1 (Mexico,D.F., 1977), p. 24. SO. Mufioz and de Oliveira, "Migraci6n y movilidad ocupacional," pp. 95-97. 51. Anson, "La relaci6n entre la politica de migraci6n," pp. 19-26. 52. The opportunity cost in each year is calculated by adjusting the migrant's last premigration rate of earningsby the averagerate of changein earningsin his zoneof origin. 53. The nominal rate of return was calculatedto be 288 percent in the first year.The way in which this was adjustedto accountfor differences in the cost of livingin the premigration residence is not spelled out in Anson's paper. The maximum possible number of years of residence of his sample in Monterreywas less than six. 54. Anson,"La relaci6nentre la politicade migraci6n,"pp. 16-17. In his summary table 1, Anson refers to the costs of migrationas the family costs, implyingthat they refer to the costs of moving the migrantand his family.Furthermort,in the text he notes that 65 percent of the migrants madethe movetogetherwith the complet. family.Sincethe averagesizeof the family unit was 4.4 persons, this total cost of moving strikes me as surprisinglylow. 55. Contreras, Estratificaci6ny movilidad, pp. 89-91. 56. Mufioz and de Oliveira, "Migraci6n y movilidad ocupacional," p. 84. 57. Unikel, El desarrollourbano, pp. 179-84. 6. Migration to the United States THE RIO GRANDE and the northern border of the states of Chihuahua, Sonora, and Baja California definethe political boundary between Mexico and the United States. Historically, however, they have not posed an effectivebarrier to the movement of population. Until the mid-nineteenth century, of course, the U.S. area that now abuts Mexico was an extension of that country, having been part and parcel of the Spanish domain before Mexican independence.Much of southwestern United Statesand Califor- nia had been settled by Spaniards, and after independence Mexicans continued to be free to move to the northern territories. But migration to these territories was in fact relatively small. Only New Mexico had re- ceived a substantial flow of settlers. In 1849, when the Treaty of Guadalupe Hidalgo ceded to the United States Mexico's northern territories, the Hispanic population was esti- mated to be on the order of 73,000-100,000. Of these, approximately 60,000 were in New Mexico, 7,500 in California, and 5,000 in Texas.' Mexicans continued to move northward throughout the nineteenth cen- tury in response to a demand for labor in mining,agriculture, and railway construction, but in little more than a trickle. Only in the decade of the 1870s did as many as 5,000 Mexicansmigrate across the border. 2 At the turn of the century, the total Mexican population in the United States amounted to only 103,393. AsBarton Clark has observed, the significance of that number lies,not in its absolute size,but in its geographicaldistribu- tion. Heavily concentrated in the Southwestand California, it definedthe region toward which most of the twentieth-century migration would be directed.3 Indeed, the great bulk, 85 percent,of the population of Mexican origin is still concentrated in the Southwest. The turn of the century marked a watershed in the course of Mexican migration to the United States, for the flow suddenly assumed major proportions. In the first decade alone, more Mexicans migrated to the United States than had done so during the entire preceding eighty years since independence. With the exception of the depression decade of the 1930s, this flow has continued unabated (seetable 6-1). Mexico did not become subject to numerical immigration quotas until the amendment of 170  MEASURES OF STOCKS AND FLOWS 171 the Immigration and Naturalization Act in 1965. The number of legal immigrants averaged around 44,000 annually during the remainder of that decade and has increased to between 60,000 and 70,000 during the 1970s. These documented immigrants, however, constitute only part of the flow across Mexico's northern border. Indeed, that legal movement has been overshadowed by a much greater flow of illegal migrants who cross the border seeking temporary employment or an extended residence in the United States. Although the impact this phenomenon has had on condi- tions in the U.S. labor market has been of considerable concern, if not controversy, this is not the primary focus here. Rather, this chapter views the flow as an extension of the process of internal migration that was discussed in Chapter 5. What will be examined are the size of the flow, characteristics of migrants, the motivation underlying the decision to migrate, the nature of their migratory experience, and some of the possible consequences for employment conditions in Mexican labor markets. Measures of Stocks and Flows of Migrants Clearly, the total number of temporary and permanent migrants across the frontier has been influenced by conditions in both countries. For example, the sharp increase in migration rccorded in table 6-1 during 1911-30 is traceable to the political and economic turmoil that character- ized the revolutionary period in Mexico. At the same time in the United States, the labor shortages experienced during World War I and the general prosperity that extended for a decade beyond that war provided attractive employment opportunities for Mexicans. Particularly with the Table 6-1. Immigration to the United States from Mexico, 1821-1977 Years Immigrants Years Immigrants 1821_1900a 28,002 19.51-60 299,811 1901-10 49,642 1961-70 443,301 1911-20 219,004 1971-75 318,075 1921-30 459,287 1975 62,205 1931-40 22,319 1976 57,863 1941-50 60,589 1977 62,205 a. No recordof Mexican immigration is availablefor 1886-93.As a result,thenumber given somewhat understates flow. the actualsizeof the nineteenth-century Source: U.S.Immigration Service,Annual Report, 1974 and 1977. 172 MIGRATION TO THE UNITED STATES termination of free immigration from Europe in 1921, Mexican workers were ideally positioned to respond to any rightward shift in the U.S. demand for unskilled labor. The reestablishment of order and stability in Mexico in the 1930s and the concurrent severe economic depression in the United States brought an abrupt end to the growing stream of registered immigrants. That decade also saw the first concentrated effort to expel illegal immigrants already resident in the United States; the goal was to alleviate some of the eco- nomic distress resulting from the widespread unemployment. Between 1929 and 1937, some 400,000 Mexicans were repatriated. In addition, the Border Patrol was more intensively employed to reduce the illegal entry of Mexicans.4 The economic expansion and labor shortages spawned by World War II again created an environment favorable to renewed Mexican migration. Immigration regulations were relaxed to permit a larger flow of legal migrants. In addition, the bracero program was adopted as a vehicle for the recruitment of temporary migrants for employment in agriculture. Although the program was intended to expire at the end of the war, pressures from agrarian interests caused the U.S. Congress to extend it to 1951 and then until 1964. During its lifetime, some 4.8 million workers were contracted to work in the United States.' Illegal Migration However, these recorded flows of immigrants and braceros did not constitute the total count of migrants to the United States. The flow of illegal migrants began to exceed that of the legal in the 1940s. Returning braceros spoke glowingly of high earnings and of their desire to return to the United States. Others were undoubtedly influenced by such reports and experienced similar desires. Since the number of Mexican workers seeking bracero contracts exceeded the certificates of need issued by the secretary of labor, all those wishing temporary employment in the United States could not be accommodated by the bracero program. Furthermore, a bribe was frequently required to ensure selection. Thus, to many and particu- larly those who had acquired a familiarity with the United States through previous entries as braceros, it must have seemed easier simply to cross the border on their own.6 The other side of the coin was the ready demand for illegal migrants among agricultural employers who preferred to avoid being bound by the conditions of employment and government supervi- sion that were imposed by the formal bracero program.7 The apprehension and expulsion of illegal migrants mounted substan- tially after the end of World War II. From 26,689 in 1944 they reached a  MEASURES OF STOCKS AND FLOWS 173 peak of 1,075,165 in 1954 when a concentrated effort, Operation Wet- back, was mounted to sweep from the interiors of Texas, Arizona, and California to the border. In the process, some of the accumulated backlog of migrants undoubtedly fell into the net. Perhapsbecauseof the successof this effort, the number of apprehensions declined over the following decade to fewer than 100,000 a year. Still, over the whole lifetime of the bracero program, 1942-64, more than 5 million illegal migrants were apprehended. 8 Since no account can be taken of multiple crossingsand apprehensions, however, this figure undoubtedly overstates the number apprehended. A largeproportion of deportable Mexican aliens-62 percent in 1974- are apprehended within seventy-two hours of their crossing into the United States. 9 Since they failed to realize the objectiveof their entry, the temptation to try again is likely to prove irresistible. Indeed, interviews with apprehended Mexicans have recorded multiple apprehensions for many. Furthermore, the efforts of the Border Patrol to control the flow of aliens have varied in intensity over time. During periods of economic expansion and tight labor markets in the southwestern United States the patrol seemsto have relaxed its vigilance,only to return to more restrictive postures when unemploymentmounted. Thus, after the intensivesweep of 1954 apprehensions may have declined at least in part becausea return to routine control procedures permitted a large flow of migrants to enter undetected and to remain undisturbed. Finally, the accuracy and significance of the reported numbers of apprehensions have been called into question. As part of its study of migration to and across the northern frontier, the Centro Nacional de Informaci6n y Estadisticasdel Trabajo (CENIEr)manned all of the sites on the Mexican sideof the border to which the Irmmigration and Naturaliza- tion Service (INS) delivered the illegal alien3 apprehended. CENIET re- searchers counted the returnees and administered a brief interview to almost all of them. It is disturbing to finda very large discrepancybetween the number of apprehensionsof deportable al].ensreported by the INS and the count of expulsionsmade by CENIET duringAugust 1978. Whereasthe CENIET enumerators registered fewer than 33,000 expulsions, the figure estimated from INS sources amounted to almost 89,000, a discrepancyof approximately 56,000.10No reasonable me-lodological basis for this large discrepancycould be advanced, which suggeststhe possibilitythat a systematic inflation of the actual number of cetentions and deportations has occurred in the INS."An unlikely alternative explanation is that more than half the apprehensions of deportable aliens do not result in expul- sions. In the absence of any effective method of enumerating the un- 174 MIGRATION TO THE UNITED STATES documented Mexican aliens in the United States the estimating procedures employed have given rise to widely disparate conclusions. Some of the more extreme estimates of the number of illegal aliens issued by the INS during the 1970s placed the number of illegal aliens between 1 million and 12 million, though the empirical basis for most estimates were weak at best. In 1976 the INS contracted Lesko and Associates to determine the number of undocumented workers. The report, which has been widely criticized on methodological grounds, estimated between 4.2 million and 11 million, with a probable number of 8.2 million of which 5.2 million were Mexicans. 12 With the accession of a new director of the INS, Leonel Castillo, another estimate was released in 1978 that placed the number between 3 million and 6 million. The controversy over the number of illegal aliens stimulated several 3 more serious efforts to resolve the issue." One such study, by Lancaster and Scheuren, placed the number of all illegals, including non-Mexicans aged 18 to 44, at between 2.9 million and 5.7 million in 1973 with a 4 suggested specific figure of 3.9 million." Members of the white race, which presumably include Mexicans, were placed at between 2.0 million and 3.7 million. The estimates were derived by application of "capture and recap- ture" techniques that are appropriate for estimating populations not subject to census counts. The authors relied on a matching by the Bureau of the Census of households enumerated in the Current Population Survey (cps) that also appeared in any one of the following three registers: Internal Revenue lists of taxpayers, contributors to Social Security, and recipients of Social Security benefits. An estimate of the population, in- cluding illegal aliens, was calculated from this matching. It was then compared with an estimate derived from an adjustment of the 1970 census figures forward to 1973; this accounted for births, deaths, and net legal migration but excluded illegal migration. The difference between these two population estimates was attributed to the presence of illegal aliens. The authors recognized the fragility of some of the assumptions under- lying the procedure and tried to compensate by allowing for a generous margin of error. Although the estimates may be subject to some question, they at least have the virtue of being based on data and procedures that are explicitly described; readers are thus allowed to make adjustments as they see fit.'5 Mexicans are presumed to form a preponderant though unknown pro- portion of illegal aliens in the United States. Although Mexicans constitute about 85-90 percent of all deportable aliens apprehended, this proportion cannot be taken to represent the true incidence of Mexicans within the total alien population. Multiple entries and apprehensions within any given period are realistically possible only for Mexicans and Canadians  MEASURES OF STOCKS AND FLOWS 175 (the latter, however, do not avail themselvescf the opportunity in signifi- cant numbers). Furthermore, since the apprehensions of Mexicans are concentrated in the border region, they are easier to effectthan are those of persons originatingin other countrieswho are dlispersed over a wider area. North and Houstoun have estimatedthat Mexicansmake up between 60 and 65 percent of the illegal alien population."6 The Flow of Migrants The measuresofferedthus far are concernedwith the sizeof the stock of illegal migrants. To appraise the impact of migration on the Mexican labor market, however, one needs the rate of change in the stock of Mexican permanent residentsin the United States and the size of the flow of temporary migrants. As in the case of the stock, the flows have been subject to widely divergent estimates. For e-ample, Secretary of Labor Marshall was cited as stating in 1977 that probably 2 to 3 million illegal aliens enter the United States each year. The iu's has offered estimatesof a similar order of magnitude. During the 1970s, the INS apprehended about 700,000 deportable aliens annually. Based o0]its presumption that, for every apprehension, two aliens escape deten:ion, the number of illegal entries would be estimatedto exceed 2 million.17 However, these appear to be rather arbitrary measuresthat do not distinguishbetween gross and net flowsor between permanent and temporary residencein the United States. Several attempts have been made to estimate the net flow of illegal immigrants from demographic statistics. Using U.S. population statistics from the 1970 census and subsequent counts from the cps, David Heer estimated the growth of the population of Mexican origin between 1970 and 1975. After allowing for undercounting of this subgroup in the census and for net documented immigration,he obtained a residual increasethat could not be accounted for by the national growth rate and was therefore attributed to net illegal immigration. He actually derived sevenestimates by employingcross-classifications of various assumptions about the ratio of net to gross documented migration and the proportions by which the population of Mexican origin was undercounted in the 1970 census. His preferred estimate of the annual rate of growth of the illegal immigrant population was 116,000 between 1970 and 1]975.18 An alternative method was employed by J. Gregory Robinson.19 He departed from an assumption that neither the population census nor the cps capture the alien population, but that reported deaths do include most aliens. If so, age-specificdeath rates, especiallythose from violent causes, should be higher in those regions in which aliens are presumed to be concentrated. In fact, he did find a persistent deviation in the death rate 176 MIGRATION TO THE UNITED STATES trend for ten states since 1960 for white males aged 20-44, but for no other group. By applying different death rates and varying assumptions regarding the proportion of illegal alien deaths that go unrecorded and the proportion of the illegalpopulation captured by the population census,he deriveda range of estimatesof the illegalalienpopulation at various points in time. Thus an estimate of the annual rate of change was provided. The two estimates of the growth considered most plausible by the author for the five southwestern states that are presumed to host most Mexican migrants are given in table 6-2. The annual rate of increasefor the decade of the 1960s would thus lie in the range of 36,000-46,000 a year and 230,000-267,000 a year during the 1970-75 interval. Robinson consid- ered these to be of the same general order of magnitude as the estimates made by Heer or those derivable from the estimates of Lancaster and Scheuren. Apparently only one demographicstudy based on Mexican censusdata has estimated the growth in the stock of undocumented migrants in the United States.20 Howard Goldberg compared the 1970 and 1960 Mexican population census and, using the residual method, concluded that 1.6 million Mexicans who were living in Mexico in 1960 could not be accounted for in 1970 by either death or legal emigration to the United States. He attributed this loss to illegalemigration. The Goldbergestimate implies an annual net flow four times greater than Robinson's. It is my beliefthat Goldberg's estimate is likelyto be biased upward. Had the legal emigrants and the presumed number of illegal emigrants remained in Mexico, the population would have been greater by some 2,045,000 in 1970, plus any offspring they would have had. Not including the latter, this would have implied a rate of population growth of 3.84 percent a year rather than the 3.45 percent actually recorded. A natural rate of increase of that magnitude is extraordinarily high even in the developingworld. Only the Ivory Coast, Kuwait, and Libya reported growth rates equal to or higher than 3.8 percent per year during the 1960-76 interval, and the first two were recipients of large flows of immigrants. 2" Table 6-2. Estimates of the Growth in IllegalAlien Population for Five Southwestern U.S. States, 1960-75 (thousands) Estimate 1960-70 1970-75 1960-75 1 363 1,136 1,499 2 464 1,335 1,799 Note: The fivestates are Arizona, California,Colorado, New Mexico,and Texas.  MEASURES OF STOCKS AND FLOWS 177 Manuel Garcia y Griego adopted an extensive estimating procedure and applied it to the flow of illegal migrants over a five-year period, 1972-77.22 This approach used as its empirical input the INS statistics on apprehensions and the migration histories of Mlexicans who were expelled to Mexico and interviewed by staff researchers of CENIET. The latter information was culled from the first of three border surveys that were undertaken by CENIET and extended over a three-week period during October and November 1977. Since the INS reports a distribution of persons expelled by the duration of their stay in the United States, it is possible to classify the migrants into cohorts according to the year of entry to the United States. Of course, the aliens apprehended are only part of the flow across the border. In addition, there are those who escape detection and return to Mexico of their own volition. The size of this flow is estimated from information derived from the migration histories of a sample of those people apprehended and delivered to the Mexican reception centers dur- ing the first survey of 1977. For every expulsion recorded, there were between 0.7 and 1.2 voluntary returns over the 1970-77 period. Slightly more than half of the sample of 3,686 persons interviewed, or 1,857, reported that they had crossed into the United States on previous occa- sions for a total of 4,687 entries.' These individuals were likewise distrib- uted into cohorts defined by the year of each entry. For each cohort, the ratio of voluntary to involuntary departures was computed and applied to the INS data on apprehensions for an estimate of the number of undetected illegal entries. A final component of the flow ol: migrants was derived from the data on documented, or legal, entries to) the United States. If doc- umented entrants reported previous periods of residence, they were as- sumed to have been illegal aliens and thus were assigned to a cohort by year of entry. Obviously, one component of the flow of unidocumented aliens cannot be accounted for by the available data: those who entered and were not detected. It is axiomatic, however, that all who entered at some point in the past and have not been recorded as exiting, must exit from the stock of aliens in the United States at some point in the future either by expulsion, voluntary departure, legalization of status, or by death. An estimate of the distribution of these future exits from each entering cohort thus provides a basis for adjusting the numbers of known arid estimated entries, as de- scribed above. It can serve further to estimate the net change in the stock of illegal migrants in the United States. The basis for this estimate is found in the distribution of aliens by duration of their stay. From the data of the INS and the employment histories of the expellecdinterviewees, it is evident that the bulk of the migrants remain in the United States for less than a 178 MIGRATION TO THE UNITED STATES year; the proportion ranges from 83 to 87 percent of those exiting between 1972 and 1976. The balance reported durations of more than a year, and the number reporting each additional year of residencedeclinedin a fairly steady sequence. A projection of the rate of change in the number of migrants with from two to seven years' residence yields a number of undetected individualswith longer stays in the UnitedStates.The figureso estimated can then be assigned to cohorts defined by year of entry. (It is assumed that the number of aliens in the United States of any entering cohort goes to zero in forty years.) This calculation is required if a full consideration of stocks and flows is to be derived. Two estimates of entries, departures, and net flow are then presented. The high approximation is based on the apprehensions reported by the INS. However, since the CENIET researchersfound such a largediscrepancy between the apprehensions reported and the expulsions recorded at the Mexican reception centers,a low estimate is offeredthat reflectsthe sizeof the discrepancyrecorded by CENIET. The method so devised thus permits an estimate of a net change in the stock of undocumentedMexican aliens in the United States. The approximate flow for the years 1972-76 can be seen in table 6-3. The estimated number of entries increased substantially over the period and averaged between 629,000 and 2,043,000 a year. Estimated exits did not increase as rapidly and gave rise to an increasingnet addition to the stock of Mexicans who were in the United States permanently or for Table 6-3. Flow of Undocumented Workers to the United States, 1972-76 (thousands) Low estimate High estimate Year Entries' Exitsb Net flow Entries, Exits Net flow 1972 465 453 12 1,497 1,458 39 1973 545 517 28 1,770 1,638 131 1974 651 632 20 2,139 2,062 77 1975 680 606 75 2,239 1,983 256 1976 799 687 113 2,571 2,286 284 Average 629 579 50 2,043 1,885 158 a. Entries referto border crossingsinto the UnitedStatesor to changesin the legalstatus of migrants that lead to their classificationas deportable. b. Exits refer to returnsto Mexico,voluntary or by expulsion,and to legalizationof one's presence in the United States. Exits through death are not consideredhere. Source:Manuel Garciay Griego,El volumen de la migraci6nde mexicanosno documenta- dos a los Estados Unidos, Estudios no. 4 (Mexico,D.F.: CENIET, 1980).  MEASURES OF STOCKS AND FLOWS 179 extended durations. The last two years suggestedan annual expansion of the stock of Mexican aliens in the United States of between almost 100,000 and 125,000. The stock of deportable aliens in the United States on January 1 of each year, from 1972 to 1977, was also calculated. The low and high approximations were 234,000 and 436,000 for 1972, and 482,000 and 1,224,000 for 1977. As of those moments in time, the stock was estimated to contain, as its largest component, individualswith less than one year's residencein the UnitedStates, cr approximately45 percent of the total. A totally different approach to calculating the number of un- documented Mexicans is based on the volum,zof remittances from indi- viduals in the United States to individualsin Mexico. Although there have been other studies adopting such an approach, only the work of Juan Diez-Canedowill be reported here. 2 4 There exists persuasiveevidencethat most individual remittances originate with workers temporarily in the United States.Permanent Mexican residentsthere have their familieswith them and only infrequentlysendmoney to Mexico. When they do, they are likely to do so by personal check. In contrast, remittances by temporary residents is by certifiedcheck and postal money orders. These instruments are recorded in Mexico so that it is possible to identify the names of the sender and recipient; the source and destination, by state, of the check; and the amount remitted. Diez-Canedoemploys the findingsof North and Houstoun to definethe proportion of Mexican aliens in the United States that does send money home and the average size of remittances made. Low, high, and medium estimates are reached by varying the values of the parameters of the estimatingequation. The results are as follows:in any month of 1975, the number of undocumented workers in the United States ranged from 235,000 to 2,900,000. Applying the "most reasonable" values of these parameters as revealedby North and Houstouri,one would reach a "more realistic" figure of 815,000. This estimate can be compared with those of Garcia y Griego. For example, if one takes r.heaverage of the latter's estimated undocumented stock for January 1 of 1975 and 1976 as indica- tive of the average stock during 1975, one gets an upper limit of 812,000 and a lower limit of 332,000. The upper limit proves virtually identicalto Diez-Canedo's most reasonable estimate. The most comprehensiveeffort to explore the migration phenomenon has beenundertaken by CENIET. Their studies includethree border surveys in which undocumented workers who were returned to Mexico by the INS have been enumerated and interviewed.Alsoincluded is a national house- hold survey designed to yield the number of Mexicans, aged 15 or older, who were currently working or looking for work in the UnitedStates and 180 MIGRATION TO THE UNITED STATES those currently in Mexico who had been to the United States with similar intentions at any time during the previous five years (1974-78).2s Some 60,000 households were polled during December 1978 and January 1979. The results of this survey provide a measure of the number of temporary migrants: 405,467 Mexicans who habitually reside in Mexico were re- 6 ported to be either working or looking for work in the United States." This measure differs from other estimates in that it does not distinguish be- tween legal and illegal aliens. Except for border communities from which a Mexican labor force commutes daily to work legally in the United States (the so-called green- card holders), no region is a significant source of legal temporary workers.27 The daily commuters are excluded from this survey measure, however, because they were not "absent from their habitual residence." It should also be noted that the survey was conducted at a time when the number of undocumented workers temporarily in the United States was likely to have been at its minimum. Because of the importance of agricul- ture as an employer of undocumented workers, there is a pronounced seasonal pattern in the migration flow that corresponds to the peaks of agricultural activity. Furthermore, the survey coincided with the Christ- mas season, which may be presumed to have further minimized the number away from home. Permanent Leakage That there is a significant permanent leakage out of the migrant stream into the United States is suggested by one additional bit of information yielded by the CENIET household survey. This is the number of persons 15 years or older and residing in Mexico at the time of the interview who had been to the United States to work or to look for work at least once at any time during the previous five years (1974-78). The preliminary number of these Mexicans is given as approximately half a million."8 If, as appears to be the case, most migration is temporary, one would expect the stock of Mexicans in Mexico with migration experience to be considerably larger than the number in the United States at any time. Consider the following. If migration were purely temporary, and the average duration of stays were six months, the maintenance of a stable group of 500,000 workers in the United States at all times over a year would require a stock of approximately 1 million Mexicans in Mexico and the United States who would rotate between the two countries. A stock of this size would require that all those in it spend six months in the United States each year and that there be no infusion of new migrants. If not all those with migration experience wished to rotate at six-month intervals,  MEASURES OF STOCKS AND FLOWS 181 and if there were a constant infusion of new migrants, then the total stock of Mexicans with migration experience could be expected to expand rapidly and to be captured by the CENIET survey. That there has been a significant infusion of new migrants is suggested by the data on apprehensions.Although first-timemigrants were probably disproportionately represented among those detained by the Border Pa- trol, they probably made up a substantial proportion of each cohort of successful entrants to the United States. For example, of the expelled Mexicansinterviewedby CENIET in its first border survey,58 percent were first-time migrants. If the probability of apprrhension were only twice as great for a firsttimer as for others, this ratio of apprehensionswould imply that first timers constituted approximately 40 percent of the flow of successfulcrossers. If the probability of getting caught were three times as great, the proportion of firsttimers to success] ul crossers would declineto about 30 percent. If migration were purely a seasonal or temporary phenomenon, this infusionof 160,000 to 200,000 new migrants into every cohort would very quickly build up the stock of workers in Mexico with migration experience to levels greater than those recorded by CENIET'S migration study. 29 An averagestay of six morths further impliesthat two cohorts of migrants would enter the United States a year. If the stock of returned migrants in Mexico •emainedrelativelyconstant at around 500,000, this would implyan annual leakage into the U.S. labor force that approximately equalled the number of new entrants into the migration stream. An estimate can be made of the absolute size of that leakage based on the availableinformation regarding the average number of workers in the United States over the year and the past rate of apprehension, as revealed by the North and Houstoun study. This can be done, however, only by assuming an appropriate differential rate of apprehension for first timers and experiencedborder crossers. Adopting 500,000 as the size of each cohort, an average probability of 0.81, and a probability of apprehension for first-timersthat is from two to four times the probability for experienced crossers would yield a permanent or long-term leakage into the U.S. labor force of between roughly 100,000 and 200,000 from eachcohort. Giventwo cohorts a year, the annual range of leakage would be twice as great. The high estimate of the net leakages offered for 1975 and 1976 by Garcia y Grieg, about 250,000-280,000, lies near the bottom of this range. If the increasingtrend of leakagesnoted by him were projected further to 1978-79, the time of the national household survey, the leakage could more nearly approximate the upper limit of the range, 400,000, suggested by my tentative calculation. The conclusion about the nature of the migrant flow that I have drawn from this large infusion of first timers departs sharply from that drawn by 182 MIGRATION TO THE UNITED STATES CENIET. After the study observed that 58 percent of the apprehended migrants interviewed in the first border survey were first timers, it con- cluded: "The large proportion of 'first-timers,' in principle, tends to demonstrate a large renewal in the composition of those who migrate to the U.S. If these data reflect the general character of the undocumented migration phenomenon, it would lend support to the hypothesis that the career of the undocumented emigrant is shorter than is usually thought."3 0 In view of the unexpectedly small size of the returned migrant stock enumerated by the household survey, the opposite conclusion would appear to be more tenable. It is, of course, possible that the CENIET household survey underenumerated the Mexicans with migration experi- ence. If this is the case, this estimate of the permanent leakage will prove to have been exaggerated. Net Outflow The CENIET survey thus offers grounds for concluding that temporary and permanent migration to the United States adds up to a significant withdrawal of labor from the market in Mexico. At the end of the decade of the 1970s, this implied the withdrawal of 500,000 man-years in the category of temporary migrants and a net leakage of up to 400,000 a year into the U.S. labor market on a longer term basis. The latter would represent the highest annual rate of all those offered here, and, in the light of the data reviewed immediately below, it would also seem to be the least plausible. Ordinarily, one would expect to derive more precise estimates of migra- tion flows for the 1970-80 decade once the 1980 census data became available. A procedure could then be adopted that would apply expected survival rates to each age cohort of the 1970 census to estimate the number of individuals that would be expected to be in the same cohort in 1980 in the absence of any migration. The comparison of this expected number with the actual enumeration of the 1980 census would then provide a measure of net migration for each age cohort. Unfortunately, the data currently available do not permit conclusions that can be advanced with much confidence. None of the available sources agree on the age distribu- tion of the population in 1970 or 1980 and, therefore, yield very different estimates of emigration. For example, the age distribution of the population as published in the official censuses of 1970 and 1980 yields the numbers recorded in table 6-4. Although the 10-14 age group is included in the table, I consider these numbers less reliable than the others because of the greater margins of error in the 1970 enumeration of infants and young children. As can be  MEASURES OF STOCKS AND FLOWS 183 Table 6-4. Estimated Net Migration by Sex and Age Group, 1970-80 (thousands) Males, Fsnmales' Age group Census CONAPO Census CO.NAPO 10-14 - (80.3) (49.0) (21.)) (98.5) 15-19 -133.7 -128.0 64.3 107.1 20-24 - 306.6 -265.0 34.) 41.7 25-29 -178.4 -278.5 - 89.3 -101.2 30-34 -32.6 -198.0 -141.1) - 95.3 35-39 122.0 - 49.8 71.3 35.5 40-44 124.9 23.0 91.4 47.4 45-49 -1.3 - 9.7 - 34.6 - 3.8 50-54 14.5 -10.1 11.3 -8.8 55-59 -9.0 58.5 -18.3 -3.3 60-64 27.1 48.7 26.9 16.8 Totalb -373.1 -808.9 15.; 43.7 a. A negative number represents net emigration, a positive number net immigration. b. Totals do not include the 10-14 age group. Sources:Secretariade Industria y Comercio,Direcci6n General de Estadistica,IX Censo generaldepoblaci6n, 1970:Resumen general(Mexico,D.F., 1972); Secretariade Programa- ci6n y Presupuesto, X Censo general de poblaci6n y ririenda, 1980: Resumen general abreviado (Mexico, D.F., 1984); Consejo Nacional de Poblaci6n, Mexico demografico (Mexico, D.F., 1982). seen, the count of the male population between the ages of 15 and 34 was 651,000 smaller than "expected"; for the age group 35 to 64, it was about 278,000 greater than expected. The result wa; a net outflow of 371,000 men. For women, this method yielded a number 15,500 greater than the expected number in the 15-64 age groups arid implied a small net im- migration of women. In view of the size of ,he legal outflow of adult women during the 1970s, this result would seem to be implausible. Thus, this set of data yields a net outflow of 357,600 people 15-64 years of age, a number smaller than the outflow of almost 40C,000 legal migrants of that age.3' This would imply a net return to Mexico of illegal migrants over the decade. An alternative estimate of the net outflow is derived from population distributions estimated for 1970 and 1980 frormcensus and other data by the National Population Council (CONAPO), which appeared in a 1982 publication, Mexico demografico. The age distributions in this publication for both years have been adjusted for a presumed undercounting of infants and young children as well as for other presumed shortcomings of the 184 MIGRATION TO THE UNITED STATES published censuses.These data yield estimates of net migration consider- ably greater than those derivedimmediatelyabove. As can be seen in table 6-4, male net emigration amounted to about 809,000, femalesrecord a net return migration to Mexico of some43,000, and the aggregatenet emigra- tion was 766,000. Since legal migration in the age group 15-64 was estimated to have amounted to approximately 390,000, this would place illegal migration at about 375,000 for the entire decade, well below most other estimates. A third sourceof population estimates is another, later publication that officialsof CONAPO consider to contain the most authoritative estimatesof Mexico's population by age group for 1970 and 1980. These calculations were the product of a joint effort by CONAPO, the Ministry of Program- ming and the Budget(spp), and the Latin American Center for Demogra- phy (CELADE). 3 3 The age distributions offered by this publication reflect adjustments for undercounting as well as a level of migration that is assumed to have taken place. Migration is assumedto be constant in each year of the 1970s at a levelof 70,000, a number that is only 12,000 greater than the annual average number of legal migrants to the United States. This would imply a net illegaloutmigration of only 120,000 Mexicans of all ages over the decade. Thus, it would seem that the attempt to employ census data for estima- tion purposes does not contribute much to resolvethe controversy sur- rounding the volume of illegal migration to the United States. This effort only serves to increase the reservations regarding the accuracy of the census enumerations that were expressed in Chapter 2 and its appendix. Not only do the Mexican sources based on the most recent censusesshow significant differencesamong themselves,but they provide little support for any of the estimates reviewedearlier in this section. The CONAPO data (37,500) would appear to come closest to the low estimate of average annual emigration (50,000) offered by Garcia y Griego for the years 1972-76 but still fall far short of the latter. But it seemsinconceivableto me that the census sources, even though deemed deficient, could fail to capture the volumeof population loss if the actual annual loss approached the high estimate of 158,000 offeredby Garcia y Griegoor evenhalf of the 400,000 figure that I suggested might be derived from the procedure applied to the CENIET household survey. After all, these rates of net outmigration would imply a net population loss over the whole decade of between 1.5 and 2 million in addition to the almost 600,000 legal mi- grants. The failure of the censusto record losses of such magnitude would suggest that perhaps the annual volume of net undocumented migration over the decade did not exceed the 50,000 lower estimate of Garcia y Griego. This would, of course, place the annual level of illegal migration below the other estimates reviewed in this section.  CHARACTERISTICS OF TIHE MIGRANT FLOW 185 Summary In summary, the various studies reviewed here exhibit a fairly wide range of estimates of both the stock of undocumented Mexican aliens in the United States and their annual rate of flow to this country. With respect to the flow, the CENIET survey results reporting over 400,000 workers temporarily in the United States at the seasonal trough would seem to be a reasonable number. The seasonal peak number of migrants is therefore likely to exceed half a million; on average over the year, the flow may represent a withdrawal from the internal Mexican labor market of some- thing on the order of a half a million work yea rs. The possible implications of this withdrawal for employment condit:ons in the domestic labor market will be explored at a later point in this chapter. With respect to the annual leakage into the lJnited States on a prolonged or permanent basis or the stock of undocumented residents, less can be said with great certainty. One conclusion that can be drawn from the various studies that measure the stock of illegal aliens in the United States, however, is that none lend much support to some of the INS "official" estimates issued during the middle of the '1970s regarding their total number. The order of magnitude appears not to have been between 8 million and 12 million, but rather fewer than 6 million and perhaps as few as 4 million. Within this total, Mexicans were a significant number, perhaps 2 million to 3 million. The annual net additions to this stock, however, are believed by some to have been increasing toward the end of the decade. It would not be surprising if this in fact were so. As the stock of resident Mexicans in the United States increases, the information costs of migration are significantly reduced as are the out-of-pocket costs. With a decline in the ratio of costs to benefits associated with migration, one would expect the size of the flow, as well as the leakages out of that flow, to 34 increase. Having said this, however, I find it difficult to offer with much conviction an estimate of the actual size of i:he net annual leakage into prolonged residence in the United States. If pressed hard, I would offer only a qualified opinion that the rate of leakage during the decade of the 1970s is unlikely to have exceeded an average of 100,000 individuals a year. Characteristics of the Migrant Flow Apparently, there have been at least some constants in the character of the migrant flow to the United States over time. One of the most striking is the stability in the relative importance of various states as a source of migrants and the disproportionate share of migrants contributed by a 186 MIGRATION TO THE UNITED STATES small number of states. The CENIET household survey found that 61 percent of the migrants originated in six states of central Mexico and another 18 percent from two of the border states. These same areas contained less than a third of Mexico's population but provided 80 percent of the migrants. Of special importance are the states of Guana- juato, Jalisco, Zacatecas, Michoacan, San Luis Potosi, Durango, Chi- huahua, and Baja California. Only the last two border on the United States. The relative importance of these states as sources of migrants has changed little over time except for the emergence of Baja California as a major point of origin. Studies dating back to the 1920s by Gamio and Foerster confirm the preponderance of the central plateau as the source of migrants.3 5 Similarly, there has been little change over time in the relative impor- tance of the areas in the United States to which the migrants flow. Over three-fourths of the migrants flow to the four southwestern states border- ing on Mexico, although California alone is the destination of fully half the migrants and Texas of another fifth. Arizona and New Mexico, however, are no more important than other states farther removed from the border. In Gamio's study, these same four southwestern states were the recipients of over half of the migrants; outside the Southwest, only Illinois 36 was a primary destination. Diez-Canedo's study of remittances provides an alternative source of information on the destination of migrants. Based on the number of remittances recorded from the various states north of the border, Califor- nia is first by a wide margin. In second place is Illinois, followed by Texas, New York, and Minnesota. That northern states do not appear as a major destination for migrants, as recorded by the CENIET surveys, suggests that the remittances may originate with the longer-term migrants or permanent legal residents, contrary to the temporary status attributed to them by Diez-Canedo.3 It would seem to be consistent with rational behavior for temporary Mexican migrants to the United States to minimize travel cost, particularly if they qualify only for low-skill employment. However, those remaining in the United States may begin to consider regional differences in earnings and may be prepared to undertake the larger investment associated with a long move. A further positive factor favoring such moves may be a reduced probability of detection, apprehension, and expulsion. But this line of reasoning does not seem to derive much support from the distribution of legal Mexican aliens in the United States. INS data show that, of the 882,606 Mexican aliens who reported under the alien address program in 1974, 52 percent resided in California, 29 percent in Texas, and 7 percent 3 in Illinois. New York was insignificant as a place of residence. "  CHARACTERISTICS OF TIHE MIGRANT FLOW 187 A similar distribution is yielded by INS data on the chosen states of residence for Mexican immigrants admitted in 1974. Of the 71,586 so admitted, 47 percent settled in California, 32 percent in Texas, 9 percent in Illinois,4 percent in Arizona, and 1.3 percent in New Mexico. No other state was chosen by as many as 1 percent of the immigrants.New York and Minnesota were chosen by only 422 and 57, respectively. 39 Thus, it would appear that migration currents, both legal and undocumented, during the 1970s retained their traditional destinations. The flow of migrants from different points in Mexico shows different patterns of distribution over the principal destinations. The states of western Mexico-Baja California, Sonora, Jalisco, and Michoacan-are more closely linked to California; those in the central and northeastern part direct a larger proportion to Texas. These movements cannot be explained solely by considerations of distance. For example, Texas is much closer to Jalisco and Michoacan than is California, yet the flows from these favor California by a better than 7:3 ratio. 4 0 Custom and the higher wage levels prevailing in the latter apparently are sufficient to overcome the obstacles of distance. The recorded personal characteristics of the legal and illegal migrants exhibit considerabledifferences.The legalimmnigrant flow is characterized by roughly equal numbers of men and women and by complete family units with young children. It is a youthful stream; 40 to 45 percent of the immigrants are under 20 years of age and most of the rest between20 and 39.41The sex and age distribution of undocumented Mexicans differs sharply from the legal flow. The first CENIET border survey of apprehended aliens returned to Mex- ico in 1977 was dominated by males of labor force age; 88 percent were males. Veryfew smallchildren were recorded; only 1.4percent of the total returnees were younger than 15. The age distributions of apprehended men and women were very similar, with the bulk, 81.9 and 77.5 percent, respectively,falling in the 15-34 age bracket. The men with rural resi- dences in Mexico outnumbered those from urban areas by only a slight margin. 4 2 Yet women were predominantly urban; 76 percent reported urban residences,mainly from the border communities. 43 The sex and age distribution of the apprehended aliens matched closelythat of workers in the United Statesas enumerated by the household survey." One exception, however, was the proportion of migrants with rural origins which, at 78.3 percent, was significantlylarger than the 48.5 percent recorded by the border interviews. Carlos Zazueta also reported the marital status of the absentee workers; they were evenly divided between single and married individuals. The educational level of workers absent from home revealed the mean 188 MIGRATION TO THE UNITED STATES years of schooling as 4, somewhat above the mean of the population 15 years and older in 1970 (3.1 years). Women however, recorded a higher level of schooling (4.7 years) than did men (3.9); this probably reflected the predominance of urban residents among the former.4 5 Although the migrants would appear to boast an average educational levelhigher than that of the Mexican adult population in 1970, the difference could be expected to have narrowed substantially by 1977-78, when the migration survey was undertaken, as educational opportunities continued to ex- pand. In fact, in view of the significantnegative associationof educational level and age within the general adult population, it may very well be the case that a comparison of age-specificeducational attainments of the migrants would revealthem to have no more schoolingthan their nonmi- grant counterparts. If a comparison were to be made with only the rural population, however, the migrants are likely to report more than the average years of schooling. The occupational experience of the undocumented migrant group reflects the low average levels of schooling and the rural origins of most migrants. A majority of those who had been employedin Mexico during the month before their departure for the United States worked in agricul- ture. The border interviewersrecorded the primary sector as the sector of employment of 52 percent of the expelled migrants; the national house- hold survey recorded approximately 65 percent. This difference is obviously a reflection of the difference in the incidence of rural origin recorded by the two surveys, which was noted above. The household survey recorded 16 percent with secondary sector jobs, roughly half of which were in manufacturing, and 19 percent with tertiary, of which one-third were in commerce. 4 6 The sector of employment in the United States of a plurality of the temporary migrant workers is still agriculture; 34 percent of the sample of returnees had worked there. Following close behind agriculture is the servicesector with 32.3 percent. These are mainly unskilled jobs in restaurants or domestic work. Manufacturing and con- struction also emerged as significant sectors of employmentfor 17.8 and 9.2 percent, respectively, of the returnees. However, the employment status referred to November 1978, a month of relatively low agricultural activity in the United States; had a month during the principal growing season been chosen as a reference, Zazueta and Mercado believea larger proportion would have been found in agriculture.4 7 The literature on undocumented migration to the United States places great emphasis on its seasonal and temporary nature. There exists a considerable body of evidencethat supports this view, such as some of the microcosmicstudies of rural communitiesundertaken by social scientists. Certainly, the years of the bracero program were so characterized.Foster,  CHARACTERISTICS OF THE MIGRANT FLOW 189 in his study of Tzintzuntzan, noted that in 1960 half of the village's men had been to the United States, many of them ten or more times-a testimony to the temporary nature of migration as it was structured under the terms of the program. 4 " A more recent study of the ejido LasTortugas, in the municipio of Puruandiro, Michoacan, found no fewerthan 70 men away, illegallyin the United States, from this communityof 165 families. While such migration had become an establishedpracticeover the preced- ing thirty years, only one person had succeededin legalizinghis presencein the United States and remaining there permanently. 4 9 In his study of nine communities in northeastern Jalisco that have a long traditon of migra- tion, Corneliusfound temporary migrants outnumbering permanent set- tlers in the UnitedStates by a margin of 8:1 during 1930-76. Furthermore, fewer than 19 percent of the returned illegal migrants expressed any 50 interest in permanent migration to the United States. The impression of short stays in the United States is also reinforced by some of the CENIET findings. For example, the first border survey of undocumentedworkers expelledfrom the United States yielded a median stay of only 1.5 days. This short duration is a consequenceof the large proportion of individuals apprehendedwithin aday or two of crossingthe border. Since 5.6 percent had been in the United States for more than a year, 588 days on average, however, the average length of stay for the apprehended group as a whole was 87 days. With the exclusionof those detained almost immediately after crossing the border, the modal group reported a stay of two to six months. However, this sample is dominated by individuals-S 8 percent-making their first attempt to enter the United States and is surely not representativeof the Mexican population illegally present at one time.5" Repeated attempts at crossing are likely to improve the evasiveskills of the undocumentedworkers. Thus, the average stay of successfulcrossers is lengthened in excessof an average that includes the unsuccessfulcrossers. 5 2 Furthermore, if the probability of detection and apprehension declineswith the length of a successfulstay, as seemsreason- able to expect, longer-term residents will be underrepresented in the sample of expelled workers. The national household survey offered two measures of the length of stays of migrants in the UnitedStates. For those who had migrated during the previous five years but were in Mexico at the time of the survey, the average length of stay was 160 days, or slightlymore than five months. Fully 68 percent remained six months or less; 13 percent remained more than a year. Among those who were currently in the United States but whose usual residence was still considered to be in Mexico, only 36 percent had been away for fewer than six months, while 32 percent had been absent for over a year.53 Omitted from conisideration, however, have 190 MIGRATION TO THE UNITED STATES been those who are in the United States and who no longer consider Mexico to be their usual place of residence.This imparts a downward bias to the averagelength of stay of illegal migrants. Unfortunately,the CENIET survey was not structured to yield information regarding the number of Mexicans who have remained in the United States for an indefinite or permanent duration. The view that most Mexican migration is only temporary and of short duration receives further support from the findings of the North and Houstoun study of illegal aliens in the United States. Their subjects are hardly a random sample of illegal aliens in the United States at any one time. The sample included few Mexicans who had been picked up in the U.S.-Mexican border states and was further limited to those who had worked at least two weeks in the United States at some time (although not necessarilyduring the last trip that ended in apprehension). By recording the migration histories of the sample, North and Houstoun were able to determine the accumulated time spent in the UnitedStates and the number of trips back and forth from Mexico. The accumulated average was 2.4 years. Since the Mexicans averaged 4.5 trips, this suggests an average duration of about six months per trip.54 A projection of such a figureto the entire population of Mexican undocumentedworkers in the UnitedStates, however, would require the assumption that the probability of appre- hension was the same for all individuals,regardlessof the duration of their stay. If, as would seem more reasonableto expect, the relationshipwere an inverse one, then the duration of stays yielded by a sample of those apprehended would yield an average duration that is downward biased. The migration experiencesof somecommunitiesstand in sharp contrast to those reported above for Las Tortugas or to the general impressionthat migration is largely temporary. Robert Shadow's study, in 1975-76, was of a community with a well-established tradition of migration to the United States, Villa Guerrero in northern Jalisco. He found a substantial number of permanent migrants among those with migration experience. Of 564 persons with such experience, 133 had withdrawn from the migrant stream because of age. The rest were almost evenly divided between temporary and permanent migrants-40 and 37 percent of the total with migration experience, respectively.The great majority of the permanent migrants of both sexes were married and livingin the United States with their spouses.55 Shadow's reported findingsin Villa Guerrero, however, would seem to stand as an exception rather than as a rule. Most microstudiesemphasize the temporary nature of migrationand report a lack of enthusiasm among potential and returned migrants for permanent residence in the United States. In so doing, they would seem to reinforce the conclusion tendered  CAUSAL FACTORS 191 in the preceding sectionthat the past net leakagesinto prolonged residence in the UnitedStates are better characterized as mnodestflowsrather than as torrents. Causal Factors in Migration Much of the professional and popular literature on migration has focusedon the relative strength of "push" and "pull" factors in motivating the movementsof populations. The literatureon Mexican migrationto the United States is no exception. Articles in the U.S. press that treat the subject rarely fail to stress the "high" levelsof unemploymentand under- employmentas the primary forcesimpellingMexican workers northward. Commenting on the causesof the increasingflcw of illegalmigrants to the United States over time, Wayne Cornelius states, "Since the late 1960s, however, 'push' factors on the Mexican side seem to have been more important [than demand fluctuationsin the U.S.].Indeed,there is no other way to explain the large increasesin illegal migration from Mexico to the U.S."5 6 The dominant long-term factor is held to be the growth of unem- ployment and underemployment, which follows from the failure of the economy to create enough jobs to keep up with the increase in the labor force. Thus, Cornelius characterizes the migration as "crisis-induced, income-maintenance migration." Since migrants' households operate at the margin of subsistence, any disaster, such as a crop failure or major illness in the family, creates a situation of short-term, acute economic necessitythat can be met most easilyby temporary migrationto the United States.57 "Push" Factors: Conditions in Mexico In a similar vein,John Evans and Dilmus Jamres write, "In our examina- tion of the factors contributingto the immigrantflow we focusupon three closely interrelated 'push' factors: the inadequate growth of productive employment opportunities, income inequality, and rapid population growth. We recognizethat other variables are (leterminantsof migration, but we are confronted by the evidence that the factors stressed are of outstanding importance"." 8 As evidenceof the dominanceof push factors, the authors also cite numerous other studies that reported that the primary motives for migration were work- or income-related. Although most writers on the subject have chosen to interpret such responsesof migrants as indicative of the strength of push factors, there is nothing in the responsesthat would preclude their interpretation as pull factors. In other 192 MIGRATION TO THE UNITED STATES words, migration to the United States for reasons of employment and wages may also reflect the relative attractiveness of job opportunities there.9 A work by Blejer, Johnson, and Porzecanski purports to provide an empirical foundation for the conclusion that unemploymentin Mexico is the principal determinant of migrationto the United States.6 0 Their econo- metric study concluded that a significant statistical relationship existed between unemploymentin Mexico and migration to the UnitedStates over the period 1960-75. Separate regressionswere estimated for legal,illegal, and total migration. The Mexican rate of unemployment was the single most powerful "explanatory" variable. The inclusion of the ratio of Mexican-to-U.S.unemploymentor of relative wages in the two countries did little to the explanatory power of Mexican unemployment alone. Although the proportion of the explainedvariancein the three measuresof migrationwas impressively high, one cannot help but view the resultswith some degree of skepticismbecause of the dubious quality of the informa- tion used for estimating the size of migration flows and for deriving a measure of Mexican unemployment. The data on legal migration, of course, represent actual counts, but they would not seem to be an appropriate measure of the response to Mexican labor market conditions in the year of migration. In the face of the new constraints on legal migration to the United States introduced by the 1965 amendments to the Immigration and Naturalization Act that became effective in 1968, the backlog of immigration applications quicklymounted. Thus, immigration in any given year represents the termination of an application process begun several years earlier. As a measure of illegal immigration,the authors used the INS figureson the number of deportable Mexicans apprehended. Implicit in the use of these figuresis the existenceof a constant relationshipbetween the number of detentions and the number of illegal migrants, something for which no empirical verificationexists. This procedure makes no allowance for the vagaries of the apprehension process. It is widely acknowledged that the INS reacted to the migration flows with different degrees of vigor and effectiveness during the studiedperiod, perhaps respondingmore to unem- ployment rates and political pressures in the United States than to condi- tions in Mexico. Finally, the unemployment data used for Mexico are not measured unemploymentbut represent a residualderivedfrom aggregateoutput and estimated labor force measures. Employment growth is estimated from sectoral growth of output under an assumption of a constant ratio of output to labor inputs. Intercensalprojectionsprovide the estimatesof the size of the labor force, and unemploymentis the differencebetween this and the employment estimate.  CAUSAL FACTORS 193 In the absenceof direct measuresof the labor force and its components, one can sympathize with the need to improvise.However, the questions that surround the correspondenceof the derived measuresto the actual values make it difficultto giveheavyweightto the findingsof this imagina- tive work. In any case,the events of the decadie of the 1970s would appear to have provided a counterexample. As has been shown in an earlier chapter, open unemployment rates appear to have been in a declining trend; at the same time, the rate of illegalmigrationis held to have been on the increase. Two more recent attempts to identify the determinants of undoc- umented migration to the United States have also concluded that condi- tions in Mexico, or push factors, seem to dominate.Walter Fogelemploys multiple regressionanalysis,with the number of apprehensions(inthe role of a proxy for the flow of migrants) as the d&-pendent 6 ' Several variable. regressions are run on the size of the labor force in Mexico, the average annual unemploymentrate in the United States,the annual averagemanu- facturing wage in the United States, the annual average manufacturing wage in Mexico, and time as independentvariables.The data were drawn from the years 1965 to 1977. Fogel interprets the results as showing that the U.S. economic variables have had relatively little influence on the migrant flow. The most important variable proves to be the growth of the Mexican labor force; this variable accounted for 98 percent of the changes in annual migration. Unfortunately, seriousreservationsremain r egarding the significance of Fogel's findings.One arises from the adequacy of the use of apprehensions as a proxy for successfulmigration. Another stems from the labor force measure that is employed. Since there are no actual counts of the labor force for the period studied, its size is simply a projected figure based on 1960 and 1970 census figures and thus is largely determinedby estimated changes in the size of the population of working age. Therefore, a larger Mexican population is associatedwith a larger flow of migrants, a finding that does not seem to be particularlyilluminat:.ng. The use of the Mexican average wage in manufacturing also may not be fully appropriate. That datum is drawn from a nonrepresentativesample of important manufac- turing firms and may or may not represent a suitable proxy for the wage opportunities faced by the migrants in Mexico. The second study citing push factors, by Alberto Davila, is intended to identify the determinants of changes in migration flowsover time, not the magnitude of such flows. 62 Davila's study covers the period 1973-82. His model includes proxies for both pull and push forces. The former include the demand for labor in the United States as representedby U.S. employ- ment in agriculture, manufacturing, and construction in the four states bordering on Mexico. The push factor purports to measure the demand 194 MIGRATION TO THE UNITED STATES for labor in Mexico. For this purpose, he adopts the ratio of employed individuals in the 20-34 age group to the population of this group in the three major metropolitan areas of MexicoCity, Guadalajara, and Monter- rey. Changing conditionsin these areas are held to be indicativeof employ- ment conditions generally. Finally, Davila examines the impact of the Mexican currency devalua- tion of 1976 as an explanatory variable. Multiple regression techniques are applied to twenty-four quarterly observations, and the estimates re- ported are in double-logfunctional form. Most of the elasticitycoefficients associated with the pull variablesare positive, as expected,although only those for U.S. agricultural employment are statistically significant.The sign of the push variable, the proportion of the population aged 20-34 employed in Mexico, is negative and the coefficientis statisticallysignifi- cant. The magnitude of the coefficientsleads Davila to conclude that migration is much more sensitiveto this proportion than it is to the pull factors. The 1976 devaluation,which had the effectof increasingthe U.S. wage defined in terms of Mexican pesos relativeto wages in Mexico, is found to have the expectedpositiveand significantimmediateinfluenceon the flow, although its influencedeclined over time. Again, the robustness of Davila's conclusionsdepend on the appropri- ateness of the data employed.The use of apprehensionsdata as measures of the flow is subject to the usual reservations. Of the independentvari- ables used, the one most likelyto be open to questionis the push factor, the ratio of employed to total population in the 20-34 age group. It would appear that, as a proxy for general employmentconditions, this ratio is inappropriate since it contains an inherent bias in the direction of "de- terioration." This is the casebecauseof the very substantialincreasein the proportion of the population of both sexes in the 20-24 year age group enrolled in school. The proportion increased by about 50 percentover this interval. Its effect is to build in a declining ratio of the employed to the population. Unlessone assumesthat enrollment in schoolis merely a form of disguisedunemployment,this ratio mayprove misleading.If Davilahad used the proportion of the nonschool population that was employed,he might have emerged with strikingly different results; this ratio either increased or remained unchanged in the three regions considered.In view of these considerations, one cannot place much confidencein the validity of his conclusions. Although no one familiar with labor market conditions faced by many Mexicans (especiallythose in rural areas) would deny that employment and income opportunities there leave much to be desired, there is ample ground for questioning whether migration to the United States is moti- vated primarily by poverty and unemployment.Indeed, an interpretation  CAUSAL FACTORS 195 that attributes migration solely to either push or pull forces is likely to provide an oversimplifiedcharacterization of the process. The CENIET household survey, for example, inquired into the employ- ment status of the apprehended and returned migrants prior to their departure for the United States. Only 3.1 percent of the migrants did not work in the month preceding their departure and were activelyseeking employment. The interpretation of this datum must be undertaken with care, however, sinceit could obscure the employmentprospects in Mexico for some of those who reported themselvesas employedduringthe month preceding their migration. If recent agricultural jobs had just ended be- cause of seasonal factors, and if the workers who choseto migrate had not done so, some may have been faced with a prospect of open unemploy- ment. A significantportion of the migrants, one out of five,were reported not to have been employed. Part of this group, of course, was made up of the openly unemployed, but the bulk did not activelyseek employment. 63 Whether or not this failure to look for work :s indicativeof discouraged unemploymentcannot be determined from this survey's reported findings. It is instructive to observethat the border surveyof 1977 reported that 26 percent of the apprehended migrants, exclusivcof those who had spent all of 1977 in the United States,had not held any j Db in Mexico during 1977. In other words, they chose to seek their first job for 1977 in the United States.' Again, the interpretation of this datum is made difficult by the absence of additional information. Traditionally,however, there is a very small amount of long-term unemployment reported in Mexico, and the migrant population is relatively young. Rather than being discouraged unemployed, migrants who did not work in Mexico during 1977 may either have departed from Mexico early in the year or be new entrants to the labor force. One factor contributingto the 7lonemployed status of this group is the very low labor force participation rate in Mexico of female migrants; female labor force members apparently are underrepresented among the female migrant workers. 65 If push factors were the dominant determinants of migration to the United States, one might expect to see that the least advantaged members of the labor force would be heavily representedin the migrant flow.This might be evidenced in a larger migration from the poorest states of the country or in higher rates of migration among the poor within communi- ties. However, the available information about the origins or migrants does not support these expectations. Some oirthe poorest states in the republic, such as Chiapas or Oaxaca, are very minor sourcesof migrants. The relatively unchangingimportance of the various Mexican states over time as sources of migrants suggests that factors other than a precarious position in the labor market may be playinga significantdeterminingrole. 196 MIGRATION TO THE UNITED STATES Even within states, migrants do not necessarily originate in those com- munities with the poorest employment prospects. El Bajio, in the state of Guanajuato, is a relatively rich agricultural region of commercial farm production that is fairly heavily capitalized. There is generous use of purchased and labor inputs per hectare, and it is widely irrigated. Several industrialized urban areas are nearby and offer possibilities for temporary off-farm employment. Yet the rural population of the region is heavily engaged in off-farm employment away from home, with the United States 66 dominating all other migration destinations. Other rural areas with income levels half those of El Bajio send no one to the United States. Similarly, within the sending communities, the characteristics of mi- grants do not suggest a selective process heavily weighted by negative considerations. Various community studies that have observed the migra- tion phenomenon have noted that it is not the poorest households that 67 provide the migrants to the United States. Nor is it the landless who dominate the flow. Ejidatarios, small landholders, and sharecroppers represented half of the CENIET border sample of migrants with agricultural 6 antecedents in Mexico. " In his study of Villa Guerrero, Shadow found that the incidence of migrants from the various landholding statuses was almost identical.69 Raymond Wiest, on the other hand, found landless laborers greatly underrepresented in the migrant flow from Acuitzio, Michoacan.7 0 Whereas they made up 43 percent of the total male labor force, they accounted for only 22 percent of the migrants. The incidence of migration was highest among the minifundistas, the small private farmers. A similar finding is attributed to Cornelius's study of Los Altos de Jalisco by Roberts.7 " United States-bound migrants were not the landless but the small private farmers and ejidatarios. That the poorest elements of rural Mexico are not a significant part of the northbound flow should not be surprising. After all, there are out-of- pocket costs associated with migration and the risk of failing to breach the frontier. Especially for first timers, the risks are likely to appear higher. One way of minimizing the risk of being caught is to engage the services of a "coyote," one who will smuggle the illegal alien across the border in return for a fee. According to the first CENIET border survey, about 20 precent of the apprehended migrants had employed the services of a coyote. Although these migrants were not classified by the number of crossings attempted, they were more heavily represented in those Mexican regions from which the largest proportions of first timers were to be found.7 2 The price of the service varied widely from less than US$50 to over US$300, for an average of US$175. The destination of the illegal migrant also figured into the amount. The price for being smuggled into California exceeded that for Texas, US$180 to US$150, respectively. The  CAUSAL FACTORS 197 differential may reflect a greater difficulty of evading the Border Patrol along the California border, the higher prospectiveearningsobtainable in that state, or both. Nevertheless, this sum represents a very significant outlay, which would be beyond the resources of the poorest rural people. Almost one-third of the apprehended migrants reported having bor- rowed to finance their trip. On average, the debt incurred amounted to US$97. 73 Migrants originating in the border areas borrowed the least, as one might reasonably expect. The largest sums borrowed were reported, not from the region most distant from the U.S.border, but from the central plateau. It has had a long-establishedtradition of migration and appar- ently also shows the highestlevel of earnings realizedin the UnitedStates. In this region, CENIET'S household survey found a significantlyhigherthan average proportion (42 percent) of borrowers in its sample; an average of $130 was borrowed. 7 4 Sincean individual's borrowing capacity is usually at least partly determined by the income or assets (or both) of his family, the poorest elementswere likelyto be the highest-riskcandidates for loans and therefore the most likely to be excluded. Thus, one might expect to finda dichotomy between those who opt for migration to the UnitedStates and those who limit their migration to internal destinations. Since the latter may involvesmaller cash outlaysand also a smallerrisk of total loss, one might expect to find the poorer residentsrnore heavilyrepresented in the domesticmigration stream or in local off-farm employment,while the less poor choose to go to the United States. Clearly, the financial returns to a successfulstay in the United States exceed those availablefrom the existing alternativesin Mexico. For exam- ple, in CENIET'S border survey, the apprehended migrants who had been employed in Mexico during 1977 reported average daily earnings in agriculture of Mex$65.7, and Mex$95 in construction and other activi- ties,for an averageof Mex$79.2. 7 5 In contrast, the legal minimumwage in the United States in 1977 was US$2.60 an hour, or US$20.80 for an eight-hour day. At the exchange rate then current (Mex$23 to US$1),this translates into Mex$480. Put another way, a m:grant employedat the U.S. legal minimum wage could earn in one day tie equivalent of a six-day week's earnings in Mexico. For agricultural workers it would be the equivalent of 7.3 days of Mexican wages. Net earnings after deductions for social security and incometaxes would reduce the size of the premium somewhat but would still leave it at a substantial levelof fiveto six times the Mexican rate of earnings. Actual earnings of migrants are available only in a preliminary form, but even these are instructive. The CENIET household survey collected earnings information as given by the returned rmigrants-at hourly, daily, or weeklyrates-without attempting to reduce them to a singleconsistent 198 MIGRATION TO THE UNITED STATES basis. Of those reporting hourly earnings, three-fourths earned less than US$4 and only 2 percent earned more than US$8. Nearly 60 percent of those reporting daily earnings earned less than US$20; most of the rest earned between US$20 and US$50. 76 Finally, half of the respondents earned less than US$100a week,while another 40 percent earned between US$100 and US$200.' 7 However, these rates of earning refer to trips completed in the past and are not time-specific.Nevertheless, although most earnings did not depart much from the 1977 legal minimum wage, they clearly representedlarge premiums over the rates of earnings current in Mexico in 1978. It is understandable that prospective wages in the United States would have a strong attraction for young mobile workers. Perhapsthe most persuasiveevidencethat migrationto the United States is not undertaken by the neediest is to be found in the way in which the accumulated earnings are spent. While some proportion may surely go to finance current consumption, substantial sums are destined for other purposes.78 Virtually all who have observed the impact of these earnings at the community levelimply that they are used more as discretionary rather than subsistenceincomes.The most common expendituresseemsto be on home improvements,ranging from the addition of rooms or installation of plumbing and flooring materials to the acquisition of furnishings and appliances. Returning migrants frequently arrive home laden with cloth- ing or appliances purchased in the United States. Other common uses to which savings are put are the purchase of land, the financing of an independent business venture, or the purchase of trucks, automobiles, or tools relating to a trade. Visibledifferencesin material levelsof well-being can thus be observed in the villages between U.S. migrant and other households. 79 Household Decisionmaking As has been indicated, the motivations underlying migration can be discussedin terms of push and pull factors. While this approach may have the virtue of simplicity, it also is likely to do violence to reality. The decision to migrate to the United States is undoubtedly not based on only one or the other of these two forces. Indeed, both push and pull forcesare frequently at work simultaneously.Amore fruitful line of analysis is likely to result from a model that considers migrationto the United Statesas only one of severaloptions facedby Mexican workers. The choiceamong those options would be governed by the perceived costs and benefits and the degree of risk aversion characterizing the individuals and their house- holds. Kenneth Roberts followsthis approach. He developsa model of migra-  CAUSAL FACTORS 199 tion in which the rural household is the decisionmakingunit, pursuing the objective, not necessarilyof income maximization, but rather of a stable minimum level of income designed to enisure the survival of the household. 8 0 In this view, the household takes stock of its manpower resources and the labor requirements of cultivating its own land and decides how much labor time can be spared for off-farm wage employ- ment. By allocating family labor over different employments,the house- hold can reduce the risk posed to its survival by unexpectedvariations in income from any one source. An added incentive for realizing cash earn- ings from off-farm employment is that it allows for household participa- tion in a network of reciprocal relationships,a systemdesignedto provide for unforeseen contingencies through loans. 8 ' While migration to the United States providesone way of earning the cash required to participate in this network, it is by no means the only one. The variety of responsesof households to the range of choicesavailable to them has been well documented. Oaxaca is one of the poorest states of Mexico, and one in which great relianceis placed on off-farm earnings. It ranks third as a state of origin for Mexican internal migration; over half of the flow is destined for Mexico City alone. Yet very few undocumented workers in the United States are from Oaxaca.Nor can distance be the sole determining factor in the decision of Oaxaoceiosto shun the U.S. labor market. Other poor communitiesmuch closer to the northern border are equally averseto choosingthe UnitedStates as a destination for temporary migration.12 The existence of an element of choice is illustrated by the experience of at least one sample of temporary migrants to the United States. Cornelius reports that 45 percent of the individuals who had illegallycrossed into the United States had worked in Mexican communi- ties other than their own and thus were farailiar with local alternative destinations.8 3 On the basis of his comparison of four contrasting zones, Roberts concludes that the key factor determining the incidence of off-farm em- ployment is the degreeof penetration of capitalistagriculturalproduction 4 By capitalist penetration, he means the degree to which the in the zone." local agricultural sector is integrated into markets for agricultural inputs and outputs and for consumer goods. Penetration appears to imply a degree of technological "modernization" that proves to be labor- displacing. The resulting labor surplus can be funneled into off-farm employment. Furthermore, the increased dependence on markets for purchased agriculturalinputs and consumergoods is offeredas a sourceof pressure that necessitatesthe search for off-farm employment." 5 Although the changing mode of production that releases labor may provide sufficientground for explainingthe incidenceof off-farm employ- 200 MIGRATION TO THE UNITED STATES ment, it is not adequate for predictingthe choiceamong work alternatives. Only one of Roberts's four zoneswas characterizedby a significantflow of workers to the United States. This was El Bajioin the state of Guanajuato, a region that has long been a source of migrants to the United States. Within that region, Roberts detected only one factor that clearly distin- guished householdswith U.S.migrant membersfrom those with none: the size of the household, or, more precisely,the number of adult males in the household. The relation between size and the incidence of international migration was direct. This is interpreted by Roberts as indicating that a large household is better able to assume the risk of sending one of its members to the United States. Given the fact that El Bajio is a relatively prosperous agricultural zone, Roberts findssupport forhis contention that "higher levels of familyincome encourage extended family relationships which permit the riskier alternative of U.S. migration.""86 While Roberts's conclusions seem plausible enough, it would appear premature to raise them to the status of generalizations. If the size and income of a household were the determining factors in U.S. migration, then it should be possible to observe them at work in other rural states. Most states have farm units that are run by households ranging in size,and in corresponding income, from small to large. Yet, if one observes the heavy concentration of U.S. migrants in only a handful of states, one is compelled to look further for variables that may affect a household's calculation of the expected income derivablefrom migration to the United States. The "Pull" of the United States Corneliusdisposesof the question of why the UnitedStates is chosen as a destination in a simple and direct manner. His sampleof migrants chose the United States because jobs were deemed easier to find there than in Mexico and becauseof the higher earnings that were obtainable there." 7 In rephrasing this response, one could state that the migrants' expected income from employmentin the UnitedStates exceededthat from alterna- tives present in Mexico. Again, however, this failsas a satisfactorygeneral explanation of why the UnitedStates was chosen. If jobs were more easily obtainable and expectedincomes higher in the United Statesfor those that had successfullymigrated there, why would this same perception not be shared and acted upon by Mexicans throughout the country? In fact, both Corneliusand Roberts mention what may be a key factor, even though they do not exploit fully its significancefor their migration models. Allof the literature points out the "long tradition" of migrationto the United States from certain areas in Mexico.88 Corneliusand Roberts, in  CAUSAL FACTORS 201 addition, emphasize the importance of networks and extended family relationships that can be built and reliedupon t:o insure survivalin the face of some unexpected disaster or interruption to income. In addition, however, tradition and networks can be interpreted as significantlyaffect- ing the perceivedcosts and benefitsof migration to the UnitedStates. That migration continues to originate in certain highly concentratedlocalitiesis at least partially explainablein terms of the reduction in information costs and, therefore, in the degree of risk incurred by a prospectivemigrant. "Tradition" implies the accumulation of a fund of information available locally and at no cost through household and communitynetworks. The information can include such vital intelligenceas how best to elude the migra,the Mexican expressionfor the Border Patrol; the relativeprospects of employment at different destinations in the United States; the mechan- ics of job search; and a whole catalogue of do's and don'ts that are critical to the successfulrealization of a migrant's goals. In other ways, too, the longstandingtradition of migrationto the United States from certain regions stands to reduce CDsts of migration. As leak- ages have occurred from the migration stream into permanent residencein the United States, the network of household relationshipsextending from communitiesin Mexico to communitiesin the UJnited Stateshas expanded. Thus, information flows about employment possibilities are facilitated, and the degree of uncertainty is reduced. Relatives or fellow-villagers in the United States may provide a temporary free refuge until a job is secured and thus reduce the investment the migrant himself must make in his employment quest. They may also represent a source of credit for the financing of other migration costs. Thus, the lexistence of an established Mexican community in the Chicago area would explain its continuing importance as a destination of temporary migrants in spite of its consider- able removal from the border. Similar commriunitiesin the U.S. border states offer these same advantages in addition to that of proximity. The greater and more prolonged the past flow of Mexicans from a particular area to the United States, the lower become the risks associated with migration of subsequent generations. The combination of greater information and reduced risk serves to increase the expected returns to U.S.-bound rnigration relative to those from internal migration. The converse applies for those regions in Mexico that did not serve as important sources of migrants to the United States in times past, when such migration was either legal or easier. The relative scarcity of information and lack of a stateside family or village network increase the risk of failure considerably and, consequently, reduce the expected returns to migration relative to returrs from internal migration. Areas from which internal migration has become traditional have also 202 MIGRATION TO THE UNITED STATES built networks of information and assistance in both the source and destination communities. These network relationships raise the expected returns to internal migration relative to those from moves to the United States. Perpetuation of past patterns of behavior is thus strengthened.A change can probably occur only gradually, as individualswillingto accept high risks in order to migrate to a destination outside the established network do so successfullyand establish a beachhead. As this foothold expands, the relative attractivenessof that destination increases, and the flow of migrants could be expectedto increase.Viewedin these terms, the persistence of migration from a rather small area of Mexico can be explained as the workers' rational response to perceived returns from either international or internal migration. It also suggests that the past pattern will extend into the future with only gradual modifications. The fact that the traditional pattern of migrationis likelyto persist tells us nothing about the level at which that migration will be sustained.The number of workers who will attempt migration to the United States will obviously be a function of developmentson both sides of the border. To the extent that the process of economicgrowth is more rapid in Mexico and is reflected in rising wage levels, the relative attractiveness of the United States will tend to decline. Giventhe very large absolute differen- tials that currently exist, however, U.S.wages are likelyto be perceivedas relatively attractive for some time to come. One study has developedan estimate of real wage differencesbetween the various minimumwage zones of Mexico and the southwesternUnited States.The real wage estimatestake into accountdifferencesin purchasing power among areas. Between 1969 and 1975, the differential declined from 5.56:1 to 4.12:1 on average.The Mexican devaluation of 1976 led to an increase in the differential to 7.43:1 and 7:1 in 1977 and 1978, respectively,a change that might be expectedto increasethe attractiveness of temporary migration. 89 This same study alsoundertook a broader comparison of certain attrac- tions in both countries that could be expected to influence permanent migration. The authors developed a socioeconomic opportunity index that included the availability of social servicessuch as education, health, housing, and so on, for both U.S. and Mexican districts.Within this index, all sixteen of the U.S. regions includedwere at least 1.75 standard devia- tions above the compositemean for both Mexican and U.S.districts. Only three Mexican districts were comparably rated and thus competitivewith those in the United States. They were the three largest metropolitan regions of the country: Mexico City, Guadalajara, and Monterrey-the chief destinations of internal migrants. Incidentally, these large urban conglomerations have been only minor sources of U.S. migrants, which confirms that they compete with the United States as destinations.  CAUSAL FACTORS 203 Conditions on the U.S. side of the border al o help determine the volume of migration. Over the past few years, the United States has become increasingly concerned about the implicatiomn of recent levels of migration for the U.S. labor force. This concern also extends to the social conse- quences of greater "permanent" migration. With the rate of unemploy- ment in the United States approaching 10 percent at the height of the 1981-82 recession, the clamor that "something be done to stem the tide" became more strident. Both the Carter and Reagan administrations have advanced a series of proposals to reduce the flow of illegal migration; the U.S. Congress currently has before it a bill intended to deal with the issue. The ultimate form and effectiveness of any measures that may be adopted cannot be foreseen nor can they be taken into consideration for purposes of this discussion. Nevertheless, it may be presumed that an effective demand for Mexican workers will continue to exist in the United States. On the basis of past experience, it is not unreasonable to expect that more legal strictures against migration might not be stringently enforced if the unemployment level declined or "labor shortages" developed. Indeed, given the demographic profile of the U.S. population and recent birth rates, it is quite possible that the next two decades may see a decline in the growth rate of the domestic labor force. The current impressions of a labor "surplus" may then convert to one of 'shortage." Clark Reynolds has estimated the supply of, and demand for, labor in the United States for 1990 and 2000. Given an annual rate of growth in the GNP of 3 percent, he adopts different assumptions concerning the rate of growth in productiv- 90 ity and labor force participation rates. Assuming a 1.38 percent annual increase in productivity (the average rate for the period 1960-77) and stable labor force participation rates, a sizable labor deficit of 5 million is forecast for the year 2000. (These estimates allow for an annual immigra- tion of 400,000.) Reynolds holds that the shortfall of supply will occur in those jobs at the bottom of the occupational hierarchy. As minority and other disadvantaged groups improve their educational and skill levels, fewer such workers will be available to fill these positions. Thus, a large and growing demand for unskilled Mexican labor will be evidenced in the United States in years to come. Reynolds's projection of a shortage of unskilled labor echoes the findings of a study undertaken by the NationaL Planning Association in the middle 1970s. That study concluded that by the latter part of the 1980s, the supply of people available to fill low-skill occupations will have grown more slowly than the demand. The absolute decline in the available supply foreseen for the 1980s is predicated on the declining size of the teenage population-a primary supplier of such labor-and on the upgrading of skills of disadvantaged groups, the other pri:icipal labor source.91 To be sure, adjustments to this perceived shortage can take many forms, includ- 204 MIGRATION TO THE UNITED STATES ing substitution and changes in relative wages. The authors of the study also foresaw a surplus of labor for the more highlyskilled jobs. Therefore, a realignment of wages across occupational groups may bring the various subsectors into a state of equilibriumwithout an allowance for immigra- tion. It would not be surprising,however, to find the low-wageemployers resisting market pressures on wages and opting for the employment of Mexican migrant labor. The Impact of Migration on the Mexican Labor Market The option of migrating to the United States in search of employment has frequently been likened to a safetyvalve. Impliedis a beliefthat, in the absence of migration, conditions in the Mexican labor market would deteriorate significantly,with greater open unemploymentand underem- ployment, lower wage levels, or both. Conversely,the persistenceof the migration phenomenon would be expectedto raisewages abovethe levels that would otherwisehaveheld. 92 Asfar as I can determine,however, there has been no attempt to quantify the changesthat would be associatedwith different levels of migration-and with good reason. In order to estimate the impact on the wage level, one would need to delineatethe spatial and occupational boundaries of the relevant labor market and would require measures of the price elasticities of supply and demand for labor. In addition, it would be necessaryto know the proportion of the migrants who would have beenin the labor force in Mexico had they not migrated. Unfortunately, all of these bits of essentialinformation remain in the realm of the unknown. Therefore, any estimate of the wage effectsof migration must be largely speculative in nature, dependent on the assignment of rather arbitrary valuesto the parameters of the market equation. Although little confidencecan be placed in such estimates,they may at least establish the general order of magnitude of the wage effect of migration. The question examined in this sectionis, by how much are wages in the Mexican labor market likely to be affected by illegal migration to the United States?The first issueto be addressed is the impact that migration has on the domestic supply of labor. The CENIET household survey esti- mated that somethingon the order of 500,000 Mexicanswere temporarily in the United States at any one time. If my estimate of the Mexican labor force in 1980, amounting to about 20 million were reasonably correct, these temporary migrants amounted to only 2.5 percent of the total. Migrants are not potential competitors of all other members of the labor force, however, sincethey lack the skillsappropriate to many occupations. Furthermore, since migrants tend to originatein relativelyfew areas of the  IMPACT ON MEXICIAN LABOR MARKET 205 country, the effects on the Mexican labor force in these originating areas might prove to be greater than in those regiens providing few migrants. For example, CENIET'S survey indicated that migrants represented 6.5 93 percent of the labor force of the eight major source states. Thus, if the migration option were suddenly to be removed, and if all the suspended migrants were to remain in their own local labor markets, the wage effects in those markets could be quite substantial. In such an event, it would seem reasonable to assume that these migration-prone individuals would substi- tute internal for international migration; thus, the wage effects would be spread more evenly over the national labor rriarket as a whole.9 4 (For the sake of convenience, I will assume that the migrants would, in fact, be greatly dispersed internally.) It has been noted that migrants are competitive in only some occupa- tions. The occupational distribution of the national labor force, as esti- mated by the national household labor force survey in 1979, is presented in table 6-5. It also offers a crude estimate of the proportion of each occupational category with which the migrants might be competitive. Given the low educational levels of most migrants, it is safe to assume that they would be negligible participants in the markets for professional, managerial, and administrative personnel. I arbitrarily assume that they Table 6-5. Percentage Distribution of the Labor Force by Occupational Group, and the Percentage of Each Group Subject to Competition from Migrants, 1979 Percentage of Percentage Percentage total distribution subject to employment of competition subject to Occupational labor from competition group force migrants from migrants Professionaland technical workers 7.6 0 0 Managers 2.7 0 0 Administrativepersonnel 8.6 0 0 Salespeople 11.0 33.3 3.7 Workers in servicesand operators of motor vehicles 16.0 50.0 8.0 Agriculturallaborers 28.6 28.6 Nonagricultural laborers 25.4 66.6 16.9 99.9 57.2 Source:Secretaria de Programaci6ny Presupuesto,Encuesta continua sobre ocupaci6n, vol. 7, 1st quarter 1979.(Mexico, D.F., February 1980). 206 MIGRATION TO THE UNITED STATES would be competitive in one-third of the sales and one-half of the service employments. Finally, migrants are assumed to be competitive in all agricultural and two-thirds of the nonagriculturallaborer jobs. The areas of effectivecompetition thus add up to approximately 57.2 percent of all occupations. According to the 1980 census,the labor market from which migrants are withdrawn would have measured some 11.4 millionpartici- pants, and migration would result in a reduction of 4.4 percent in that force. For purposesof this estimatingprocedure, a 5 percent reduction will be adopted as a rough approximation of the likely actual change. Assigningvalues to the price elasticitiesof demand for, and supply of, labor is not as easily done. Therefore,an expedientmethod of calculating the wage effect is to use differentvalues for these elasticitiesdrawn from a range within which the true valuesare likelyto occur. For the elasticityof demand, three values are adopted, ranging between - 0.5 and - 1.5, and three for supply between 0.2 and 0.7. Each of the three demand elasticities is combined with each of the threesupply elasticities to yielda total of nine estimatesof the change in wages associated with a 5 percent reduction in the supply of labor. The percentage changein the wage is calculated using the formula w = h E- m where w is the percentage change in the wage, h the percentage change in the supply of labor (representative of a horizontal shift in the supply schedule),E is the price elasticity of the supply labor, and ' is the price elasticity of the demand for labor.95 Table 6-6 presents the results of the various computations. The esti- mated changes in the wage range from a low of 2.27 percent, when the highestabsolute values are assignedthe two elasticities,to 7.14 percent at the other extreme, when both elasticitiesare assumedto be at their lowest values. Most of the combinationsof the elasticitiesyieldestimated changes on the order of 3 to 4 percent. This interval would appear to be based on elasticities most likely to approximate the actual ones. The recent esti- mated flow of migrants would thus implya wage levelin low-skilloccupa- tions 3 to 4 percent higher than would otherwise prevail. Conversely,a sudden suspensionof migration would have a once-and- for-all negative impact of the same magnitude. A decline of this order is clearly much smaller than that which most students of the migration phenomenon seem to assume when they contemplate the possibilitythat migration might be impededor halted. To place a declinein the domestic wage level of this magnitude in perspective,it should be pointed out that it is approximately equal to the averageannual rate of increasein real wages  IMPACT ON MEXICAN LABOR MARKET 207 Table 6-6. Percentage Change in the Wage Level in Labor Markets Subject to Competition from Migrants, Give,i Different Price Elasticities of the Supply of, and Demand for, Labor Elasticity Elasticity of demand of supply -0.5 - 1.0 - .S 0.2 7.14 4.17 2.94 0.5 5.00 3.33 2.50 0.7 4.17 2.94 2.27 of workers in the smallest enterprises in the industrial, service, and com- mercial sectors between 1960 and 1975.96In other words, the reintegra- tion of the migrants into the domestic labor market during the 1970s would have implied a fall in wages about equal to a single year's rate of advance; real wages would thus have been left unchanged for about a year. This estimate of the size of the change in wages that might follow a suspension of opportunities for migration to the United States was predi- cated on a wide dispersal of the would-be migrants over the entire country. To the extent that the labor force in some of thc traditional source areas is not dispersed widely but remains concentrated in local or regional labor markets, the wage effect in those markets might be expected to be greater. Such a decline in real wages could be expectecl to spur out-migration to areas boasting more favorable employment conditions, however, and the local depression of wages would thus be moderated over time. The estimated wage effect was predicated on the assumption that all of the U.S.-bound migrants would enter the labor market in competition with other workers, but in fact this might not be so. F7orexample, if migrants to the United States were to have land of their owrn that was left uncultivated during their absence, loss of the migration opportunity could lead some simply to self-employment rather than to entry into the wage labor mar- ket. Indeed, a substantial number of migrants apparently do leave their plots uncultivated. Cornelius's sample found that 19 percent of the mi- grants who were ejidatarios or small landholders left their plots unculti- vated, a surprising revelation in view of the household and community 97 labor surplus conditions that are presumed to characterize rural areas. He also cites work by Richard Mines that suggests that productive re- sources are not fully exploited in areas given to heavy migration to the United States as the area under cultivation declined. The estimates of the wage effect of migration offered above were based on the assumption held by CENIET that migration is largely a circular 208 MIGRATION TO THE UNITED STATES phenomenon with insignificantpermanent leakage out of the Mexican labor force. If a permanent leakage does occur, as the discussionearlier in this chapter suggests, then the above estimates understate the positive impact that migration has had on wage levels in Mexico. A major differ- ence between temporary (or circular) and permanent migrationis that the former has only a once-and-for-all impact while the latter impact is exponential. For illustrative purposes, let us assume that the elasticitiesof supply of, and demand for, labor are 0.5 and - 1.0, respectively,values that may be regarded as plausible. Furthermore, out of each cohort of 500,000 mi- grants, 100,000 leak out permanently into the U.S. labor market. Since two cohorts are assumed to migrate to the United States each year, this implies a permanent leakage of 200,000 workers and a constant circulat- ing group of 400,000 in the United States. Under the formula introduced above, the once-and-for-all effect of the withdrawal of the circulating 400,000 workers would then raise wages in the relevant Mexican labor market by 2.57 percent. The first-year impact of the permanent with- drawal of 200,000 workers would be an additional 1.28 percent, for a total increase of 3.85 percent. In each subsequent year, however, the leakage of an additional 200,000 will have a further impact on the wage level that is cumulative. At the end of fiveyears, for example, the annual leakage of 200,000 will have resulted in a wage level 7.3 percent higher than that associated with a constant circulating stock of 400,000. If, instead of a leakage of 100,000, there were one of 200,000 from each cohort of 500,000 going to the United States, the longer-term impact on the Mexican wage level would, of course, be greater. With the same elasticities, the once-and-for-all increase associated with a constant 300,000 stock of circulating workers in the United States would be 1.9 percent, while the first-year impact of the permanent withdrawal of 400,000 workers would be on the order of 2.56 percent. After fiveyears, the Mexican wage level in the relevant market would be 14.5 percent higher than it would have beenhad there been a constant circulatingstock of 300,000 in the United States. As a general rule, the larger the proportion of each cohort of migrants to the United States that remains there, the greater the long-run impact migration will have on the wage levelin Mexico. Obviously, a suspension of migration to the United States, including that of permanent illegal migrants, would alsohave a cumulativeimpact on the Mexican wage level since the domestic supply of labor would expand more rapidly. If an annual leakage of 400,000 workers were halted, the rate of real wage increasein the relevant part of the Mexican labor market would probably fall below 1 percent, givenrates of economicgrowth of the same order of  IMPACT ON MEXICAN LABOR MARKET 209 magnitude as those that held over the 1960-75 interval. If the actual leakage were closerto the 200,000 level,absorption of this number in the Mexican labor market might imply a reducticn in the annual rate of increase in real wages to perhaps 2 percent. Clearly, these estimates of changes in wages associatedwith different levels of circulating and permanent illegal migration should not be taken to represent anything more than illustrative and crude measures.At best, they may offer measuresof a general order of magnitude under plausible assumptions regarding the elasticitiesof supply and demand and the size of migration flows. Implicit in these estimates was the maintenance of sustained and high rates of economicgrowth such as those characterizing the Mexican economy over the past four decades. Naturally, a significant decline in the rate of growth might be expected to produce less favorable conditions in labor markets generally, although the marginal effect of migration on wage levels might not prove to be much different. In addition to the favorable consequencesof migration for wage and employment conditions in Mexican labor markets, there are the conse- quences for the Mexican balance of payments.Two recentestimatesof the approximate total contribution of migrant remittancesto the flow of U.S. dollars to Mexico in the second half of the past decade are surprisingly similar: approximately US$315 million. One of these is derived from Diez-Canedo's study of checkscleared through ;Mexican banks; the other was a product of CENIET'S household survey. More visible,of course, is the impact of migration on the incomelevelof households with migrating members. Given the wide disparity between wage levels in the United States and Mexico, eventemporary employment in the United States makes possible a substant,al improvement in living standards over those that would otherwisebe achievable.Of lesserimpor- tance are likely to be the additions to human capital associated with migration. By and large, migrants are employecd in occupations requiring little skill or training. Sincemost migrants do not remain in their U.S.jobs for more than a few months, they are unlikelyto become candidates for training for more highly skilled jobs. Whether rmigrants return to Mexico with attitudes that may render them more produlctiveis open to question. Corneliusobserves that psychologicalprofilesof migrants revealthey have a greater need for achievement than nonmigiants, and some evidence exists that this trait is enhanced by experiencesin the UnitedStates." Such individualsmight alsobe more likelyto perceivegreater long-run opportu- nities for advancement in the U.S. labor market and, as a result, form a disproportionate share of the permanent leakage from the migration flow. If that were the case, Mexico would suffer a loss of highly motivated producers that it could ill afford.'°° 210 MIGRATION TO THE UNITED STATES Summary This chapter has reviewedthe presentstatus of our knowledgeand understanding of the migration phenomenon of Mexicans to the United States.Whileconsiderable gapsexistin quantitativedata,at leastit canbe said that the severalattemptsto estimateempirically the flowof workers to the UnitedStateshavenarrowedthe rangewithinwhichthe estimated measures lie.In contrastto off-the-cuff estimatesof government officials and others that the UnitedStateswas beinginundated by a flood of Mexicanworkers,themoreseriouseffortsof researchers suggest flowsof muchmore modestproportions.The availableestimatesof gross flows acrossthe northernborderwouldseemto be morereliablethan those of net flows. A proper assessment of the long-runimpactof migration on the Mex- ican labormarketdependsheavilyon the extentof permanent leakageof labor out of the circularmigratoryflowand into the UnitedStates.The cumulative effectsof such a leakage can be considerable. My attemptto assess the impactof circularmigration on wagelevels in the relevant parts of the Mexicanlabor marketsuggests that, whilepositive,it is likelyto have beenquite modest,raisingthe wagelevelby no more than 3 to 4 percent.Eventhis conclusion mustbe treatedwith considerable care,forit is basedon a rather arbitrary,althoughplausible, definition of the "rel- evant labor market" and on arbitraryelasticity coefficients. It islikelythat theU.S.labormarketwillcontinueto appearattractive to Mexicanworkersfor sometimeto come.Thelargewagedisparities that currentlyexistwillprobablysurvive wellintothenextcentury.Indeed,if the U.S.domesticsupplyof laborto low-skill occupations declines during therest of this century,a fasterincreasein wagesfor thesejobsrelativeto the general wage level may well occur; thus, the differentialbetween unskilled wageratesin the UnitedStatesand Mexicomaybe maintained or evenenhanced. to foresee It is difficult thestepsthat maybe takenin theUnitedStatesto stemthe illegalmigrationflowfromabroadand especially fromMexico. Althoughthemeasures ultimately adoptedmayserveto reducetheflow,it is hard to believe that theywillstop it. Aslongas the gainsto prospective migrantsloomlargeand the advantages to U.S.employers of employing Mexicanworkersremain,therewillbe workersand employers willingto bear the increasedriskof detection. Indeed,one mayhypothesize that an attempt to closethe border willgiverise to more "permanent"illegal migrationrelativeto circularsincesuccessfulborder crossersmay be reluctantto test their luckagain.  NOTES 211 Some part of the clamor in the United States that the authorities do something to reduce the flow is surely relatedI to the high levels of unem- ployment that were characteristic of the U.S.labor market during the early 1980s. Should the U.S. economy revert to a state of moderately full employment,the pressures to act on the migration issueor to enforce any new legislation might also subside. Further down the road, if actual shortages of unskilled labor were to materialize in the United States, we might see a de facto relaxation of barriers to inigrationor even an outright encouragementof migration through an orgainized program similar to the bracero program of the mid-centurydecades.It would thus appear that the United States would be unlikelyto severcomrpletely, or for long, the links that bind the two countries' labor markets. Notes 1. Carey McWilliams,North from Mexico: The Sianish-Speaking People of the U.S. (New York: Greenwood Press, 1968), p. 52. 2. U.S. Immigration and Naturalization Service (INs', Annual Report 1974 (Washington, D.C.: Government Printing Office, 1975), pp. 56-58. 3. Barton M. Clark, "Mexican Migration to the Unit-d States," in Mexican Migration, ed. T. Weaver and T. E. Downing (Tucson: University of Arizona, Bureau of Ethnic Research, 1976), p. 52. 4. Ibid., p. 54. 5. Julian Samora, Los Moiados: The Wetback Stcry (South Bend, Ind.: University of Notre Dame Press, 1971), p. 19. 6. Ibid., pp. 44-45. 7. Paul R. Ehrlich, Loy Bilderback, and Anne H. Ehrlich, The Golden Door (New York: Ballantine Books, 1979), pp. 211-14. 8. Samora, Los Mojados, pp. 44-45. 9. INs, Annual Report 1974, p. 10. 10. Carlos H. Zazueta, "Consideraciones acerca de los trabajadores mexicanos in- documentados en los Estados Unidos: Mitos y realidades" (Mexico, D.F.: Secretaria de Trabajo y Previsi6n Social, Centro Nacional de Informaci6n y Estadisticas del Trabajo [cENIETI, March 27, 1979; processed), pp. 18-24. Monthly data on expulsions were not available for 1978 from the INS, but the total for the y ear was. For 1972-77, however, the proportion of the annual total attributable to August of each year varied within a narrow range of 8.2 to 10.2 percent. Zazueta applied the extreme percentage values to the 1978 annual total of apprehensions to obtain a high and low estimate for August of that year, 97,219 and 77,604 respectively. The value cited in :he text is simply the mean of these extreme values. 11. Ibid., pp. 25-31. A similar large discrepancy was reported between the number of expulsions reported by the INS and by CENIET over a three-week interval, October 24- November 13, 1977. Although some conceptual or juridical differences could give rise to a discrepancy between "illegal aliens apprehended" and expulsions, these are not likely to suffice as explanations. An extended and careful discussion of this issue appears in Manuel Garcia y Griego, El volumen de la migraci6nde mexicanosno documentadosa los Estados 212 MIGRATION TO THE UNITED STATES Unidos (nuevas hip6tesis), Estudios no. 4 (Mexico,D.F.: CENIET, 1980), pp. 215-44. A discussion of the conceptual differences of measuresemployed by the INS and their signifi- cance is contained in Charles B. Keely, "Counting the Uncountable: Estimates of Un- documentedAliensin the United States," Populationand DevelopmentReview, vol.3, no. 4 (1977), pp. 473-81. 12. Lesko and Associates,"Final Report: BasicData and GuidanceRequired to Imple- ment a Major Illegal Alien Study during Fiscal Year 1976," Contract no. CO-16-75 (Washington,D.C.: Officeof Planningand Evaluation,U.S.Immigrationand Naturalization Service,1975; processed).The estimateof the total number of illegalalienswas based on the personal opinions of seven "experts" canvassedby Lesko.The estimateof Mexican aliens was derivedfrom an assumedand constant ratio of apprehensionsto entries.Furthermore,it was assumedthat only 2 percent of the successful entrants returned to Mexicoof their own free will. No empiricalfoundations for these proportions were offered. 13. Ideally, the number of illegal aliens in the United States could be derived by an enumeration based on census or other samplingprocedures.One such attempt was made under contract from the INS in 1976byJoseph Reyesand Associates.Thepracticaldifficulties confronting this effort apparently proved overwhelming,however, and the project was abandoned after two years. (Garciay Griego, El volumen de la migraci6n,pp. 31-34.) 14. Clarise Lancaster and Frederick J. Scheuren,"Counting the Uncountable Illegals: Some Initial StatisticalSpeculationsEmploying Capture-RecaptureTechniques,"1977 Pro- ceedingsof the SocialStatisticalSection of theAmericanStatistical Association(Washington, D.C., 1978), pp. 31-34. 15. For a fuller critique of this study, see Ehrlich, Bilderback,and Ehrlich,The Golden Door, pp. 182-90; and Garcia y Griego, El volumen de la migracidn,pp. 45-47. 16. David S. North and Marion F. Houstoun, The Characteristics and Role of Illegal Aliens in the U.S. Labor Market (Washington,D.C.: Linton, 1976), pp. 40-45. 17. David M. Heer, "What is the Annual Net Flow of UndocumentedMexican Immi- grants to the United States?" Demography, vol. 16, no. 3 (1979), pp. 417-23. 18. Ibid. This estimate assumes a 10 percent undercount in the 1970 census and a 70 percent net migrationrate among documentedimmigrants.The sevenestimatesranged from a low of 426,000 to a high of 1,201,000 over the entire five-yearperiod. 19. J. Gregory Robinson, "Estimatingthe Approximate Sizeof the IllegalAlienPopula- tion by the ComparativeTrend Analysisof Age-Specific Death Rates," Demography, vol. 17, no. 2 (1980), pp. 159-76. 20. Howard Goldberg,"Estimates of Emigration from Mexico and IllegalEntry into the UnitedStates, 1960-70 by the ResidualMethod" (Washington,D.C.: Centerfor Population Research, GeorgetownUniversity,n.d.), cited in North and Houstoun, Characteristics and Role of Illegal Aliens, p. 72. 21. World Bank, 1978 World Bank Atlas (Washington,D.C., 1978), p. 8. 22. Garcia y Griego, El volumen de la migraci6n. 23. An additional 162 crossings were reported but proved unusable because certain information was lacking.Only the latestfifteenentriespreviousto the most recent onewere recorded. 24. Juan Diez-Canedo,"La migraci6n indocumentadaa EstadosUnidos; un enfoque," Documento de investigaci6nno. 24 (Mexico,D.F.: Bancode Mexico,July 1980). 25. CENIET, "Analisisde algunosresultados de la primera encuestaa trabajadores mexi- canos no documentados devueltos de los Estados Unidos, octubre 23-noviembre 13 de 1977" (MexicoD.F., n.d.); Carlos H. Zazueta, "Consideracionesacercade los trabajadores mexicanos indocumentados en los Estados Unidos; mitos y realidades" (Mexico,D.F.: CENIET, 1979; processed);CarlosH. Zazueta, "Mexican Workersin the UnitedStates:Some  NOTES 213 Initial Results and MethodologicalConsiderationsof the National Household Survey of Emigration (ENEFNEU)," paper prepared for the Working Group on MexicanMigrants and U.S.Responsibility,Centerfor Philosophyand PublicPolicy, University of Maryland,March 1980; Cesar Zazueta, "Investigaci6nrecientesobre migraci6n mexicanaindocumentadaa los Estados Unidos," (Mexico,D.F.: CENIET, February 19, 1980; processed). 26. Cesar Zazueta, "Investigaci6nreciente sobre migraci6n," p. 79. 27. The dailycommutershold immigrantstatus that is enewableeverysix months as long as they continue to be employed in the United States. At the end of fiscal 1978, these numbered 49,290 (U.S. Senate, Committee on the Jud ciary, 96th Congress,Temporary Worker Programs: Backgroundand Issues[Washington,D.C.: Government PrintingOffice, 19801, p. 82). In addition, seasonal commuters with a legal (H-2) status are admitted, primarily during the peak agricultural season. These workers were estimated to number approximately 8,000 in 1974 (Anna-StinaEricson, "The Impact of Commuters on the Mexican-American Border," in MexicanWorkers in the UnitedStates,ed. GeorgeC. Kiser and Martha W. Kiser [Albuquerque:Universityof New Mexico Press, 1979], p. 215). 28. Cesar Zazueta, "Investigaci6nreciente sobre migraci6n," p. 79. 29. This number is derived as follows. If the average size of each cohort of successful crossers into the United States were approximately500,000 workers, the number of first timers mightrange from 30 to 40 percentof that number,or 160,000-200,000. Of course,if the probability of apprehension for first timers were actually more than three times that of experiencedmigrants,then the absolute number of first timerswould be smaller.Thereis no way of knowingwhat theseprobabilitiesare. Anaverageprobabilityover repeatedsuccessful entriescan be derived from North and Houstoun's data on Mexican migrants (Characteris- tics and Role of Illegal Aliens, pp. 86-87). Their respoiidents reported an averageof 4.5 successful trips to the United Statesand 0.8 apprehensions,for a 0.18 averageprobabilityof apprehension.If the probability of detentionwere three timesgreater for first timersthan for repeaters,this would imply a probability of 0.374 for the former and 0.124 for the latter. 30. CENIET, "Analisis de algunos resultados," p. 41. 31. INS, Annual Report, various years. 32. Consejo Nacional de poblaci6n, Mexico demografico,breviario 1980-81 (Mexico, D.F., 1982), pp. 38-39. 33. Consejo Nacional de Poblaci6n(CONAPO), Secretariade Programaci6ny Presupuesto (srp), and Centro Latinoamericano de Demografia (CELADE), Mexico: estimaciones y proyeccionesde poblaci6n, 1950-2000 (Mexico,D.F., 1982), p. 18. 34. However, the benefits derived from temporary migration, measured in Mexican pesos,should havebeendecliningfrom 1978through the leginningof 1982as a consequence of increasingovervaluation of the peso.Conversely,the sharp devaluationsof 1982 and the deep recessionthat followedin Mexico should have had the opposite effect. 35. Manuel Gamio, Mexican Immigrationto the United States (Chicago: University of Chicago Press, 1930); Robert F. Foerster, The Racial ProblemInvolved in Immigration (Washington,D.C.: GovernmentPrinting Office, 1925). 36. Carlos Zazueta, "Mexican Workers in the United States," p. 50. 37. Diez-Canedo,"La migraci6nindocumentada." Diez-Canedospeculatedthat remit- ters are more likely to be temporary or undocumentedmigrants.Long-term or permanent migrants were held to be more likelyto have their immcdiatefamilieswith them and their financial responsibilitiesto family members in Mexico correspondinglyweakened. The CENIET household survey limited its inquiry to the werk force in the United States that nonethelessmaintainedits customaryresidence in Mexico.Thus, longer-termresidentsin the United States would be less likely to be recorded. 38. INs, Annual Report, 1974, p. 104. 214 MIGRATION TO THE UNITED STATES 39. Ibid., pp. 51-55. 40. CENIET, "Analisisde algunos resultados," p. 25. 41. INS, Annual Report, 1974, pp. 5 1-52. 42. Rural localities are defined by these studies as those with fewer than 20,000 inhabi- tants. 43. CENIET, "Analisis de algunos resultados," pp. 26-34. 44. Carlos Zazueta, "Mexican Workers in the United States," pp. 34-44. 45. Ibid., pp. 41-43. 46. Ibid., pp. 58-59; CENIET,"Analisis de algunos resultados," pp. 35-36; Cesar Zazueta and Fernando Mercado N., "El mercado de trabajo norteamericano y los trabajadores mexicanos: algunos elementos teoricos y empiricos para su discusi6n," paper read at the Mexico-United States Seminar on Undocumented Migration, September 4-6, 1980, Uni- versity of New Mexico, Albuquerque, New Mexico, pp. 20-22. 47. Zazueta and Mercado, "El mercado de trabajo norteamericano," pp. 20-22. 48. George M. Foster, Tzintzuntzan: Mexican Peasants in a Changing World, rev. ed. (New York: Elsevier, 1979), p. 297. 49. Luis M. Fernandez 0. and Maria T. de Fernandez, "Capitalismo y cooperaci6n: el caso de un ejido en Michoacan," in Capitalismo y campesinado en Mexico (Mexico, D.F.: Instituto Nacional de Antropologia e Historia, 1976), p. 216. 50. Wayne A. Cornelius, "Mexican Migration to the United States: Cause, Conseqences, and U.S. Responses" (Cambridge, Mass.: Center for International Studies, Massachusetts Institute of Technology, July 1978; processed), pp. 25-26. 51. CENIET, "Analisis de algunos resultados," pp. 40-46. 52. Cornelius reported that first timers had a much higher rate of apprehension than repeaters ("Mexican Migration to the United States," p. 24). 53. Zazueta and Mercado, "El mercado de trabajo norteamericano," p. 24. 54. North and Houstoun, Characteristics and Role of Illegal Aliens, pp. 84-8 6. 55. Robert D. Shadow, "Differential Out-Migration: A Comparison of Internal and International Migration from Villa Guerrero, Jalisco (Mexico)," in Migration Across Fron- tiers: Mexico and the United States, ed. Fernando Camara and Robert V. Kemper (Albany: State University of New York at Albany, 1979), pp. 71-72. 56. Cornelius, "Mexican Migration to the United States," p. 36. 57. Ibid., pp. 38, 29. 58. John S. Evans and Dilmus D. James, "Conditions of Employment and Income Dis- tribution in Mexico as Incentives for Mexican Migration to the United States: Prospects to the End of the Century," International Migration Review, vol. 13, no. 1 (1979), p. 5. S9. Indeed, even Cornelius, perhaps unwittingly, illustrates the role of pull forces in his discussion of short-term economic push factors: "But the recent economic development which has undoubtedly had the greatest impact on migration to the United States was the devaluation of the Mexican peso in September, 1976-the first such devaluation in 22 years. Before September 1, 1976, if a migrant working in the U.S. sent one U.S. dollar to his family in Mexico, it brought 12.50 pesos. With U.S. dollars now yielding nearly twice as much as before, rapid inflation, and a high unemployment rate in Mexico, a large increase in illegal migration during 1977 was inevitable." ("Mexican Migration to the United States," pp. 36-38.) One would have thought that a sudden increase in the attractiveness of U.S. employment would constitute a pull rather than push factor. 60. Mario 1. Blejer, Harry G. Johnson, and Arturo C. Porzecanski, "Un analisis de los determinantes economicos de la migrad6n mexicana legal e ilegal hacia los Estados Unidos," Demografia y economia, vol. 11, no. 3 (1977), pp. 326-40. 61. Walter Fogel, "Twentieth-Century Mexican Migration to the United States," in The  NOTES 215 Gateway: U.S. ImmigrationIssuesand Policies,ed. BarryR. Chiswick(Washington,D.C.: American Enterprise Institute, 1982), pp. 193-221. 62. Alberto E. Davila, "EconomicDeterminants of IllegalMexican Immigrationto the U.S.," Economic Review, Federal ReserveBank of Dallas, May 1983, pp. 13-23. 63 Carlos Zazueta, "Mexican Workers in the United States," pp. 56-59. 64. cENiET, "Analisisde algunos resultados," pp. 35-36. 65. Cesar Zazueta, "Investigaci6nreciente sobre migraci6nmexicana," pp. 91-92. 66. KennethRoberts,AgrarianStructureandLaborMigrationin RuralMexico: The Case of CircularMigration of UndocumentedWorkers to the United States (Austin: Institute of Latin American Studies,Universityof Texas at Austin, 1980), pp. 266, 248-49. 67. Ibid., p. 266; Shadow, "Differential Outmigration," pp. 72-75; Raymond E.Wiest, "Implications of InternationalLabor Migrationfor Mexi:an Rural Development,"in Migra- tion Across Frontiers:Mexico and the United States,ed. Fernando Camara and Robert V. Kemper (Albany:State Universityof New York at Albany, 1979), p. 87. 68. CENIET, "Analisis de algunosresultados," p. 36. 69. Shadow, "Differential Outmigration," p. 72. 70. Wiest, "Implications of International Labor Migration," pp. 88-90. 71. Roberts, AgrarianStructure and Labor Migration,p. 75. 72. CENIET, "Analisis de algunos resultados," pp. .56-61. Cornelius reports that 30 percent of the migrants in his sample employedcoyotes on their first attempt to cross the border; 41 percent of his sample had used them on their most recent trip ("Mexican Migration to the United States," p. 23). 73. CENIET, "Analisis de algunos resultados," p. 63. 74. Cesar Zazueta,"Investigaci6nrecientesobre migzaci6nmexicana,"pp.88-90. Male migrants were reported to be much more likely to financetheir trip by borrowing than were women, the respectiveratios of borrowers to total migrants being45 to 20 percent. In part, this may reflectthe relativelygreaterfemalemigrationfrom the border citiesrather than from the interior and, subsequently,a lesser need for a cash reserve. 75. CENIET, "Analisisde algunos resultados," pp. 37-39. 76. There appears to have been no attempt to distinguish between the gross rate of earnings and the net rate after payroll deductions. Bcth were probably reported by the respondents. 77. Zazueta and Mercado, "El mercado de trabajo norteamericano,"p. 23. 78. Apparently, there are some households for which migrant earnings constitute the "only source of income" (Fernandezand Fernandez,"Capitalismoy coperaci6n," p. 216). 79. Shadow, "Differential Outmigration," p. 75; Wiest, "Implications of International Labor Migration," p. 88; Michael Belshaw, A Villagc Economy (New York: Columbia UniversityPress, 1967),p. 31; P. LamartineYates,Mexico'sAgriculturalDilemma(Tucson: Universityof Arizona press, 1981), p. 245. 80. Roberts, AgrarianStructure and Labor Migration,pp. 68-73. 81. In his discussion of household networks, Roberts relies on the findings of Ina R. Dinerman in her study of rural householdsin Huecorio,MIichoacan, "Patterns of Adaptation among Households of U.S.-Bound Migrants from Michoacan, Mexico," International Migration Review, vol. 12, no. 4 (1979), p. 496. 82. Shadow, "DifferentialOutmigration," p. 70; Roberts, AgrarianStructureand Labor Migration,pp. 129-34. 83. Cornelius, "Mexican Migration to the United States," p. 22. 84. Roberts, AgrarianStructure and Labor Migration.The four zones studied were Las Huastecas, San Luis Potosi; Mixteca Baja, Oaxaca; Valsequillo,Puebla; and El Bajio, Guanajuato. 216 MIGRATION TO THE UNITED STATES 85. Why the increasedrelianceon purchased farminputs should lead to a greater needto earn off-farm cash income in order to finance those purchases is not clear. If a move away from peasant and toward commercialagricultureincreasestotal output and the cashincome from farming, as it apparently does, then the earnings from farming should suffice to cover the costs associated with that mode and still leave the household better off. Otherwise, there would be no incentive to shift from subsistence to commercial agriculture. 86. Roberts, Agrarian Structure and Labor Migration, pp. 263-67. 87. Cornelius, "Mexican Migration to the United States," p. 22. 88. Cornelius makes much of the persistence of tradition within families as an intergenera- tional phenomenon. In 62 percent of the families having some history of migration to the United States, members of two generations have gone there. An additional 8 percent reported participation in the migratory flow through three generations. Eighty-seven percent of the migrants also reported having brothers or sisters working in the United States. He also cites other studies confirming this intergenerational pattern. ("Mexican Migration to the United States.," p. 21.) 89. Michael E. Conroy, Mario Coria S., and Felipe Villa G., "Socio-economic Incentive for Migration to the United States: Cross-Regional Profiles, 1969-1978," paper presented at the 1980 Annual Meeting of the Population Association of America, April 11, 1980, Denver, Colorado, pp. 15-30. 90. Clark W. Reynolds, "Labor Market Projections for the United States and Mexico and their Relevance to Current Migration Controversies," Food Research Institute Studies no. 17 (Stanford, Calif.: Stanford University, 1979; processed). 91. Harold Wool, "Future Labor Supply for Lower Level Occupations," Monthly Labor Review, vol. 99, no. 3 (1976), pp. 22-31. 92. In localities where a large proportion of the labor force migrated to the United States, the wage and employment effects have been found to be quite visible. Wiest observed in Michoacan that the large wage differential in earnings between the United States and the local community raised the reservation price of local workers and created labor "shortages" in agriculture. Local labor requirements for the sector were satisfied by recruiting workers from outlying hamlets. ("Implications of International Labor Migration," p. 91.) 93. Cesar Zazueta, "Investigaciones recientes sobre migraci6n mexicana," p. 80. 94. There is some empirical foundation for the assumption that internal and international migration are substitutes for each other. A finding of the CENIET household survey was a lower frequency of internal migration involving a change in residence among those who had experienced migration to the United States. 95. This formula is derived from the reduced form equations of supply and demand as discussed in Micha Gisser, Introduction to Price Theory, 2d ed. (Scranton, Penn.: Interna- tional Textbook, 1969), pp. 378-80. 96. See Chapter 7. The three sectoral annual rates of increase were 4.5, 3.6, and 3.2 percent, respectively. 97. Cornelius, "Mexican Migration to the United States," p. 48. 98. Diez-Canedo, "La migraci6n indocumentada;" Carlos Zazueta, "Mexican Workers in the United States," pp. 51-54. 99. Cornelius, "Mexican Migration to the United States," p. 20. 100. For a discussion of the presumed benefits and costs to Mexico of illegal migration to the United States, see Allen R. Newman, "Some Theoretical Considerations for Mexican Illegal Migration to the United States," paper presented at the annual meeting of the Rocky Mountain Council for Latin American Studies, February 14,1981, Las Cruces, New Mexico. 7. Wages in Mexico, 1935 to the Present ANYONE WISHING to chronicle the course of wages over the past fifty years in Mexico is faced with a formidable challenge. Systematic collection of wage data has still to be undertaken in Mexico. Thus, the data that have been available originate from different sources, are not comparable, and certainly not representative. Compounding the problem of wage data scarcity is that of converting nominal wages into real wages. The course of real wages varies significantly depending on the choice of a deflator, particularly for the years prior to 1960. An evaluation of the evidence on the course of wages will be attempted in this chapter. Much of the discussion on the course of wages in Mexico takes as its point of departure changes that have occurred in the level of legal mini- mum wages. Since these do represent an observable datum, it is under- standable that they should be cited often. It is almost universally recog- nized, however, that legal minimums have had only a very limited effective coverage and that departures of actual wages from the minimums have been significant and of different magnitudes and signs over time. Never- theless, since the legal minimums have served as a sort of benchmark for the course of wages, they do deserve attention. History of Minimum Wages Minimum wages have a relatively long his tory in Mexico. The Constitu- tion of 1917 provided for their establishment, but it was not until 1934, during the administration of Lazaro Cardenas, that this provision was implemented. The law provided for a very complex structure of mini- mums. The country was divided into 111 zones for which regional boards were to define two separate minimums, one for the agricultural sector and another for all other wage-earning employment. In addition, the regional boards were empowered to establish occupational wage minimums for unorganized workers; eighty-six were in effect in 1983. The number of 217 218 WAGES IN MEXICO covered occupations has been increasing over time. As recently as 1974- 75, only sixty-fiveoccupational rates were in force.' The original statute provided for a revision of rates every other year. With the accelerationof inflation, however, an interim adjustment was made near the end of the 1971-73 biennium, and thereafter minimumshave been revisedat least once a year. The decentralization of the minimumwagesetting processwas intended to allow each zone to determinerates that, to some extent, reflectedlocal labor market conditions. As a result, a structure of minimum wage rates emerged that was marked by large disparities.For example, in the early 1940s, the highest rural minimum wage was 4.3 times the lowest. That differential almost doubled over the rest of the decade. Sincethen, it has steadily narrowed and by the end of the 1970s stood at a ratio of approx- imately 2:1. The nonagricultural structure of minimums has undergone similar changes over time. During the early 1940s, the highestrate stood in a ratio of 4.8:1 to the lowest. After wideningto 7.7:1 at the end of that decade, this ratio had shrunk to 2.3:1 by 1979. The highest minimums have generallyobtained in the regions borderingthe United Statesand the large metropolitan areas, while the lowest were assigned the depressed zones of the southern peninsular and Pacific regions and parts of the central plateau. The ratio of urban-to-rural rates has also varied sharply over time. The initial urban minimummeasured only 6 percent above the rural, a differential that increased to 26 percent by 1950-51, only to narrow gradually thereafter to under 10 percent in 1979. A significantchange in the policy orientation for wages emergedduring the administration of PresidentL6pezPortillo. Over the precedingtwenty- fiveyears, adjustmentsin the minimumwage had been designedto prevent a declinein its purchasing power and to advance its real value gradually over time. Toward the end of the decade of the 1970s, minimum wage administration was investedwith an expandedpurpose. One of these was redistributive in nature. A narrowing of minimum wage differentials among zoneswas advocated as a means of raisingthe lowestwage incomes of the coveredpopulation in order to achievegreater degree of equality in the distribution of income. Therefore, the national commissionon mini- mum wages began adjusting the lowest urban and rural minimums at an accelerated rate, narrowing the differentials between urban and rural minimums within, as well as across, zones. A second rationale offered for the narrowing of differentialswas the expectationthat it would reduce the incentives for "improper," that is, excessive,rural-urban migration.' Beginning with the determinations for 1978, the number of different minimum wage rates has been reduced rapidly-first from 144 to 59, by 1980 to 18, and to only 4 in 1983. This sharp reduction was facilitatedin  REAL WAGES, 1940-55 219 1981 by the consolidation of the rural and urban rates within each zone; thus, one of the differentialsthat had characterizedthe system ever since its inceptionwas eliminated.The wide relativedifferentialthat had existed between the highest and the lowest minimurm rate was also sharply nar- rowed, from a ratio of 3.28:1 as recently as 1977, to only 1.4:1 in 1983. The course of the legal minimumsover time from their inceptionto the present is set forth in table 7-1. For most years, the nominal and real minimums are presented and indexed in two ways. The first is a simple arithmetic average of minimumsacross all zones; the second presents the minimums as weighted averages,with the size of each zone's labor force serving as the weight. As can be seen, the weighted average minimum is consistently larger than the simple average, a reflection of higher wage levelsin more denselypopulated zones. The nominal rates are deflated by the consumerprice index for Mexico City sincea national priceindex does not become available until 1968. No sooner had the initial legal minimums been declared for the bien- nium 1934-35 than they began an uninterrupted declinein real terms as increasesin the general price leveloutpaced acljustmentsin the minimums. The urban minimum bottomed out during the 1946-47 biennium at a level 48 percent below the initial 1934-35 level; the rural minimum reached its lowestpoint in 1950-51,56 percent below its initiallevel.Both rates then began to recover, but the minimurnsdid not exceed their initial real values until the 1964-65 biennium. 3 Over the next twelve years to 1977, the weighted minimums increased by about 60 percent, peaking in 1977 on a full-yearbasis. Thereafter, in the face of acceleratinginflation, the urban real minimum wage lost ground. During 1983, it averaged between two-thirds and three-fourths of the peak 1977 level.In contrast, the rural minimum, which underwent a more rapid rate of increaseuntil it was finally merged with the urban, continued to increase in real terms through 1981, after which it moved together with the urban minimum. The Course of Real Wages, 1940-55 The question that has troubled students of Mexican economicdevelop- ment is the extent to which the courseof the rninimumsaccurately reflects that of actual wages paid. Particularly troublesome has been the sharp decline, noted above, in the real minimumsbetween 1934 and the end of the 1940s.In general,the consensusseemsto be that actual real wagespaid also fell, though by a narrower margin than the minimums.Little agree- and the proportion of the ment exists on the extent to which real wages Fell labor force affected by the decline. for those years have Unfortunately, the wage statisticsthat are a.vailable 220 WAGES IN MEXICO Table 7-1. Average Nominal and Real Legal Minimum Wages, 1934-35 to 1983 (Mexican pesos) Real Nominal unweighted Real unweighted average Nominal weighted average minimum weighted average minimum wage index average wage index wage (1978 = 100)' minimum wage (1978 = 100)- Years Urban Rural Urban Rural Urban Rural Urban Rural 1934-35 1.15 1.09 53.2 57.8 n.a. n.a. n.a. n.a. 1936-37 1.31 1.21 52.3 ss.4 n.a. n.a. n.a. n.a. 1938-39 1.46 1.31 47.0 48.3 n.a. n.a. n.a. n.a. 1940-41 1.52 1.30 46.8 45.9 n.a. n.a. n.a. n.a. 1942-43 1.52 1.35 34.7 35.3 n.a. n.a. n.a. n.a. 1944-45 1.90 1.65 29.4 29.3 n.a. n.a. n.a. n.a. 1946-47 2.48 2.05 27.9 26.4 n.a. n.a. n.a. n.a. 1948-49 3.01 2.40 29.3 26.8 n.a. n.a. n.a. n.a. 1950-51 3.35 2.66 28.2 25.7 n.a. n.a. n.a. n.a. 1952-53 5.35 4.55 37.4 36.5 n.a. n.a. n.a. n.a. 1954-55 6.34 5.26 39.5 37.8 n.a. n.a. n.a. n.a. 1956-57 7.25 5.99 39.0 37.0 n.a. n.a. n.a. n.a. 1958-59 8.13 6.86 37.7 36.4 n.a. n.a. n.a. n.a. 1960-61 9.89 8.83 42.8 43.8 n.a. n.a. n.a. n.a. 1962-63 12.44 10.92 52.6 52.9 n.a. n.a. n.a. n.a. 1964-65 16.00 13.47 54.9 62.6 17.33 14.78 64.0 61.3 1966-67 18.69 15.72 70.3 67.7 20.01 17.42 68.6 67.0 1968-69 21.58 18.32 77.5 75.4 23.12 20.12 76.0 74.0 1970-71 24.91 21.20 81.1 79.1 26.99 23.48 80.1 78.2 1972-73 29.29 24.94 83.6 81.6 31.93 27.73 83.1 81.0 1973b 34.56 29.43 88.3 86.2 37.68 32.73 93.1 90.8 1974' 39.38 33.52 89.4 87.3 43.42 37.79 87.6 85.6 197 4 - 75 d 48.04 40.90 92.5 90.2 52.97 46.10 91.4 89.3 1976' 58.68 49.87 98.9 95.1 64.74 56.66 96.2 94.4 19761 72.18 61.34 107.3 104.5 79.63 69.55 107.8 105.8 1977 79.37 67.45 102.5 99.8 87.56 76.48 103.1 101.1 1978 90.55 78.99 100.0 100.0 99.37 88.50 100.0 100.0 1979 105.80 96.26 99.2 103.5 116.02 106.81 99.1 102.5 1980 124.53 121.33 92.3 103.1 136.62 134.16 92.3 101.7 1981 167.00 167.00 95.1 101.7 178.87 178.87 93.8 105.3 1982g 224.00 224.00 87.3 100.1 240.00 240.00 85.2 95.7 1982h 291.20 291.20 80.9 92.8 312.00 312.00 79.0 88.7 1983i 364.00 364.00 78.2 89.7 390.00 390.00 76.4 85.8 1983i 400.40 400.40 67.2 77.0 429.00 429.00 65.6 73.6 n.a. Not available. a. Deflated by the consumer price index for Mexico City. (Notes continue on the following page.)  REAL WAGES, 1940-55 221 serious deficiencies.The only officialdata derive from the annual October wage surveys of selected manufacturing firras. These companies are not necessarilyrepresentativeof the whole sector nor even of the industriesof which they form a part.4 The surveyedfirms are describedsimplyas being among the "most important" in Mexico. Prior to 1964, establishmentsin only six cities were surveyed;since then, twelve citieshave been surveyed. Usuallythe included firms are among the largest in their respectiveindus- tries. In 1939, probably no more than 43,000 workers were employed in the surveyedestablishments.By 1950, the number of workers coveredhad increased to 54,000, or by 25 percent, distributed over approximately 600 establishments. Since employment in the modern industrial sector had virtually doubled in the interveningdecade, the full impact this expansion may have had on wage levelsis not captured by the survey.' Furthermore, since no employment weights are available, the average wage can be computed only as a simple average of the wage paid in the included industries. Sinceonly the average daily wage appears, no account can be taken of the impact on earnings of variaticns in hours or days worked. Utilizing these data, L6pez Rosado and Ncyola Vazquez estimated that real wages in the covered industries had fallen by approximately 27 Table 7-1 (notes continued) b. EffectiveSeptember 17 to December31, 1973. c. Effective January 1 to October 7, 1974. d. EffectiveOctober 8, 1974, to December31, 1975. e. EffectiveJanuary 1 to September30, 1976. f. EffectiveOctober 1 to December31, 1976. g. Effective January 1 to October 30,1982. In addition to the adjustmentrecordedhere, a further graduated general wageincreasewas decreedby the secretaryof labor on March 19. All wages under Mex$20,000 per month were increased by 30 percent, those between Mex$20,000 and Mex$30,000 by 20 percent, and sll higher wages by 10 percent. The increaseswere made retroactive to February 18. Sincethe highest minimumwage in effect yieldeda monthlywage of lessthan Mex$10,000,the dlecree raised all workers at effectively the minimum wage by 30 percent in February. The legal minimums were not officially increased by 30 percent until November 1. If account is taken of the decreed increase,the weighted index of urban minimumswould have averaged 107.0 over the January-October interval; the rural, 120.1. The weighted minimums for the year are approximate only. h. EffectiveNovember 1 to December31, 1978. The weightedminimums are approxi- mate only. i. Effective January 1 to May 31, 1983. The weighbied minimumsare approximate only. j. Effective June ito December31,1983. The weigh ted minimumsare approximateonly. Sources: Comisi6n de los Salarios Minimos, Salarios minimos, various issues; prices 1934-35, James W. Wilkie, Statistics and National Policy (Los Angeles: Universityof California, L.A.,Latin American Center, 1974), p. 2.!9; prices 1936-60, Nacional Finan- ciera,S.A.,50 anosde revolucidnmexicanaen cifras (Mexico,D.F., 1963); prices 1961-83, Bancode Mexico, Informe anual, various issues. 222 WAGES IN MEXICO percent between 1939 and 1947, although these were believed to have recovered slightly over the following two years.' A rather different estimate of the drop in real wages is obtained from another officialsource cited by these same authors that surveyedtwenty- four industrial groups with roughly one-sixth of the total industrial labor force in 1949. Accordingto this source, real wagesfell by a lesser amount, 18 percent between 1939 and 1948. Two additional groups for which data were presented also revealed substantial declines:railway workers by 14 percent and public (federal) employees by 31 percent over the decade 1939-49.'While substantial, these declinesappear to have beenless severe than those for the urban minimum wage. Two other analysesof the course of wages deserve mention since they attempted a less aggregativeapproach.John Isbisterconcludesthat annual earnings of blue-collar workers in the manufacturing sector fell by 11 percent between 1940 and 1945 and then remained relativelystable until 1955.8 During that same interval, the real valueof the legalminimumwage for urban workers had fallen by 40 percent. StylianosPerrakis examined the course of earnings of unskilledworkers in Mexico Citymanufacturing industries.9 Between 1940 and 1944, the legal minimum applicableto the Federal District and actual earnings fell together by about 27 percent. Thereafter, unskilled earnings leveled off even as the legal minimum continued its decline to 1951. As discussedin Chapter 2, the decade of the 1940s was one of substan- tial economic growth. Gross national product grew at an average annual rate of 6 percent, while per capita income rose at a 3.7 percent rate. Output per worker also increased at an average annual rate of 3.2 percent, although the intersectoral disparities were large: from 3.6 percent in agriculture to 0.5 percent or less in the secondary and tertiary sectors. A decline in real wages in such an expansionary context is somethingof an anomaly and has led to numerous attempts to analyze and interpret these findings.'"The anomaly is heightened by Adolph Sturmthal, who traced changes in the functional distribution of national income and concluded that the weighted real average wage must have risen by at least 10 to 12 percent over the decade of the 1940s." One possible explanation offeredby both L6pez Rosado and Sturmthal is the changingweight in the structure of employment.Although wages in the canvassed sectors may have been falling in real terms, a shift of labor from even lower absolute wage sectors could have resulted in a rising average wage. The unweighted wage data do not fullyreflectthe impact of probable changes in the occupational structure within the industrial sec- tor, which may have been moving toward more skill-intensiveemploy- ments. And to the extent that individual workers experienced upward  REAL WAGES, I940-55 223 occupational mobility within the expanding industrial sector, the decline in average earnings might provide a misleading guide to the course of earnings of individualsemployedin the sector over a period of years. Both of these sources also hold out the possibility that daily wage rates may underestimateearnings if hours or days worked increasedover the decade. While this possibility was not empiricallytested, Sturmthal believessuch increaseswould seem plausiblein viewof the substantial industrial expan- sion of the decade. Alternatively,the decline in real wages is viewedas a consequenceof a lag in the adjustment of nominal wages to changes in the general price level.While it is not unusual for wages to lag prices at the beginningof an inflationary period, one would expect that, if the real wage prior to the onset of inflation were equal to the market price of labor, the nominal wage would soon adjust so as to restore the market clearingreal wage.The revisedexpectations of workers would lead to an increase in the reserva- tion price of labor expressed in nominal terms that would be sufficientto restore the real wage. In the Mexican case, the accelerated rate of price increasesdated from the mid-1930s and continued through the 1940s. On the one hand, it might have been expected that, if the base year wage level had beentruly indicative of the market price of labor, the initial declinein the real wage would have led to an earlier recovery than could be observed from the available data. On the other hand, it has been observedthat the declinein real wage rates was accompanied by increases in fringe benefits, social services,and probably in hours worked that would haveoffset at least part 2 of the welfare loss occasioned by decliningwage rates.' Finally, all the students of this issue concur in the view that the wage data that are availableare drawn from a very narrow sample of firms and therefore cannot be taken to be representativeof the industrial sector as a whole or of the urban labor market. Somequestion of the reliabilityof the data themselvesmay also be raised, although the trend they establish is so pronounced and pervasive that it is hard tD dismiss them as completely misleading for the sampled firms. There may be an alternative explanation for the observeddeclinein the wage rates of the sampled firms. First, this sample is drawn from among the largest firms in each of the industries sampled. These firms in the "modern" sector are those most likely to observe the legal minimum requirements of employment. Furthermore, since they are also the best candidates for organization,the trade unions would servean enforcement function to ensure the payment of wages at least equal to the minimum wage. Next, it can be inferred that the minimftum wages established in the 224 WAGES IN MEXICO initial 1934-35 years were well in excess of the market price of labor. For example, the output per worker in 1934 in the economy as a whole can be estimated to have been equal to approximately Mex$567 in current pesos. The average urban minimum wage defined in annual terms (365 x Mex$1.52) was equal to Mex$420, or 74 percent of GDP per worker. The annualized average rural minimum was only slightly smaller: Mex$398, or 70 percent of the average output per worker in the economy as a whole. Since agricultural output per worker in that period was less than half of the national average, the annualized rural minimum exceeded the value of output per worker. The share of wages and salaries in GDP for a proximate year, 1940, has been estimated at 29.7 percent, and it can therefore reasonably be inferred that the minimums had been set well above the market price of labor."3 Given the strong pro-labor sympathies of the Cardenas government, it would not be surprising to find the estab- lishment of minimum wages reflecting such sympathies, with little refer- ence to actual market wage conditions. Thus, it may be surmised that the adoption of wages at least equal to the legal minimums in the large, modern sector firms yielded substantial economic rents to the workers employed there. Subsequently, as the real value of the minimum wage rates began their decline in 1936 and as the nominal minimums were adjusted at rates below the rate of inflation, these firms were under no market compulsion to adjust the actual wages paid by more than the adjustment in the nominal legal minimum. Real wages could therefore fall at an equal rate with the legal minimum until the latter was a more accurate reflection of the market supply price of unskilled labor. As Perrakis pointed out, this occurred in Mexico City in 1944, when the real wage paid for unskilled labor stabilized even though the real minimum continued to decline for several additional years.'4 Thus, it is entirely possible that the course of neither the legal minimum wage nor the wages in these surveyed firms provides a true reflection of what was happening to wages generally in either the urban or rural labor markets. In the case of the legal minimum, it has always been widely accepted that it serves as an effective wage only in the larger establish- ments of the modern urban sector or in the modern large-scale commercial agricultural sector. Such employers probably accounted for no more than 20 percent of total employment in the late 1930s or early 1940s. Since the surveyed firms were not representative of jobs outside the effective range of the minimum wage, there is no way to conclude from these wage data what the general course of wages may have been. Wages for urban unskilled labor beyond the scope of legal minimum wage coverage could have been rising in real terms at the same time that wages in the surveyed sector were declining.  REAL WAGES, 1940-5 5 225 Indeed,there are severalbits of information that, taken together,would lead one to treat with caution, if not skepticism,the conclusionthat urban wages suffered a general decline during the decade of the 1940s. It is an axiom of economic theory that the price of a factor of production to any particular use is determined by its opportunity cost. In the case of urban unskilled labor in a developingeconomy, its opportunity cost is generally taken to be definedby forgone earningsin the rural or agricultural sector. Thus, a clue to the course of unskilled urban wages might be provided by the course of incomes or earnings in the rural sector. Urban-Rural Wage Differences The firstdevelopmentthat would influencethe reservationpriceof labor is the changing land tenure relationships in agriculture. One might expect that the conversion from landless laborer to landowner would have re- sulted in an increase in the income of the land recipient, particularly since there is no evidenceavailablethat would suggesta declinein output on the plots transferred to their new owners. In the late 1930s, the distribution of almost 18 million hectares of land might have been expected to induce workers to remain in the sector in hopes of becomingthe beneficiariesof further distributions. Indeed,there are refererces in the literature to urban residents returning to their rural communitiesof origin in order to estab- lish and pursue their claim to land. The decade of the 1940s saw a continuing process of distribution. While not acquiring the scope of the process during the Cardenas regime (1934-40), the land reform did distribute an additional 8 millionhectares during 1941-49.15 The transfer of land might therefore be viewed as resultlingin a once-and-for-all in- crease in the returns to rural labor or as creating an anticipationof such an increase. In either case, the reservation price of labor to the urban sector would not be expected to decline under such conditions. Although land redistribution effected an immediate change in the level of income of many rural laborers, the trend of income over time would also be relevant to the wage trend for unskilledwork in the urban sector. Data reviewed in Chapter 2 indicate that oatput per worker in the agri- cultural sector was increasingat a faster rate than the rest of the economy during the 1940s and by a very considerablemargin, 3.6 percent a year to less than 0.5 percent. To be sure, this is ]1otsufficient evidence of an increasein incomes among those strata of the rural population most likely to view urban unskilled employment as an alternative source of work. One may come closer to approximating the income trend of these groups, however, from the output data of the agriculturalsector classified by size or form of land tenure. Presumably,the relevant group is the small landholder and his family. Although there are no income data for farm 226 WAGES IN MEXICO families, the value of output per plot is available for 1940 and 1960. Measured in constant Mexican pesos, output on plots of less than five hectares increased at a rate of 4 percent per year, while on ejido plots it increased at a rate of 3.9 percent. A very significantpart of this growth, and perhaps most of it, was realizedduring the decade of the 1940s; Sergio Reyes Osorio and his coauthors hold that output on the smallest farm units remainedvirtually unchanged between 1950 and 1960.'"Most of the increase in output per plot is traceable to an increase in productivity of approximately 2.7 percent a year; the balance is attributable to a growth in the average size of plots. Furthermore, the increasesin the value of output duringthe 1940s were accompanied by an improvement in the rural-urban terms of trade. Agri- cultural prices rose during the decade by 279 percent. The wholesaleprice index of Mexico City increased by 191 percent and the consumer price index by 228 to 254 percent, depending on which source is cited.' In the face of such increases in rural income, one would not expect to find the reservation price of rural labor to the urban sector falling. The contribution that rural-urban migration has made to the evolution of intersectoral wage differentials is also at issue here. For example, Perrakis attributes the decline in the real minimum wage to "a delayed effect of the land redistribution program of the late thirties and the subsequent massive immigration into the cities."" Implied is the domi- nance of push factors in the rural zones, which propelled people into the urban areas and resulted in a depression of wages for unskilledlabor. It is certainly true that the decade of the 1940s saw a sharp increasein the rate of growth of the urban population, defined here as people residing in communities with at least 15,000 inhabitants. Although this segment of the population had grown at a rate of only 3.1 percent a year during the 1930s, it acceleratedto 6.2 percent in the succeedingdecade.'9 An alterna- tive hypothesis could be that the accelerated flow to urban areas was a response to an increase in the relative attractiveness of urban employ- ments. Had the flow been simply the product of push factors, one could reasonably have expected that the "excess" supply would have encoun- tered difficultiesin finding acceptable employment,that open unemploy- ment would have become more prevalent, and that the wage level would have been lowered. The evidenceon unemployment,however, offers little support for such a scenario. According to data reviewed in Chapter 3, unemployment in 1940 was estimated at about 3.1 percent of the labor force, while the 1950 census reported a declineto 1.3 percent. Most of the literature on Mexican internal migration supports the view that it has beenstrongly motivatedby differentialincomeopportunities, as  REAL WAGES, 1940-55 227 has been documented in Chapter 5. If unskilledwages had been declining in the urban centers during the 1940s, this. would have implied that migration accelerated as the relative attractivenessof urban employment declined. Of course, urban wages apparently still remainedabove those in the rural sector. It may therefore be argued that the existence of an absolute differentialwas the critical condition givingrise to migration and that the narrowing of the urban unskilled-rural wage differentialneed not have served as a disincentiveto a high rate of migration. In my opinion, however, the data that elicitskepticismabout the alleged general decline in urban earnings are those that trace the trend in con- sumption and vital statisticsduring the years between 1940 and 1960. The prevailing view is that averagereal wages for :ndustry declined duringthe 1940s and into the early 1950s, reaching their lowest level in 1952. They recovered to the 1940 level and fluctuated within 10 percent above that level until 1958 when the beginningof a sustainedrise in averageearnings began. In the caseof the legal minimumwageand the earnings of unskilled workers in the large-plant sector in Mexico City, the 1940 real wage level was not achieved until 1964 and 1962, respectively.Yet the data on consumption levels seem to indicate increasei of substantial proportions during this period. Consumption and Income Table 7-2 provides per capita measures oi apparent consumption for three agricultural commodities that might be expected to vary with changes in income. As can be seen, consumption of these commodities, regarded as staples in the diet of lower-incDmegroups, increased in a continuous fashion in each five-yearinterval since 193S-39. Somemight interpret the increase in corn and bean consumption as support for the conclusion that wages declined during the 1940s and early 1950s. Esti- mates of the income elasticity of demand, calculated from cross-section studies of households at different income levels,suggestthat it is negative, in other words, that consumptionwould increasein the face of a declinein income. While cross-sectionstudies apparendlydo invariably yield nega- tive values for the income elasticity of demand for corn and beans, time series data suggestpositive values.Furthermore, a negativevalue over all income groups is not necessarilyinconsistent: with positivevalues within the lower income groups. In any case,the continued rise in per capita consumption,even after real wages surpassed their 1940 levels, casts doubt on the presumed negative values of the income elasticityof demand. In the caseof sugar, the income elasticity of demand is consistently considered to be positive. As can be 228 WAGES IN MEXICO Table 7-2. Apparent Annual Per Capita Consumption of Selected Agricultural Commodities, 1935-39 to 1960-64 Consumption in kilograms, five-year averages Years Corn Beans Sugar 1935-39 91.4 6.2 15.7 1940-44 100.8 7.2 18.8 1945-49 108.4 8.0 22.3 1950-54 134.3 11.4 24.6 1955-59 158.9 15.1 28.1 1960-64 199.2 18.0 28.2 1965-69 171.3 18.5 35.4 Note: Apparent internal consumption is estimated from production plus imports minus exports. Source: Secretaria de Agricultura y Recursos Hidraulicos, Direcci6n General de Economia Agricola, Consumos Aparentes de Productos Agricolas, 1925-1978 (Mexico, D.F., 1979). seen in table 7-2, large increases in consumption were recorded in each successive subperiod. Estimates of the caloric intake of the population suggest a very significant increase between 1934-38 and 1960. Data provided by Reyes Osorio indicate an average increase of 47 percent in per capita caloric intake over the period, from 1,800 to 2,654.20 There are other products whose use was suggestive of rising incomes. A large increase was recorded in the production of beer. From 83 million liters in 1935, production, and apparent consumption, expanded to 180 million in 1940,353 million in 1945, 572 million in 1953, and 745 million in 1957.21The proportion of the population without shoes also fell steadily in the intercensal period 1940-60, from 27.6 percent in 1940 to 19.1 and 14.3 percent in subsequent census years. The ratio of radios to population increased sharply from 1:52 in 1940 to 1:16 in 1960.22These data suggest a growth in the discretionary income available to households that is not consistent with a decline in real wages. Of course, if the number of employed household members had increased during this period, then an upswing in consumption in the face of a decline in real wages would still have been attainable. However, the data on labor force participation offer little support for such a possibility: the male rate proved constant between 1940 and 1950. While the female participation rate doubled over the same interval, from 4 to 8 percent, the absolute increase that this implies is far too small to account for the substantial expansion in consumption we have observed.  REAL WAGES, 1940-55 229 Vital statistics for the period are suggestiveof improved living stan- dards. For example, infant mortality, which is very sensitiveto consump- tion levels among the lower-incomegroups, declined during the twenty- year period 1940-60. In 1940,there were 123 infant deaths per thousand births; the two subsequent census years recorded declinesto 106 and 74. Over the same period, the proportion of deaths attributable to parasitic 23 and intestinal diseases dropped from 43 percent to 36 and 27 percent. Thus, a plausible alternative can be offered to the view that wages fell generallyduring the 1940s and into the 195 Os.Those wages in the tradi- tional large-scalemanufacturing sector that had beenraised well above the market price of labor through institutional interventionin the wage deter- mination process may, indeed, have declined during that period. But it would be reasonableto posit that, in the faceof a rising opportunity costof urban labor, wages in that part of the labor market unaffected by institu- tional intervention rose, and the differential rate of earnings within the urban manual labor force was thus narrowed. The narrowing that is documented below for the years after 1960 would then appear to be a continuation of a processthat had begun earlier.Although wages in large, long-establishedfirms mayhave been fallingin real terms, we have no data indicatingthe courseof wagesin the new or growing firmsof the industrial sector that were responsiblefor the impressivegrowth of industrial output during those years. Narrowing of Wage Differentials Although it may be thought that unskilled labor abounded at a constant wage, it is not at all clear that the supply of workers with the minimal educational requisites for industrial employment was perfectly elastic. After all, evenas late as 1950, 42.5 percent of the population over 15 years of age was classifiedas illiterate, a proportion somewhat smaller than that of the population with no formal schooling. 2 4 The 1960 census recorded the median level of education of this same population at only 2.4 years. Sincethe educational attainment of the aduli:population in 1940 was even more modest, it is quite possible that the shift in the demand for urban labor resultedin increasesin real wages. At t:hevery least, unskilled wages may have shown little change, but it is nct likely that they could have declined as sharply as the limited available data would suggest. As one scholar has observed, "One cannot believethat political stability could have so easily endured if real wages had halved at a time of such rapid overall growth."2 5 There are fragmentary data reinforcing :he belief that the legal mini- mum wages declinedto levelsbelow the market wage for unskilled labor 230 WAGES IN MEXICO during the 1940s and remained there until the 1960s.These data originate in community studies, usually by anthropologists, that record the nature of economic activityby the various classesof residents.The most common source of information is interviews with individuals during which work histories are compiled. Although such information may not be completely representative or may be distorted by faulty memories,it may be indicativeof the closeness of actual wages paid to the prevailing minimums. For example, Lourdes Arizpe reports that young men from the rural districts of the northwestern part of the state of Mexico would migrate to participate in harvestingand other peak season activitiesin the fieldsaround Xochimilco,in the Federal District. In 1945, the going wage for such work was on the order of Mex$25 a week, or Mex$4.16 a day. At the time, the legal minimum for rural workers was only Mex$2.45 a day. 26 The same source reports that men from the same rural district seeking casual employment in Mexico City in the early 1950s could work nights as stevedores in La Merced market and earn on the order of Mex$30-40 a week. 2 7 During 1950-51, the daily legal minimum for the Federal District was on the order of Mex$4.50, although it was increased sharply during the following bien- nium to approximately Mex$6.70. In 1955, masons (albaniles)in con- struction were earning Mex$ 15 per day, roughly 60 percent more than the minimum wage; this differential was to narrow over the subsequent decade to 40 percent in 1965. Rural wages in some areas also seemedto exceed the legal minimums. For example,L6pez Rosado citesdata for 1948-49 that permit a compari- son of regional daily wages with legal minimums. 28 Wages paid in the northern border states in 1949 were approximately 50 percent above the legal minimum; in the central and coastal regions,they were about 40 and 100 percent higher, respectively.In 1955, Arizpe reported daily wages in agriculture in the Mazahua region in the northwestern part of the state of Mexico as Mex$7 while the rural daily minimumwage for the zone was Mex$3.31. Even in Chiapas, one of the poorest states of the republic, wages above the minimum have been reported for 1957. In the center of the state, not far from Santiago de las Casas, wagesduring the corn harvest season were Mex$30 plus maintenance for a six-day week,while the legal minimum was Mex$3.48 a day. Furthermore, it was noted that the larger farmers of the region typically faced an upward-sloping supply schedule for labor and generallyhad to offer more than the wage prevailing among small landholders in order to recruit a sufficient number of workers. 30 Beginning in 1960, the pace of adjustment in legal minimums quick- ened. The firstthree bienniumsproduce rates of increasein the unweighted national urban average minimum wage of 21.6, 25.8, and 28.6 percent; the rural minimumswere adjusted upward by similarproportions. In view  WAGE DATA, I960-75 231 of the impressivestability in consumer prices,which rose only 10 percent between 1960 and 1965, these raises in the rninimumsrepresented large increasesin their real values. Clearly,had the minimums been effectiveat the beginning of the decade, such rates of increasewould have propelled the whole wage structure in the coveredsectors upward at a rate inconsis- tent with the observed stability of prices. Instead, these increases first served to narrow the gap between the minimums and the higher market wage. In the Chiapas region mentioned above, the minimum and the market wage reached a virtual equality by 1963. Thereafter, the going wage lagged behind adjustments in the minimum. By 1966, the nominal cash wage had reached Mex$7.50 a day, only 3.5 percent above the 1957 level in real terms. However, the legal minimum had increased to Mex$13. Not even the addition of the cost of mainte- nance-estimated to range from Mex$2.30 to Mex$3.85 a day-sufficed to close the gap. Payment of wages in kind (namely, corn) was also common. Indeed,90 percent of the employingfarmerspaid wages, at least in part, if not entirely, in kind. According tc Frank Cancian, payment in kind had increased over the same interval by a larger proportion than the cash wage, by about 13 percent. 31 In the Mazahua area, the rural wage by 1965 had increased to Mex$10 a day but now was below the legal minimum of Mex$11.50. During peak seasons, however, the daily wage would rise to equal the legal minimum. Indeed,by 1968,local shortagesof agricultural labor began to appear in the region and the daily wage rose to Mex$15, slightly above the legal minimum. 32 A similar changing relationship between the legal minimum wage and actual daily wages of urban unskilledworkers in the modern manufactur- ing sector has been documented by Isbister. 3 3 In 1950, actual wages stood in a ratio of 1.45:1 to the legal minimurn in the six industrial cities surveyed. By 1960, the ratio had shrunk to 1.16:1 and reached parity in 1964. An additional six cities surveyedin the latter year yielded a similar ratio. Sincethese actual wagedata were drawn from larger establishments, which normally tend to pay wages above the minimum, the achievement of parity suggeststhat, in other urban employments(the so-calledinformal sector), actual wages may have been surpassed by the legal minimum. Perrakis's observation of earnings in small establishments during the middle of the decade led him to concludethat the legal minimumhad once again become an effective minimum. 34 Wage Data from Economic Censuses, 1960-75 The course of wages since 1960 can be charted by data appearing in the economiccensusesthat are taken every five years.For table 7-3, wage data 232 WAGES IN MEXICO Table 7-3. Average Real Annual Total Remunerations in Services, Industry, and Commerce by Establishment Size, 1960-75 (1960 pesos) Average total remunerations' Percent Sector and size of change establishment 1960 1965 1970 1975 1960-75 Establishments with no paid employees Services 6,315 5,625 6,918 8,245 30.6 Industry 2,710 5,173 5,102 7,058 160.4 Commerce 4,059 4,089 7,501 7,768 91.4 Establishments with paid employees Small Services, 1-2 workersb 5,397 n.a. 9,799 8,655 60.4 Industry, 1-S workers 4,973 6,005 6,855 9,610 93.2 Commerce, 1-2 workers 6,279 7,197 7,973 10,613 69.0 Medium Services, 3-8 workers' 7,869 n.a. 10,437 11,096 41.0 Industry, 6-25 workers 8,119 8,940 10,534 12,977 59.8 Industry, 1-25 workers 6,644 7,945 9,308 12,003 80.7 Commerce, 3-8 workers 9,465 9,941 10,957 14,505 53.2 Large Services, 9 or more workersd 16,470 n.a. 18,637 18,044 9.6 Industry, 26-100 workers 10,272 11,969 14,419 17,185 67.3 Industry, 101-500 workers 11,854 15,058 18,148 22,116 86.6 Industry, over 500 workers 12,877 18,241 23,225 28,236 119.3 Commerce, 9 or more workers 14,259 16,611 16,595 20,135 41.2 National average urban minimum wage' 3,610 5,328 6,978 7,617 110.0 n.a. Reliable data for 1965 are not available. a. Nominal remunerations are deflated by the consumer price index for Mexico City. b. 1-3 workers in 1960. c. 4-10 workers in 1960. d. 11 or more workers in 1960. e. Since a weighted average minimum wage was not available for 1960, the unweighted average is used throughout to maintain comparability. Sources: Secretaria de Industria y Comercio, Direcci6n General de Estadistica, Censo de servicios, 1961, 1966, and 1971; Censo industrial, 1961, 1966, and 1971; Censo comercial, 1961, 1966, and 1971. Secretaria de Programaci6n y Presupuesto, Censo de servicios, 1976; Censo industrial, 1976; Censo comercial, 1976. Comisi6n Nacional de los Salarios Minimos, Salarios minimos, various issues.  WAGE DATA, I960-75 233 from the censuses for the industrial, commercial, and service sectors have been extracted; the industrial census includes manufacturing, construc- tion, and electric power generation and distr,bution. The data are avail- able in most cases by size of establishment; the refore, this basis for analysis has been chosen to avoid the bias inherent in the wage data drawn from only the large-firm manufacturing sector. It is generally accepted that, to the extent that they lie above the market price of labor, the legal minimum wages are observed only in larger establishments. Small employers are more likely to respond to supply conditions in the labor market in setting the wages they pay. Thus, the observation of the course of wages in the small establishments of the three sectors ought to provide a better guide to the market-determined wage for low-skill labor than the data originating in large manufacturing firms. Annual total remunerations per worker are expressed in real terms. Before a discussion of the trends revealed by the data, a few remarks are in order about the data themselves. First, although data for very small establishments are included in the table, it should be kept in mind that these are likely to be subject to wide margins of error. As is generally recognized, such establishments keep poor r-cords, and caution is advis- able in accepting their data at face value. Second, there is little consistency in the manner in which employment is reported. In some of the earlier censuses, the only employment figure given was one as of a given date, for example, June 30. In others, it appeared as an average of two points in time. In the 197' census of manufactures, the employment figure is an average of three observations. But the remunera- tions datum refers to total wage payments made during the year. Thus, some unknown proportion of the changes ir. measured annual remunera- tions per worker may be attributable to t:hese differences in the way "annual" employment is measured. A third difficulty encountered in estimating remunerations of workers in small establishments was the absence of sufficient detail in the industrial censuses since 1960 to permit the separation of unpaid workers from paid employees. According to the 1960 census, over 98 percent of the unre- munerated workers (including proprietors, the self-employed, and unpaid family workers) were to be found in establishments with twenty-five or fewer workers. In subsequent censuses, there was the problem of allocat- ing the unremunerated workers over the various size strata in order to compute the remunerations of paid personnel. The solution was to follow rather closely the distribution of unremunerated workers among establish- ments of different sizes that appeared in th- 1960 census, with an added constraint that the bulk of this class of worker had to fall within the two smallest establishment categories. 234 WAGES IN MEXICO As a result of this allocation, it is quite possible that the average remuneration in the size categories of 1-5 and 6-25 workers may be somewhat distorted. The overallocationof unpaid workers to one of these two categories would have the effect of raising the average remuneration of the remaining paid workers in that category and lowering that in the other smallestablishment category.For this reason, the averageremunera- tion for the composite categoryof establishmentswith 1-25 workers has also been included, which should provide a very close approximation of the average remuneration of paid employeesfor the entire group of small establishments. Fourth, the concept of remunerations used includes all wage and salary payments plus fringe benefits.The latter tend to be proportionately larger in the larger establishments. Since so many small establishments employed very few or no paid workers, it seemed desirable to try to estimate the returns to unremuner- ated labor, particularly in light of the common belief that a self-employed or unpaid familylabor status represents a last refugein a jobs-scarcelabor market and, therefore, one that is characterizedby extremelylow income. It is possible to arrive at such an estimate for those establishmentswith no paid employees.This was done by deducting from net value added an imputed return to capital.35 The latter, in turn, was calculatedby applying a rate of interest to the capital employed equal to the return on a highly liquid asset that would have been easily accessible to individuals with modest means. (For this purpose, I adopted the interest rate payable on hipotecarios ordinarios, a form of mortgage bond that is availablein units as small as Mex$100; in 1970 the interest payable on these bonds was 8 percent.)36 The residual of net value added, less the imputed returns to capital, was then divided by the number of unremunerated persons re- ported as working in the establishmentswith no paid employees. To the extent that unremunerated workers are likely to include a larger proportion of part-time workers than is true of the remunerated labor force, the estimated return to the former may be understated. In any case, while the absolute value of the returns to the former may contain a significant margin of error, it may not be unreasonable to hope that the trend in these returns does reflecttheir actual course. With these caveatsin mind, let us now examine the course of remunerations over time. Increasesin Wages by Size of Establishment The data in table 7-3 reveal substantial increases in the real average remunerations of employeesin all size strata of the establishmentsand in 37 The total increases in the last column of the table all three sectors.  WAGE DATA, I960-75 235 translate into annual rates of growth in real wages in excessof 3 percent for most strata. Furthermore, a notable narrowing can be observedin the differentialrate of earnings between the smallestand the larger establish- ments within sectors. Only in the industrial ,ector does a widening occur between the smallestand the largest establishmentstrata, but narrowing is again evident between the smallest and all the other large establishment strata. Among the small and medium-sizefirmns, intersectoral differentials appear to have narrowed as well. The one observation that stands out as a significantdeparture from the general observable tendenciesis that for the industrial establishmentswith 500 or more employees.Whereas in 1960 such firmspaid wagesthat were lower than those in "large" establishmentsin either commerceor services, by 1975 earnings there had surpassed wages in the other two sectors by a large margin. Indeed, they recorded the largest rate of increase of any observation within the set of data originating in firms with paid em- ployees. This rapid growth in reported earnings may be attributable to several possible factors. First, it could represent an increasingweight of techno- logically more complex industries that employ a larger proportion of skilled manual workers; thus the large increasein averageearnings would reflect a change in the occupational structure in largefirms.Second,it may be a reflection of a growing structure of managerial and administrative personnel in modern industrial firms. Sinc, the reported remunerations include all personnel and not only wage earners, changing weights be- tween manual and nonmanual employeescould be a contributing cause of the high rate of change in average earnings. A third possibility is that wages in large firms have been increased as a result of collectivebargain- ing. Trade unions are heavily concentrated in the large-firm sector. (The effectiveness of this kind of interventionin the wage determinationprocess will be discussedlater in this chapter.) Fourth, the faster rate of increasein large firmsmay reflectmovementalong an upward-slopingsupply curve of labor. Dependingon how "high" the hiring requirementsof this sector are relativeto the average characteristicsof the manual and nonmanual labor force, the notable expansion of employment there may have required a faster rate of increase in offered wages than in the rest of the large-firm sectors. (In commerceor servicesa "large" firm is one with only nine or more workers, rather than 500 or more.) Although notable gains in real earnings can be observed among the small establishment strata, these may, in fact, understate the actual gains realized by workers over time. One woulcL not expect that the establish- ments in the stratum with the smallest number of employees in 1975 would necessarily be the same ones that were recorded in 1960. To the 236 WAGES IN MEXICO extent that firmswere "successful," they may have grown and moved into larger size strata in subsequent censuses.Since earnings per worker and establishment size are so highly correlated, one mightexpect that the rate of salary increase in a panel of small firms followed in a longitudinal manner would exceed that recorded within any of the smaller size classes. It is reassuring to note that the course of the imputed earnings of unremunerated labor (includingthose of the self-employedand unpaid family labor) resembles that observed for paid labor. The increases re- ported for this classof labor in the industrial and commercesectors ranked among the largest recorded for any of the strata; thus, the differentialin earnings between wage and nonwage labor was narrowed. Furthermore, the same narrowing of intersectoral earnings differentials that was observed for wage labor is present within the category of firms with no paid employees. it was observed above that the legal minimumwage had ceased to be an effective wage from the latter half of the 1940s until sometime in the 1960s. The earnings data from the small establishment strata (table 7-3), which are most likely to reflectlabor market conditions, seem to confirm this. In 1960, the average total remunerations were well in excess of the legal minimums, which in annual terms amounted to Mex$3,610.3 3 The lowest rate of earnings is that for the smallestindustrial establishmentsat Mex$4,973. Even if one allows for the various labor charges that render total remunerations greater than wage payments to employees,it is clear that the going wage comfortably exceededthe legal minimum in1960. By 1965, however, the margin had narrowed significantly. The annual- ized legal minimum expressed in 1960 pesos is Mex$5,328. Earnings in the two sectors for which data are available for 1965, industry and commerce, were well in excess of the legal minimum. By 1970, annual remunerations in the smallest industrial establishments, at Mex$6,855, had slipped below the legal minimumof Mex$6,978. It may be surmised that a substantial proportion of workers so employed were receiving wages below the legal minimum. Averageearnings in smallestablishments in the other two sectors continued to exceed the minimum. Between 1970 and 1975, however, earnings surged and again surpassed the legal mini- mum in the smallestestablishmentsof all three sectors. Throughout these years, wages in the larger establishments lay above the minimum in all sectors. Productivity of Labor The advances in real wages that occurred during the 1960-75 inter- val had their foundations in increasesin the productivityof labor. Tables 7-4 and 7-5 present severalrelevant measuresper worker for the industrial  WAGE DATA, I960-75 237 sector. The preferred measureof productivity,r.et valueadded per worker, increased significantlyin all strata. Only in one intercensalinterval, 1960- 65, were there widely disparate rates of increase among the various size strata; the rate of increaseduring that interval proves to be directly related to establishment size. Thereafter, however, the rates of change tended to converge,and in the 1970-75 interval these are slightlyinverselyrelated to size. Underlying this growth in productivity are increases in the amount of capital per worker employed.These were parti,cularlylarge, in real terms, between 1960 and 1970. For the 1970-75 interval, the increasein invested capital per worker progressed at a rate slower than the consumer price index; this impliesthat, in real terms, investmentwas negative.In fact, it is difficultto believethat this happened, especiallyin light of the increase in net value added per worker that exceededthe surge in prices by a substan- tial margin. It may be that the 1975 census valuation of capital reflectsan original cost rather than a current replacemeni: value. Thus, the amount of capital per worker in 1975, as well as its rate of increase, may be greatly understated. Of interest is the relationship between the rate of growth in productivity and remunerations in the three intercensalintervals. Expansion in produc- tivity outran wage increasesby a considerable margin between 1960 and 1965. The difference narrowed significantly during the next five-year period. Finally, during 1970-75, remunerations increased more rapidly than productivity. This changing relationshipis consistentwith a growing relative scarcity of labor in the market. A rather mixed pattern of change can be observed within the service sector (see table 7-6). Unfortunately, data for net value added are not available for the census periods before 1970. A substitute measure of productivity that can be employed is the volume of sales per worker. During the 1960-70 interval, remunerations outran by a considerable margin the increases in sales receipts per worker in the two larger estab- lishment strata; they did so again over the 1970-75 interval, but by a smaller margin. In the case of the smallest establishments,remunerations increased faster than receipts per worker in 1both periods as well, but the margin of increase was greater in the latter period. For the 1970-75 interval, it is possible to compare changes in net value added and remu- nerations per worker. As in the case of sales per worker, increases in remunerations far outdistanced those in value added. In table 7-6, the observation of rates of :.ncreasein remuneratio-nsin excess of those in productivity in both the service and industrial sector would seemto implya redistribution of incorme in favor of wage and salary recipients, particularly during the 1970-75 interval. Indeed, the national income accounts for the decade do indicate a growth in the share of wages 238 WAGES IN MEXICO Table 7-4. Output, Net Value Added, Capital Employed, and Remunerations Per Worker in the Industrial Sector by Size of Establishment, 1960-75 (currentpesos) Year and Value of Total establishment production Net value Capital remunera- size per added per per tion (number of employed employed employed per paid workers) person person' person employee' 1960 No paid employees 9,351 3,243 6,663 2,710 1-5 22,681 9,223 11,037 4,973 6-25 46,478 12,521 37,046 8,119 1-25 32,913 10,643 22,232 6,644 26-50 60,867 13,642 52,221 10,272 51-1001 101-250 66,065 15,641 62,726 11,854 251-500 501 and over 76,540 16,535 74,959 12,877 Annual minimum wage' 3,610 1965 No paid employees 12,168 6,027 4,346 5,680 1-5 28,154 12,227 10,489 6,593 6-15 49,833 19,826 31,488 8,824 6-25 57,716 22,167 38,327 9,816 1-25 44,919 17,864 26,276 8,724 26-100 82,897 29,797 61,247 13,142 101-500 104,500 39,8S2 87,406 16,534 501 and over 124,620 48,744 104,776 20,029 Annual minimum wage, 6,325 a. For 1960and 1965,net value added figuresrepresentan estimatederivedby subtracting from the total value of production and receiptsthe cost of all purchasedinputsand contracted services. b. For establishmentswith no paid employees,the returns to labor were estimated by subtractingfrom net valueadded the imputedreturns to capital. For all other establishments, total remunerations include all wage and salary payments plus fringe benefits and payroll taxes. and salaries in the national product. From 35.7 percent of GDP in 1970, their share increasedto 38.1 percent in 1975. A further rise to 40.3 percent was recorded for 1976. The interpretation of this increase is clouded by cyclicalconsiderations,however, since it coincideswith a sharp slowdown in the rate of expansion of the economy.In any event, this last proportion was not maintained, for by 1978 the share had declined back to the approximate level of 1975, to 37.7 percent. 39  WAGES SINCE 1975 239 Year and Value of Total establishment production Net value Capital remunera- size per added per per tion (number of employed employed employed per paid workers) person person' person employee' 1970 No paid employees 17,739 7,329 8,074 6,683 1-5 46,672 19,336 20,383 8,980 6-15 71,745 28,937 46,233 12,499 6-25 84,822 33,149 54,279 13,799 1-25 69,823 27,718 40,952 12,193 26-100 120,256 44,382 86,756 18,889 101-500 155,953 60,027 118,893 23,774 501 and over 187,674 74,546 142,637 30,425 Annual minimumwage' 9,851 1975 No paid employees 46,112 17,639 13,982 16,311 1-5 98,576 39,468 29,110 22,208 6-15 152,006 58,324 61,743 27,449 6-25 169,026 63,621 71,873 29,990 1-25 143,613 54,908 56,447 27,739 26-100 243,579 88,505 110,861 39,715 101-500 319,061 121,630 175,446 51,110 501 and over 360,420 144,440 227,803 65,252 Annual minimum wage' 19,334 c. The annual minimumwage for 1965-75 is a weightedaverageof all urban zone rates. Eachzonerate isweightedbythe zone's labor force.Weightedaverageswere not availablefor 1960. The 1960 wageis a simplearithmetic averageof allthe urban minimums.As a rule, the weightedaverage tends to exceed the simple average by approximately 8-10 percent. Sources:Same as for table 7-3. Wages since 1975 An alternative source of wage data for the years since 1968 is the Banco de Mexico. These data, however, are limited to a sample of large firms primarily in the manufacturing sector. Table 7-7 presents an index of nominal and real earnings, including fringe benefits, from 1968 through 1982. An index of the weighted average, real arban minimum wage is also included for comparison purposes. As can be seen, the course of the two wage series is quite similar. That for earnings is somewhat smoother than 240 WAGES IN MEXICO Table 7-5. Rates of Change in the Value of Production, Net Value Added, Invested Capital,and Total Remunerations Per Worker in the Industrial Sector by Establishment Size, 1960-75 (percent) Factorand establishmentsize (number of workers) 1960-65 1965-70 1970-75 1960-75 Value of production per worker No paid employees 30.1 45.8 159.9 393.1 1-5 24.1 65.8 112.1 334.6 6-15 n.a. 44.0 118.7 n.a. 6-25 24.2 47.0 99.3 263.7 26-100 36.2 45.1 102.6 300.2 101-500 58.2 49.2 104.6 382.9 501 and over 62.8 50.6 92.0 370.9 Net value added per worker No paid employees 85.8 21.6 140.7 443.9 1-5 32.6 58.1 104.1 328.0 6-15 n.a. 46.0 101.6 n.a, 6-25 77.0 49.5 91.9 408.1 26-100 118.4 48.9 99.4 548.8 101-500 154.8 50.6 102.6 676.6 501 and over 194.8 52.9 93.8 773.5 Invested capitalper worker No paid employees -34.8 85.8 73.2 109.8 1-5 -5.0 94.3 42.8 163.7 6-15 n.a. 46.8 33.5 n.a. 6-25 3.5 41.6 32.4 94.0 26-100 17.3 41.6 27.8 112.3 101-500 39.3 36.0 47.6 179.7 501 and over 39.8 36.1 59.7 202.9 Total remunerationsper worker No paid employees 109.6 17.7 144.1 501.9 1-5 32.6 36.2 127.3 346.6 6-15 n.a. 41.6 119.6 n.a. 6-25 20.9 40.6 117.3 269.4 1-25 31.3 39.8 127.5 317.5 26-100 27.9 43.7 110.3 286.6 101-500 39.5 43.8 115.0 331.2 501 and over 55.5 51.9 114.5 406.8 Average legal minimum wage 62.0 55.7 96.3 395.7 Consumerprice index 9.8 19.3 76.4 131.1 n.a. Not available. Sources: Secretaria de Industria y Comercio, Direcci6n General de Estadistica, Censo industrial,1961, 1966, and 1971; Secretariade Programaci6ny Presupuesto,Censo indus- trial, 1976.  NZAGES SINCE 1975 241 Table 7-6. Service Sector Receipts, Value Added, and Remunerations Per Worker, 1960-75 Total Year and Annual wages remuneration establishment size Receipts per Valuead(ied and salaries per (number of workers) worker per worker per worker worker' 1960 Current pesos No paid employees 14,540 n.a. - 6,315 1-3 workers 20,796 n.a. 5,259 5,397 4-10 35,485 n.a. 7,260 7,869 11 and over 69,368 n.a. 14,911 16,470 1970 No paid employees 21,144 13,249 - 9,062 1-2 workers 48,412 33,816 11,484 12,837 3-8 45,389 29,689 12,043 13,673 9 and over 89,058 58,509 20,817 24,415 1975 No paid employees 37,781 21,390 - 19,055 1-2 workers 68,797 43,402 18,197 20,002 3-8 82,399 49,127 22,427 25,643 9 and over 128,955 73,304 34,656 41,678 Percentagechange 1960-70 No paid employees 45.4 n.a. - 43.5 1-3 or 1-2 132.8 n.a. 118.4 137.9 4-10 or 3-8 27.9 n.a. 65.9 73.8 11 or 9 and over 28.4 n.a. 39.6 48.2 Consumer price index, 31.0 1970-75 No paid employees 78.7 61.4 - 110.3 1-2 42.1 28.3> 58.5 55.8 3-8 81.5 65.A 86.2 87.5 9 and over 44.8 25.3 66.5 70.7 Consumer price index, 76.4 1960-75 No paid employees 159.8 n.a. - 201.7 1-3 or 1-2 230.8 n.a. 246.0 270.6 4-10 or 3-8 132.2 n.a. 208.9 225.9 11 or 9 and over 85.9 n.a. 132.4 153.1 Consumer price index, 131.1 n.a. Not available. - Not applicable. a. Seetable 7-4, note b. Sources:Secretariade Industria y Comercio,Direcci6n General de Estadistica,Censo de servicios,1961, 1966, 1971; Secretariade Programaci5ny Presupuesto,Censo de servicios, 1976. 242 WAGES IN MEXICO Table 7-7. Index of Nominal and Real Average Remunerations of Production Workers in Manufacturing, 1968-82 (1978 = 100) National Index of Index of consumer Index of real weighted nominal price real averageurban Year earnings index earnings minimum wage 1968 24.2 29.7 81.5 78.6 1969a 25.6 30.7 83.4 76.1 1970 26.8 32.2 83.2 84.4 1971a 29.4 34.0 86.5 79.9 1972 31.3 35.7 87.7 90.0 1973 35.2 40.0 88.0 85.1 1974 44.2 49.5 89.3 93.1 1975 53.5 57.0 93.9 93.5 1976 66.7 66.0 101.1 104.4 1977 86.5 85.1 101.6 103.5 1978 100.0 100.0 100.0 100.0 1979 117.2 118.2 99.2 98.8 1980 144.7 149.3 96.9 92.1 1981 194.0 191.1 101.5 94.2 1982 303.6 303.6 100.0 100.7b Note: Remunerations include wages plus fringe benefits of production workers. a. Years in which no adjustment in minimum wages was made. b. Takes into account the increase decreed in March (see note g of table 7-1). Sources:Remunerations in manufacturing:Direcci6nGeneral de Estadistica,Estadistica industrialmensual (Mexico,D.F., various issues);minimumwages: Comisi6nNacional de los SalariosMinimos, Salariosminimos (Mexico,D.F., various issues). that for the minimum wage and reflects the discontinuous adjustment of the latter in the face of a changing price level. Real earnings continued to increase throughout the early and middle years of the 1970s despite an 40 acceleration in the rate of inflation. Beginning in 1978, however, the rate of increase in earnings lagged behind that in prices, so that by 1980 real earnings had fallen 4.6 percent below their 1977 level. The real minimum wage declined 10.5 percent over the same interval. These declines in real wages coincided with an accelera- tion of the rate of inflation from 17.5 percent in 1978 to 26.3 percent in 1980. Not all forms of wage payment nor all groups of workers were equally affected. For example, during 1978, the real wages of salaried white-collar workers suffered a greater decline in average earnings than did those of manual workers in industry. During this interval, a somewhat steeper decline in wage rates was partially offset by an increase in fringe  MrAGES SINCE I975 243 4 benefits.1 As will be seen below, fringebenefitpayments increasedrelative to wage rates in 1979 and 1980 as well. The course of real earnings was reversed in 1981 as earnings climbed back to their 1977 level.They continued to riseduring the firstfivemonths of 1982 eventhough the pace of inflation was quickening. During the rest of the year, however, real wages again declined as the Mexican economy slid into a serious economic crisis with output falling and inflation accelerating.On average,earnings were only slightlybelowthe 1981 level but were clearly deteriorating from midyear on, a course that was to extend into 1983. The declinesin real earnings reported for 1979 and 1980 are surprising in view of the strong expansionary forces that were in evidencein those years. Industrial output expanded at a 9 percent rate in each year, and industrial employment increased by an estimated 7.3 and 6.7 percent. Labor shortages were reported, not only of skilled workers, but even of 4 2 Increasesin unskilled in severalsectors and urban centers of the country. total employmentwere held to have outrun the growth of the labor force. Under these conditions, one would not ordinarily expect to observe real wages falling. These divergent trends in economic activity and employ- ment, on the one hand, and in real wages,on the other, would seem to raise important questions about the determinants of wages in this part of the manufacturing sector. Attention might also b., directed to the relationship between these wages and those in the broader labor market. These issues will be addressed later in the chapter. Wages of Domestic Servants In an earlier section, which traced the evolution of wage payments in three major sectors by size of establishment,we observed that real earn- ings appeared to have risen substantially in all size strata, even in those in which the impact of institutional forces might be expected to be weak or absent. Those findings suggestedthat labor market forces of supply and demand were effectivelypushing wage levels upward in such firms. This observation may be checked against the behavior of wages in another market which is widely acknowledged to lie beyond the influence of institutional factors-the market for domestic servants.Within this mar- ket wage changes over time might be expectedto reflectchanges in supply and demand conditions in the broader market for low-skilled female labor. Wage data for femaledomesticsin Mexico City for the years 1963 through 1975 are available in a study appearing in a publication of the labor ministry.43 A wage series was constructed from the monthly wages offered in advertisements in two of the leading newspapers of the city, 244 WAGES IN MEXICO Excelsiorand El Universal. Between 1963 and the early 1970s, the real cash wage of domestic servants in Mexico City was estimated to have risen on the order of 25-35 percent (see table 7-8). While this rate of increase lagged behind the rise in the real legal minimum wage for Mexico City of about 50 percent by 1970, it did not compare unfavorably with the rate of increase in earnings in consumer goods manufacturing of about 36 per- cent. Nor is it much different from the advance recorded between 1960 and 1970 for the small-establishment strata of the commerce and industry sectors in table 7-3. The course of real cash wages, however, took an abrupt turn following 1972 as inflation began to accelerate. By 1975, real cash wages had fallen by about 19-27 percent.4 Most of the decline occurred in 1974 and coincided with the more than quadrupling of the rate of inflation in that year compared with the rates characteristic of the first three years of the decade. The decline in real cash wages may simply reflect the lack of experience with inflation in a society that had enjoyed two decades of relative price stability. Lacking sophistication or any organizational sup- port, domestic servants may not have adjusted their reservation price sufficiently rapidly to prevent a decline. The fact that they usually receive a very substantial proportion of their remuneration in kind (that is, room and board) may also have tended to blunt their awareness of the full extent of the decline in the purchasing power of their cash wage. The pattern of domestic servant remunerations since 1975 is not re- corded in any statistical source in Mexico. As a first approximation to the measure of the change that has occured since 1975, samples of cash wages offered for live-in household help had to be culled from the same two newspapers employed in the survey cited above. The samples were drawn from various issues in the months of July and August 1979, July 1980, and May 1981. They provided evidence that the loss suffered earlier in the decade had not only been recouped but that substantial further gains in real earnings had been made. The offers observed for 1979 averaged 70 percent above those for 1975 and almost one-third higher than the pre- vious peak of 1970. A year later, the average real wage offered had increased again. It was double that of 1975 and 55 percent greater than 1970. Lest the sample not be representative of the actual average offered wage but rather biased upward, the lowest frequently quoted wage offer was also compared with the past recorded averages. A cash wage of $2,500 was adopted as the minimum for 1980. It was at the bottom of the Mex$2,500-3,000 interval within which just 50 percent of the observa- tions fell. (Only six of the seventy-eight offers were for less than Mex$2,500 per month.) Even this wage, in real terms, proved to be 50 and 19 percent greater than the 1975 and 1970 averages, respectively.  WAGES SINCE I975 245 The sharp upward trend in real wagescontinued into 1981.The average of sixty-ninewage offers was Mex$4,862 per month, or 89 percent greater in real terms than the previous high of 1970. The most frequently cited minimumof Mex$3,500 was 40 percent abov.-the preceding year's mini- mum and 14 percent higher in real terms. A notable feature of the 1981 advertisementswas that there was a much wider range of offers than in earlier years. The coefficientof variation mea;ured 29.1 percent as com- pared with 22.1 and 23.2 percent in 1979 and 1980, respectively. If one were to judge by the incidence in 1981 of extremely high salary offers, those between $6,500 and $8,000, market pressureson wages must have continued to be strong. Table 7-8. Average Monthly Cash Wages Offered to Domestic Servants, Mexico City, 1963-8:1 (current pesos) M ethod II Method I Index of e.;timated Index of monthly real wages monthly real wages Year wage' (1963 = 100) wage" (1963=100) 1963 277 100.0 258 100.0 1964 288 101.8 283 107.3 1965 320 109.0 308 112.6 1966 349 114.1 333 117.0 1967 360 114.5 358 122.3 1968 363 113.9 383 128.9 1969 356 107.4 408 132.2 1970 447 128.2 433 133.5 1971 427 116.1 458 133.7 1972 478 123.6 483 134.1 1973 511 118.7 508 126.8 1974 532 101.1 533 108.7 1975 615 99.8 558 97.3 1979, July 2,164 152.8 n.a. n.a. 1980, July 3,260 198.8 n.a. n.a. 1980, July' 2,500 152.4 n.a. n.a. 1981, May 4,862 242.4 n.a. n.a. 1981, May' 3,500 174.5 n.a. n.a. n.a. Not available. a. Method I is a simple average of the sample offers in each year. b. Method I estimated an annual wage from a regression line fitted by least squares to the sample averages. Both methods are reported as published. c. Minimum wage offered. Sources: "Analisis del mercado de los servicios dom63ticos en Mexico," Cuadernos de empleo, no. 1 (Mexico, D.F.: Secretaria de Trabajo y Prcvisi6n Social, 1976), pp. 55-114. The data for 1979-81 were culled from issues of Excelsior and El Universal. 246 WAGES IN MEXICO That this was the case should not be surprising.Domestic serviceis not considered a preferred form of employment, and the widespread labor shortages, even of unskilled labor, should have meant a wider range of alternative employments open to women at higher wages. For example, even the most casual empiricistcould not have failed to notice the sharp increase in the number of women employed in construction projects between 1979 and 1981 in Mexico City. Not only did their number increase, but also the variety of tasks to which they were assigned. What is notable about this wage series is that it stands in such sharp contrast to that recorded in recent years for the large-firmsegment of the industrial sector in table 7-7. Real wages for domesticservantsevidenceda strong upward movement at the same time that those in the large-firm stratum of the industrial sector appear to have been falling. In the absence of additional information about wages in other parts of the labor market, these cross-currentsare difficultto interpret. However,past wage trends in the market for domestichelp do not suggestthat that market was isolated from the broader market for low-skilled labor. If, indeed, these two markets are closely integrated, then it is reasonable to expect that other unskilled wage rates that are largely market determined may also have been rising generally. The possibility cannot be ruled out that the sharp advance in the real wages of domesticswas a unique occurrence and hardly representativeof the course of other wages for low-skill jobs. After all, it is not ususual to observe money wages laggingprice increasesin the context of accelerating inflation. What sets off these years in Mexico from other inflationary episodesin Latin Americais the very strong expansionaryprocessthat was unfolding at the same time. With open unemploymentat very low levels, the large increases in the demand for labor may very well have over- whelmed the common tendency for money wages to lag prices. Industrial Wages Conversely, if real wages were rising throughout most of the labor market, then the lackluster performance of industrial earnings, as repre- sented by the Banco de Mexico wage index, would appear to be the anomaly, somewhat reminiscentof their behavior during the decade of the 1940s. At least two possible explanations of their relatively poorer per- formance during 1978-80 can be offered. The firstto be consideredhere is the institutional setting within which wages in the large-firm industrial sector are determined; the second is a possible change in the composition of the work force. The large-firmsector is also one in which workers are most likelyto be  WAGES SINCE 1975 247 organized and in which wage levels are su':ject to determination by collectivebargaining. All collectivebargaining agreements in industries falling under federal jurisdiction must be registered with the Direcci6n General de Conciliaci6n.In casesin which the parties are unable to agree to terms, the conciliation serviceintervenes in an attempt to facilitate an agreement.It is at this juncturethat the governnent is in a position to exert its influenceon the course of wages in the organized sector. 45 At the end of each year, the administration formulates guidelinesde- fining the desired rate of wage adjustment for the coming year. The adjustment of the legal minimum wage, which becomes effective at the beginning of each year, signals the target rate of adjustment that the government considers appropriate for the private and public sector wage contracts. In the past, these guidelines seem to have been arrived at by a process of discussionand negotiation within the leadershipof the govern- ment party, the Partido RevolucionarioInstitucional (PRI). Since the labor movement and the private sector are representedat the highestlevelsof the party, the definitionof guidelinesacceptableto the principal political and 46 economic interest groups could be achieved before their promulgation. In recent years, however, the guidelines appear to have taken on an increased significance.In view of the high rates of inflation experienced since the middle of the 1970s, the guidelines have been formulated with two objectives in mind. One is the defense (preservation)of the real wage to protect the standard of living of the wage-earningclass, the other to restrain the inflationary impact of wage adjustments on costs. Thus, the annual wage adjustments have beengreatly influencedby the expectedrate of inflation during the ensuing year. In 1979, when inflationwas expected to approximate a rate of 15 percent, a guidelinewas adopted allowing for a 13.5 percent adjustment. In fact, the price levelrose at a faster rate than predicted, and wages that did conform to the guidelinestended to suffer declines in real terms, as did the minimum wage. In 1980, the guidelinesat the beginningof the year defineda 19 percent rate of adjustment. For the first time since the declaration of the interim emergencywage adjustment of 22 percent, following the devaluation of 1976, the government revised its guidelines as the year progressed. For contracts negotiated from March through June, a 20 percent rate was set, 22 percent for July and August, 25 percent for Septemberthrough Novem- ber, and 27 percent for December. 47 Sincethe price level rose by over 26 percent during 1980, those bargains concluded earlier in the year were subject to attrition in real terms to the extent that their terms adhered closely to the guidelines.Since the guidelines adopted were based on the expected rate of inflation for the coming year and did not attempt to recoup any loss in real wages that might have occurred during the preced- 248 WAGES IN MEXICO ing year, an underestimationof the rate of inflationwould tend to lead to a progressivedeclinein the real wage. This was observed in the wage index of the Banco de Mexico. In view of the shortfall of the guidelinesrelativeto the rate of inflation, actual wages would decline in real terms only to the extent that they actually conformed to the guidelines.In fact, of course, the guidelinesdo not constitute a rigid ceiling on the size of wage increases.Rather, they definedthe floor on the sizeof adjustments acceptableto the trade unions. Indeed, some large-firm managers expressed the view that their wage bargains were expected to exceed by 1-1.5 percentage points the adjust- ment in the minimum wage which also served to define the guideline. An extensivestudy of the terms of settlementof a sample of agreements concluded in 1977 and 1979 has been undertaken by Cesar Zazueta and his associatesat CENIET.48 These provide a basis for comparing the actual rates of adjustment with those of the guidelinesfor those years. The 1977 guidelinewas set at a modest 10 percent increase,coming as it did on the heels of the major emergencyreadjustmentof 22 percent in October 1976. In addition, a new administration had just entered officewith the declared objective of reducing the rate of inflation. Given the economy's decline into recession and the general air of crisis that prevailed following the devaluation, the government was able to gain a widespread commitment to wage adjustments conforming to the guidelines.Zazueta's data reveal that departures from the guidelinesdid occur in 1977, but that these were relativelyinfrequent and were held within narrow bounds. Beforea review of his findings,a few comments are appropriate on the way in which his data are organized. The organization correspondsto the two fundamentalkinds of contract negotiation that occur under the Mexican industrial relations system. Basic contracts are negotiated for a two-year term, and negotiations may treat not only basic salariesand fringe benefitsbut other nonwageterms of employment as well. A reopening clause permits renegotiation only on basic wages during the term of the contract, but it is possible to seek adjustments in particular occupational rates as well as general wage changes. Zazueta's data are drawn from thirty industrial groups that have been designatedas lying within the jurisdiction of the federal authorities. These include the largest manufacturing industries as well as electric power generation, construction, transportation, and communicationsin- dustries. Public as well as private sector firms are included. The covered sectors account for the bulk of the trade union members of the country. 49 For each of the covered industries, the average rate of change in basic wages is provided separately according to the nature of the settlement, contract renegotiation, or wage reopening.  WAGES SINCE 1975 249 Thus, in 1977, there were fifty-one observations of rates of wage change distributed over the twenty-seven industrial groups that were involved in negotiations and conciliation during the year. Each of these observations represents an average of all bargains reached within the industry with the participation of the conciliation service. In all, 396 separate agreements were so concluded. Of the fifty-one average change rates, fully forty-two provided for adjustments in basic wages that averaged within plus or minus one percentage point of the 10 percent guideline established for the year. An additional six fell between 11.1 and 12 percent; only three rates exceeded 12 percent. No trend in the size of settlements is discernible over the year. Given the air of economic crisis that characterized much of the year and the honey- moon period that any new administration ordinarily enjoys, it may have been easier to persuade the parties to the collective bargaining process to practice the restraint the government was urging. In addition, the recent grant of the emergency adjustment may have removed some of the urgency of larger adjustments during 1977. That adjustment proved to have effec- tively assured workers of no loss of real earnings, on average, over the year. In the case of those parties negotiating a new contract, restraint on basic wages could be offset, at least in part, by more generous concessions on fringe benefits. Indeed, this seems to have occurred in 1977. Whereas the average increase in basic wages of 10.05 percent, agreed upon in the renegotiation of contracts, was virtually identical to the official guideline, the increase in total remunerations averaged 11.16 percent. When adjust- ments in remunerations accorded both in contract reopenings and in renegotiations are considered, the overall average increase amounted to 10.94 percent. This proves to be smaller than the actual increase in remunerations recorded in the large-establishment sector by the wage index of the Banco de Mexico. According to that index, the November 1977 level of remunerations stood 13.2 percent above that of the previous November.50 Whether the latter figure is the product of wage drift within the covered sector, higher rates of increase in firms that reached accords without benefit of conciliation or that are nonunion, or a general preva- lence of large overtime payments cannot be determined on the basis of the information at hand. The close adherence of contract settlements to the guidelines observed during 1977 was not duplicated in 1979. The guideline for the latter year was set at 13.5 percent. Zazueta's data, disaggregated on a monthly basis, reveal a very close adherence in those settlements concluded in January. Basic salaries were adjusted by 13.76 and 13.95 percent in the renegoti- ated and the reopened contracts, respectively. As the year progressed, 250 WAGES IN MEXICO however, deviations from guidelinesbecome increasinglyevident. During the last three months of the year, wage adjustments were averaging between 15 and 16.5 percent. Much less restraint was shown in the adjustment of fringe benefits, perhaps as the price of closer adherence to the guidelinesof adjustments in the more visiblebasic wages. This can be observed in the settlements achievedin contract renegotiations. Whereas basic wages were adjusted upward by 14.36 percent on average over the year, total remunerations increased by 18.75 percent. The average of the contracted changes in total remunerations over both contract reopenings and renegotiations amounted to 17 percent, well in excess of the guide- lines. But even this change failed to equal the rate of inflation of over 18 percent. This negotiated increase in remunerationsvirtually matched the increase of 16.3 percent recorded by the Banco de Mexico's wage index (table 7-7). The experience of 1979 revealedthe impracticability of promulgating a single guideline for the year in the face of accelerating inflation. It had already become evident that the rank and file of organized labor was becomingincreasinglyimpatient with the lag of wages behind prices and increasinglyunwillingto acceptthe limitationson wage increasesposed by the guidelines.Furthermore, 1980 saw a continuation of the strong expan- sionary trends in output and employment of the previous year. This continued buoyancy of the economy might reasonably be expected to reduce employers' resistance to the wage demands made by organized labor. A recognition of these conditions led to successiveincreasesin the guidelinesas 1980 progressed and it became obviousthat the actual rate of inflationwould surpass the 19 to 20 percent rate that had beenexpected to prevail at the beginning of the year. A review of the contractual changes agreed upon with the intervention of the conciliation service during 1980 reveal a close adherence of basic wage changes to the guidelinesof each subperiod during the year. Depar- tures from the guidelines in most months averaged no more than 1 percentage point above the guideline.Only in July and Augustdid actual wage adjustments exceed the guideline by more, 1.6 and 4 percentage points, respectively.The averagerate of adjustment for all contract settle- ments over the year proved to be 22.5 percent, as compared with the 26.3 percent rate of increasein the consumerprice index. Only the settlements effectedin Decemberprovided for a wage adjustment greater than the rise 5" As in 1979, however, concessionsmade in the form of in the price index. fringe benefitsincreased the rate of adjustment in total remunerations to 52 25.6 percent, less than 1 percentage point short of the change in prices. Thus, on average over the year, contractual arrangements provided for wage adjustments that maintained real remunerations at a levelvirtually constant with respect to 1979.  WAGES SINCE 1975 251 In view of these contractual adjustments, the declineof 3.1 percent in average real remunerations reported by the Banco de Mexico wage index comes as a surprise. Of course, the contractual adjustments reported above refer only to those that were achieved under the aegis of the conciliation service. An unknown number oil additional wage bargains were concluded without the intervention of che service. It is highly un- likely, however, that any of those bargains settled for less than the guide- lines-the union could be assured of getting at least the guidelines by taking the dispute to the conciliation service. More likely,those bargains concluded independently of the servicegranted wage concessionsequalto or greater than those obtainable through conciliation. The second possible explanation of the reported decline in average remunerations may lie in a change in the compositionof the work force. In both 1979 and 1980 employment in the industrial sector increased by about 7 percent. Employersfacedwith expandinglabor requirementshave two options available. They can hire workers at the entry levelwage and, after the probationary period of three months, change their status to permanent employees,whereuponthey becometenured and dischargeable only at heavy cost to the employer. Alternatively,if employersare uncer- tain about the permanence of the higher staffing levels occasionedby the economicboom, they are more likelyto opt for the hiring of eventuales,or temporary workers, on short-term renewab e contracts of up to three months' duration. Such workers are generallyemployed at wages below those of the permanent work force and enjoy none of the fringes accorded the latter. Thus, a sizable expansion of the industrial labor force based on temporary workers would tend to bias downward the level of average earnings of the work force as a whole. The decline in the average levelof real remunerations in the large-firm sector reported by the Banco de Mexico wage index is therefore not necessarily inconsistent with the maintenance of stable real wages for each of the various componentsof the industrial labor force. The relative stability, or possibly the smal] decline, in real earnings in this large-firm industrial sector in the face of rapidly expanding employ- ment suggeststhat wagelevelsthere may have been abovethe market price of the classes of labor employed. Thus, some attrition in the real wage, relative to wages elsewhere,could occur without increased turnover rates or difficultiesin hiring. Also mitigatingan immediateturnover response to a deteriorating relativewage position is the valueof vestedproperty rights associated with permanent employment status. It does not follow that the stability or small decline in real wages observedin the organized large-firmsector extended to other segmentsof the labor market. Indeed, in an environmentof vigorous economicexpan- sion, one might expectto findwage differentialsnarrowing as lower-wage 252 WAGES IN MEXICO firms are the first to feel the impact of a tightening labor market. If the increases recorded in the wages offered to domestics are indicative of market conditions for low-skilledlabor generally,then such a narrowing may well have occurred as real wages rose in those parts of the market in which institutional constraints on wageswere not operative.Unfortunate- ly, the paucity of timely and representativewage data make it impossible to determine the courseof wages for various subsectorsduring these boom years. The trends in wages in the large-enterprisesectorcould be interpreted as indicating that the trade union movement in Mexico has had relatively little independent power to affect the monetary terms of employment. Whatever influence it has is derived from its institutional role within the structure of the dominant political party, the PRI. Alternatively,it mightbe arguedthat the organized sector has had bargainingpower independent of government support but has chosen not to exerciseit in deferenceto the economic policy objectives of the government. The organized sector would have had an opportunity to influence these objectives through bargaining within the inner councils of the party. Thus, once a policy has been adopted, the labor movement, as representedby the dominant labor federation, the Confederaci6nde Trabajadores de Mexico (CTM), fallsinto line. However, some of its affiliateswith a greater than average degree of independence and power may hold out for settlements that exceed the agreed-uponguidelines.In either event, the governmentwould emergeas a formidableinfluenceon the course of wages in many large industrial firms. Effects of the Minimum Wage Since the government's discretion in defining the money terms of em- ployment, through the minimum wage mechanismand its intervention in collectivebargaining, is so broad, one might appropriately pose a question regarding the result of its decisions.Has government intervention tended to produce substantial distortions in actual wages compared with the market price of labor? By and large, if the wage policies of Mexican government administrations were viewed over the long run, it would not seem that these have had serious distortive effectseventhough wages have occasionallybeen peggedwell above the market price of the relevant class of labor. For example, the Cardenas administration, which expressed strong sympathies for the objectives of organized labor, encouraged and supported trade unions in their pursuit of high wages. These same con- cerns for the status of wage earners were reflectedin the establishment in the middle 1930s of legal minimumwages that were well in excessof the  EFFECTS OF THE MINIMUM WAGE 253 opportunity cost of labor. Nevertheless, even the Cardenas administration was unable, or proved unwilling, to protect: the gains made in the early years of the administration. Real wages in large firms and the real mini- mum wage had already begun to decline during the last half of Cdrdenas's administration. As was observed earlier, the fall in both wage measures was to continue through the 1940s and into the early 1950s before they stabilized. Apparently, organized labor lacked the independent power to resist the decline. By the late 1960s, the legal minimum wage had once again been raised to a level at which it became an effective minimum. However, the distor- tive effects of a minimum that lies above the market price of unskilled labor would appear to be fairly minimal in Mexico since evasion is so widespread throughout the small-enterprise sector. The limited effective- ness of the legal minimum has been demonstrated by a survey of 2,500 households in Mexico City that reported that 26.9 percent of the em- ployed and remunerated labor force between the ages of 21 and 60 in 1969 53 earned less than the legal minimum wage then in force. (The respondents included self-employed as well as salaried workers, however, and the survey did not distinguish part-time from full-time earnings.) Particularly among low-skilled occupational groups characterized by low levels of formal education there were large proportiDns that earned less than the minimum wage. For example, 36 percent of the unskilled and 27 percent of the semiskilled production workers, 65 percent of the unskilled and 24 percent of the semiskilled construction workers, and 35 percent of the retail clerks reported earnings less than the niinimum wage (see table 7-9). Virtually no occupational group, including technical, semiprofessional, and managerial classes, was without individuals reporting subminimum earnings. To be sure, some of these low earnings were undoubtedly associated with part-time employment, but a majority probably reflected earnings for a full work week. What is noteworthy about the data is that the average earnings of occupational groups with substantial numbers of subminimum earners lie well above the minimum wage. This suggests a large variance in the earnings distribution within each group. Only the unskilled construction workers and unskilled service workers reported average earnings below the minimum. The latter, however, include domestic workers who receive substantial income in kind for which no accDunting is made. Also worthy of note is the close direct relationship between earnings and years of schooling. Whether the legal minimums have had substantial distortive effects on wages in the modern industrial sector is more difficult to determine with any degree of confidence. In practice, most large firms treat the legal 254 WAGES IN MEXICO Table 7-9. Earnings and Educational Levels of Selected Occupational Groups in Mexico City, 1970 Proportion earning Average less than the Average monthly minimum years earnings wage of schooling Occupation (pesos) (percent) completed Professionals 5,679 0.0 15.4 Technicians 3,293 9.0 11.6 Semiprofessionals 2,542 8.8 11.5 Upper- and middle-level public employees 2,643 6.6 10.7 Owner-managers Construction 2,538 55.2 5.5 Manufacturing 3,225 13.0 7.1 Commerce 3,187 22.4 6.2 Services 3,056 20.8 8.0 Office employees 1,890 8.6 9.1 Mozos and office boys 1,134 22.2 6.1 Retail clerks 1,376 34.9 5.9 Peddlers 1,113 62.4 3.2 Service workers Skilled 1,374 27.3 5.2 Unskilled 605 77.9 3.4 Drivers 1,608 9.8 5.0 Production workers Skilled 1,631 18.1 6.0 Semiskilled 1,215 26.7 4.7 Unskilled 1,016 35.5 4.1 Construction workers Skilled 1,198 20.7 4.8 Semiskilled 1,122 24.3 2.8 Unskilled 810 64.9 2.2 Legal minimum wage 847 Note: In 1970 the exchange rate was 12.5 pesos to the U.S. dollar. Source: Humberto Mufioz Garcia, Orlandina de Oliveira, and Claudio Stern, "Migraci6n y marginalidad en la ciudad deMexico," El perfildeMexico en 1980, vol.3,6th ed. (Mexico, D.F.: Siglo Veintiuno Editores, 1979), pp. 354-57.  EFFECTS OF THE MINIMUM WAGE 255 minimum as an entry-levelwage for unskilledworkers, but most low skill occupations are reimbursed at rates above the minimum. Although the wage paid for unskilled labor may thus exceed that paid in the informal labor market, the quality of labor demanded by the larger establishments may also be quite different. To the extent that the relevant labor market for more skilled occupations is "internal," that is, skilled positions are filled by the progressive advancement of workers from less skilled jobs rather than from outside the firm,newly hir-d workers enteringunskilled job positions may be expected to have the qualifications required for advancement. Clearly, the reservation price of such workers must be higher than that of workers unlikely ever to qualify for advancement beyond the ranks of unskilledlabor. Had the legal minimumbeen substan- tially higher than the market wage for the quality of workers recruited by the large-firm sector, one might haveexpect,d a continuingslidein wages in this sector as the real minimumdeclined.As can be seen in tables 7-1 and 7-7, however, wage declinesmatch those in i:helegal minimumfrom 1976 through 1980. Thereafter, earnings stabilized while the minimum con- tinued its downward course. To test whether the minimumwage has substantiallydistorted the price of unskilled labor to the modern sector relative to its market price, one might observe the course of occupational differentialswithin the modern sector after the legalminimumhad begun to increasein real terms and had once again assumed its role as an effectivewage. One might hypothesize that, if the minimum were increasing over time relative to the market- determined price of unskilled labor, this would tend to produce a narrow- ing in skill differentialsas unskilled rates were pressedupward. Unfortu- nately, reliable data on occupational earnings are scarce, and studies of occupational wages are understandably alsD scarce. One effort by Jesus Reyes Heroles traced the course of differentialsin manufacturing in Mexico City for a very reducednumber of occupational titles from 1940 to 1976.54 His findings suggest a fairly small degree of compressionfrom the early 1950s,when the legalminimumwas belowthe market wage, to 1976. The ratio of a mechanic's wage to that of entry- level occupational titles, such as mozo or pe6n, fell from 2.1:1 to 1.8:1.5 All of the narrowing occurred between 196Cand 1968, an interval during which the real urban minimumwage was increasedby about 80 percent. It is not clear that even this contraction can be attributed to a legal minimum wage peggedat above the market level.Rey-s Heroles's tables implythat the mozos' wages and, by exension, other unskilled wage rates were consistently above the legal minimum. 5 6 This can be inferred from data showing the ratio of the mechanic's wage to the legal minimum to be 256 WAGES IN MEXICO substantially greater than that of the mechanic's wage to the mozo's; during the 1970s, the former ratio fluctuated between 2.3:1 and 2.7:1 in contrast to the 1.8:1 ratio of the latter.57 Apparently, skill differentials of all kinds were markedly narrowing during the decades of the 1960s and 1970s. ReyesHeroles followed the ratio of the highestpaid manual occupationaltitle to the mozo's wageover time and found the ratio decliningcontinuously from about 3.3:1 in the early 1950s to about 2.4:1 in the middle 1970s. 5 8 White-collar occupa- tional earnings are also shown to have declined relative to blue-collar wages. ReyesHeroles charts the ratio of various officesalariesto those of mechanics over the period 1962-75. Earnings for a bilingual secretary declined from about 1.8:1 to 1.4:1; those for a cashier fell from 1.7:1 to 1.2:1.59 Since the wage of a mechanic had declined relative to unskilled rates, this implies that the white-collar earningshad also declinedrelative to the low-skill manual occupational groups. It is quite possiblethat part of the narrowing that occurred during the decade of the 1960s was traceable to the rapid increase in the real value of the minimum wage. However, that this contraction can be observedat most points within and between the manual and officeworker wage structuressuggestscontribut- ing factors other than the legal minimum. One such factor could be the large expansion of the supply of individuals with greater formal educa- tional and vocational training. In any event, over the decadeof the 1970s, it would appear that any gap that might have existed between the legal minimumand the market price of urban unskilled labor must havediminishedsubstantially.It was noted earlier in the chapter, for example, that, in contrast to the decade of the 1960s when wage-level changes in the small-enterprisesectors lagged behind changes in minimum wage, during the first half of the 1970s remunerations in those sectors advanced at a faster rate than did the minimum wage. In real terms, the former increased almost three times more rapidly, 29 to 11 percent. If the rapid rise of domestics'earnings at the end of the decade is indicative of conditions in the unskilled labor market generally,market-determinedwages must have been outstripping the advances in the minimum, especiallysince the latter were fallingin real terms after 1976. Although minimum wage policy may appear to have had only limited distortive effects on the price of industrial labor, can the same be said for rural labor? As was mentioned in an earlier section, a sharp change in policy occurred during the late 1970s with respectto the rural minimums. The minimum wage was invested with new purposes: to effect a redis- tribution of income in favor of the lowest-wageworkers and to reduce the  EFFECTS OF THY MINIMUM WAGE 257 income incentives for migration from rural to urban areas. As a result, the differential between rural and urban minimunmwages was rapidly reduced as rural and urban minimums were merged within zones and the number of different minimums was sharply reduced. On the surface, the results of this procedure would not appear to have had a serious distortive effect. After all, the real rural minimum wage was increased between 1976 and 1981 by only about 9 percent on average. However, the increases in the lowest wage zonies in Chiapas, Oaxaca, and Queretaro were more on the order of 89 percent. Were these to represent the effective increases in wages paid in such traditionally poor and low- wage areas, one might expect some substantial labor-saving adjustments to follow. This might obtain especially in commercial agriculture, which is a major employer of rural labor-and in which it may be more difficult to evade the legal minimum. In the small landholder sector, the new higher legal minimums are unlikely to have any more impact than they have had in the past. Nor is commercial agriculture in the northern regions likely to have been greatly affected by the accelerated change. There, since rural minimums and actual wages paid have always been relatively high, the impact of the policy change is likely to have been minimal. The efficacy of the legal minimum wage for achieving these expanded policy objectives was neither tested nor questioned by the authorities responsible for their promulgation. In order to realize the objective of income redistribution, employment of the poorest wage earners must first be assured. Higher minimum wages will do little to increase the incomes of those who may find fewer days of employment available to them in rural areas. Agriculture is a sector in which substitutions in production are relatively 60 easy to make. As was pointed out in Chapter 4, this substitution may take the form of machinery for labor, or less labor -intensive cropping patterns, or both. The result of such substitutions could be a redistribution of income within the low-income rural population that could leave some displaced workers and their families even worse off than they were. Furthermore, higher rural minimum wage, are not likely to be effective deterrents to rural-urban migration. To the extent that they result in a reduction in wage-earning opportunities in agriculture, the differential between expected annual earnings in the country and the city may very well increase and serve to stimulate a greater flow of migrants than would otherwise have occurred. Indeed, the very rationale of the migration objective may be open to question. The authors of the policy assume that migration has been "excessive" and, therefore, undesirable. Yet the weight of the evidence examined in Chapter 5 seems to indicate that, from the 258 WAGES IN MEXICO point of view of the welfare of the migrants, migration has had favorable consequences.Whether the social costs of migration have exceeded the benefitsis a question that has not yet been addressed in Mexico. 6" If legal minimumwages havehad only a limitedimpact on actual wages in most of the labor market, what then is the significanceof the govern- ment's wage policy as effected through its intervention in collectivebar- gaining?The studies of the CENIET group seem to suggestthat the govern- ment's guidelines,which tend to parallel its minimumwage policies,may be effectivein gainingcomplianceof that part of the organized sector that is subject to the conciliation procedures of the government. The relevant question that remains, then, is, how far does the impact of the guidelines extend? We have observed that the rate of change in wages in the large- firm stratum of the industrial sector seemed to parallel the changes in the legal minimummore closelythan did wages in the small-enterprisestrata. Is this to be interpreted as evidenceof the effectivenessof the government's guidelines over a substantial part of the modern sector? If the guidelines, in fact, exercised a wide influence over firm wage adjustments, one would expect, not only that the average wage would move in tandem with the legalminimum, but alsothat the individualrates of change would be clusteredwithin narrow bands around the mean rate of change. Whether or not the wage structure shows signs of stability over time can be empirically tested by reference to the available data. The annual October surveys of the industrial sector provide average straight-time hourly earnings of the manual labor force in a number of industries in twelve urban centers. 6 2 Followingthe extension of the areas surveyed in 1964, the sample encompassedsome 800 firms; the number has grown gradually to 867 firms in 1975, which employ some 255,000 workers. I have calculated the rate of changein earningsover the 1965-75 period for twenty-fiveindustrial groups that appear to be identical in the 6 3 The simple arithmetic rate of change first and last years. over the decade would be 204 percent, virtually identical to the change in the legal mini- mum wage of 205 percent. However, underlying that average is a con- siderable dispersion in the rates of change among industries.The changes ranged from 122 percent in electric generation to 385 percent in tire manufacturing. The standard deviation was 63.2, which indicates that fullya third of the rates of changelay beyond plus or minus 63.2 percent- age points of the mean. This dispersion hardly bespeaksa close adherence to guidelines. Unfortunately, the survey does not provide other information that might serve as a basis for explainingdifferentialrates of increase.One can only observe that the rate of change was inverselyrelated to the wage of  EFFECTS OF THE MINIMUM WAGE 259 the base year. The simple coefficientof correlation between the base year wage and the rate of change in the wage was - 0.42, significantat the 5 percent level. It can also be observed that, as expected, the degree of dispersion in hourly earnings in the terminal year 1975 was somewhat smaller than in 1965. The coefficientsof variation were 37.6 and 34.4 percent for the first and last years, respectively. Given the nonrepresentativenessof the data provided by this survey,one can only speculate about the significanceof these observed trends. A narrowing of the interindustrywage structure could be a consequenceof a progressive tightening of labor markets. An increasing scarcity of labor may be expected to be felt firstand foremost by firmsand industriesat the lower end of the wage structure. Thus, the faster rate of increasein wages there may reflect an attempt to remain competitivein the recruitmentof labor. The behavior of wages in this rather selectsample of firmswould thus appear to mirror that which was observedir. table 7-3; there, wage data drawn from the industrial censuses found wage differentials narrowing between the lower-wagesmallenterprisesanc.the higher-wagelarge firms. One possible inference could be that, althoujghgovernment constraints have had some effect on the higher-wage industries and firms, market forceshave beenmore important determinants of wages in the lower-wage sectors. The faster rate of increase in the low-wage industries may have been achievedwithin the framework of collectivebargaining as well. As long as employers consider it desirable or necessary to surpass the in- creases definedby the guidelines,agreementscould be reached without an impasse in negotiations and without submissionto the officialmediation procedures. Since the guidelinesare enforceable largely in the context of the mediation process, agreementsreachedoutside that processare at least partially insulated from their influence.64 It would seem to follow that resort to the conciliation process would tend to be more common during periods of economicrecessionor reduced profitability. If employers offer a wage acljustment smaller than that implied by the guidelines,the union would have every incentive to create an impasse and carry the negotiationsto the conciliationboard in hopes of securing support for a settlement more closcly approximating the guide- lines. Conversely,an employer confronted by a demand in excessof these would follow a similar strategy. During such periods, the guidelinesmay therefore representtentative minimumand mn aximum sizesof adjustments viewed from the perspective of unions and employers, respectively. Of course, the guidelinesserveto define these limits in timesof expand- ing economic activity as well. In such times, however, it may be in the 260 WAGES IN MEXICO interests of both parties to reach an accord independently of the concilia- tion process, particularly if employersconsider the guidelinestoo restric- tive in view of their staffing requirements. One might also expect the bargaining power of the unions to be en- hanced during prosperous times as the credibility of a strike threat may appear greater. In fact, it is not clear that posing such a threat has gained for unions substantially better terms than those assured by the guidelines. The study of negotiated settlements by Zazueta and his associatesdistin- guished between those achievedwithout the threat of a strike and those concluded after an actual strike or the threat of one. In 1980, settlements concluded without a strike threat yielded increases in remunerations almost 5 percentage points greater than those in which a strike was threatened but not effected,and 2.6 percentage points greater than those in which a strike preceded the settlement.65 If these findingsare representa- tive, they would tend to reinforce the judgment offered above that Mex- ican trade unions generallyhave only limited independentpower to affect the monetary terms of employment. Summary Sustained growth of the Mexican economy from 1940 to 1982 appears to have carried with it widespread improvementsin the wage conditions of employment.Somecontroversysurrounds the courseof real wages during the period 1940-55, which saw a steep declineand gradual recoveryin the real wages earned in the modern industrial sector. However, reasons and limited evidencehave been offered in support of the view that even that period was one of steadilyadvancingreal wages in those parts of the urban labor market that lay beyond the purview of the legal minimumwage and collectivebargaining. For the period since 1960, improved data contribute to the conclusion that increases in real wages were widely distributed over the urban econ- omy. Between 1960 and 1975, earnings data drawn from the economic censusesindicate an annual rate of increasein real earningsapproximating 3 percent. Furthermore, a narrowing of earnings differentials was observed between the low- and high-wageclasses of establishments, not only within sectors but across them as well. The course of wages, at least through 1975, would not seem to be consistent with the existence of a large volume of surplus labor overhanging the labor market. On the contrary it would appear that the demand for labor in the nonagricultural sectors shifted to the right more rapidly than did supply and thus accounted for the persistent increase in real wages. While institutional intervention in the labor market in the form of legal  SUMMARY 261 minimumwages and collectivebargaining enjoyshigh visibility,it would not appear to have played a decisive role in determining the course of wages in most of the labor market. As was observed, the relationship of legal minimum wages to market-determinedwages has evinced frequent and substantial changes at differenttimesduring the past, and increasesin the former were not a necessaryprecondition for raisingreal wage levelsof those employed toward the lower end of the urban wage structure. With respect to the period since 1975, the limited availability of data makes it difficult to offer conclusions invested with a high degree of certainty. Earnings data for the large, modern industrial enterprises indi- cate a peakingof real earningsin 1977 and a subsequentdecline, in spiteof an unprecedented rate of economicexpansion from 1978 through 1981. In this sector, institutional interventionwould appear to have exercised a restraining influenceon the courseof wages.With respect to the rest of the economy, there are no officialwage statisticsavailable.A sharp increasein the offering price for domestic serviceswa; noted, and it was suggested that this might be indicativeof similar changesin other market-determined wages for low-skilled labor. Certainly, the widespread reports of labor shortages in1980 and 1981 would lead one to expect equally widespread rising real wages. But the accelerationof inflationin recent years mayhave overwhelmedincreases in moneywages. To the extent that inflation is not fullyanticipated or that workers, in the face of risingnominal earnings, do not readily perceive their decliningpurchasing power, it is quite possible that a rapid expansion of employmentcoulcltake place in an environment of generally falling real wages. I am skeptical, however, about the lasting influence of such mis- apprehensions. While money wages might lag price increasesin the early stages of an inflationary episode, I would not expect them to continue to do so indefinitely,especiallyin the contexi: of a rapid expansion in real output and employment. Only when the economic censusesfor 1980 are published will it become possibleto determinethe net change in earnings of various groups over the 1975-80 interval. Even those figures,however, will not shed light on the year-to-yearvariaiions that might have occurred within that period. Of course, the entire panorama and the outlook for real wages has changed drastically since 1981. High rates of inflation combined with a widespread deterioration in general economicconditionsdo not create an environment favorable to a continued increase in real wages. Indeed,it is certain that part of the impressivegains made by workers generally over the past decadeshas been lost. The concludingchapter will include a brief discussionof recent changes in the Mexican labor market conditions and their possible implications.  262 WAGES IN MEXICO Notes 1. Comisi6n Nacional de los SalariosMinimos,SalariosMinimos,various issues. 2. Ibid., 1979 and 1980. 3. Clark W. Reynoldsdeflatedthe nominalminimumwagesby the wholesalepriceindex for Mexico City and reported a much faster recoveryof real minimums.Accordingto his method, real minimums had recovered most of their loss by the middleof the 1950s. The Mexican Economy (New Haven, Conn.: Yale UniversityPress, 1970), pp. 84-85. 4. Secretaria de Industria y Comercio, Direcci6n General de Estadisticas, Trabajo y salariosindustriales(Mexico,D.F.),variousissues.Since1975,the surveyhas beenpublished bythe Secretariade Programaci6ny Presupuesto(spp),Coordinaci6nGeneraldel Sistemade Informaci6n. 5. Diego G. L6pez Rosado and Juan F. Noyola Vazquez,"Los salarios realesen Mexico, 1939-50," Trimestre econ6mico, vol. 18, no. 2 (1951), p. 204; John Isbister, "Urban Employment and Wages in a DevelopingEconomy: The Case of Mexico," Economic Development and Cultural Change, vol. 20, no. 1 (1971), p. 44. 6. L6pez Rosado and Noyola Vazquez,"Los salarios reales," p. 206. 7. Ibid., pp. 204-206. 8. Isbister,"Urban Employmentand Wages," p.36. How Isbisterderivedannualearnings is not explained in his article. For several years beginningwith 1950, he cites the annual October wage survey, Trabajo y salariosindustriales.However, that publication provides only hourly and weekly earnings data. The source cited for the observations of the 1940s, Revista de estadistica,is not available to me, and I know neitherthe form in which the data appeared nor their origin. 9. StylianosPerrakis,"The SurplusLabor Model and WageBehaviorin Mexico," Indus- trial Relations, vol. 11, no. 1 (1972), pp. 80-95. 10. Ibid.; Isbister, "Urban Employmentand Wages;" L6pez Rosado and Noyola Vas- quez, "Los salarios reales"; Horacio Flores de la Pefia and Aldo Ferrer, "Salarios reales y desarrollo econ6mico," Trimestre economico, vol. 18, no. 4 (1950), pp. 617-28; Adolf Sturmthal,"EconomicDevelopment,IncomeDistribution,and Capital Formation,"Journal of PoliticalEconomy, vol. 63, no. 3 (1955). pp. 183-201. 11. Sturmthal, "Economic Development,IncomeDistribution," pp. 187-88. 12. Ibid., pp. 189-90. 13. Reynolds,Mexican Economy, p. 84. 14. Perrakis, "The Surplus Labor Model," p. 88. 15. Sergio ReyesOsorio,Rodolfo Stavenhagen, SalomonEckstein,and Juan Ballesteros, Estructuraagrariay desarrolloagricolaen Mexico (Mexico,D.F.: Fondo de Cultura Econ- 6mica, 1974), p. 50. 16. Ibid., p. 213. 17. Ibid., table 11-22; Nacional Financiera, S.A., 50 arios de revoluci6n en Mexico (Mexico, D.F., 1963), p. 109; and World Bank data. 18. Perrakis, "The Surplus Labor Model," p. 91. 19. Luis Unikel,El desarrollourbano de Mexico, 2d ed. (Mexico,D.F.: El Colegio de Mexico, 1978), table 11-A3. 20. Reyes Osorio and others, Estructura agraria,pp. 90 and 101-102. 21. Nacional Financiera, 50 ariosde revolucion en Mexico, p. 83. 22. World Bank data. 23. World Bank data.  NOTES 263 24. Organizationof AmericanStates,Americaen cifras 1977, vol. 3 (Washington,D.C., 1979), p. 102. 25. Timothy King, Mexico: Industrializationand Trade Policiessince 1940 (London: Oxford UniversityPress, 1971), pp. 26-27. 26. LourdesArizpe,Migraci6n,etnicismoy cambioecon6mico(Mexico,D.F.: El Colegio de Mexico, 1978), p. 75. 27. Ibid., p. 80. 28. L6pez Rosado and Noyola Vasquez, "Los salarios reales," pp. 201-02. 29. Arizpe,Migraci6n,etnicismo y cambio econ6mico,p. 86. 30. Frank Cancian, Changeand Uncertaintyin a PeasantEconomy: The Maya Corn Farmer (Stanford, Calif.: Stanford UniversityPress, 1972), pp. 50-130. 31. Ibid., pp. 129-30. 32. Arizpe,Migraci6n,etnicismo y cambio econ6nico, pp. 118-19. 33. Isbister, "Urban Employmentand Wages," p. 38. 34. Perrakis, "The Surplus Labor Model," p. 85. 35. The census before 1970 did not include a measure of net value added. I therefore estimated it by subtracting from total receipts the value of all purchased inputs. 36. Banco de Mexico, IndicadoresEcon6micos,vol. 4, no. 8 (1976). 37. Between 1970 and 1975, the real earnings of vsorkers in the smallestsize of service establishmentsrecorded a significant decline in contrast to all other categoriesand sectors. The significanceof this declineis difficultto gauge.It mayflow from data deficiencies that assignedto 1970 an unexpectedlyhighvalue to earningsin this stratum. Note that the 1970 earningsfiguresfor servicesdepart from their previousrelation to those in commerce,onlyto return to the earlier relation in 1975. 38. The annualizationof the minimumwage is cornputedaccordingto the provisionsof the labor code, which states that workers are entitled to a normal day's pay for Sundays. Thus, the annualizedminimumwage is computed as the daily rate times 365. Whetherthis provision is normally observed by small employers,however, is open to question. As indicatedin the notes to table 7-3, the unweightedaverageminimumwage is recordedthere in order to retain comparabilitywith 1960,for whicha weightedminimumwasnot available. Had the weightedaveragebeenused,the averagesfor 1965and subsequentyearswould have been between 8.3 and 10 percent higher, or Mex$5,771, Mex$6,978, and Mex$8,391. However,the rate of change in the legal minimumis virtually the same,regardlessof which measure, the weighted or unweighted,is used. 39. sp', Sistemade cuentas nacionalesde Mexico, vol. 1 (Mexico,D.F., 1981). 40. The increasesin real earnings yieldedby this inclexbetween 1970 and 1975 approxi- mates that derived from the largest-firmstratum of the industrial census, 17.7 and 21.7 percent, respectively. 41. Banco de Mexico, Informe anual 1978 (Mexico,D.F., 1979), pp. 62-64 42. Banco de Mexico, Informe anual 1980 (Mexico,D.F., 1981), pp. 32-35. 43. "Analisisdel mercadode los servicios domesticosen Mexico," Cuadernosde empleo, no. 1 (Mexico,D.F.: Secretariadel Trabajo y Previsi65n Social,1976), pp. 51-114. 44. The rangeof increasesand decreasesin remunesationsreflectsthe use of two different methods to estimate the rate of change.The lower values are derived from the data as recorded in each year of the survey.Sincethe annual valuesshowedrather sharp and irregular year-to-year variations, however, a trend line was fitted to the nominal wage data by least-squaresmethod,and a seriesof estimatedannualr emunerationswas derived.Thelarger values appearing in the text are attributable to the wage series so derived. 45. Richard U. Miller,"Labor Legislationand MexicanIndustrial Relations," Industrial Relations, vol. 7, no. 2 (1968), pp. 171-82. 264 WAGES IN MEXICO 46. It was impossibleto obtain from governmentofficials a definitiveview of how the guidelinesare formulated.The processdescribedhere was consideredto be the mostplausible one in the opinion of several well-informed individuals both in government and in the private sector. 47. Cesar Zazueta, Jose L. Vega, and Jaime Rozenel, Comportamiento de la negociacion de salarios contractuales, Mexico 1977 y 1979 (Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, CENIET, 1981), p. 13. 48. Ibid. 49. Cesar Zazueta and Simon Geluda, Poblaci6n, planta industrial y sindicatos, Serie Estudios 7 (Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, CENIET, 1981). 50. November was chosen as the month of year-to-year comparison because of the prevalence during December of irregular year-end bonuses and other payments that might distort the change in regular wage payments. The December-to-December change was 12.8 percent. 51. Cesar Zazueta, Jaime Rozenel, and Jose Luis Vega, Reporte sobre la negociaci6n de salarios en Mexico, 1980 (Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, CENIET, 1981), p. 61. 52. Ibid., p. 68. 53. Humberto Mufioz Garcia, Orlandina de Oliveira, and Claudio Stern, "Migraci6n y marginalidad en la ciudad de Mexico," El perfil de Mexico en 1980, vol. 3, 6th ed. (Mexico, D.F.: Siglo Veintiuno Editores, 1979), pp. 354-57. There are no data for a later year with which to compare this ratio of people earning less than the legal minimum. 54. Jesus F. Reyes Heroles, "Welfare Effects of Short-run Macroeconomic Policies in a Dual Economy: The Case of Mexico," Ph.D. dissertation, Massachusetts Institute of Tech- nology, 1980. 55. Ibid., p. 111. The occupational title mozo refers to an entry-level position similar to that of an office boy who runs errands, carries internal communications from office to office, serves coffee or other beverages, and performs similar functions. The wage rate assigned mozos was virtually identical to that reported for pe6n, or laborer. Reyes Heroles used the former only because it was more frequently reported than pe6n. 56. Table 7-9 shows that, in 1969 aswell, earnings of mozos were about34 percentabove the legal minimum wage. In fact, just half of the individuals in this job category earned an amount at least 20 percent greater than the minimum wage. 57. Reyes Heroles, "Welfare Effects of Short-run Macroeconomic Policies," p. 116. 58. Ibid., p. 114. 59. Ibid., p. 122. 60. The empirical measures of the elasticity of substitution of capital for labor that are available refer only to the short run, that is, to a single crop year, or to two at most. These have been estimated to be on the order of 0.25 to 0.4 on average. Given a longer period of adjustment, however, one would expect substantially higher coefficients to hold. Hunt Howell, "Machinery-Labor Substitution in Mexico's Central Plateau," in The Book of CHAC: Programming Studies for Mexican Agriculture, ed. Roger D. Norton and Leopoldo Solis M. (Baltimore, Md.: Johns Hopkins University Press, 1983), pp. 375-411; Luz Maria Bassoco and Roger D. Norton, "A Quantitative Framework for Agricultural Policies," in The Book of CHAC, ed. Norton and Solis, pp. 127-35. 61. While the sharp increases in rural minimums thus had the potential for introducing great distortions in the price of rural labor to some employers, subsequent developments have reduced their probable impact, at least for the time being. The lag in the adjustment of the legal minimums to recent high rates of inflation has, in effect, reduced their real value substantially (see table 7-1). A tour of rural areas in four states in central Mexico taken by me  NOTES 265 in May 1985 revealedthat, in many areas,the legal minimumwas not an effective minimum. Small farmers in some zones were paying wages 40 percent above the legal minimum. In others, wages more closelyapproximating the legal minimumwere beingpaid. Only infre- quently was a farmer found offering less than the legal minimum. 62. Trabajoy salariosindustriales,1965 and 1975. 63. The industriesselectedfor thiscomparisonincludemeat packingand canning,milling, baking,edibleoilsand shortenings,beer,nonalcoholicbeverages, syntheticfabrics,footwear (exceptplastic and fabric),plywood,wood furniture,paper, boxesand containers,printing and publishing, leather tanning, tires, soaps and detergents, matches, insecticides,glass containers, cement, foundries, automobile assembly,automobile parts, construction, and electric generation. 64. Unions confronted with an offer in excess of the guidelineswould have a strong incentiveto settlewithout a dispute sincedisagreementwould then requiresubmissionto the mediation process. Given the guidelines,the union could not count on improving on the employer's offer by such a submission. 65. Zazueta, Rozenel,and Vega, Reporte sobre la negociaci6n,p. 72. 8. Conclusion IN CHAPTER 1, it was argued that an appreciation of Mexico's employ- ment "problem" could be gained only by identifyingthe variablescritical to evaluating the performance of the labor market and then charting their course over time. The key or critical variables can be determined by referenceto various models of labor markets that have been advanced in the economic literature. Furthermore, these modelsprovide an analytical framework for guiding empirical inquiries into labor market performance since they offer a statement of the relationship among variables and the outcomes that can be expected to follow from their interaction. By com- paring the Mexican experiencewith the outcomes predictedby the various models, I hoped to identify the model most closelycorresponding to the conditionsprevailing in the Mexican labor market. This can now be done. The most widelyheld perception of the Mexican labor market was one that conformed most closelyto the Todaro variant of the surplus-labor,or dual-market, model. The distinguishingfeatures of that model included a compartmentalized, or dual, urban labor market. Institutionally deter- mined wage levels in the "modern" or "formal" market were greatly in excessof those in the "informal." An "excessive" flow of labor to urban areas was to be found, as was the formation of queues of unemployed or "marginally" employed awaiting access to preferred employments in the formal, high-wage sector. Consequently,extremelydepressed wage levels would prevail in the informal labor market-wages equal to or evenbelow those available from rural employment.Implicithere is a perfectly elastic supply of labor to the urban sector at the wage prevailing in the formal sector. Indeed, given the existence of "surplus labor" in the agricultural sector, the supply of labor to the urban sector would be perfectlyelastic at any wage above the average product of rural labor.' Does such a model provide a "close fit" to the Mexican experience? If the observations of the performance of the labor market over a forty-year period, as recorded here, are even approximatelyvalid, one would have to answer an unqualified "no." 266  REVIEW OF LABOR M/.RKET CONDITIONS 267 Review of Labor Market Conditions Consider the characteristicsof Mexican irternal migration. The empir- ical evidencethat was brought to bear on this issue represents the antith- esis of the conditions of the Todaro model. Rather than flowing into a queue to await the opening of improved ernployment opportunities, mi- grants moved quicklyand easilyinto employmentsin both the formal and informal sectors at wages above those ea:,ned prior to migration. As Anson's study of migrants to Monterrey showed, positive rates of return were forthcoming during the first year rather than in some more distant future. Over time, improvedearnings were not contingent on intersectoral mobility within the urban economy, that is, from informal to formal; earnings levels rose continuously in all of the principal sectors of the nonagricultural economy. Of course, it is this recorded behavior of real wages that renders any of the surplus labor models inapplicable to the Mexican case. In the face of a perfectly elastic supply schedule of labor to the urban economy, one would expect wages to remain stable in that part of the urban economy in which wages were market determined until the labor surplus has been absorbed. Yet, there has beenin Mexico a steady upward trend in wages in enterprises that are likely to be representative of the informal sector as well as in the domestic service market. Since institu- tional intervention is absent or ineffectivein these markets, the upward trend in real wages is explainableonly in terns of the forcesof supply and demand. Consistent with this behavior of wages is the inferencethat the growth of employment in the tertiary and informal sectors has not been simply supply determined,but rather has representeda market responseto increases in effectivedemand for the output of those labor services. I could findno convincingevidenceof a large labor surplus overhanging the urban labor market. Recorded open unimployment rates in Mexico have been low by developing-countrystandards, and although the quality of these measures for the most part is less than ideal, it is hard to detect a significantincreasingtrend. Much of the qualitative evidencesuggeststhat a large part of the urban unemployment may be frictional in character. Both the frequency of voluntary separatior. from employments and the short period of job search recorded bespeak a labor market in which job scarcity is not perceivedto be a problem. Underemploymentin the form of less than full-time employmentis largely a voluntary phenomenon rather than evidence of market failure. 268 CONCLUSION Even the agricultural sector, which is frequently held to be a repository of large quantities of surplus labor, seems not to live up to this character- ization. While the rural labor force may be surplus to the agricultural sector, it need not be surplus to the economy as a whole. This follows from the close integration of rural and urban labor markets (discussed in Chapter 4) and the easy movement of labor from one to the other. Other indications that labor has been a scarce commodity in rural areas are the prevalence of hired labor on even small farm units, the reported resistance of small peasants to technological changes that require larger inputs of labor, and the widespread trend toward mechanization and less labor- intensive cropping patterns. At least one rural household study established that a far smaller quantity of rural family labor was effectively available for employment than was estimated by the rather simple and crude methods customarily used. Thus, virtually all of the elements that have conventionally added up to a pessimistic appraisal of labor market performance seem to have been laid open to question by the available empirical evidence. Rather than an economy that has failed to provide a "sufficient" volume of employment, Mexico emerges from this account as one that has coped remarkably well. Not only has a rapidly growing labor force been absorbed in productive employment, but the terms of employment have shown a steady improve- ment throughout the course of forty years. The labor market has proved to be an efficient mechanism for reallocating labor resources in response to changes in sectoral and regional demands for labor. By avoiding institu- tional interventions that seriously distorted the price of labor, Mexico has also avoided the misallocation of labor characteristic of the Todaro model, a misallocation that would be evidenced by high rates of open urban unemployment and the queuing of labor at the entry points to the formal sector. Because the labor market was allowed to price labor at its oppor- tunity cost, labor-intensive productive activities have been able to thrive alongside a more capital-intensive productive sector that requires a sub- stantially different class of labor. The flexibility of the labor market thus facilitated a high level of employment and utilization of the labor force. In turn, the relatively full employment has implied that the steady growth of the economy since 1940 has exerted an equally steady upward pressure on the general wage level as new or expanding sectors had to attract labor from other employments. As the price of labor has risen, a gradual adjustment of producers in labor-intensive activities was also observed. Capital-labor ratios and labor productivity also rose gradually to support the higher wage levels. In short, the period 1940-80 witnessed a steady transformation of the Mexican economy along a relatively smooth path with the benefits of growth widely distributed over a growing labor force.  THE ECONOMIC CRISIS 269 No claim is made here that the allocatio:n of investment responded solely to market forces or that it was not influenced by distortions intro- duced by institutional intervention. For example, government trade, credit, and investment policies obviously played an important role in influencing the allocation of capital resources. Since such intervention was not necessarily guided by market criteria, i: is quite possible that the growth path followed by the economy was not the most efficient one potentially available. However, once the parameters that guided the allocation of capital were defined, the labor market responded to the resulting pattern of demand for labor with a high degree of efficiency. If this account of the development process is a reasonably accurate portrayal of what has actually unfolded, how can one explain such a great preoccupation with an "employment problem?" This question was addressed in Chapter 3, in which concepts of labor underutilization were discussed. It was noted there that what is commonly labeled an employ- ment problem can more appropriately be called a concern with the exis- tence of low incomes and poverty. Specifying the object of concern accu- rately is important; the conditions associated with or "causing" it can then be identified and policy measures appropriate to the amelioration of that condition can be formulated. For example, recognition that the concern is with low incomes rather than with employnrment opens up a much wider range of alternative policy responses. Measures can be framed that can address the problem directly and with a greater degree of efficiency than would more generalized responses, such as accelerated employment crea- tion. There is no reason to presume that the principal beneficiaries of the latter would be those who form the intended target population, the poorest strata of society. The importance of the proper specification of the problem to be attacked will become apparent in the discussion that follows of recent developments in Mexico. The Economic Crisis Recent employment conditions in Mexico would appear to belie the rosy scenario of the past that has been spelled out in the preceding chapters. In mid-1982 Mexico began a slide into a severe recession that left rising open unemployment and declining real wages in its wake, a clear and sharp interruption of the trends that extend back to 1940. While the crisis is very real, it should be emphasized that it in no way represents an inevitable consequence of the growth pattern that held sway in Mexico in the years before the latter part of the 1970s, Nor can it be ascribed to a failure of the labor market. The origins of the crisis and the changes that it 270 CONCLUSION has spawned are examined below. More important for the future, how- ever, are the lessonsto be drawn for policy initiativesfrom an understand- ing of the labor market and the manner in which it has functioned in the past. The crisis of the Mexican economy was the result of several factors, including a misreadingof the employmentproblem of the country, faulty management of the economy, and serious miscalculations regarding the future price of petroleum and the prospective foreign exchange earnings that would be forthcoming from petroleum exports. The government of L6pez Portillo appears to have been persuaded that Mexico faced both a short-term and a longer-term employmentproblem. The short-term prob- lem stemmed from the reduced rate of growth in the first year of the administration, a consequenceof the program of austerityinitiated by the new government in an effort to stem inflation and to reduce the fiscal deficitand its reliance on external financing.As noted in Chapter 3, open unemployment in the large urban areas did rise in 1977 as economic activity continued at the rather depressed levels of 1976, although the increasecan hardly be viewedas alarming.Unemploymentrates in 1977 in the three major metropolitan areas of Mexico City, Guadalajara, and Monterrey averagedonly 0.7, 1.2, and 1.5percentagepoints, respectively, above their 1973-75 average. 2 Nevertheless,according to Leopoldo Solis, the government decided to accelerate growth in 1978 to close the breach that was presumed to have appeared betweenthe growth in the labor force and the level of employment. It was estimated that a 7.5 percent growth rate would be required to absorb the increase in the labor force and to reverse the deterioration in employment conditions associated with the recession of the two preceding years. 3 All of the studies of the employment situation also pointed to the existence of a long-term employment problem in the form of serious underemployment.Thus, the governmentapparently opted to go beyond a limited response to the cyclical problem and to force a more prolonged accelerated rate of growth. Expanded investment in import-substituting activitiesin both the industrial and agricultural sectors were advanced to further the twin goals of employmentcreation and an increased degree of self-sufficiency. The administration also sought to address the social issue of income distribution by introducing or increasing the subsidization of basic foods and services.Sincethese decisionscoincidedwith a rapidly increasingflow of petroleum from the new southeastern fields and with an expanding availabilityof credit from abroad, the resources required for the ambitious program of the government appeared to be clearly in hand. Furthermore, the governmentchose to ignore the past volatilityin the price of petroleum  THE ECONOMIC CRISIS 271 and to assume that oil prices would continue to rise at a rate high enough to sustain the higher level of government expenditures and to serviceits foreign debt. (In fairness to the Mexican government, it was not alone in forecasting a monotonic upward trend in petroleum prices.) Unfortunately, the requirements of expanded investmentand consump- tion quickly outran the resources available to government. Unable to divert additional resources to itselfby increasedtaxation, the government resorted to massive deficitfinancing. In each of the final four years of the administration, the deficitincreased as a proportion of GDP until, in 1982, it amounted to 18 percent of aggregateoutput. The deficitwas financedby an accelerated rate of monetary creation and by foreign credits. The private sector, as well, was encouraged by an increasingly overvalued exchange rate to turn to foreign sources of funding. The economy re- sponded to these stimuli with four years of rapid growth. The boom collapsed followingthe worldwide glut of petroleum and the subsequent fall in its price. A severe balance of payments disequilibrium ensued as export revenues faltered, the cost of servicingthe burgeoning external debt mounted, and capital flightintensifiedas the inevitabilityof devaluation of the peso became increasinglyapparent. A further extension of foreign credits and the renegotiation of existing debt was made con- ditional on the adoption of policiesdesignedto restrain domesticinflation, reduce the fiscal deficit, and address the disecluilibriumin the balance of payments.The impacton economicactivitywas quicklyfelt. After peaking early in 1982, national output declined duringthe second half of the year, cancelingout the entire growth of the first half. Over the entire year, GDP declined by 0.5 percent from the level of 1981, and there was further contraction in 1983 as GDP fell by 4.7 percent. Construction activity was an early casualty of the restrictive measures and was sharply curtailed during the second half of 1982. Manufacturing output also fell, with the production of capital and consumer durable goods leading the way. On balance, industrial output fell by 2.7 percent in 1982 and by a further 8.3 percent in 1983.4 By the end of 1983, the contractionary forces had run their course and the economy appeared to have stabilized. As would be expected, the recession had a visible impact on open unemployment. By the end of 1982, unemployment in the three major metropolitan areas had risen to an average o:fover 8 percent, more than double the rate during the first quarter of the year.5 The unemployment rate appears to have stabilized at about that levelduring 1983 and early 1984. Real wages in the large-enterpriseportion of the industrial sector began to slip during 1982 as the pace of inflation accelerated. Reflectingthe 272 CONCLUSION leftward shift in the demand for labor and government restraints on advances in nominal wages, real earnings declinedsharply during 1983. By midyear they were some 25 percent below their averagelevel of 1981.6 The adjustments made in legal minimumwages were a further manifesta- tion of the government's policy of wage restraint. As can be seen in table 7-1, through the first half of 1983 the real value of the minimumwage had declinedby almost 20 percent from its averagevalue in 1981. In spite of a midyear adjustment, it continued to slideduring the latter half of the year, averaging 70 percent of its 1981 value. If the trends in modern industrial sector earnings and in legal minimum wages are representative of the course of remunerations in the labor market at large, these declineshave completelywiped out the gains realizedover the decade of the 1970s and have carried real wages to levels prevailing during the latter half of the 1960s. While the recessionappears to have had a severeimpact on real wages, its impact on the level of unemployment has been less drastic than one might have expected, in view of the severityof the decline in output. An estimate of the expectedlevelof unemploymentcan be derivedby applying a rather simple procedure. This would take into account the normal growth in the labor force and the employmentelasticityof output over the two-year interval 1982-83 and would yield a prediction for urban unem- ployment of over 13 percent by the end of 1983.7If the actual rate were closer to the 8 percent level reported, this would appear to have been a major achievement. Confirmation of the limited impact that the recessionhad on employ- ment is provided by data from another source-the social securitysystem, which extends over most of the modern sectors of the economy. After adjustments for changes in coverage, the number of registrants in 1983 was only 0.36 percent below that of 19 8 2 .' Also consistent with this observation is the precedent recorded during the 1976-77 recession. Although that recession was not as deep as the current one, it will be recalledthat the response of the open unemploymentrate must still appear to have been mild. These experiencessuggestthat the economy and labor market are quite flexibleand adaptive to cyclicaldisturbances,just as they have been to secular changes. In the current environment, the absence of wage rigidity is likely to have been a major factor in maintaining employ- ment at a higher than expected level.9Had real wages remained at their prerecessionlevels, the employmenteffects of the economicdeclinemight have been considerably more severe. It is not possible to chart empiricallywith precision how markets have accommodated the labor force in productive employment over the past two years. I can only suggest,in a general way, some of the adjustments that appear to be taking place.  THE ECONOMIC CRISIS 273 The immediate implication of a decline in real wages is that some productive activitiesthat were not worth uncLertaking in an environment of expanding employment at higher wages will now be viewed as viable. For example, rural workers, who had abando nedcultivationof their plots in response to higher returns to wage labor in urban construction and manufacturing activities, would now be expected to return to agricultural production as urban employment opportunit-iesshrink. The surprisingly large increasein agriculturaloutput in 1983 over 1982 of 8 percent would be consistent with this expectation.'°The declinein real wages could also induce producers who enjoy considerable short-run scope for substituting labor for machinery to opt to employlabor more intensively.For example, small farmers may substitute lower priced labor for the rental of farm machinery. Small contractors in the construction industry may enjoy similar options. Furthermore, one might expect a change in the output mix between labor- and capital-intensivegoods. The price cf the former should decline relative to the latter, thus encouraging substitution in consumption in favor of labor-intensive goods. While empirical data do show much sharper declines in production levels of consumer durable and producer goods relativeto nondurables, it is not possibleto distinguishthe contribu- tion to this shift of the declinein incomes from the pure substitution effect. Within agriculture, one might expectto see a reversalof trends in cropping patterns as recorded in Chapter 4. Forage crops might give way to the cultivation of more labor-intensive subsisterce or commercial crops. In fact, this does appear to have occurred. The production of corn, the principal subsistencecrop, exceededthe predicted output for 1983 by 35 percent; at 13.5 million tons, it was almost double that of 1982.11 Data on exports of agricultural commodities also show large increases in labor- intensive fruits and vegetablesas well as of zotton. In view of the large devaluation of the peso in 1982 and the subsequent continual adjustment in the exchange rate, one would alsoexpect to see an increase in exports of other nonpetroleum products, although with a lag. Again, labor-intensivegoods would appear tc enjoythe greatest compara- tive advantage and to be prominentlyrepresentedin the export basket. In 1983 there was in fact a notable increasein such exports, especiallyto the United States: exports of manufactured goods increased more than 21 percent over 1982, and there were large increases in processed foods, textile products, machinery, and automotive products." 2 This expansion extended into early 1984 as well. During the firsttwo months of the year, manufactured exports were running 74 percent ahead of the previousyear, while agricultural exports were up by 37 percent." 3 Significant increases of less labor-intensive goods have also been re- ported, as an expanded national capacity in the production of petroleum 274 CONCLUSION derivatives and automobiles has emerged. The labor-intensive tourist industry has also enjoyed a resurgence;incoming tourism advanced by a record 24 percent in 1983 over 1982.14 The devaluation of the peso also had the effect of promoting a substitu- tion of domesticgoods for those imported. Of course, the sharp reduction in imports was attributable in considerablepart to administrativerestric- tions, but the unprecedentedchange in relative prices made itself felt not only in the consumption of imported goods but also of services, like tourism abroad. Numerous examples of substitution have been recorded, beer bottles for imported aluminum cans, wood for imported plastics in furniture, services for repairing machinery or replacing parts previously imported, and so on.15 For the firsttime in many years,the consumption of domestic goods exceeded that of imported goods in the states bordering 6 the United States.1 Finally,the devaluation increasedvery sharply the peso value of wages earned in the United States relative to those earned domestically. An expected consequencewould be an increase in the flow of undocumented workers to the United States, a phenomenon that might be a factor accountingfor the reported increasein apprehensionsby the Border Patrol in 1983 and the early months of 1984. Thus it is possible to observe a reallocation of resources taking place that will raise Mexico's average labor-output ratio. As long as conditions favorable to this reallocation are not affected by policy interventions that would distort real wages and exchange rates, there is no reason to believe that it will not continue. In short, past experiencein Mexico suggeststhat the cyclical increase in unemployment may prove to be just that, a rela- tively short-term phenomenon. Of course, the longer the economy remains in a depressed state, the less likely it becomes that the process of absorption of people displaced from employmentby the recessionand those new entrants to the labor force can be achievedwithout depressingwagesstill further.Nor will a return to full employment imply an early return to the levelsof labor productivity and real wages that held before the onset of the crisis. Although even a restoration of reasonably full employment might be considered a major achievement, the conventional measures for determining the degree of underutilization will yield an uncomfortablylarge proportion of the labor force in the "underemployed" category; the open unemploymentequiva- lent of this proportion will also appear "unacceptably high." A recent economicreviewof the Mexican economy reported that 50 percent of the labor force is underemployed,a proportion little changed from what has been advanced periodically over the past decade." 7  RECEN1- DEVELOPMENTS 275 Recent Developments Before these numbers become a basis for public policy measures de- signed to accelerate the absorption of the "underemployed," it would be well to pause and recall the lessons that recent experience offer. The large measure of underemployment will not represent an accurate measure of the idle labor resources available for immediate employment in newly created vacancies. Indeed, even under the present conditions of economic recession, there are regions that report "shortages" of particular classes of labor. For example, the annual report of the Banco de Mexico for the year 1983 is cited as pointing to labor shortages in the northwestern and 8 northcentral provinces." The World Bank in January 1984 likewise re- ported that "the 1983/84 cotton production estimate for Mexico was lowered by 10,000 tons in November on evicence that an acute shortage of harvesting labor in Mexicali has reduced the prospective yield.""9 Interviews with small private farmers and ejidatarios during a tour of rural areas of central Mexico in the spring of 1985 also revealed a preoccupation with labor shortages. In sev.-ral regions, farmers com- plained of such shortages in spite of wages offered that were well in excess of the legal minimum. Furthermore, they held that they and other mem- bers of their communities were able to obtain as much employment as they desired. In these areas, at least, it did not appear that the recession had resulted in a surplus of available workers in the labor market according to 20 potential employers. Whether this perception was fully shared by land- less workers who were dependent on wage employment for all their earnings could not be established. Although a process of adjustment to the sharp changes in the param- eters of the Mexican economy is clearly undeir way, it is not characterized by an instantaneous movement to a new equilibrium position. Once resources are engaged in a particular pattern, they adapt to a new set of changed conditions only with a lag. When the economy resumes a pattern of growth it will again require reallocation cf labor resources. Historically, the supply of labor responded to shifting patterns of de- mand with a lag sufficientto exert a continuous upward pressure on the general wage level. Although labor may be disposed to respond somewhat more quickly to changes now than in the past, it would not seem reason- able to expect it to abandon completely its traditional caution. Certainly the experience of 1978-82 would counsel against an attempt to "resolve the employment problem" rapidly by vigorously accelerating the rate of expansion "to make up for the employment growth forgone during the 276 CONCLUSION recession." Such acceleratedgrowth could very quicklyexhaust the ranks of the employables among the ranks of the open unemployed, and labor shortages could again become the order of the day. Even if full employmentis restored, a question remains concerningthe real wage levelat which the labor force will be employed and the prospects for the future course of wages. In the context of a declining national product in both 1982 and 1983, the absorption of a labor force growing on the order of 3.8 percent a year could be achievedonly at decliningreal wages. The expected growth in 1984, on the order of 1 percent, will not suffice to reverse the course of real wages. Giventhe expected growth in the labor force, real wage constancyis likelyto require a rate of growth in 21 GDP approximating 4 percent. A faster rate of growth could be expected not only to absorb a growing labor force but also to do so at risingaverage real wages as labor shifts from low- to higher-productivityemployments. Past experiencehad demonstrated that an annual rate of growth averaging about 6 percent sufficed to raise average real wages at a 3 percent rate during the 1960-75 interval. 2 2 In view of the faster current rate of growth of the labor force, by approximately 1 percentage point, a return to a 6 percent rate of growth in GDP would be expectedto yield a smaller rate of increase in real wages unless substantial changes were to occur in the structure of production toward a greater intensivityin the employmentof labor. Short-Term Prospects It is not within the scopeof this study to predict how soon the Mexican economy will regain its past momentum; this will depend, in part, on an external environment that liesbeyond the control of the Mexican govern- ment and that is difficultto foresee. The future price of petroleum, Mex- ico's ability to tap foreign savings, and the degree of openness of foreign markets will all influencethe country's command over foreignexchange- a critical variable in the determination of the rate of investment. Although external factors will play a contributory role in determining the course of Mexican economy, one should not minimize the role of domestic policies in shaping the manner in which productive resources, including labor, will be employed.Sincethe state, directly (and indirectly through parastatal institutions) accounts for a large share of the domestic product, its investment and other expenditures will play a critical role in the evolution of economic activity in general and in determining the quantity and quality of the new employmentopportunities.Bystructuring the incentives to which private investment decisionsrespond, the govern-  SHORT-TERM PROJECTS 277 ment can influencethese same employmentoutcomesin the private sector. By intervening in the determination of minimum conditions of employ- ment, it can influencethe pattern of investmentamong alternativeproduc- tive activities, as well as the choice of factor proportions within any particular activity.Thus, the responsibilityof governmentdecisionmakers is very great. Particularlysince resourcesavailable for investmentare likely to be relatively scarce during the next few years, the country can ill afford to squander resources on activitiesthat do not maximizeyields in output and employment. It is easier to outline the actions that government should avoid than to suggest what they should do in the furtherance of efficient economic growth and favorable employment conditions. Measures that subsidize the price of capital will tend to favor investment in capital-intensive productive activitiesand the substitution of capital for labor. Unlessthe price of capital is allowed to reflectits true opportunity cost, it is unlikely to flow to its most efficient applications. The policies that influencethe price of labor to the modern or formal sector of the economy should avoid distorting that price. A price in excessof its opportunity cost will tend to encourage responses similar to those flowing from the subsidization of capital. Although a rising real wage level is a laudable social objective, it has been demonstrated here that acting on the price of labor through, for example, the legal minimum wage, is not likely to be an effectivemecha- nism for improving the wage conditions of most of the lowest-wage earners. More important for raisingthe wagrs of those at the lower end of the wage structure is the achievement and maintenance of a high and sustained rate of economicgrowth and a pattern of growth that minimizes the variance in the productivityof labor of any givenquality. Avoidanceof a dualistic pattern of developmentargues heavilyin favor of pricing both capital and labor resources in accordance with their opportunity costs. The exchange rate is another critical price that will affect the internal allocation of resources. Maintenance of an overvalued exchange rate favors consumption of foreign over domesl:icgoods, lowers the price of imported capital goods relative to the price of domestic labor, and limits the growth of output and employment in -he production of goods and services in which Mexico is likely to enjoy a comparative advantage in international trade-goods and servicesthat tend to be labor-intensivein production. In short, these prescriptivemeasures reduce to an application of efficiencycriteria for guiding resource a [location. Although these criteria are acknowledgedto be important, governments ordinarily pursue multiple policy objectives,some of which conflict with others. For example, a great emphasis is being placed in Mexico on 278 CONCLUSION increasing the self-sufficiency of the country in agricultural as well as in intermediate and capital goods. The rationale that is offered for the pursuit of this goal is that Mexico will then be freed from the vagaries of foreign supplies and the country will economize on the foreignexchange needed to service the large external debt. To some extent, pursuit of this objective may involve a cost in the form of forgone employment. For instance, self-sufficiencyin agricultural goods refers primarily to basic grains. Yet their production requires fewer labor inputs than do the principal export commodities of the sector. 23 To the extent that agricul- tural resources are diverted from export to production for domesticcon- sumption, employment of labor and real income will be sacrificed. The promotion of intermediate and capital goods production is also likely to conflict with the goal of maximizing the employment effects of new investment since these sectors are frequently relatively capital- intensivein nature. And, to the extent that suchproduction ispromoted by protective barriers against imports, efficiency and employmentare further compromised by the creation of secure domestic monopolies. To judge from the experienceof most countries that have pursued vigorous import- substitution policies, the irony of these measures is that they proved ineffectivein achievingone of their principal stated goals,the conservation of foreign exchange. Furthermore, the resource cost of saving a dollar of foreign exchange has all too frequently exceededthat of earning a dollar, with a consequent loss of economicwelfare.However, sincethe economic costs associated with such measures are rarely clearly discernibleand the measures themselvesappear politically attractive,it is too much to expect Mexico or any other government to abstain completely from such in- tervention. At best, one may hope that an awareness of the existence of costs associated with them will serve as a restraint on their proliferation. The next few years will continue to be difficult ones for Mexico. Policymakerswill be faced with enormous pressuresto respond to various individual and collectivehardships associatedwith the economiccrisis and its aftermath. In particular, the deteriorating employmentconditions dur- ing 1982 and 1983 have intensifiedpublic and officialconcern over the "employment problem." The familiar characterizations of the past have carried over to the present. As noted above, underemploymentnow (asin the past) affectshalf of the labor force. Whilethe languagedescribingthe employmentproblem is unchanged, one can only hope that the interpreta- tion or significanceassigned by policymakersto such a statistical artifact will reflect learning gained from the labor market experienceof the recent period of accelerated growth. An appreciation of that experience, as well as of the secular developmentsdocumentedhere, would provide a persua- sive basis for altering past perceptions about the way in which the labor  NOTES 279 market has functioned in Mexico. It would also place in perspectivethe significanceto be attached to conventionalmeasuresof labor force under- utilization. A modification of those perceptions would give Mexicans a greater sense of pride in the successof their past developmentefforts, as well as provide a sounder basis for charting responsesto the cyclicaland secular challengesthat currently confront the country. Notes 1. Surpluslabor, as used here, refersto the theoreticalconceptof labor employedat a zero or near-zero marginal product. 2. See table 3-2. 3. LeopoldoSolisM., "Reflexionessobre el panorama generalde la economiamexicana," in El sistema econ6mico mexicano (Mexico,D.F.: PrerniaEditora, 1982), p. 348. 4. BancoNacional de Mexico (Banamex),Reviewol the EconomicSituation of Mexico, vol. 60, no. 700 (March 1984), p. 85. 5. The Mexico City and Guadalajara metropolitanareas reported 8 percent; Monterrey, with its concentration of heavy industry hard hit by the recession,suffered 12 percent. Its experienceis likely to be less typical of urban areas generallythan that of Mexico City and Guadalajara with their more diversifiedproduction bases. 6. Monthly earnings statistics for 1983 were obtained from the Banco de Mexico. 7. Thiswould result from an initiallevelof unemploy mentof 4 percent at the beginningof 1982, an annual rate of growth of 3.8 percent in the labor force, and declinesin GDPof 0.5 and 4.7 percent times 0.6, the output elasticityof employmentthat held during the 1970s. Sincemanufacturingand constructionactivitywere muchharder hit than agricultureand the other major sectors, applyingthe averageoutput elasticityof employment for the economyas a wholeto changesin GDP probably underestimates the expectedimpact on urban unemploy- ment. 8. "La economia mexicana en 1983," Comercio exterior,vol. 34, no. 4 (April 1984), p. 360. This article presents a summary reviewof the annual report issuedby the Banco de Mexico. 9. Some firmsin the modern sector may have been inhibitedfrom reducing their perma- nent work forcesby the largeseverancepaymentsthat are mandatedby the labor code.These legal provisionsdo not apply to temporary employees,however,and are not observedin the informal sector of the economy. 10. Economic Intelligence Unit, Ltd., Quarterly Economic Review of Mexico, no. 1 (London, 1984), p. 15. 11. Ibid., p. 13. 12. Ibid., p. 10. 13. Comercio exterior,vol. 34, no. 5 (May 1984), pp. 462-66. 14. Reviewof the EconomicSituationof Mexico,vol. 60, no. 698 January 1984),p. 39. 15. Quarterly Economic Review of Mexico, no. 2 (1984), p. 8. 16. Comercio exterior, vol. 34, no. 4 (April 1984), p. 360. 17. Quarterly EconomicReview of Mexico, p. 12. 'Vhether the proportion of the labor force definedas underemployedhas increasedor decreasedduring the recessioncannot be predicteda priori. For example,if theminimumwageis Ldopted as the benchmarkseparating the underemployedfrom the fullyemployed,it ispossibL.for the proportion to have declined. 280 CONCLUSION This would follow if actual earningsof low-wageworkershavefallen by lessthan the decline in the legal minimum. 18. Comercio exterior,p. 360. 19. World Bank, "Quarterly Reviewof CommodityMarkets" (Washington,D.C.,Janu- ary 1984). 20. It would be difficultto offer with confidencea full interpretation of such reports allegingregional shortages of labor in the agricultural sector.They may reflecta declining willingnessof the youthful rural population to enter and remain in the agriculturallabor force; as my informants observed, higher levels of education have led to a decreased willingnessof rural youth to performthe heavy manual work demandedby current agricul- tural practice on smallholdings.The alleged shortages may also stem from an increased demand for agriculturallabor in response to higherreal guaranteedprices recentlypromul- gated by the governmentfor manybasic agriculturalcommodities.In the face of a relatively fully employed rural labor force, such a shift in demand for labor could give rise to perceptions of shortages. None of the regions visitedhad a tradition of relying on migrant labor for manual work. Most of them, however, happened to be traditional exporters of labor to the United States on both a temporary and permanent basis. The decline in nonagricultural real wages in Mexico should have had the effect of increasingthe relative attractiveness of work in the United States and may have induced a larger than usual migration. 21. If the aggregateproduction processis approximated by a Cobb-Douglasproduction function, the averagewage level is definedas W = otX/L whereW is the averagewagelevel,X is total output, and cLis the coefficientof the labor term in the production function. Convertingto log form and differentiating,the rate of changein wages will be approximated by AWIW- Ax/a + AXIX - ALIL. If a is constant in the short run, then the increasein output required to yield no change in averagewages is equal to the rate of change in the sizeof the work force.Over time, (xmay change in response to structural changes in production, for example, toward more labor- intensiveprocesses of production. In that case, W might be expectedto increase. 22. See Chapter 7. 23. Secretariade Agricultura y RecursosHidraulicos, Centro de Estudiosen Planeaci6n Agropecuaria,El desarrolloagropecuario de Mexico, vol. 6, El empleode mano de obra en las actividadesagropecuarias (Mexico,D.F., 1982), pp. 124-25. Bibliography Books and Articles Aleman, Eloisa. Investigaci6n socioecon6mica d,recta de los ejidos de San Luis Potosi. Mexico, D.F.: Instituto Mexicano de Inxestigaciones Econ6micas, 1966. Altimir, Oscar. "La distribuci6n del ingreso en Mexico, 1950-1977." In Distribu- ci6n del ingreso en Mexico: Ensayos. Vol. 1. Mexico, D.F.: Banco de Mexico, November 1982. . "La medici6n de la poblaci6n economicamente activa de Mexico." De- mografia y econom'a, vol. 8, no. 1 (1974), pp. 50-83. "Analisis del mercado de los servicios domesticos en Mexico." Cuadernos de empleo, no. 1 (Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, 1976), pp. 51-114. Anson, Ricardo. "La relaci6n entre la politica de rrnigraci6ninterna y la realidad de las consequencias personales de la migraci6n ilterna: El caso de Monterrey." Monterrey: Nuevo Leon: Instituto Tecnologico de Estudios Superiores de Mon- terrey, n.d. Processed. Arizpe, Lourdes. Migraci6n, etnicismo y cambio econ6mico. Mexico, D.F.: El Colegio de Mexico, 1978. Balan, Jorge, Harley L. Browning, and Elizabei:h Jelin. Men in a Developing Society. Austin: University of Texas Press, 19 73. Barbosa-Ramirez, A. Rene. El Baj(o. Mexico, D.F.: Centro de Investigaciones Agrarias, 1973. .Empleo, desempleo y subempleo en el sector agropecuario. 3 vols. Mexico, D.F.: Centro de Investigaciones Agrarias, 1975, 1977, 1979. Belshaw, Michael. A Village Economy. New York: Columbia University Press, 1967. Bergsman, Joel. Income Distribution and Poverty in Mexico. World Bank Staff Working Paper no. 395. Washington, D.C., June 1980. Blejer, Mario I., Harry G. Johnson, and Arturo C. Porzecanski. "Un analisis de los determinantes econ6micos de la migraci6n m2xicana legal e ilegal hacia los Estados Unidos." Demografia y economia, vol. 11, no. 3 (1977), pp. 326-40. Butterworth, Douglas S. "A Study of the Urbanization Process among Mixtec Migrants from Tilantongo in Mexico City." In Peasants in Cities, ed. William Manglin. Boston, Mass.: Houghton Mifflin, 1970. 281 282 BIBLIOGRAPHY . Tilantongo: Comunidad mixteca en transici6n. Mexico, D.F.: Instituto Nacional Indigenista, 1975. Cabrera A., Gustavo. "Poblaci6n, migraci6n y fuerza de trabajo." In Mercados regionales de trabajo. Mexico, D.F.: Instituto Nacional de Estudios del Trabajo, 1976. . "Selectividad por edad y por sexo de los migrantes en Mexico." In Conferencia regional latinoamericana de poblacidn (Actas 1). Mexico, D.F.: El Colegio de Mexico, 1972. Camara, Fernando, and Robert V. Kemper, eds. Migration Across Frontiers: Mexico and the United States. Albany: State University of New York, 1979. Cancian, Frank. Change and Uncertainty in a Peasant Economy: The Maya Corn Farmer. Stanford, Calif.: Stanford University Press, 1972. Cervantes G., Jesuis A. "Mexico: Analisis de la distribuci6n de ingreso-Aspectos metodologicos." Comercio exterior, vol. 32, no. 1 (1982), pp. 43-50. Clark, Colin. The Conditions of Economic Progress. 3rd ed. London: Macmillan, 1957. Conroy, Michael E., Mario Coria Salas, and Felipe Vila Gonzalez. "Socio-eco- nomic Incentives for Migration from Mexico to the United States: Cross- Regional Profiles, 1969-1978." Paper presented at the 1980 Annual Meeting of the Population Association of America, April 11, 1980, Denver, Colorado. Contreras S., Enrique. Estratificaci6n y movilidad social en la ciudad de Mexico. Mexico, D.F.: Universidad Nacional Autonoma de Mexico, 1978. Cornelius, Wayne A. "Mexican Migration to the United States: Causes, Conse- quences, and U.S. Responses." Cambridge: Massachusetts Institute of Technol- ogy, Center for International Studies, July 1978. Corona V., Rodolfo, and Crescencio Ruiz C. "Migraci6n interna e internacional." Paper presented at the Simposio Nacional sobre Migraci6n y Distribuci6n Espacial de la Poblaci6n, December 5-9, 1979, Guadalajara, Jalisco. DeWalt, Billie R. Modernization in a Mexican Ejido. London: Cambridge Uni- versity Press, 1978. Diez-Canedo, Juan. "La migraci6n indocumentada a Estados Unidos: Un en- foque." Documento de investigaci6n no. 24. Mexico, D.F.: Banco de Mexico, July 1980. Diez-Canedo, Juan, and Gabriel Vera. "La importancia de la escolaridad en la determinaci6n del nivel de ingreso." In Distribuci6n del ingreso en Mexico: Ensayos. Vol. 2. Mexico, D.F.: Banco de Mexico, 1982. Dinerman, Ina D. "Patterns of Adaptation among Households of U.S.-Bound Migrants from Michoacan, Mexico." International Migration Review, vol. 12, no. 4 (1970), pp. 485-501. Economic Intelligence, Ltd. Quarterly Economic Review of Mexico. London, 1984. Ehrlich, Paul R., Loy Bilderback, and Anne H. Ehrlich. The Golden Door. New York: Ballantine Books, 1979.  BIBLIOGRAPHY 283 Ericson, Anna-Stina. "The Impact of Commuters on the Mexican-American Border." In Mexican Workers in the United States, ed. George C. Kiser and Martha W. Kiser. Albuquerque: University of New Mexico Press, 1979. Eriksson, John R. "Wage Change and Employnient Growth in Latin American Industry." Research Memorandum no. 36. 'Williamstown,Mass.: Williams College, Center for Development Studies, June 1970. Processed. Evans, John S., and Dilmus D. James. "Conditions of Employment and Income Distribution in Mexico as Incentives for Mexican Migration to the United States: Prospects to the End of the Century." InternationalMigration Review, vol. 13, no. 1 (1979), pp. 4-24. Fei,John C. H., and Gustav Ranis. Development of the Labor Surplus Economy: Theory and Policy. Homewood, Ill.: Irwin, 1964. Felix, David. "Trickling Down in Mexico and the Debate over Long Term Growth-Equity Relationships in the LDCs." St. Louis, Mo.: Washington Uni- versity, n.d. Processed. Fernandez O., Luis M., and Maria T. de Fernandez."Capitalismo y cooperaci6n: El caso de un eiido en Michoacan." In Capitaiismoy campesinadoen Mexico. Mexico, D.F.: Instituto Nacional de Antropotogia e Historia, 1976. Finkler,Kaja. "From Sharecroppersto Entrepreneurs:Peasant HouseholdProduc- tion Strategiesunder the Ejido Systemof Mexico." EconomicDevelopmentand Cultural Change, vol. 27, no. 1 (1978), pp. 103-20. Flores de la Pefia, Horacio, and Aldo Ferrer. "Salarios reales y desarrollo eco- n6mico." El trimestre econ6mico, vol. 18, nc. 4 (1950), pp. 617-28. Fogel, Walter. "Twentieth Century Migration to the United States." In The Gateway: U.S. Immigration Issues and Policies, ed. Barry R. Chiswick. Washington, D.C.: American Enterprise Institute, 1982. Foerster, Robert F. The Racial Problem Involved in Immigration. Washington, D.C.: Government Printing Office, 1925. Foster, George M. Tzintzuntzan: MexicanPeasantsin a ChangingWorld. Rev.ed. New York: Elsevier, 1979. Foster, GeorgeM., and Robert V. Kemper,eds. Anthropologistsin Cities.Boston, Mass.: Little, Brown, 1974. Fuentes M., Olac. "Ensefianza media basica en Mexico, 1970-76." Cuadernos Politicos, no. 15 (1978), pp. 91-104. Gamio, Manuel. Mexican Immigration to the Unzited States. Chicago: University of Chicago Press, 1930. Garcia, Brigida. "La participaci6n de la poblaci6n en la actividad econ6mica." Demografia y economia, vol. 9, no. 1 (1975), pp. 1-31. Garcia y Griego, Manuel. El volumen de la migraci6n de mexicanos no documentadosa los Estados Unidos(nuevashipotesis).Estudiosno.4. Mexico, D.F.: Centro de Informaci6n y Estadisticas del Trabajo (CENIET), 1980. Gisser,Micha. Introduction to PriceTheory, 2d ed. Scranton,Penn.:International Textbook Co., 1969. 284 BIBLIOGRAPHY Goldberg, Howard, "Estimates of Emigration from Mexico and Illegal Entry into the United States, 1960-70 by the Residual Method." Washington, D.C.: Center for Population Research, Georgetown University, n.d. Gomez O., Luis. "Crisis agricola, crisis de los campesinos." Comercio exterior, vol. 28, no. 6 (1978), pp. 714-27. Greenwood, Michael J., and Jerry R. Ladman. "An Economic Analysis of Migra- tion in Mexico." The Annals of Regional Science, vol. 12, no. 2 (1978), pp. 16-31. Gregory, Peter. "Legal Minimum Wages as an Instrument of Social Policy in Less Developed Countries, with Special Reference to Costa Rica." In The Economics of Legal Minimum Wages, ed. Simon Rottenberg. Washington, D.C.: American Enterprise Institute, 1981. Heer, David M. "What Is the Annual Net Flow of Undocumented Mexican Immigrants to the United States?" Demography, vol. 16, no. 3 (1979), pp. 417-23. Hernandez L., Enrique, and Jorge C. C6rdova. "Estructura de la distribuci6n de ingreso en Mexico." Comercio exterior, vol. 29, no. 5 (1979), pp. 505-20. Hewitt de Alcantara, Cynthia. Modernizing Mexican Agriculture: Socio-economic Implications of Technological Change, 1940-70, Report no. 76.5. Geneva: United Nations Research Institute for Social Development, 1976. Isbister, John. "Urban Employment and Wages in a Developing Economy: The Case of Mexico." Economic Development and Cultural Change, vol. 20, no. 1 (1971), pp. 24-46. Keeley, Charles B. "Counting the Uncountable: Estimates of Undocumented Aliens in the United States." Population and Development Review, vol. 3, no. 4 (1977), pp. 473-81. Keesing, Donald B. "Employment and Lack of Employment in Mexico, 1900- 1970." In Quantitative Latin American Studies: Methods and Findings, ed. James W. Wilkie and Kenneth Ruddle. Los Angeles: University of California L.A., Latin American Center, 1977. King, Jonathan. "Interstate Migration in Mexico." Economic Development and Cultural Change, vol. 27, no. 1 (1978), pp. 83-102. King, Timothy. Mexico: Industrialization and Trade Policies since 1940. London: Oxford University Press, 1971. Lancaster, Clarise, and Frederick J. Scheuren. "Counting the Uncountable Illegals: Some Initial Statistical Speculations Employing Capture-Recapture Tech- niques." 1977 Proceedings of the Social Statistical Section of the American Statistical Association. Washington, D.C., 1978. Lesko and Associates. "Final Report: Basic Data and Guidance Required to Implement a Major Illegal Alien Study During Fiscal Year 1976." Contract no. CO-16-75. Washington, D.C.: Office of Planning and Evaluation, U.S. Im- migration and Naturalization Service, 1975. Lewis, W. Arthur. "Economic Development with Unlimited Supplies of Labour."  BIBLIOGRAPHY 285 Manchester School of Economic and Social Studies, no. 22 (May 1954), pp. 139-91. Lomitz, Larissa A. Networks and Marginality: Life in a Mexican Shantytown. New York: Academic Press, 1977. L6pez Rosado, Diego G., and Juan F. Noyola Vazquez. "Los salarios reales en Mexico, 1935-50." El trimestre econ6mico, vol. 18, no.2 (1951), pp.201-209. Maldonado Lee, Gabriel. La mujer asalariada en el sector agricola. Mexico, D.F.: Centro Nacional de Informaci6n y Estadisticas del Trabajo (CENIET), 1977. Martinez G., Ger6nimo. "El mercado de trabajo v las interrelaciones econ6micas entre Mexico, los Estados Unidos, y el Canada. " Paper presented at the second meeting of the Overseas Development Council's Working Group on U.S.- Mexican Border Area Issues, December 3-5, 1981, Tijuana, Mexico. McFarland, Earl L., Jr. "Employment Growth in Services: Mexico 1950-69." Ph.D. dissertation, Columbia University, 1974. McWilliams, Carey. North from Mexico: The Spanish-Speaking People of the U.S. New York: Greenwood, 1968. Miller, Richard. "Labor Legislation and Mexican Industrial Relations." Industrial Relations," vol. 7, no. 2 (1968), pp. 171-82. Munoz, Humberto, and Orlandina de Oliveira. "Migraci6n, oportunidades de empleo y diferencias de ingreso en la ciudad de Mexico." Revista Mexicana de sociologia, vol. 38, no. 1 (1976), pp. 51-83. . "Migraciones internas y desarrollo: Algunas consideraciones sociol6gi- cas." Demografia y econom a, vol. 6 (1972), pp. 248-60. Mufioz, Humberto, Orlandina de Oliveira, and Claudio Stern. Migraci6n y de- sigualdad social en la ciudad de Mexico. Mexico, D.F.: El Colegio de Mexico, 1977. . Migraciones internas a la ciudad de Mexico y su impacto sobre el mercado de trabajo. Temas de la Ciudad, no. 8. Mexico, D.F.: Delegaci6n del D.D.F. en Venustiano Carranza, 1978. Mufioz Izquierdo, Carlos. "Educaci6n, estado y sociedad en Mexico (1930-76)." Revista de la educaci6n superior, no. 34 (1980), pp. 5-56. Newman, Allen R. "Some Theoretical Considerations for Mexican Illegal Migra- tion to the United States." Paper presented to i:he annual meeting of the Rocky Mountain Council for Latin American Studies. February 14, 1981, Las Cruces, New Mexico. North, David S., and Marion F. Houstoun. The Characteristics and Role of Illegal Aliens in the U.S. Labor Market. Washington: Linton, 1976. Norton, Roger D., and Leopoldo Solis M. The Book of CHAC: Programming Studies for Mexican Agriculture. Baltimore, Md.: Johns Hopkins University Press, 1983. Pedrero N., Mercedes. "Corrientes migratorias internas de Mexico (1950-60)." ln Conferencia regional latinoamericana de pobhlacion (Actas I). Mexico, D.F.: El Colegio de Mexico, 1972. 286 BIBLIOGRAPHY . La participaci6n feminina en la actividad economica y su presupuesto de tiempo. Serie Avances de Investigaci6n no. 3. Mexico, D.F.: Centro Nacional de Informaci6n y Estadisticas de Trabajo (CENIET), 1977. Perrakis, Stylianos. "The Surplus Labor Model and Wage Behavior in Mexico." Industrial Relations, vol. 11, no. 1 (1972), pp. 80-95. Randall, Laura R. "The Process of Economic Development in Mexico from 1940 to 1959." Ph.D. dissertation, Columbia University, 1962. Rend6n, Teresa. "Utilizaci6n de mano de obra en la agricultura mexicana." Demografia y economia, vol. 10, no. 3 (1976), pp. 352-85. Reyes Heroles, Jesus F. "Welfare Effects of Short-Run Macroeconomic Policies in a Dual Economy: The Case of Mexico." Ph.D. dissertation, Massachusetts Institute of Technology, 1980. Reyes O., Sergio, Rodolfo Stavenhagen, Salomon Eckstein, and Juan Ballesteros. Estructura agraria y desarrollo agricola en Mexico. Mexico, D.F.: Fondo de Cultura Econ6mica, 1974. Reynolds, Clark W. "Labor Market Projections for the United States and Mexico and Their Relevance to Current Migration Controversies." Food Research Institute Studies no. 17. Stanford, Calif.: Food Research Institute, 1979. . The Mexican Economy. New Haven, Conn.: Yale University Press, 1970. Roberts, Kenneth. "Agrarian Structure and Labor Migration in Rural Mexico: The Case of Circular Migration of Undocumented Workers to the United States." Austin: Institute of Latin American Studies, University of Texas, 1980. Processed. Robinson, J. Gregory. "Estimating the Approximate Size of the Illegal Alien Population by the Comparative Trend Analysis of Age-Specific Death Rates." Demography, vol. 17, no. 7 (1980), pp. 159-76. Samora, Julian, Los Moiados: The Wetback Story. South Bend, Ind.: University of Notre Dame Press, 1971. Silvers, Arthur L., and Pierre R. Crosson. Rural Development and Urban-bound Migration in Mexico. Washington, D.C.: Resources for the Future, 1980. Singer, Morris. Growth, Equality, and the Mexican Experience. Austin: University of Texas Press, 1969. Solis M., Leopoldo. "Reflexiones sobre el panorama general de la economia mexicana." In El sistema econ6mico mexicano. Mexico, D.F.: Premia Editora, 1982. Stavenhagen, Rodolfo. "Los jornaleros agricolas." Revista del Mexico agrario, vol. 1, no. 1 (1967), pp. 164-66. . "Marginalidad, participaci6n, y estructura agraria en America Latina." Demografia y economia, vol. 4, no. 3 (1970), pp. 267-91. Stern, Claudio. "Migraci6n educaci6n y marginalidad en la ciudad de Mexico." Demografia y economia, vol. 8, no. 2 (1974), pp. 171-86. Stoltman, Joseph P., and John M. Ball. "Migration and the Local Economic Factor in Rural Mexico." Human Organization, vol. 30, no. 1 (1971), pp. 47-56.  BIBLIOGRAPHY 287 Sturmthal, Adolf. "Economic Development, Income Distribution, and Capital Formation in Mexico." Journal of Political Economy, vol. 63, no. 3 (1955), pp. 183-201. Thorbecke, Erik, and Everardus J. Stoutjesdijk. Employment and Output: A Methodology Applied to Peru and Guatemala. Paris: Organisation for Eco- nomic Co-operation and Development, 1971. Todaro, Michael P. "A Model of Labor Migration and Urban Unemployment in Less Developed Countries." American Economic Review, vol. 59, no.1 (1969), pp. 138-48. Trejo R., Saul. "Desempleo y subocupaci6n en Iv[6xico." Comercio exterior, vol. 22, no. 5 (1972), pp. 410-16. . "El desempleo en Mexico: Caracteristicas generales." Comercio exterior, vol. 24, no. 7 (1974), pp. 730-38. Unikel, Luis. El desarrollo urbano de Mexico, 2d ed. Mexico, D.F.: El Colegio de Mexico, 1978. Urquidi, Victor L. "Empleo y explosi6n demogrdifica." Demografia y economia, vol. 8, no. 2 (1974), pp. 141-53. Urquidi, Victor L., and Jose B. Morelos, eds. Po llaci6n y desarrollo en America Latina. Mexico, D.F.: El Colegio de Mexico, 1979. Weaver, Thomas, and Theodore E. Downing, eds. Mexican Migration. Tucson: University of Arizona, Bureau of Ethnic Research, 1976. Wilkie, James W. The Mexican Revolution: Federal Expenditure and Social Change since 1910. Berkeley: University of California Press, 1968. . "New Hypotheses for Statistical Research in Recent Mexican History." Latin American Research Review, vol. 6, no. 2 (1971), pp. 3-18. .Statistics and National Policy. Los Angelk s: University of California at Los Angeles, Latin American Center, 1974. Witte, Ann D. "Employment in the Manufacturing Sector of Developing Econo- mies." Ph.D. dissertation, North Carolina StmteUniversity, 1971. Wool, Harold. "Future Labor Supply for Lower Level Occupations." Monthly Labor Review, vol. 99, no. 3 (1976), pp. 22-31. Yates, P. Lamartine. Mexico's Agricultural Diletmma. Tucson: University of Ari- zona Press, 1981. Zazueta, Carlos H. "Consideraciones acerca de los trabajadores mexicanos in- documentados en los Estados Unidos: Mitos y r ealidades."Mexico, D.F.: Secre- taria de Trabajo y Previsi6n Social, Centro de Informaci6n y Estadisticas del Trabajo (CENIET), 1979. Processed. ."Mexican Workers in the United States: Some Initial Results and Meth- odological Considerations of the National Household Survey of Emigration (ENEFNEU)." Paper prepared for the Working Group on Mexican Migrants and U.S. Responsibility, Center for Philosophy and Public Policy, University of Maryland, March 1980. 288 BIBLIOGRAPHY Zazueta, Cesar. "Investigaci6nrecientesobre migraci6nmexicanaindocumentada a los Estados Unidos." Secretariade Trabajo y Previsi6nSocial,CENIET, 1980. Processed. , and Simon Geluda. Poblaci6n,planta industrialy sindicatos.Serie Estu- dios no. 7. Secretaria de Trabajo y Previsi6nSocial, CENIET, 1981. and Fernando Mercado. "El mercado de trabajo norteamericano y los trabajadores mexicanos: Algunoselementoste6ricosy empiricospara su discu- si6n." Paperpresented at the Mexico-UnitedStatesSeminaron Undocumented Migration, September4-6, 1980, University of New Mexico, Albuquerque, New Mexico. , and Jose L. Vega,Salarioscontractuales-vs-coyunturaecon6mica,1977 y 1979. Serie Estudios no. 9. Mexico, D.F.: Secretaria de Trabajo y Previsi6n Social, Centro de Informaci6n y Estadisticas del Trabajo (CENIET), 1981. , Jose L. Vega, and Jaime Rozenel. "Comportamiento de la negociaci6nde salarios contractuales, Mexico 1977 y 1979." Mexico, D.F.: Secretaria del Trabajo y Previsi6n Social, CENIET, 1981. Processed. Official Documents and Publications Banco de Mexico. Distribuci6n del ingresoen Mexico: Ensayos. 3 vols. Mexico, D.F.: Subdirecci6nde Investigaci6nEcon6mica, 1982. Indicadores econ6micos.Mexico, D.F. Monthly. Informe anual. Mexico, D.F. Annual. Banco Nacional de Credito Rural, S.A. Fideicomiso para Estudios y Planes de Desarrollo Agropecuarioy Programasde Credito Agricola. Principales indica- doresdel estudio del empleo,subempleoydesempleo rural.Mexico,D.F., 1983. . Empleo, subempleo y desempleo en el sector rural, por subregiones econ6micas. 31 vols. Mexico, D.F., 1981-83. Banco Nacional de ComercioExterior. Comercioexterior.Mexico,D.F.Monthly. Banco Nacional de Mexico (Banamex).Encuestade subempleados.Mexico, D.F.: Estudios Sociales,1983. . Review of the Economic Situation of Mexico. Mexico, D.F. Monthly. Centro de InvestigacionesAgrarias. Estructuraagraria y desarrolloagricolade Mexico. 3 vols. Mexico, D.F., 1970. Centro de Investigacionespara el Desarrollo Rural (CIDER). Mercado de trabajo del algod6n. Mexico, D.F., n.d.. Comisi6n Consultivadel Empleo.Programanacionalde empleo 1980/82,sintesis. Mexico, D.F.: Direcci6n del Empleo, November 1979. Comisi6n Nacional de los Salarios Minimos. Salarios minimos. Mexico, D.F. Annual. Comite Interamericano de Desarrollo Agricola. Estructuraagraria y desarrollo agricolade Mexico. 3 vols. Mexico, D.F., 1970.  BIBLIOGRAPHY 289 . Tenencia de la tierra y desarrollo socioocon6mico del sector agricola: Guatemala. Washington, D.C.: Pan American Union, 1965. Consejo Nacional de Poblaci6n. Mexico demogreifico: Breviario 1980-81. Mex- ico, D.F., 1982. - Secretaria de Programaci6n y Presupuesto and Centro Latinoamericano de Demografia. Mexico: Estimaciones y proyecciones de poblaci6n, 1950-2000. Mexico, D.F., 1982. El Colegio de Mexico. Dinamica de la poblaci6n en Mexico. Mexico, D.F., 1970. Gobierno Constitucional de los Estados Unidos Mv[exicanos and Gobierno Consti- tucional del Estado de Oaxaca. Programa de desarrollo rural integral de las Mixtecas OaxaqueFias Alta y Baja, 1984-88. Mexico, D.F.: Secretaria de Presupuesto y Programaci6n, 1984. Grupo de Estudio del Problema del Empleo. El problema ocupacional en Mexico: Magnitud y recomendaciones. Versi6n preliminar. Mexico, D.F., n.d. Nacional Financiera, S.A. El mercado de valores. - 50 aiios de revoluciori mexicana en cifras. Mexico, D.F., 1963. Organization of American States. America en cifras. Washington, D.C., 1977. Programa Regional del Empleo para America Latina y el Caribe. Dindmica del subempleo en America Latina. Estudios e informes de la CEPAL no. 10. Santiago: United Nations/International Labour Office, 1981. - Employment in Latin America. New York: Praeger, 1978. Secretaria de Agricultura y Recursos Hidraulicos. Consumos aparentes de produc- tos agricolas, 1925-78. Mexico, D.F., 1979. - Centro de Estudios en Planeaci6n (CESPA). El desarrollo agropecuario de Mexico. 13 vols. Mexico, D.F., 1982. Secretaria de Educaci6n Publica. Estadistica basica del sistema nacional de educa- cion tecnol6gica, 1979-80. Preliminar. Mexico, D.F., August 1980. Secretaria de Industria y Comercio, Direcci6n General de Estadistica, IV censos agricola-ganadero y ejidal, 1960: Resumen general. Mexico, D.F., 1965. V Censo agricola-ganadero y ejidal, 1970. Mexico, D.F., 1975. IV Censo comercial, 1961. Mexico, D.F., 1965. V Censo comercial, 1966. Mexico, D.F., 1968. VI Censo comercial, 1971. Mexico, D.F., 1975. IV Censo de servicios, 1961. Mexico, I).F., 1965. V Censo de servicios, 1966. Mexico, D.F., 1967. VI Censo de servicios, 1971. Mexico, D).F., 1974. VII Censo industrial de 1961. Mexico, D.F., 1965. VIII Censo industrial, 1966. Mexico, I).F., 1967. IX Censo industrial, 1971. Mexico, D.F., 1972. VII Censo general de poblaci6n, 6 de junio 1950. Mexico, D.F., 1953. VIII Censo general de poblaci6n, 1960. Mexico, D.F., 1962. 290 BIBLIOGRAPHY . VIII Censo general de poblaci6n, 1960: Poblaci6n economicamente activa (Rectificaci6n de los cuadros 25, 26, y 27 del Resumen general ya publicado). Mexico, D.F., 1964. . IX Censo general de poblaci6n, 1970. Mexico, D.F., 1972. - Trabajo y salarios industriales. Mexico, D.F. Various issues. Secretaria de Programaci6n y Presupuesto. VII Censo de comercio, 1976. Mexico, D.F., 1979. . X Censo general de poblaci6n y vivienda, 1980. 32 vols. Mexico, D.F., 1983 and 1984. . X Censo general de poblaci6n y vivienda, 1980: Resultados preliminares a nivel nacional y por entidad federativa. Mexico, D.F., n.d. . VII Censo de servicios, 1976. Mexico, D.F., 1979. - X Censo industrial, 1976. Mexico, D.F., 1979. - Encuesta continua sobre ocupaci6n. Mexico, D.F. Quarterly. - Encuesta industrial mensual. Mexico, D.F. Monthly. . Encuesta nacional de ingresos y gastos de los hogares, 1977. Mexico, D.F., n.d. . Informaci6n basica sobre estructura y caracteristicas del empleo en las areas metropolitanas de las ciudades de Mexico, Guadalajara y Monterrey. Mexico, D.F., n.d. . Sistema de cuentas nacionales de Mexico. Vol. 1. Resumen general. Mexico, D.F., 1981. . Sistema de cuentas nacionales de Mexico, 1978-1980. Mexico, D.F., 1982. Secretaria de Programaci6n y Presupuesto and Secretaria del Trabajo y Previsi6n Social. La ocupaci6n informal en areas urbanas, 1976. Mexico, D.F., 1979. Secretaria del Trabajo y Previsi6n Social. Ley federal del trabajo. Mexico, D.F., May 1978. Secretaria del Trabajo y Previsi6n Social, Centro de Informaci6n y Estadisticas del Trabajo (CENIET). Analisis de algunos resultados de la primera encuesta a trabajadores mexicanos no documentados devueltos de los Estados Unidos, octubre 23-noviembre 13 de 1977," Mexico, D.F., n.d. . Indicadores para el estudio de la poblaci6n economicamente activa basa- dos en lainformaci6n censal de 1970. Serie Estudios no.4. Mexico, D.F., 1977. Secretaria del Trabajo y Previsi6n Social, Direcci6n General del Servicio Publico del Empleo. Cuadernos de empleo, no. 1. Mexico, D.F., 1976. Secretaria del Trabajo y Previsi6n Social, Instituto Nacional de Estudios del Trabajo. Mercados regionales de trabajo. Mexico, D.F., 1976. U.S. Agency for International Development. Spring Review of Land Reform, June 1970. 2d ed., vol. 7. Washington, D.C., 1970. U.S. Immigration and Naturalization Service. Annual Report. Washington, D.C.: Government Printing Office, various years.  BIBLIOGRAPHY 291 U.S. Senate, Committee on the Judiciary, 96th Congress. Temporary Worker Programs: Background and Issues. Washington, D.C.: Government Printing Office, 1980. World Bank. Mexico: Manufacturing Sector, Situation, Prospects and Policies. Washington, D.C., March 1979. - . Quarterly Review of Commodity Markets. Washington, D.C. Index Agricultural sector, 10, 14-18, 76; earn- Banco de Mexico, 27, 95-96, 107, 239, ings in, 135, 161; employment in, 3, 246, 248, 249, 251, 275 15, 27, 123-25; history of, 101-07; BANRUtRL (Banco Nacional de Credito and labor requirements, 114-15, 117- Rural), 119, 123, 127 18, 119-22; labor shortage in, 275, Barbosa-Ramirez, A. Rene, 112, 119-20, 276, 280n. 20; and labor surplus, 90- 123-24, 127, 129 91, 266, 268; land distribution in, 101, Blejer, Mario 1., 192 102-04; migrants in, 145, 188; and Blue-collar occupations, 256 migration, 158; minimum wage in, 121, Bracero program, 109, 144, 157, 172, 230; occupational mobility of, 162; and 173, L88-89 off-farm employment, 109-13, 122, Browning, Harley L., 109, 154, 156, 162 125, 126, 133, 134, 135, 199; produc- Butterworth, Douglas S., 160, 161 tivity of, 15, 32, 33, 102, 103, 104-05, 106, 225-26, 257; seasonal employ- Cancian, Frank, 231 ment in, 82-84; shifts of labor from, CArdenLs, Lazaro, 14, 102, 217, 224, 48; since World War II, 14, 26, 109; 225, Z52, 253 size of holdings in, 102-04, 105, 110; Castillo, Leonel, 174 underutilization of labor in, 119, 122- CENIET (Centro Nacional de Informaci6n 25, 138-39; wage labor in, 127-30, y Estadisticas del Trabajo), 173, 177, 133; wages in, 102, 108, 141n. 34. See 178, 179, 180, 186, 189, 195, 204-05, also Land reform; Rural labor force 207-08, 248, 258 Agriculture: cropping patterns in, 115, Census, 18, 19, 25-26, 67-68, 231-34; 116-18, 119, 131, 132, 257, 268; em- errors, 50, 58; and migration data, ployment conditions in, 114; labor re- 146-47, 151-52, 153 quirements in, 114-15, 117-18, 119- Centro de Investigaciones Agrarias, 112 22; land-labor ratio in, 132, 133; CESPA (Centro de Estudios en Planeaci6n mechanization in, 114, 115-18, 268 Agropecuaria), 114-15, 116, 117, 118, Aliens. See Illegal aliens 119, 120, 121, 123 Altimir, Oscar, 19, 25-26, 50, 51, 56, 59, Clark, Barton, 170 60, 61-62 Colegio de Mexico, 50, 51 Anson, Ricardo, 154, 158, 159, 160, 162, Collective bargaining, 235, 247-52, 259, 163, 267 260 Arizpe, Lourdes, 230 Comisi6n Consultiva del Empleo, 2 Artisans, 102, 158, 159 Commercial sector, 111; earnings in, 161; Automation. See Mechanization employment rate in, 34, 35-37, 47-48; and migration, 158, 188; minimum Balan, Jorge, 109, 154, 156, 161-62 wage in, 47; since 1940, 26; wage in- Balance of payments, 209 creases in, 236 293 294 INDEX CONAPO (National Population Council), ci6n), 18-19, 21, 27, 62n. 8, 68, 69, 183-84 73, 74, 75, 79, 83, 86, 87, 94 Construction industry, 3, 27, 109, 110, Education, 49; and domestic workers, 40; 111, 158; earnings in, 161, 253; mi- and earnings, 23, 87-89; of illegal grants in, 159, 160; in 1980s, 271, 273; aliens, 187-89; and income level, 23- since 1940, 26; wage levelsin, 230 24, 87-88; and literacy rate in Mexico, Consumption, 14, 227, 274, 277 229; and migrants, 15S-57, 162-63; Contreras, Enrique, 154 and migration, 152, 154, 155-57; and Corn, 116, 227, 230, 273 school attendance, 22, 23, 63n. 18; Cornelius, Wayne A., 189, 191, 196, 200, among underemployed, 86; and the un- 207 employed,75-76, 79, 80 Cotton, 108, 117, 273, 275 Ejidatarios, 25, 26, 58, 63n. 19, 103, 106, "Coyotes," 196-97 108, 109, 110-11, 112, 133-34, 159, Cropping patterns, 110, 115, 116-18, 196 119, 131, 132, 257, 268 Ejido system, 102-03, 105, 106, 107, 109, 110-11, 134, 189; and productiv- Davila, Alberto, 193-94 ity, 226; and sources of income, 111; Devaluation of peso, 273, 274 and use of machinery, 115; and wage Diaz, Porfirio, 101 labor, 128 Diaz Ordaz, Gustavo, 102 Elderly, 22, 24-25, 49, 65n. 54, 87 Diez-Canedo,Juan, 179, 186, 209 Employment, 6, 32; among agricultural Discrimination, in employmentof women, workers, 3, 15, 27, 123-25; in commer- 21-22 cial sector, 34, 35-37, 47-48; condi- Disguisedunemployment, 73, 94-96 tions, 2-4, 73, 114; determinants of, Domestic service,35, 37-40; demand for, 38-39; in food and beverage services, 38; and educational level,40; migrants 40-44; growth of, 63n. 26; in industrial in, 40, 188; wages in, 35, 37-40, 80, sector, 229, 243; length of, 160; mar- 243-46, 252, 253, 267 ginal, 34-37, 40, 41, 266; of migrants, Dual labor market model, 8 159-64; open, 92-93; opportunities, 40, 73, 114; quality of, 48-49; shifts Earnings, 15-18, 243; in agricultural sec- in, 25-33; since 1940, 268; structure, tor, 135, 161; in commercialsector, 222; in tertiary sector, 4-5, 34-48, 49, 161; in construction industry, 253; dif- 267. See also Part-time employment; ferentials, 135-37, 157, 161, 202, 236; Temporary employment; Women education and, 23, 87-89; farm versus Evans, John S., 191 off-farm, 132; and growth in productiv- Eventuales, 251 ity, 237-39; in industrial sector, 258- Exchange rate, 273, 277, 278 S9; in informal sector, 86-87; male ver- Exports, 2, 14, 102, 273 sus female, 164; in manufacturing, 161, 222, 231, 242; of migrants, 160-64, Factory workers, 158 197-98; before and after migration, Family workers. See Unpaid family 162-65; off-farm, 199-200, 267; and workers poverty, 90; by state, 136-37; of teen- Fogel, Walter, 193 agers, 87-88, 96; in tertiary sector, Food and beverage services,40-44, 109- 161; of unskilled labor, 222; women's, 10, 188 87, 88, 89. See also Wages Forster, Robert F., 186 Earnings criterion, 85-87, 89-90, 94 Foster, George M., 188-89 Economic growth, 2-3, 131-32; future, Frictional unemployment,69, 71, 158, 278-79 267 ECSo (Encuestacontinua sobre ocupa- Fringe benefits,86, 95, 239, 242, 243  INDEX 295 Gamio, Manuel, 186 86; earnings in, 86-87; wages in, 255, Garcia y Griego, Manuel, 177, 179, 181, 266, 267 184 Irrigation, 14, 103, 105-06, 107, 110-11, Goldberg, Howard, 176 112, 115, 116 Gross domestic product (GDP), 16-17, Isbister, John, 222, 231 38-39, 43, 222 Grupo de Estudio sobre el Problema del James, Dilmus D., 191 Empleo, 2, 85, 93 Jelin, Elizabeth, 109, 154, 156, 162 Job creation, 2, 22 Hacienda system, 102, 103, 108 Job satisfaction, 158 Handicrafts, 111-12, 113 Job search, 75, 110, 126 Heer, David, 175, 176 Johnson, Harry G., 192 Hidden unemployment,21-22, 66, 71, 72, 77-81, 158 Kemper, Robert V., 157, 161 Household decisionmaking,198-200 Houstoun, Marion F., 175, 179, 181 Labor demand, 114-15, 117-18, 119-22, Human capital theory, 145-46, 152 144,193-94, 203-04, 206-08 Labor force. See Rural labor force; Urban Illegalaliens, 144; apprehensionof, 172- labor market 73, 174-75, 176-77, 178, 181-82, Labor force participation rates. See Partic- 186-87,188,189-90, 192,193,194, ipation rates 196-97, 274; characteristicsof, 187- Labor market: age and, 19-21, 87; condi- 91, 196-98; remittances from, 179; un- tions, 4-6, 68, 267-79; data, 25-34; detected, 177 impact of migration on, 204-09; in- Illegal migration, 109, 144; costs of, 196- tegration of migrants in urban, 159-64; 98, 201; estimates of, 182-85; flow of, models, 7-10; performance of, 268; 175-80; history of, 171, 172-75; im- sectoral distribution of, 4, 25-34; size, pact of, in Mexico, 204-09; and leak- 18-25; studies of, 4-7; women in, 3, age rates, 180-82, 201, 208-09; and 21-22. See also Participationrates length of stay, 188-90, 202; reasons Labor requirements, 129, 251; in agri- for, 191-204 cultural sector, 114-15, 117-18, 119- Immigration and Naturalization Service 22 (U.S.), 173, 174, 175, 177, 178, 179, Labor shortage, 3, 131, 259; in agricul- 186, 187,192 tural sector, 275, 276, 280n. 20; in the Import substitution, 14, 32, 274, 278 United States, 171, 172, 203-04; in Income: differentials,18, 135-37, 164- 1970s, 243, 246 65, 218-19; and migrants, 152, 162- Labor supply: and demand, 275-76; esti- 63, 164-65; as related to education, 23, mate; of, 123-25; rural, 122 24, 87-89 Labor surplus, 3, 6, 8, 34, 47, 128, 131, Income distribution, 15, 18, 48, 65n. 44, 135, 137; in agricultural sector, 90-91, 85, 92, 135-37, 222, 225, 237-38 138-39, 266, 268; in the United States, Industrial sector, 102; earnings in, 258- 203-04; in urban market, 267 59; employmentin, 229, 243; migrants Lancaseer, Clarise, 174, 176 in, 160; in 1940s, 26, 222-23; in Land distribution, 14, 101, 102-04 1980s, 271; occupational mobility in, Land-labor ratio, 132, 133, 135, 158 162, 222-23; productivity of, 15-18, Land reform, 101-04, 106-07, 158, 225 243; wages in, 221, 236, 237-38, 246- Literacy rate, 229 52, 271-73 Livestock, 131, 132 Informal sector, 41-43, 46, 47, 48, 145; L6pez Mateos, Adolfo, 102 characteristics of, 87-88; definitionof, L6pez Portillo,Jose, 2, 69, 218, 270 296 INDEX L6pez Rosado, Diego G., 221-22, 230 Minifundista, 91, 99n. 31, 104, 196 Minimum wage, 3, 8, 65n. 41, 93; adjust- McFarland, Earl L., 58, 59, 60, 61-62 ment of, 247; and agricultural workers, Manufacturing, 3; earnings in, 161, 222, 121, 230; in commercial sector, 47; dif- 231, 242; and migrants, 159, 188; in ferentials, 218-19; and domestic work- 1980s, 271, 273; wage determinants in, ers, 244; effects of, 252-60; in food 243; wage differentials in, 255-56; and beverage services, 43; history of, wages in, 37, 239 217-19, 223-25, 229-30, 230-31, Marginal employment, 34-37, 40, 41, 236; male versus female, 164; and mi- 266 grants' earnings, 108, 160-61; in mis- Mechanization, 257; in agricultural sector, cellaneous services, 44, 47; in 1980s, 15, 102, 114, 115-18, 268; in ejido 272, 275; for off-farm employment, system, 115; resistance to, 129, 268; 110; and the underemployed, 85; in the and unemployment, 130-31 United States versus Mexico, 197-98; Migrants, 108-09; in agricultural sector, urban, 222; during World War ll, 32 188; characteristics of, 153-59, 185- Miscellaneous services, 44-47 91, 196-98; in construction industry, Mobility, 10, 11, 149; in industrial sector, 159, 160; distribution of, 149, 205-06; 162, 222-23; intersectoral, 132-33, earnings of, 108, 160-64; educational 267; of migrants, 145, 162; resource, level of, 155-57; employment opportu- 165. See also Migration nities for, 159-64; flow of, to United Mortality rates, 158, 176-77, 229 States, 175-80, 187; in industrial sec- tor, 160; and integration in urban labor National Population Council (CONAPO), force, 159-64; in manufacturing, 159, 183-84 188; occupational mobility of, 162; Neoclassical labor market model, 7 self-employment of, 160; unemployment Non-native population, 146-49, 155 of, 157-59; women as, 163-64. See North, David S., 175, 179, 181 also Illegal aliens Noyola Vazquez, Juan F., 221-22 Migration, 5-7, 8-9, 10, 18; and com- mercial sector, 158, 188; costs of, 163; data, 146-47, 151-52, 153; determi- Occupational distribution, 205-06 nants of, 152, 154-55, 158; directions Off-farm employment, 109-13, 122, 125, of, 146-51, 153, 154; distances, 152, 126, 133, 134, 135, 199; and earnings, 154; and domestic service, 40; and 132, 199-200, 267; importance of, earnings, 162-65; and education, 152 110, 112; proportion of, to agricultural 154, 155-57; history of, 170, 171-72; employment, 110, 112-13; types of, impact of, on Mexican labor market, 111-12 204-09; and income, 152, 164-65; in- Open employment, 92-93 ternal, 146, 150, 226-27, 267; and Output. See Productivity landownership, 106; rate, 157; reasons for, 145, 158, 191-204; research on, Participation rates, 49, 50-58; and educa- 151-64; of rural workers, 109, 196; tion, 24; among the elderly, 22, 24-25; scope of, 146-51; seasonal, 15, 144; in labor force, 4, 19-21, 22, 24, 71; for temporary employment, 109-10, among men, 26-27, 73; in 1940s, 228; 113, 141n. 31, 144, 251; to the United regional differences in, 71-72; in the States, 10, 109, 113, 144, 157, 170- United States, 203; among teenagers, 209; of women, 153. See also Illegal 22, 24; among women, 50, 51, 56, 57, aliens; Illegal migration; Rural-urban 68, 71, 72-73, 80, 228 migration; Urban-rural migration Part-time employment, 81-82, 86, 234, Mines, Richard, 207 253; among agricultural workers, 82-  INDEX 297 84; low earnings in, 87; voluntary, 83, among, 15, 18; self-employed in, 110; 86, 87, 267; among women, 82, 83, 89 underemploymentof, 92, 268; urban Perrakis, Stylianos,222, 224, 226, 231 employmentopportunities for, 273; Petroleum, 2, 270, 271, 273-74, 276 wages, 110 Plan Puebla program, 129 Rural-urban migration, 5-7, 8-9, 18, 106, Porzecanski,Arturo C., 192 114, 151, 152, 153, 154, 226, 266; and Poultry, 131, 132 integration into urban work force, 159- Poverty, 5, 6, 269; definitionof, 62n. 6; 64; un,employment and, 75; and wages, and low earnings,90; and migration, 257 145, 158, 194-95; rural, 15, 18 PREALC (Regional EmploymentProgram for Latin America and the Caribbean), Scheuren, Frederick J., 174, 176 2, 85, 90, 91, 92, 94 Seasonal unemployment, 15, 76-77, 82, Productivity, 10, 80; agricultural, 14, 15, 124-25, 144, 160 32, 33, 102, 103, 104-05, 106, 225- Sectoral distribution, 25-33; and census 26, 257; and earnings,237-39; of problems, 25-27; changes in, 29-33, ejidos, 226; in food and beverage ser- 101; cf employment, 58-62; of labor vices, 41-42; industrial, 15-18, 243; in market, 4; since 1970, 27 manufacturing, 271, 273; in miscel- Self-employment,35, 84-85, 92, 112, laneous services,45; in 1940s, 222; af- 233, 234, 236; correlation between ter 1940, 268, 271, 277; in 1950s, 32; earnings and education among, 96; as in 1960s, 32; in tertiary sector, 36-37, disguis;ed unemployment,95; of farm- 237; and underemployment,85; and ers, 134; in food and beverageservices, wage increases, 236-39 40; of migrants, 160; in miscellaneous Professional sector, 76, 159, 242, 256 services,46; returns to, 41, 47, 94-95; "Pull" factor, in migration, 157-58, 191- of rural workers, 110 92, 194, 198, 200-04 Separations, reasons for, 76-77 "Push" factor, in migration, 157-58, Servicesector.See Tertiary sector 191-98, 226 Shadow, Robert D., 190, 196 Sharecropping, 106 Recession,69, 269-74 Sistema Alimentario Mexicano, 107 Regional Employment Program for Latin Skilled labor, 64n. 30; migration of, 158, America and the Caribbean. See PREALC 159; n miscellaneousservices,44- Remunerated labor, as hidden unem- 46; o.:upational mobility of, 162; in ployed, 81 off-farm employment, 111, 112; short- Rend6n, Teresa, 114-15, 116, 117, 118 ages, 3 Repair services, 44, 46, 64n. 30 Small bksiness,41-43, 233-34, 235, 236 Retirement benefits, 22, 86 Social security,49, 272 Revolution, 102-03 Solis, Leopoldo, 270 ReyesHeroles, Jesuis,255, 256 Students, as hidden unemployed,79-80 ReyesOsorio, Sergio, 121, 129, 130, 226, Sturmtlhal,Adolph, 222, 223 228 Sugarcane, 108, 116, 227 Reynolds, Clark, 203 Roberts, Kenneth, 196, 198-99, 200 Robinson, J. Gregory, 175-76 Teenagers: as disguisedunemployed,96; Rural labor force, 5-6, 8, 10, 14-15, 25, earnings of, 87-88; as illegal aliens, 108-13, 127-30, 131, 133; in domestic 187; in low-skilloccupations, 203; par- service,37, 40, 109; migrants in, 108- ticipation rates of, 22, 24; as part-time 09; and migration, 109, 196; and off- workers, 80; school attendance of, 22; farm employment, 109-13; poverty unemploymentamong, 73-75 298 INDEX Temporary employment, 109-10, 113, of, 191, 192, 194-95; in 1970s, 69-71, 141n. 31, 251. See also Seasonal unem- 270; in 1980s, 27, 269, 271, 274; open, ployment; Underemployment 66-77, 79, 126-27; rates, 6, 67, 68-71, Tertiary sector, 1, 15, 27, 32, 33, 101; 267; secular ttends in, 67-68, 69-71, earnings, 161; employment in, 4-5, 34- 75; by sex, 67, 71-74; teenage, 73-75, 48, 49, 267; migrants in, 159-60, 188; 80; urban versus rural, 71; among migration and, 145; productivity in, women, 68, 74, 76 36-37, 237; since 1940, 26; unskilled Unikel, Luis, 18, 38, 165 labor in, 160; wages in, 64n. 29. See United States: bracero program in, 109; also Commercial sector; Domestic ser- illegal migration to, 171, 200-04; labor vice; Food and beverage services; Mis- shortage in, 171, 172, 203-04; labor cellaneous services surplus in, 203-04; migration to, 10, Todaro, Michael, 8 109, 113, 144, 157, 170-209; partic- Todaro model of labor market, 8-10, ipation rates in, 203 266, 267, 268 Unpaid family workers, 35, 40, 64n. 37, Tourism, 274 84, 85, 86, 92, 107, 129, 233, 234, Trade unions, 86, 223, 235, 247-52, 253, 236, 268; as disguised unemployed, 95 259, 260 tUnremunerated labor, 41, 45, 46, 233-34, Treaty of Guadalupe Hidalgo, 170 236. See also Unpaid family workers Trejo, Saul, 93 Unskilled labor, 8, 34-35, 229; earnings of, 222; market conditions for, 252, Underemployment, 9, 81-94; of agricul- 253; and migrant workers, 159, 160, tural workers, 119, 120; definition of, 162, 209; and migration, 158; occupa- 81, 91; and educational levels, 86; gov- tional mobility of, 162; shortages in, 3; ernment reactions to, 40, 270-71; and in tertiary sector, 160; wages of, 224 use of income criterion, 85-87, 89-90, Urbanization, 25-26, 37, 38, 75, 102 94; measuring, 81-82, 83, 90-94, 138; Urban labor market, 5-6, 8, 10, 87, 159- and migration, 114, 191; and produc- 64, 267 tivity, 85; regional variations in, 93; Urban-rural migration, 102, 151, 157, seasonal, 82-84; urban, 86; visible, 82- 273 85; voluntary, 83, 86, 87, 267; among women, 68. See also Part-time employ- Vendors, 95, 109 ment Underutilization of labor, 2, 10, 81, 122- Wage labor, 108, 127-30, 131, 133, 268 30, 269; in agricultural sector, 119, Wage(s), 11; actual versus legal, 230-31; 120, 122-25; compared with unem- in agricultural sector, 102, 108, 141n. ployment, 6; and sectoral distribution, 34; changes in, 34-35; in commercial 101. See also Underemployment; Unem- sector, 37, 236; in construction indus- ployment try, 230; differentials, 225-27, 230, Undocumented worker. See Illegal aliens; 234-36, 255-56; of domestic workers, Illegal migration 35, 37-40, 243-46, 267; in dual mar- Unemployment, 2, 4, 9; among agricul- ket model, 8; in food and beverage ser- tural workers, 125-27; characteristics vices, 40-44; future levels of, 276, 277, of, 75-76; definition of, 68; disguised, 278; and growth rate, 6; history of, 73, 94-96; duration of, 67-68, 69, 76, 219-29, 231-34, 245; impact of migra- 158; and educational levels, 75-76, 79, tion on, 204-09; increases and produc- 80; frictional, 69, 71, 158, 267; hidden, tivity, 237-39; in industrial sector, 221, 21-22, 66, 71, 72, 77-81, 158; in labor 236, 237-38, 246-52, 271-73; in in- market model, 7; mechanization as formal sector, 255, 266, 267; in kind, cause of, 130-31; migration as cause 231, 244; intersectoral, 226; in labor  INDEX 299 market model, 7; in manufacturing,37, 77, 78, 79-80; as illegal aliens, 183, 239; and migration, 193; in miscel- 187, 1E 8; in labor force, 3, 21-22; laneous services,44, 45; in 1980s, 276, migration of, 153, 163-64; and mini- 277; and off-farm employment, 110; of mum wage, 164; participation rates of, self-employed,95-96; since 1940, 219- 22, 50, 51, 56, 57, 66, 68, 71, 72-73, 25, 268; since 1975, 239-43; in tertiary 80, 228; rural, 62n. 12; in school, 22- sector, 64n. 29; in Todaro model, 9; of 23; as inderemployed, 82; unemploy- unskilled labor, 34-35; urban, 132. See ment a.nong, 68, 74, 76 also Earnings; Minimumwage World War 11:agricultural sector since, Wiest, Raymond, 196 14, 26; bracero program and, 109; Wilkie,James, 67 minimum wage during, 32; and produc- Women: in construction industry, 246; tivity, 14, 222, 268, 271, 277 discrimination in employmentof, 21- 22; in domestic service,37, 40, 243-44; Y earnings of, 87, 88, 89, 164; employ- Yates, P. Lamarte, 105 ment opportunities for, 21, 72-73, 77, 246; as hidden unemployed,21-22, 66, Zazueta, Carlos, 187, 248, 249 Peter Gregory is professor of economics at the University of New Mexico and a consultant to the World Bank. The most recent World Bank publications are dcscribed in the annual spring and fall lists. The latest edition is available free of charge from Publications Sales Unit, Department B, The World Bank, Washington, D.C. 20433, U.S.A. I  THE WORLD BANK This book takes sharp issue with the prevailing perception that Mexican employment conditions have not improved or have even deteriorated over time. It reveals a steady and substantial improvement in the earnings of workers at the bottom of the nonagricultural wage structure and showvs that rural-urban migration has caused migrants' earnings to rise significantly and has not prevented the rise of urban unskilled wages in general. The findings indicate that estimates of the underutilization of labor are not only grossly exaggerated but also misleading for the formulation of employment policy. The author uses new information to estimate the fLow of migrant labor to the United States. His examination of Mexican labor markets shows the unexpected importance of nonagricultural labor for rural household incomes. Finally, he assesses the impact on employment of the recent economic crisis and draws on lessons of the past to advance employment policy. prescriptions for the future. Peter Gregory is professor of economics at the University of New Mexico and a consultant to the World Bank. The Johns Hopkins University Press Baltimore and London Cover design by George Parakamannil ISBN 0-8018-3343-4