Peru - The Decline of the Value of Labor in Peru : Causes and Remedies

Document of The World Bank FOR OFFICIAL USE ONLY FILE coPy CONFIDENTIAL: Report No.: 11247-PE Type: (SEC) CONFIDENTIAL Title: THE DECLINE IN THE VALUE OF LA Report No. 11247-PE Author: IGLESIAS, F Ext.:38673 Room:17029 Dept.:LA1CO PERU THE DECLINE OF THE VALUE OF LABOR IN PERUI CAUSES AND REMEDIES February 26, 1993 Country Operations Division Country Department I Latin America and the Caribbean Regional Office Th ,docurent has:a84estricted distribotioh end-may-be used byrecipients only in the pefoameof thei o efi6ial duties. its contents may ototherwise b disclosed without World Bank authorzation. CURRENCY EQUIVALENTS (As of December 31, 1992) Currency Unit - Nuevo Sol (SI.)' US$1.00 - S/. 1.63 S/.1.00 - US$0.61 FISCAL YEAR January 1 - December 31 ABBREVIATIONS AFP Administraci6n de Fondo de Pensiones (Pension Administration Fund) CGTP Confederaci6n General de Trabajadores del Pero CIDA Canadian International Development Agency CL Comunidades Laborales CPI Consumer Price Index CTS Compensation for Time of Service EAP Economically Active Population FONASA Fondo Nacional de Salud (National Fund for Health) FONAVI National Housing Fund GTZ Deutsche Gesellschaft fUr Technische Zusammanarbeit (German Technical Cooperation Agency) INEI Instituto Nacional de Estadistica e InformAtica INP National Planning Institute IDB Inter American Development Bank IPSS Instituto Peruano de la Seguridad Social (Peruvian Institute for Social Security) ISAPRES Private Health Institutes LFP Labor Force Participation MTPS Ministerio de Trabajo y Promoci6n Social (Ministry of Labor and Social Development) OSS Private Health Care Providers PLS Peruvian Living Standard Survey SENATI National Industrial Training System TECSUP Instituto Superior Tecnol6gico (Higher Institute of Technology) USAID United States Agency for International Development (hO July 1, 1991, a new monetary unit, the Nuevo Sol (SI.), was. introduced at a conversion factor of S/. 1.00 - I/m.1.00. The so-called Inti Mill6n (I/m.) -- equivalent to one million Intis (1/.1,000,000) -- was introduced on December 16, 1990 to simplify accounting and as a means of transition between the Inti and the Nuevo Sol. PERU THE DECLINE OF THE VALUE OF LABOR IN PERU: CAUSES AND REMEDIES Table of Contents Page PREFACE . . . . . . . . . . . . . . . iii EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv PART 1: POLICY CONCLUSIONS AND RECOMMENDATIONS . . . . . . . . . . . . . 1 I. MAIN POLICY CONCLUSIONS AND RECOMMENDATIONS . . . . . . . . . . 2 A. Introduction . . . . . . . . . . . . . . . . . . . . . . . 2 B. Main Features of Labor Market Rules and Labor Policies . . 3 C. Impact of these Regulations . . .. .. . . . . . .. .. 9 D. Protection by Trade Unions and by Public Enterprises . . . 12 E. Need to Reduce the Discretionary Power of the Authorities . ....... . . . . . .0. . . . . 13 F. Wages and Income Policies . . . . . . . . . . . . . . . . 14 G. Industries with Special Regimes . . . . . . . . . . . . . . 17 H. Employment Ridigity . . . . . . . . . . . . . . . . . . . . 20 II. REFORMING THE FORCED SAVINGS SCHEMES . . . . . . . . . . . . . 22 A. Compensation for Time of Service . . . . . . . . . . . . . 22 B. Social Security and other Payroll Taxes . . . . . . . . . . 23 C. Main Recommendations .. . ....... . .. . . . . . 31 PART 2: THE BASIC ISSUES . . . . . . . . . . . . . . . . . . . . . . . 37 III. DEMOGRAPHIC AND ECONOMIC BACKGROUND . . . . . . . . . . . . . 37 A. Demographic Trends . . . . . . . . . . . . . . . . . . . . 38 B. Economic Background . . . . .. . . . . . . . . . . . . . . 40 C. Characteristics of the EAP . . . . . . . . . . . . . . . . 45 IV. MAIN ANALYSIS OF LABOR ISSUES . . . . . . . . . . . . . . . . 47 1. LABOR MOBILITY . . . . . . . . . . . . . . . . . . . . . . 47 A. Labor Force Participation and Employment Structure . . 47 B. Job Mobility . . . . . . . . . . . . . . . . . . . . . 52 - ii - Table of Contents (Continued) Page 2. UNEMPLOYMENT . . . . . . . . . . . . . . . . . . . . . . 55 A. Unemployment During the 1980s: General Trends . . . . 55 B. Structure and Trends of Unemployment by Age and Education . . .. . . . . . . . . .. . 56 C. Analysis of the Determinants of Employment . . . . . . 58 D. Duration of Employment . . . . . . .. . . . . . . . . 59 3. WAGE DISTORTION . . . . . . . . . . . . . . . . . . . . . . 61 A. Trends in Wages and Salaries . . . . . . . . . . . . . 61 B. Determinants of the Value of Labor . . . . . . . . . . 65 C. Empirical Analysis: Main Results . . . . . . . . . . . 67 REFERENCES . . . . . . . . * . . . . . . . . . . . . * . . * . 73 APPENDIX A: TECHNICAL EDUCATION IN PERU . . . . . . . . . . . . . . . 77 APPENDIX B: EXPERIENCE WITH THE PRIVATE ADMINISTRATION OF PENSION PLANS IN CHILE, AND ITS RELEVANCE FOR THE PERUVIAN CASE . . . . . . . . . . . . . . . . . . 83 APPENDIX C: NOTES, TABLES AND FIGURES FOR CHAPTER 4 . . . . . . . . . 90 APPENDIX D: ANALYSIS OF THE DETERMINANTS OF FEMALE LABOR FORCE PARTICIPATION . . . . . . . . . . . . . . . . . . 105 - iii - PREFACE This report was written while the Fujimori administration was undertaking deep economic reforms to stabilize the Peruvian economy and to resume economic growth. In consequence it is focused on the analysis of the impact of those reforms on labor markets in order to identify remaining obstacles to the restoration of labor productivity and the scope for further reforms which will both help during the stabilization process and improve the prospects for sustained economic growth. Feliciano H. Iglesias is the principal author of the report. The first two chapters draw on background papers by Ricardo Paredes and Alejandra Cox-Edwards (Consultants) respectively. The third chapter was drafted by Aitor Ezcurra (Consultant). The last chapter was written by the main author with the assistance of Aitor Ezcurra and it incorporates the analysis of the Peruvian Living Standard Measurement (PLS) surveys carried out by Indermit H. Gill (Consultant). Appendix A was prepared and written by Izumi Ohno. Robin Gaster has edited the manuscript. Marlene Sims provided secretarial support. The report is based both on findings of a Bank mission which visited Lima in October/November 1991 and on analysis of the two main data sources: the Peruvian Living Standard Measurement surveys and the Peruvian Ministry of Labor surveys. The mission was composed of Feliciano H. Iglesias (mission leader), Izumi Ohno, Alejandra Cox-Edwards, Ricardo Paredes and Aitor Ezcurra. The analysis of the survey data was carried out by Feliciano H. Iglesias and Indermit H. Gill with the assistance of Aitor Ezcurra. - iv - EXECUTIVE SUMMARY In the past 20 years, Peru's labor laws and policy environment have been incompatible with economic growth. State-led development created a large (and largely inefficient) public sector and an intricate network of controls in virtually every sector of the economy. Wage trends in the formal and informal sectors have diverged sharply. Real wages fell by about 60 percent in the formal sector between 1980 and 1990, while they increased in the informal sector by more than 50 percent. These trends have in turn produced large responses in male and female labor supply, as well as shifts in the structure of the labor force. The upshot has been sizeable inefficiencies in the allocation of human resources to the economy. What are the most important issues to be addressed to avoid further declines in labor productivity? Although previous assessments have suggested that the biggest obstacle to efficient labor markets in Peru was inadequate labor laws, the empirical evidence in this study reveals that it has not been the sole cause of deterioration of labor market conditions, and the decline in labor productivity. A. Policy Issues and Recommendations Employment Rigidity and Real Wage Fluctuations For Peruvian labor policy, the main issue is how to reduce employment rigidity and fluctuations in real wages in the formal sector. As inflation accelerated in recent years, real wages in the informal sector increased because of built-in indexation; in the formal sector real wages were drastically reduced. These extremely low real wages (in the formal sector) are incompatible with efficient labor markets. The size and nature of informal employment (and mobility) means that unemployment is unlikely to depend on the downward flexibility of wages in the formal sector. Thus, better wage indexation in the formal sector would have created a more stable wage structure. Reduced price distortions and increased wage stability were unlikely to have created general unemployment. Thus, fluctuations in real wages in the formal sector stemmed from poor indexation of labor contracts, which was extremely distortionary in times of hyperinflation. The policy implications are clear. Labor contracts should be freely negotiated and temptations to carry out heterodox stabilization programs, which distort labor market signals through incomes policies, should be avoided. Legal Restrictions on Labor Mobility and Wages Trends in employment, mobility, and wages show that legal restrictions in labor contracts -- except for poor indexation -- are less - v - distortive in Peru than might be thought. The legal framework in Peru is similar to many other Latin American countries but labor market outcomes have been quite different. Real wages collapsed in the formal sector and grew in the informal sector. Moreover, while legal restrictions were nation-wide, labor markets were highly sensitive to geographical location. Appropriate legal reforms are needed - especially to improve labor relations which would help to introduce more efficient methods of production. It would be wrong, however, to believe that such reforms alone would eliminate distortions in labor markets. Employment rigidity in the formal sector has been largely induced by policies aimed at maintaining high employment. This has distorted wages and demand for labor and reduced the share of the male labor force employed in the formal sector. Forced Savings Schemes Other distortions in labor markets derive from forced savings schemes. Recent reforms of most schemes, if implemented, will reduce the distortionary impact of payroll taxes used to finance these schemes. Particularly important are Social Security reforms, similar to those introduced some years ago in Chile. However, given the urgent need to increase employment in the formal sector, deeper reforms are needed in the payroll tax system to increase the efficiency of formal labor markets. Compensation for time of service (CTS) in the event of job loss has been reformed 'to eliminate its worst effects, which stemmed from poor indexation of entrepreneurs' liabilities. Under the new system - periodic payments into special accounts owned by employees - no indexation is needed. Even so, there is no justification for CTS. Compensation-for job lose cannot be greater than its costs for most employees of the formal sector. When job loss (or separation) is voluntary, insurance is not needed. Where it is not, CTS, which depends on length of service, is not a substitute for unemployment insurance given the structure of unemployment and job mobility. The only payroll taxes that should be maintained are those that finance (reformed) pension schemes, health insurance, and apprenticeships. There are enormous potential gains from developing institutions that reinforce entrepreneurs' efforts to increase worker productivity. Privatization Significant wage segmentation in labor markets derives from strongly unionized public enterprises, where wages are much higher than in other sectors of the economy. Moreover, even with existing technology, at least one third of the work force of public enterprises is redundant. To improve labor markets efficiency, privatizations should be implemented as soon as possible. - vi - B. Labor Market Trends Labor Force Participation Between 1985 and 1990 (and contrary to expected long-term trends) the female labor force participation fell from 52 to 44 percent of working age women. The sectoral structure of the labor force also changed substantially. Among women, self-employment declined from 53.7 percent of the female labor force to 45.0 percent; among men, it increased from 33.6 to 35.5 percent. Female labor force participation increases with education, except at low educational levels. Age generates a bell-shaped profile. Participation is higher in rural than in urban areas, and (within urban areas) is lower in Lima. Location does not, however, affect labor force participation of educated women. As expected, participation is lower for married and cohabiting women and for those with young children. The husband's income has strong negative effect on participation. Cross-section estimates predict a decline in female participation where both wife and husband face wage reductions. Witness the retirement of women from the labor force between 1985 and 1990, in response to the dramatic fall in real wages. Job Mobility Two main variables determine job tenure - geographical location and education. In rural areas, tenure is twice as high as in urban areas. Moreover, in urban areas (other than Lima) tenure is as high in the informal sector as in public and formal private employment. In Lima, schooling is the main determinant of tenure; this reflects the effect of formal education on the ability to learn through experience. The informal sector shows a more heterogeneous pattern of job mobility. However, investments in human capital cause lower job mobility in both the formal and'the informal sectors. Schooling affects the choice of sector and more schooling implies a higher probability of belonging to the formal sector, where mobility is lower. Nominal restrictions from labor markets regulations have less effect on mobility than might be expected. Job attachment varies with geographical location, although labor market regulations are the same in all areas. Thus, mobility in Peru fits with economic rationality. Unemployment Unemployment in Lima has been fairly constant - 9.9 percent in 1985 and 9.5 percent in 1990. Male unemployment declined, while female unemployment increased slightly. In 1985, unemployment was high among young workers (both men and women) and among those with average education. Unemployment in Lima for those in the 14-25 age bracket with .10 years of - vii - schooling was 31.9 percent for women and 21.5 percent for men. Between 1985 and 1990 unemployment increased for the less educated workers. Statistical analysis reveals that among men, the risk of unemployment is higher for young workers, and decreases with age until about 45 years, thereafter remaining constant. Unemployment decreases with education, but, is higher in urban than in rural areas. Married and cohabiting men, as well as heads of families, are less likely to be unemployed. Likewise, among women, the likelihood of unemployment decreases with age until roughly 40 years, remaining constant thereafter. Unemployment increases with schooling, but the effect of education is not noticeable among women with higher education. Unemployment also tends to be lower in rural areas, except for highly educated women and for women between 46 and 65 years old. As with men, married and family heads are less frequently unemployed. The heterogeneity of the risk of unemployment is measured by persistence indicators. If persistence is high among a group of workers, most are never unemployed while a few are very frequently unemployed. In contrast, if unemployment tends to hit uniformly all workers in a group, measured persistence is low. In Peru, as in other countries, as workers become more experienced their unemployment becomes less frequent and also more persistent. For men, persistence of unemployment is 1.5 times that of women. In rural areas unemployment is less frequent but also more homogeneous (less persistent). Heterogenity of the risk of unemployment also increases with education. Periods of unemployment are usually short. More than half of male unemployed workers are re-employed before the 13th week, and only 11 percent remain jobless, after 36 weeks. Around 97 percent of such workers are unemployed for less than a year. Unemployment spells become shorter with age - until about 45 years, becoming longer again afterwards. Schooling has no discernible effect on the length of unemployment. However, unemployment spells are shorter for rural workers and for married and cohabiting workers. Wages and Salaries Wages and salaries in Peru's formal sector plummeted in the 1980's. In Lima, real wages fell by roughly 60 percent between 1980 and 1990. The fall, however, was erratic. It followed the trends of other South American countries until 1987. Then, real wages in Peru were about the same as in 1980. Since 1987, however, formal sector wages have seen a dramatic and sustained decline unlike most other Latin American countries. In the private sector, salaries and wages decreased by 61 and 64 percent respectively but, in the informal sector, real incomes increased by more than 50 percent. The legal minimum wage fell by more than 75 percent in real terms and real remuneration in public administration declined even more - by 80 percent. - viii - An analysis of the determinants of earnings in 1985 shows that returns to schooling were high - roughly 11 percent, and even higher for more educated workers. That is high enough to make educational policies the single most important component of any labor market strategy. For the same education and experience, earnings were more than 20 percent lower in rural than in urban areas, but returns to education were similar. This suggests, that migration was highly profitable, which has powerful implications for the allocation of public resources and for the design of a sound strategy for economic development. Moreover, while policies to educate the rural population are the most appropriate for economic development, they also increase the wealth of those living in the poorest areas. In all geographical areas informal sector employees earned less than formal sector employees or self-employed workers. Incomes of self- employed workers are higher than those of formal sector employees only in Lima. In the private sector, unions had little effect on earnings, although earnings were much higher in union-dominated public enterprises. Regression analysis indicates that, for the same human capital variables, workers are paid at least 30 percent more in public than in private enterprises. Analysis of the determinants of earnings in 1990 shows that in Lima returns to schooling were fairly high (more than 12 percent), that the self- employed earned 30 percent more than other workers, and that men earned 19 percent more than women. Between 1985 and 1990, the earnings advantage of the self-employed increased significantly. For employed men, returns to schooling were high and increased with education. For men with 10 or more years of schooling, the rate of return implied by the estimates was 15 percent, while it was 8 percent for those with less than 10 years. All those effects are obtained with both the 1990 and 1985 data, and the estimates in both years are similar. Unexplained gender-related differences in earnings exist only in the informal sector in 1985 and 1990. Among formal sector employees, there is no evidence of earnings differential by gender that cannot be explained by specified human capital variables. Among informal employees, male earnings are higher than female earnings after correcting for schooling and experience. Among self-employed workers, female earnings were lower in 1985 but not in 1990. This change took place while women withdrew from self-employment to other sectors, and to non-market activities. PART 1: POLICY CONCLUSIONS AND RECOMMENDATIONS Most observers of the Peruvian economy are worried by the dramatic and persistent decline of labor productivity and employment in the formal sector. Although an inadequate legal framework is frequently identified as the main cause of the undesirable labor market trends, this causality has not been put to test. It is however, quite clear that there is a non negligible risk of missing the most important objective of any reform which is to increase the efficiency of labor markets if the true nature and determinants of the present distortions are not properly identified. The main objective of this report is to provide evidence of the impact on trends in employment and wages of labor regulations and policies with special emphasis on the labor markets response to macroeconomic policies in recent years. The analysis of this evidence is used to derive policy recommendations aimed at restoring productivity growth in formal labor markets and increasing the overall efficiency of labor markets. The report is in two parts. The first one contains the final conclusions and recommendations deriving from the main analysis carried out in the second part. The policy recommendations are spelled out in two chapters. The first one deals with policy conclusions unrelated to the forced savings schemes. These are: adequacy of the basic labor laws, policies towards trade unions and public enterprises, policies to reduce the uncertainty derived from the discretionary power of the authorities, policies to correct distortions derived from mistaken macroeconomic policies, policies to avoid the distortionary impact of general regulations on specific industries, and finally the key policy recommendations aimed at reducing employment rigidity in the formal sector. The second chapter, focused on the forced savings schemes which earmark payroll taxes to finance health and pension benefits, includes an evaluation of recent reforms and identifies further measures aimed at reducing the distortionary impact of the forced savings schemes and increase the incentives for expansion of employment in the formal sector of the economy. - 2 - I. MAIN POLICY CONCLUSIONS AND RECOMMENDATIONS A. Introduction 1.1 The efficiency of labor markets depends mostly on the adequacy of the institutional framework that determines labor relations in general and the nature of labor contracts in particular. In the long run, most countries are likely to end up with institutions that are endogenously determined by the real needs of their labor markets. It is difficult to find rigorous historical evidence of persistent departures of existing institutions from those required by the real needs of the economy, except in modern times when the most extreme degree of political interference with private decisions has become possible. 1.2 Most labor market institutions and even public interventions in labor markets tend to be similar across countries, even if they must be adapted and enforced to different degrees, depending on the country's level of development. The necessary adjustment of institutions to the realities of each country happens mostly as an unplanned endogenous process. Peru is no exception. Its legal framework is essentially the same as in many other countries, where economic growth is however much higher. In most cases, the existing labor laws are simply not used when they are dysfunctional; conversely, they are a legal recognition of desirable practices when they are in fact applied. The implication is that until recently, Peru experienced social and economic transformations not unlike those of other countries in Latin America. 1.3 In general terms, there is no evidence that existing regulations, as they are applied in Peru today, represent very serious obstacles to economic rationality. In chapter 4 there is ample evidence that despite the nominal distortions that the legal framework could potentially impose on labor relations in Peru, the structure of unemployment, mobility, and earnings does not show many important departures from what is desirable and observed in other countries. Real wages in Peru followed until 1988 trends not essentially different from the benchmarks provided by the most relevant Latin American countries. 1.4 This analysis leaves no doubt that the key features of Peruvian labor markets are associated with geographical location. In Lima, all the evidence.suggests that labor markets perform their functions much more efficiently than in any other area of the country. Efficiency seems clearly determined mostly by the degree of economic development, and not by the distortions induced by labor regulations. The most important evidence for this, in terms of the markets ability to promote changes in productivity and economic growth, is the complementarity of education with learning through experience. This effect is clearly much stronger in Lima than in other areas. Obviously, few policy changes can be sensibly recommended to eliminate these geographical differences. For the purposes of this report, it has to be accepted as a fact that markets in Lima are simply more efficient. - 3 - 1.5 Of course, this is not meant to suggest that less investment should take place in areas where labor markets are less efficient. It simply means that other benefits of those investments will have to compensate for the lower efficiency of labor allocation in those areas. In rural areas in particular, real wages appear to be substantially lower for workers with the same human capital indicators than in urban areas. This reflects not irrationality, but lower efficiency and compensatory reductions in labor costs. This market response is better than that created by any regulation. 1.6 This and many similar considerations lead to the conclusion that the suffering of the Peruvian labor force and the decline of its productivity cannot be easily remedied through the simple correction of nominal distortions supposedly produced by specific regulatory constraints. Those obvious distortions simply do not exist. The cause of the decline in labor productivity seems to have deeper roots. The following sections offer an explicit analysis of the impact of the legal framework on labor markets in general and in some specific industries, and provides policy recommendations to reverse the anomalous trends in labor force participation and restore labor productivity in formal markets. B. Main Features of Labor Market Rules and Labor Policies 1.7 During most of the last two decades, Peru has suffered from a set of misguided policies that were conceived at the end of the 1960's, inspired by a strategy of development that has been called industrial fundamentalism (Schultz, 1964), which aimed at fostering economic growth and social change through the development of public enterprises. As a result, the Peruvian economy is today burdened with a heavy legacy of institutions, regulations and policies -- not all in the,realm of labor -- that together have badly damaged labor productivity. 1.8 In general terms, the legal framework inherited by the present administration is based on the old idea that distributive policies can favor workers without much harm to industrial growth if a high volume of investments is maintained by forced savings schemes, which also keep the cost of capital low. This is an old aspiration of well-intentioned economists, who since the time of J. S. Mill or before thought that distribution of total value added in the economy could be substantially altered through public intervention without adverse effects on investment and economic growth. 1.9 The facts have stubbornly shown that most redistributive policies tend to be self-defeating. They produce bad economic incentives. Severe adverse economic effects can be expected when enterprises do not have incentives for the efficient allocation of resources in general and of human resources in particular. Usually, total investment in the formal sector declines. If it is artificially maintained, returns drop and economic stagnation follows. This is true not only in developing countries but also in the most advanced economies like the postwar U.K. or even Sweden in recent times. Inefficient labor markets impose very heavy long-run costs no matter the economy's level of development. 1.10 This strategy reduces the scope for private decision-making by accumulating distortions in general and those of labor markets in particular, while undertaking increased public investments financed with forced savings schemes. It has been a failure wherever it was tried, and Peru is no exception. More than 20 years of adjustment of the economy to an institutional framework based on this strategy has produced a dramatic fall in the productivity of the labor force, as shown in trends examined in this study. 1.11 It is appropriate to underscore here that the public sector could have played a quite different role, even with fairly interventionist policy guidelines, if public investment had been complementary with private investment rather than a substitute. The example of various other countries in both Latin America and Europe, with traditions and institutions not so different from those of Peru (Brazil, Colombia, Spain) proves that high growth is possible along with a reasonable degree of public intervention, but only if wage dispersion is not too heavily distorted by inadequate regulations and wage floors. The key for sustained growth was the maintenance of a business environment conducive to private investment in enterprises that continuously increased workers productivity. In all these successful cases of rapid economic change, policies relied on the provision of an economic environment conducive to spontaneous change in the organization of formal enterprises, and that led to increased productivity. 1.12 In these cases, labor markets were still distorted in various ways, including restrictions on firing similar to those prevalent in Peru. The difference is that public intervention was predictable, and formal enterprises remained free to determine the nature of labor contracts as a separate component of selection mechanisms based on true merit and productivity. In short, even when static distortions existed, incentives for private investments were maintained, and these increased productivity by reallocating labor resources and expanding demand for skills. The basic strategy was to provide opportunities for private investment by not interfering with the basic principles of private enterprise in the formal sector of the economy. 1.13 Labor market regulations and policies in Peru were in contrast inspired by a naive approach aimed at improving the lot of employees, while ignoring the costs in terms of foregone investments in both public and private firms. The underlying model was mostly a typical Keynesian approach, where private investments were assumed to be determined by the availability of financial resources at subsidized prices. Official credit alone was expected to provide for enough investment, preferably in public enterprises. Labor policies were optimistically devised on the assumption that productivity would increase rapidly as a result of investments in physical capital. The evidence examined in this study indicates that the long term effect of these policies has been a dramatic decline in labor productivity. -5- History of the Legal Framework!' 1.14 The Peruvian Constitution of 1933 contains a number of articles related to labor matters. Constitutional guarantees include the basic right to freely engage in any profession, industry, or occupation that does not violate moral or health standards or public security. Congress is given wide power (and corresponding obligations) in the areas of labor and social welfare. The state is obliged to enact laws governing collective bargaining. It must also promote a profit-sharing system for wage and salary earners, and enact laws to protect labor in general. The state must set safety standards; guarantee health and hygiene; and fix working conditions, severance pay, and workmen's compensation. 1.15 While labor legislation in Peru dates back to 1873, until the 1960's this legislation consisted primarily of basic laws lacking in detail, while regulations frequently overlapped or were outdated. That increased the importance of judicial interpretation in delineating the law. In 1963, a commission began to codify Peruvian labor law. This code, which took more than five years to develop, followed the International Labor Code of the ILO while incorporating customs indigenous to Peru. During the 1960's, numerous fundamental labor laws and several thousand executive decrees were enacted. These regulations provide much of the basic framework for current labor legislation. 1.16 In October 1989, the APRA Government submitted a revised labor code to Congress for urgent action. These statues have not been ratified. Since the administration of President Alberto Fujimori has taken office, a number of Presidential decrees related to labor matters have been issued while administrative interpretations of existing regulations have also changed. The overall impact of these changes has been to provide increased flexibility in the labor market, and to reduce uncertainty and frictions in industrial relations. The system of payroll taxes to finance social security and other benefits has also been modified. Employment Stability 1.17 The existing legal framework was mainly introduced when general economic strategies were very interventionist and labor policies fairly paternalistic. As a result, they are dominated by concerns about employment stability. Many administrative changes in the labor code have been justified as efforts to promote and preserve employment stability. In examining legislative areas which are obviously related to labor stability (such as length of probationary employment, restrictions on temporary and part-time employment, and dismissal procedures) it is important to consider both the direct and indirect impact on stability. Additionally, nearly all labor issues (such as unionization, forced savings programs, social security schemes) affect employment stability. 1/In this section, the source for historical information prior to 1968 is: Labor Law and Practice in Peru, United Sates Department of Labor: Bureau of Labor and Statistics (BLS Report No. 338; 1968). - 6 - 1.18 Probationary Employment. Prior to 1985, the probationary period of employment was three years. After individuals were employed at a firm for more than three years, they were considered permanent employees. In late 1985, soon after the APRA administration took office, the period of probation was reduced to three months. Recently, the new Job Promotion Law was passed which allows a more flexible probationary period. Under this new legislation, the initial period of three months is extendable to a total of six months, provided the work requires a training and adaptation period. In the case of management and other positions of trust, the probation period can be extended to one year. 1.19 Temporary Employment. When the time the probationary period was three years, there were severe limitations on the number of temporary employees that a firm could hire. In mid-1986, soon after the probationary period was reduced, many of these restrictions were removed. Instead a program permitting massive temporary employment (PROEM) was introduced. 1.20 The PROEM program was a transitory measure designed to last two years. Under it, any firm employing at least one permanent employee could hire an unlimited number of temporary workers for a maximum length of tenure of two years. Employees kept on after the two year limit were considered permanent. In effect, the PROEM program created a probationary period of two years rather than the legislated three months. The program was widely utilized with nearly all new workers employed during 1986 and 1987 hired initially as temporary workers. While the PROEM program was eventually extended in 1988 for an additional two years, the transitory nature of the legislation aggravated business uncertainty and caused large-scale dismissals prior to the decision to extend the program. 1.21 After the PROEM program ended, a stable institutional environment providing for the employment of temporary employees was created with a system of Fixed Term Contracts. The net effect was comparable. Firms can now hire any number of workers under fixed term contracts, so long as a given individual's tenure with a firm does not exceed two years. Any employment generated by new or expanded economic activity, or because of technological change, can be filled through fixed term contracts. Recently, temporary employment was authorized for initiation or increase in workload, changing labor market needs, or for retraining of companies. 1.22 Individual Dismissals. In October 1991, a new decree changed some of the norms that regulate the basic labor law: the "Ley de Estabilidad Laboral." The changes in the decree are more formal than substantive, but they represent nonetheless an improvement in the legal framework to the extent that they define more precisely how to apply the law. According to the decree, employers are not allowed to dismiss workers unless they have committed a serious offense. Serious offenses include: refusal to work, absenteeism, unjustified refusal to take a medical examination when medical conditions affect production, drunkness, stealing, and illegal strikes. Any of these causes ultimately has to be proven in court. In case of lack of proof, the judge may reestablish the worker in the job, and the worker may choose a compensation equivalent to 12 months salary. The law affects all workers in the private sector, as long as they work at least four hours per day for a given employer and have worked for the employer for more than three years. - 7 - 1.23 Kass Dismissals. Existing rules permit mass dismissals of employees should such measures be necessary on economic or technological grounds. In practice, approval for mass dismissals must be granted through administrative channels. This procedure has not been used, mostly because employers prefer to pay their employees for their voluntary resignations. This practice is now very common, even in the public sector. While it continues, it will be very difficult for the Ministry of Labor to approve any application for a mass dismissal in the private sector, especially if that means that workers will lose their jobs without compensation. Minimum Wages 1.24 Minimum wage legislation was first established in 1962. Minimum wages were to be set by a board with representatives from the government, employers, and employees. The level of the minimum wage varied with the sector of employment, regional cost of living, and nature of the occupation. This process continued until August 1985, when minimum wages were unified, with the government unilaterally determining the level. Throughout the 1980's, the real value of the minimum wage declined substantially. As the minimum wage is very low, it is tempting to conclude that the legislation has had minimal effects; that the wage conditions that result are not binding. However, the level of minimum wages is, in fact used in indexing compensation, and ceilings and floors of payroll taxes. Of course, while the current level of minimum wages is very low now and has been low in recent years, this fact does not mean that the minimum wage will have no adverse effects in the future should its real level increase. 1.25 In August 1990, minimum wage procedures reverted to a system similar to the original. A board representing the government, employers, and employees determine, by sector, the minimum wage level. The increased sectoral flexibility in minimum wage levels is beneficial, reducing the negative effects of such legislation. Organized Labor 1.26 During a long period in Peruvian history, strikes were used for political rather than economic objectives.- During the administration of former President Alan Garcia, unions became large monopolistic entities with substantial capacity to wage political strikes. The current administration has attempted to modify regulations governing unionization so as to neutralize some of this political power. Unions can now be organized with as few as twenty members. These unions are to be formed along craft lines rather than on industrial or geographical lines (such as citywide unions representing diverse occupations). These rules aim to restrict unions to economic rather than political objectives. 2/The proceeding analysis is based upon the interpretation provided by: Johnson, Nancy R. The Political, Economic, and Labor Climate in Peru. Industrial Research Unit, The Wharton School, University of Pennsylvania, 1978. - 8- Profit Sharing and Employee Ownership 1.27 The basic labor laws define the concept of "Comunidad Laboral" (CL) with the aim of using profit sharing as an instrument to promote employee participation in management and ownership. Recent regulatory reforms have maintained the principles of the basic laws but have changed important aspects of the system: the balance has now shifted toward restricted profit sharing and away from employee ownership. 1.28 Previous regulations sought to use profit sharing to fund the CLs, which fostered the development of partially or totally labor managed firms. In practice, however, only 4 percent of all workers in Metropolitan Lima have part of their income paid in the form of shares in the CL. Although the fraction is small, it is heavily concentrated in mining and utilities, where 30 percent of the workers report income in the form of CL shares. 1.29 Under the recent reforms, profits are not distributed to individuals entirely through the CL, as under the previous system, but are calculated using a well-defined formula as a function of wages and days worked. With the disappearance of this important function, the CLs lose much of their rationale. Also, the previous mandatory funding of the administration services of the CLs has been eliminated (with 1.5, 0.5, and 1.2 percent of profits in the industrial sector, mining, and fishing industries respectively). As a result, worker participation in the ownership, management, and income of private enterprises is now an option for entrepreneurs rather than an obligation imposed by the regulatory framework. Mandatory profit-sharing as a component of employees compensation is maintained with rates of 10 percent of profits in industry, telecommunications, and fishing, 8 percent in mining, and 5 percent in other enterprises. These rates are somewhat lower and more uniform than they have been historically. 1.30 These reforms are a desirable improvement in the legal framework of labor market relations. There is no evidence in the literature of strong improvements in productivity derived from employees sharing in profits, so although profit sharing is quite common, it is not expected to produce increased efficiency if imposed by law. On the contrary, there are good reasons to expect that inadequate definition of property rights in enterprises will likely have severe adverse effects when labor-managed firms are artificially created by public regulation. The Peruvian labor laws were based on the opposite assumption, that employee ownership would create incentives for increased labor productivity. The facts have proved that this assumption was unrealistic. 1.31 The regulatory framework in place after the recent reforms still allows for profit-sharing schemes in the private sector if they seem superior to alternative compensation schemes. In the public sector, however, it is difficult to make a good case for the private appropriation of public property. In principle, eliminating employee ownership of public enterprises is undoubtedly a natural component of any sound reform of the public sector. Even simple profit-sharing schemes may be problematic in public enterprises. After all it is difficult to find a good reason for the existence of profits in many public enterprises which provide public services as monopolies. Yet, if the industry is not a monopoly, the most desirable policy for public - 9 - enterprises will surely be privatization. Once privatized, entrepreneurs and employees are free to choose the best compensation system, which in some cases may well be profit-sharing or even employee ownership. C. Impact of these Regulations 1.32 The adequacy or otherwise of Peru's legal and regulatory environment for labor markets has to be judged by its real effects in the economy, and not on the basis of a priori assumptions. The evidence in chapter 4 shows that the formal sector has not been able to adapt to changing circumstances, and has in recent years lost productivity and total employment relative to the informal sector. In many industries at the margin of the formal sector, an unprotected informal sector developed where private arrangements outside the legal framework were possible and profitable. In the long run, this dual system has produced outcomes which are contrary to the stated objective of the industrialization policies of the 1970's. The formal sector's share of total employment has declined. In fact, the need to bypass the legal framework is so strong in Peru that it has even triggered an ideological response in the emergence of a somewhat naive extreme economic liberalism, which views the further development of informal labor relations as a desirable strategy. 1.33 Table 1.1 shows the evolution of the industrial structure of employment from 1980 to 1988. Employment growth was well above the growth of value added, which actually declined at an average annual rate of 0.7 percent. This simple fact illuminates crucial trends in Peruvian labor markets. As the amount of work decreased, the number of individuals employed increased. There is obviously little economic rationality in this. The natural response of an efficient system would have been to reduce employment, releasing the less productive workers, instead of distributing a reduced volume of production among a growing number of workers. 1.34 Figures in Table 1.1 clearly show that employment trends have been dominated by the expansion of public employment. The growth of this component of the formal sector did not translate into increased productivity of the labor force. On the contrary, value added declined both for the economy as a whole and in the formal sector. The substitution of informal activities for formal jobs is the final outcome of this process, as revealed by the decline of employment in manufacturing. Sectors which expanded employment more than the average are those where informal jobs are likely to exist, such as commerce and services. 1.35 Direct evidence on the share of the informal sector in each industry can be obtained from the data on Social Security coverage. Table 1.2 shows that in Lima, Social Security coverage is widely dispersed across industries. There is also evidence that the share of the informal sector is high, but has not increased during the most recent years. - 10 - Table 1. IMPLOYED INDIVIDUALS SY ECONIC ACTIVITY (thousands) Anna S1980 1988 Growth .-Mining:111.7 172.9 5.6% Whyfature648 742.2 17 Elet.jGas1 and Water 16.8 21. Construction 217.9 259-4 2.2Z Commerce 731.8 1080.8 5.0% TraAspo.rt4ton 245.8 317.1 3.2% Financial Services 1397 180.1 3.2% sevces 1251.4 1923.9 5.% Total Employment 5586.5 7205.5 of Mvch Public Sector 545.6 885.1 6.2% Pi61~ eEterprise 90.7 172.1 8.3 Othe Public 454.9 713 5.8% SOURCEf INEI and INP 1.36 Wage trends examined in Chapter 4 reflect the increasing difficulty of maintaining the strategy of expanding public employment, as well as the effects of this failed strategy on the wage structure. Wages in the public sector have indeed shown significant downward flexibility. Minimum wages have also declined enough to become a non-binding constraint on the wage distribution in the formal sector. A striking feature of these trends is the relative stability of real wages in the informal sector. 1.37 These trends in wage and in employment lead to the conclusion-that the problem with the formal sector is not one of lack of wage flexibility. It may sound paradoxical, but more rational adjustment in the formal sector would have included less flexibility in real wages and more flexibility in employment. The stability of wages in the informal sector is proof that the rigidity induced by the way the regulatory framework was administered in Peru is the ultimate cause of the decline in the value of labor in the formal sector and the consequent increase of men's labor supply to the informal sector revealed by the analysis of labor mobility in the second part of this report. - 11 - Table 1.2 PERCENTAGE OF LABOR' 1ORC WITHOUT SOCIAL SECURITY 1986 98. 1989 . 1990 Aioture 60,6% 778 72.7. 78.6 MininS 9.5% 0.0% 30.4%1.1 .~~Band~. (Conomip. Goods) 63.4% 55.3% 56.4% 58.9% Manuf. (Ind.Goode) 40.01 37.6Z 25.0 39.5Z Elect., Ga.& Water 23.1. 10.0% 12.02 22.7. Construttion 61,3' 60.7% 62.6% 67. Comerce 76.61 75.02 7341 % 7847% Financial Firms .30.8 26.31 31.61 35.11 Personal Serwtces 79.8% 83.3% 74.0Z 788 Non-Personal Services 31.61 34.2% 29.3% 28 7% Bouseverk $3.22 87.42 79.12 83.31 All 597 79 71 05 SOURCE: Ministry ot Labor. Eneeste de ogaree 1.38 It is therefore a fact that the sector most directly protected by the legal framework and by the presence of unions is also the sector that has suffered the greatest decline in productivity. Most of the adjustment during this period of economic decline has come through the reduction in the value of labor, and not through fluctuations in employment in existing industries. The labor policy followed has imposed high costs of adjustment in terms of employment rigidity. Consequently, adjustment had to come mainly through reduced wages. To be sure, this response did not succeed in maintaining the formal sector share of employment in the long run. In the end the informal sector experienced both faster growth than the rest of the economy and also maintained per capita value added. 1.39 The sizeable costs imposed by the legal framework, in terms of employment rigidity, is visible through the data in Table 1.3, which records the costs of voluntary dismissals in firms belonging to various industries. Clearly, firms found it extremely expensive to attain the desired factor-mix by reducing employment, despite the severance payment mechanism provided by the CTS. The costs of employment reduction vary widely across industries, reflecting the specific regulations which apply in each case. - 12 - Table 1.3 -OSTS OF VOLUNTARY COMPENSATIO #OR QUITS (1991) 1DDUSTRY PERCENT ISMISSED COST Fisheies (D.b*c- .5' S$2.5Q0-3.oo Ports (Public2 10-15 uses monthly vage Tetile 1.5 2-5 tees monthLy wage Oil indu18 6-10 times monthly wage Be i$ng 12 48 times monthly wage Chemical 18 20-14 times monthly vage Commerce 15 12 times montely aage SOURCE' tove t irms management! D. Protection by Trade Unions and by Public Enterprises 1.40 There is a general perception in Peru that trade unions have been strong, and have been able to substantially increase the cost of labor in certain industries. The strength of trade unions derives to a great extent from the legal framework. Some legal rules, as for example the need to maintain in successive collective negotiations the advantages obtained in the previous ones, are of a general nature. Others create special regimes in certain industries. In two industries, construction and textiles, trade unions are particularly strong. These are precisely the industries where collective negotiation covers the whole sector and not just specific firms. 1.41 Despite the long tradition of strong unions in Peru, the continuous deterioration of economic conditions in recent years has much reduced their bargaining power. The authorities have also become more flexible in their policies toward firms that need to reduce their labor force, and also in matters of wages. Wages have clearly shown substantial downward flexibility. In collective negotiations trade unions have accepted safeguard clauses which meant growth in nominal wages far below inflation. The difficulties of the trade unions were revealed in the analysis of their effect on wages in the private sector. Evidence in Chapter 4 reveals that, already in 1985, there was no significant effect of unions on earnings in the private sector. 1.42 The highest distortion of earnings functions are determined by employment in public enterprises. Since unions tend to be stronger in sectors where public enterprises are predominant, it can only be said that unions per se do not have a distortionary impact on wages. Segmentation of labor markets induced by the existence of strongly unionized public enterprises is then one clearly identified and important source of labor market inefficiency. Moreover, high earnings in public enterprises is accompanied by excess employment. Rough estimates by managers of public enterprises indicate that - 13 - more than one third of their employees may be redundant. Hence it is necessary to conclude that it is highly desirable to allocate financial resources to carry out highly profitable privatizations of most of public enterprises as soon as possible. E. Need to Reduce the Discretionary Power of the Authorities 1.43 One of the most important problems that employers apparently face in matters of industrial relations is the arbitrariness of decisions taken by the special Labor Courts. These Courts frequently declare dismissals unjustified, even in the presence of the circumstances that the law explicitly enumerates as causes for justified dismissal. The Courts have used the concept of justified cause for dismissal in a very restrictive sense, largely because the law is itself ambiguous. 1.44 The legal framework generally gives considerable discretionary power to the administration and the government. This has frequently resulted in the political determination of conflicts. The framework has also created incentives to make the negotiation process itself more prone to conflict. 1.45 The discretionary power of both the administration and the Special Courts covers: a) recognition of the existence of true reasons for justified dismissal; b) arbitration of conflicts during the process of collective bargaining; and c) the power to allow mass dismissals and to close plants. 1.46 These discretionary powers apply only to the.private sector. In the public sector, the authorities play an even more direct role in the resolution of conflicts because there are no established formal procedures to handle them. The authorities have the mandatory task of arbitration. 1.47 The authorities have generally used their discretionary power according to the spirit of the basic legal framework of industrial relations in Peru, which is aimed at making dismissals expensive and at protecting employees from the consequences of economic shocks, on the principle that employers tend to adopt short-term responses to shocks. In recent times, the authorities seem to have taken decisions with a better understanding of the need to adjust labor demand during times of crisis. 1.48 The discretionary powers of the administrative authorities should be curtailed to reduce the risk of political interference with the process of industrial conflicts resolution. If the authorities are currently using their discretionary power to help in the adjustment process, this is a good thing. However, a temporary favorable evolution in the use of the legal framework is not enough to create a more stable permanent framework for industrial relations. What is required is a better procedure to deal with collective bargaining and the resolution of industrial conflicts: such a framework would, as much as possible, need to eliminate the incentives to substitute volatile political decision-making for genuine collective bargaining between the parties concerned. 1.49 At the end of the last presidential term, new legislation Was passed which introduced mandatory collective bargaining at the industry level. - 14 - This legislation has not yet been applied. If it were applied, it would generalize the regulation that currently covers the textiles, construction, and banking industries. Private entrepreneurs are opposed to this regulation, because they fear that it would imply more intervention by the authorities to arbitrate conflicts and consequently more uncertainty in the bargaining process. F. Wages and Incomes Policies 1.50 During the current period of adjustment in search of economic stability, some macroeconomists may think that bargaining at a more aggregated level would be useful for implementing an incomes policy. The underlying idea is to minimize the employment effect of the adjustment process. Most so- called heterodox approaches to stabilization are variations on this theme. The truth is that wage controls of any kind are potentially very harmful policy during stabilization periods. Wages are essential price signals under any circumstance but particularly critical during stabilization periods. Repression of inflation through wage controls is undesirable no matter how it is carried out. 1.51 In the specific case of Peru, it is clear that labor markets have already been repressed sufficiently to prevent firms from shedding labor as needed. Wages have if anything been all too low in the formal sector, certainly too low to serve as incentives for the required adjustment of employment by firms. The result has been overstaffed industries and the long- term decline in the employment share of the formal sector. As often happens with economic policy, the outcome was the contrary of stated objectives, which in this case were presumably to maintain the highest possible level of employment in the formal sector by implementing a policy of wage flexibility. 1.52 Trends in real wages and earned incomes in the formal and in the informal sectors provide important evidence on this point. It is clear that the reduction in the value of labor was mostly a phenomenon specific to the formal sector. Moreover, the decline in the value of labor has taken place not as a gradual phenomenon but as a sudden lack of adjustment of nominal wages when inflation accelerated. This indexation failure in the formal sector contrasts with the substantial indexation of informally earned incomes. This is the most salient feature of recent trends in the Peruvian labor markets. 1.53 What consequences for policy design flow from this evidence?. Some economists blame indexation of nominal wages in Latin America for a number of catastrophic events similar to the dramatic depression in Chile, when the path of the nominal exchange rate was guaranteed by the government and domestic prices were expected to be anchored to international prices. So-called heterodox approaches to stabilization blame indexation for these evident policy failures, and rely on incomes policy to reduce the growth of nominal wages. This is not the place to examine the merits of this approach. Still, one of the most important recommendations in this report is the need to reduce as soon as possible the present distortion of real wages in the formal sector originally caused by unanticipated inflation in the presence of poorly indexed labor contracts. - 15 - 1.54 The recent Peruvian experience leaves little doubt that economic depression becomes even deeper if real wages, for one reason or another, lose substantial real value over short periods of time. More generally, it is difficult to see what is to be gained by wild fluctuations of real wages that do not obviously correspond to true variations in real productivity. This is the main issue in the analysis of Peruvian labor markets which any sound program for economic recovery will have to address. The recommendation in this study is to reverse the trend in real wages in the formal sector or at least not to design macroeconomic policies on the assumption that present real wages are adequate nor sustainable. Failure to anticipate a substantial increase in real wages in the formal sector as a necessary outcome of the required adjustments in labor markets can be a severe mistake in the design of macroeconomic policy. 1.55 Efficiency in labor markets requires reasonable price signals, and these are certainly not provided by real wage declines of 60 percent in less than two years. This is in fact a reason for real concern: labor productivity cannot be expected to increase while real wages are maintained at their current levels. A much better indexation of wages which prevents their excess volatility is a necessary component of desirable labor market institutions. 1.56 Proper indexation of labor contracts in the private formal sector does not mean incomes policies of any kind. On the contrary, their purpose is precisely the opposite of that of incomes policies. It aims at reducing the incentives that distortions in real wages derived from money illusion create to maintain old forms of organization. These inadequate incentives are a fundamental obstacle to economic recovery. 1.57 The analysis of economic development has revealed that economic progress depends on the adoption of scientifically based new technologies and not on the accumulating of old forms of capital (Kuznets 1971). Economies may remain for long periods stagnating even though they reveal remarkable levels of efficiency in the use of existing resources simply because the incentives that entrepreneurs face do not lead to the adoption of new technologies. Economic retardation happens when market signals make the transformation of traditional methods of production unprofitable. 1.58 Economic growth is a matter of shifting continuously towards new methods of production that allow higher levels of productivity. The discarding of traditional methods to be replaced by new ones is a process that is biased against unskilled labor. Economists call this phenomenon capital- skill complementarity because the new forms of capital are indeed increasingly complementary with the human capital of workers. This is reflected in increasingly higher levels of labor productivity and consequently higher wages and lower labor/output ratios. In order to maintain an environment conducive to economic growth it is essential that real wages conform to this gradual change in the relative productivity of various forms of labor and physical capital. These are policy implications of modern economic growth in which heterogeneous factors of production are used with increasing efficiency through the replacement of old techniques by new ones in a continuously moving disequilibrium process in which entrepreneurs are induced to invest in new forms of capital. (Schultz 1981). - 16 - 1.59 Economic growth cannot be obtained by inducing investments in old labor intensive forms of capital. The employment rigidity and low real wages that prevail in Peru today go a long way in the suppression of incentives for the transformation of the formal sector of the economy and the consequence is its stagnation. The present market signals that employers face are not those required to induce investments in new forms of capital that allow for reduced labor/output ratios and increased productivity. The new forms of capital create a skilled-intensive derived demand for labor which endogenously generates efficient wages well above those observed today in the formal sector. 1.60 The evidence examined in the second part of this report reveals that in Peru, from 1985 to 1990, real wages in the formal sector declined dramatically both in absolute levels and relative to wages in the informal sector. This downward wage flexibility did not allow the formal sector to expand. The share of the formal sector in total male labor force declined substantially during this period. Low real wages in the formal sector have turned out to be a self-defeating policy in terms of employment generation. Existing price signals are consistent with high labor/output ratios but they are not conducive to modernization and growth. 1.61 The literature on efficiency wages has revealed the limitations of policies aimed at reducing real wages in order to attain higher labor intensity in production processes. These policies derived from a simplistic view of labor markets that ignored the subtle mechanisms required to attain efficiency in the organization of production with an increasingly heterogeneous labor force. It is efficient for employers in the formal sector to maintain real wages well above those prevailing in the informal sector because productivity declines at lower wages. Wages and compensation systems are quite different in formal enterprises and in the informal sector. Efficient wages are endogenously determined by the needs of productive organizations in formal labor markets. Severe distortions of those wages entail great damage to efficiency and economic stagnation. 1.62 The new views on labor markets are radically different from the old models that explained unemployment as a result of reduced demand for labor caused by wage rigidities and institutional barriers to market clearing at full employment. The fundamental finding of the new labor economics is that efficiency is what matters and not the total number of jobs. Efficient organization of production in formal enterprises generates wages above market clearing levels. As a result, unemployment coexists with formal and informal employment. 1.63 It is now clear that this new view of labor markets fits with the realities of observed trends. Economic progress entails increasing the quality of jobs and not the quantity of total hours worked during the life- cycle. Modernization is clearly associated with a decline of both hours worked per unit of output and a concentration of the working years in a shorter period of the life-cycle when workers are most productive. This is a necessary trend imposed on labor markets by the increasing capital-skill complementarity of new methods of production. 1.64 The use of incomes policies to force real wages to market clearing levels is viewed from the perspective of new labor economics as highly - 17 - distortionary. It may be effective in maintaining higher levels of unemployment for a while but the price in terms of efficiency may be very high. Although the advances of modern labor economics are only beginning to discover the subtle mechanisms at work in labor markets, they suggest that there is much to be lost if policies based on mistaken ideas about labor market mechanisms lead to distortions in the market signals severe enough to hinder economic progress. There is a clear presumption that incomes policies would be harmful under most circumstances and particularly perverse if used as components of programs of economic recovery which aim at triggering investment in new forms of capital highly complementary with skilled labor. 1.65 This report cannot examine the most important specific policy recommendations for avoiding severe fluctuations in real wages. Ideally these policies must be designed in such a way that prediction of inflation becomes of little interest to individual investors. This is the first best policy in matters of indexation: to render it unnecessary. If, however, indexation is necessary, in the face of persistent and high inflation rates, it should be as perfect as possible. To surprise economic actors with unanticipated inflation in order to reduce the real wage bill is not good public policy under any possible scenario. This strategy has been very effective in Peru in reducing the real wage bill, but the outcome has been the.reduction of incentives for gains in productivity, and increased inefficiency in formal labor markets. 1.66 In Peru today, it would be a serious mistake to bargain collectively at an aggregated level in order to keep wages low and employment in the formal sector high. In fact, the appropriate strategy is precisely the opposite: trade unions should negotiate with employers to produce conditions under which employment can be restructured so that productivity and wages increase drastically in the formal sector. This is not only possible but necessary for economic recovery. The adjustment must take place through reduced employment per unit of output, and not through further wage reductions. Incomes policy is absolutely inadequate in Peru's current circumstances. G. Industries with Special Regimes 1.67 Some industries are constrained by special regulations, or they suffer more than others from the adverse effect of general regulations. MinIng is one of the most important of the industries with very special regimes. Various factors help to make labor costs relatively high in mining firms. First, there are strict safety and health regulations. Given the nature of this industry these regulations are obviously necessary. However, some of the services that existing regulations require enterprises to provide go well beyond the concerns for safety and health. When mines are more than one hour away from an urban center, for example, the enterprise must provide housing, a hospital, and schooling. The costs of these mandatory services have continuously increased through collective bargaining to the point of becoming an unbearable burden for small enterprises. The main factor explaining the particularly privileged position of miners is that this sector is heavily unionized. Miners' unions are the strongest and oldest in the Peruvian labor movement. - 18 - 1.68 Mining companies need a more flexible employment level because the industry is particularly vulnerable to price fluctuations. Volatile international prices make the level of activity in the sector fairly erratic and under these circumstances, laws that enforce stability imply high cost of unused labor. Recently, the need to adjust employment has been recognized, and mining companies have been authorized to contract out to miners cooperatives certain tasks, allowing a much higher degree of flexibility. 1.69 The relative advantages of miners over other salaried workers are reflected in the very long list of potential hires maintained by the mining companies. The mining companies also keep records of previously hired personnel because after three months of total employment, even over separate periods, the workers become,permanent employees under the stability laws. Miners wages, which are higher than those of most other industries, are only a fraction of total compensation. Data from private mining companies indicate that non-pecuniary compensation is 170 percent of wages. 1.70 Special labor regulations also exist in the construction industry, where the special labor regime is mostly due to the nature of the activity involved. Construction firms have to organize labor to carry out activities in a very random way both in space and in time. Economic cycles also affect this industry more than any other. Labor laws recognize these factors, and general labor regulations do not apply to this industry. Workers are hired for particular tasks and are not subject to the stability laws. Firms have a core of permanent employees who are protected by the general regulations, but they can manage the rest of their labor force with as much flexibility as demand requires. 1.71 While employment is flexible, compensation is subject to strict regulation. This in principle aims at adjusting for the erratic nature of the income earned by workers. The wage itself is regulated at a level that is today about ten times the general minimum wage. A number of pecuniary compensations with varied nominal purposes are added to the negotiated wages. In fact, most of these compensations have little to do with their stated purpose and are simply an instrument used by the unions to bargain for better conditions. 1.72 Despite nominally rigid regulations, firms manage to avoid their worst effects by a number of practices derived from their ability to bargain with workers at the individual level. Moreover, since employment is flexible, firms are able to pay in many cases in the form of productivity premiums, more than the regulated rates. As a result, the regulated constraints become non- binding. Moreover, there is also a non-negligible although difficult to measure amount of informal labor relations in the sector. 1.73 In general terms, employment flexibility is basically adequate in this industry because workers.are hired for specific tasks. Rigidity of employment for the core of permanent employees is as derived from the general stability laws. 1.74 Another piece of legislation introduced before the end of Alan Garcia's presidency would, if applied, give the unions the power to designate 25 percent of the workers to be hired in the construction industry. This so called "bolsa de trabajo" is of course strongly opposed by entrepreneurs. If - 19 - applied, this regulation would represent a serious threat of union interference with hiring, and would certainly not help to boost the allocational efficiency of labor markets. Apparently, it is not very likely that this legislation will ever be applied. 1.75 As expected, total earned incomes per hour worked, for the equivalent level of skills, tend to be much higher in this industry than in others, because workers must be compensated for the irregular flow of income. Moreover, the equivalent permanent income of construction workers should be higher than that of workers in other industries to compensate for uncertainty. Although unions are very important in this sector and have succeeded in regulating many aspects of labor contracts the nature of the relationships between firms and their workers make it fairly easy for employers to find ways to pay compensation according to productivity. 1.76 In general terms, there are no severe distortions in this particular labor market. Managers of firms in the sector have indeed confirmed that, even under the present difficult economic circumstances, there is not much evidence of idle workers because they shift easily between formal and informal jobs. Evidence from the PLS confirms that unemployment incidence is high in this industry but duration of unemployment spells much shorter than average. 1.77 Textiles has strong and old unions. Collective negotiations with the unions are carried out for the whole industry. As a result, wages and salaries are completely indexed. This rigidity of real compensations was not very harmful when it was introduced about 40 years ago because inflation was low, but its effects are quite different with today's inflation rates. The minimum wage in textile firms is twice as high as in other industries. This wage is also enforced down-stream, for employees of retail shops, where the same qualifications generate twice the earnings of employees in other retailing sectors. This effect is not compensation for the particular hardship of the job, but stems simply from the bargaining power of the industry unions. 1.78 The rigidity induced by the labor regulations in this sector have made it difficult for firms to survive without special exemptions. Production for export, for instance, can be undertaken with a labor force that is exempt from the stability laws. Frequently, firms export more than half of their production. Much work is also carried out by workers under a piece rate payment system. However, piece rate contracts are not the optimal compensation system for many jobs in modern industrial organization. As processes become more skill intensive, piece rate contracts become increasingly inefficient and should be replaced by other labor contracts that save on monitoring costs and induce labor force attachment. Hence excessive regulation is undesirable because its goal -- setting binding wage floors -- is not attained, while the firm's resource use is distorted by incentives for suboptimal labor contracts. These costs should not be underestimated. In the long run they may be the main obstacle to increased productivity. They are similar to the costs of informalization, which have more widely meant rapid decline in the value of labor in Peru. 1.79 Finally, in the banking industry too, collective bargaining is binding for the whole industry. The strength of the unions in this sector is - 20 - revealed by the fact that most leaders of the main trade union, the CGTP, come from this industry. Firms must pay two year leaves for union representatives. 1.80 Employment rigidity is higher in banking than in other industries both because of union power and also due to the nature of the business which makes justified dismissals hard to prove. Even the most obvious faults are not likely to result in dismissals. For example, even when money is missing from a cashier's station, justified dismissal cannot be proved. It is necessary to prove that the fault is a repetitive event. 1.81 Alternatives to the costs of labor regulation are also fairly costly, as the industry needs experienced and reliable permanent employees. Very few jobs can be carried out by temporary employees without great efficiency losses. Subcontracting has been a frequent alternative. This practice is nothing but another instance of informalization, to the extent that direct market contracting substitutes for regular employment. 1.82 In sum, general labor regulations do not have the same impact in all industries, for two reasons. First, there are in many cases specific regulations imposed by the bargaining process where unions play an important role. In sectors where union membership is important and unions are very strong, the specific regulations represent important constraints both in terms of barriers to entry and exit, and in terms of wage floors and a number of non-pecuniary compensations. 1.83 Second, different industries respond differently to the same general labor regulations. In some cases, the industry cannot easily substitute subcontracting for its own labor force; nor can it hire temporary labor without great efficiency losses. These industries are usually more unionized, and the distortionary effect of regulations for them is high. In other cases, the industry does not require a strongly attached labor force and can easily use temporary labor or subcontracting to avoid the worst effects of regulations. 1.84 The most important special regulations appear where collective bargaining applies to the whole industry: in textiles, banking, and construction, and where the nature of the industry requires a special regime as in mining or fisheries. In the first case, the special regime tends to segment the labor market. In the second case, however, regulations, if freely negotiated, are desirable because they save costs both by replacing individual transactions with general agreements, and by reducing frictions in industrial relations allowing for the internalization of a number of effects that are external to individual enterprises but internal to the industry. H. Employment Rigidity 1.85 This review of the effect of regulations on specific industries provides further evidence that there is substantial employment rigidity in Peruvian labor markets. There is little doubt that the employment rigidity has become particularly harmful during the period of hyperinflation as it has been necessarily accompanied by an extreme distortion in real wages in the formal sector. This main distortion explains the rapid growth of male labor - 21 - supply to the informal sector while the share of women in the formal sector increased. The inefficiency that this evidence entails must be underscored. 1.86 It could also be possible to read the evidence on these labor flows as simple rational response of labor markets to the wage distortion which would be itself exogenously generated by inflexibility of nominal wages. This raises the question of the determinants of the wage inflexibility and makes this interpretation rather unconvincing. 1.87 The important policy implication is that, whatever the nature of the process that generates the observed outcomes, employment rigidity is a very undesirable feature of labor markets, particularly during periods of deep economic adjustment that entails industrial restructuring. There can be little doubt that everything must be done to restore real wages in the formal sector to much higher levels than those observed today and this will be impossible as long as the labor/output ratio is not drastically reduced. 1.88 Policies are therefore needed to reduce employment rigidity. Regulations must be changed to accommodate the real needs of the economy during the economic recovery. Although employment rigidity does little harm when it applies to tenured employees with long experience, it is very harmful for young workers and for industries which require restructuring. A special emergency legal framework must therefore be introduced to reduce the costs of restructuring enterprises and to increase the incentives for employment of young workers in the formal sector. 1.89 The cost of inducing workers to quit voluntarily will decline if new labor policies make it easier to carry out substantial employment reductions on economic or technical grounds. This is one of the most needed changes in the legal environment which now constrains the labor demand of employers. With little change in the legal rules but a drastic change in the views of policy makers the present trends could be reversed. The change in views entails the rejection of heterodox approaches to stabilization which aim at maintaining artificially high labor/output ratios. Real wages in the formal sector cannot be expected to be kept severely distorted while industrial restructuring takes place. This fundamental inconsistency must be understood. 1.90 The current regulatory framework has been made substantially less rigid through the recent introduction of fixed term labor contracts. Further improvement will require more direct incentives for hiring young workers through the elimination of wage floors for them, and the closely related introduction of effective apprenticeship programs which in other countries have been very successful in improving workers productivity and their adaptation to labor markets (see Jimenez and Kugler, 1987). These programs must be financed with earmarking schemes which guarantee their continuity and proper internalization of costs and benefits as discussed in the next chapter. - 22 - II. REFORMING THE FORCED SAVINGS SCHEMES 2.1 In Peru, regulations of a clearly fiscal nature have introduced forced saving schemes through labor contracts. These schemes are payroll taxes earmarked for specific uses. One is a severance payment scheme which provides a lump-sum payment at the time of job separation. Two finance pensions and health insurance under the umbrella of the Social Security system. Two others finance training and housing. A. Compensation for Time of Service 2.2 The Compensation for Time of Service (CTS) is a form of forced savings scheme whose ultimate purpose seems quite poorly defined. In principle, it would be unnecessary if an unemployment benefit scheme were in place and the pensions system worked properly. Under current conditions of Peru, the CTS is a very imperfect substitute for both. 2.3 Before December 1990, CTS regulations meant that in case of job separation, be it quit or lay-off, employers had to pay workers a lump sum equal to the accumulated savings of the employees. These savings were one monthly salary per year based on the most recent monthly earnings. 2.4 In principle, this system implies a certain capitalization of employees savings, but in practice it is a very poor indexation and capitalization scheme. The CTS could not be seen as a genuine savings scheme precisely because the relationships between the original savings and the final liabilities of employers was very remote and subject to great uncertainty. 2.5 The scheme was potentially very distortionary. By indexing actual payments to the most recent wages, the CTS implied different obligations for employers depending on the steepness of life-cycle wage profiles. Moreover, for all industries, in a very inflationary environment the system is very sensitive to fluctuations in the real value of wages. As is well known, these fluctuations are very important in all economies but particularly in those with high inflation. The CTS added to the volatility of real earnings and to the uncertainty of employer obligations. 2.6 The system obviously created large incentives for employers to use practices aimed at avoiding its worst effects. These included using non-wage compensation to reward increased productivity and consequently, the relative share of non-wage benefits must have attained very distortionary proportions in some industries. On the other hand, regulators attempted to redefine earned income comprehensively, in order to calculate contributions both to Social Security, and to the CTS itself. 2.7 The system allowed employees to borrow from the accumulated savings under specific circumstances. Loans in these cases were to be repaid according to rules defined by the legal framework which again used very poor indexation schemes. As a result borrowers could repay their loans at real values much lower than the principal borrowed, particularly during periods of rapidly accelerating inflation. - 23 - 2.8 The December 1990 reform brings more transparency to the CTS. Now, upon job separation employees receive a lump sum determined by the accumulated savings which have actually been paid by the employer every semester into an account owned by the employee. All banks and financial institutions under the surveillance of the Banking Superintendency are eligible to operate the CTS accounts. Workers use the bank of their choice, and may also leave the savings with the employer. 2.9 After these reforms, the system is still one of forced savings with the same objectives, but its worst effects have been eliminated. Employers are now released of their obligations every semester, and consequently the uncertainty of these liabilities has been eliminated. For employers, the new system is just a way of defining total annual employee compensation, with a fraction earmarked for the CTS accounts. The employees also have much better knowledge of the real value of the accumulated savings. Moreover, competition for the CTS funds should encourage more efficient management. CTS also provides the financial system with an important source of long term savings; it is today practically the only source available in Peru. 2.10 The new law also includes a system for the payment of the CTS accumulated before December 1990. Employers can pay immediately if they wish, but have ten years to deposit the total of past CTS in the employees accounts. Obviously if employers can obtain higher returns from these resources than the expected growth of the real value of employee's wages, they have an incentive to postpone payments. However, uncertainty about the evolution of real wages makes delaying payment of past CTS obligations less attractive. Given that wages are currently extremely low, in most cases it would be very risky to postpone these payments. Of course, for less tenured young workers, future wage growth is expected to be much higher than average, and employers will pay these CTS obligations first. B. Social Security and other Payroll Taxes Introduction 2.11 Social Security contributions represent 18 percent of wages. This contribution is divided in equal parts into components for pensions and health care insurance. Employers are directly responsible for 2/3 of the contributions and are authorized to make a payroll deduction of 6 percent as the employee contribution. Independent workers and the self-employed can also be voluntary contributors if they pay the entire 18 percent tax. 2.12 There are three other payroll taxes. One is the employer premium for Life and Accidents Insurance to the IPSS which applies to the entire payroll. The rates vary between white and blue-collar workers, and also among blue collar workers depending on the job risk. A second contribution is a six percent tax on wages to finance the National Housing Fund (FONAVI). With an indexed ceiling, the employer pays five percent of wages, and is authorized to deduct 1 percent of wages directly from the worker's salary. Finally, a third tax of 1.5 percent of wages is paid by employers in manufacturing, repairs, installation, and maintenance activities to finance a training program (SENATI) (see Appendix A). - 24 - Table 21 AVERAGE PENSION (14 fmillion of Dec 1990) AVERAGE MiININ$UN YEAR PENSION PENSION 1980 Z06,89. 1981 173.61 1982 146.77 1983 143.87 1984 12,83 1985 134.68 1986 11.9.28 1987 133.95 1988 164.69 1989 (June) 25.8 5.42 .SOURCEi Rueda et al. (1989) 2.13 The first two contributions are found in some form in the regulatory framework of most Latin American countries. Some countries also have a payroll financed training system. 2.14 The pensions system is of the pay-as-you-go variety. It started in 1936 with the so-called ISeguro Social Obrero". Legally, the system is a partly funded benefit system but the norms-that should regulate the invested funds have not been enforced. One of the system's main shortcomings is that while contributions have stagnated, beneficiaries have increased in number. In 1980, there were 18 contributors for each beneficiary. By 1989, that ratio was only 11. The IPSS has maintained a balance between income and expenditure by increasing the ceiling for contributions while reducing the average level of benefits. Tables 2.1 and 2.2 provide evidence of this trend. 2.15 Until October 1990, the 18 percent Social Security contribution had a ceiling on maximum insurable earnings, which translated into a ceiling on contributions. The ceiling was defined periodically as a multiple of the Lima minimum wage (LMW). There is also a legal floor for contributions. Through time the floor has been reduced in real value because it was indexed with the minimum wage while the ceiling has been redefined to increase the contribution of the highest paid workers. In 1989, as shown in Table 2.3, in 1989, 31 percent of reported salaries were at or below the minimum, 40.5 percent were between 1 and 2 minimum salaries, and less than 1 percent reported salaries above the ceiling of 20 times the minimum wage. Because many of the benefits are independent of contributions, the system creates incentives for employers - 25 - Table 2.2 MAXIMUM AND MINIMUM SOCIAL SECURITY CONTRIBUTION (I./ million of Dec. 1990) (1) (2) (3) (4 (5) DATE dAXIMUM RATIO MINtIUM RATIO LIMA ?WAGE (1):1(5) WAGE (3)/(5) MIUWNU: WAGE Jan. 1980 734.85 5 146.97 1 146.97 S>.~:~.;Ja. 1988 163$7.4 20' 81.87 1 iT81.87 t 1990 -- no 31.55 1 31.55 Notes Column (1) shows the "Remuneraci6n MAxima Asegurable" or maximum salary over which the contribution may be calculated. Column (2) shows the "Remerci6n Minima Asegurable" or minimum .salary over which the contribution is calculated." This limit only applies to the employer contribUttion, or 12 percent of the wage. There is no limit to the "Remuneraci6n Maxima Asegurable" for the calculation of the employee's contrthution (6 percent of the wage). SURCE: Legislaci6n de Seguridad Social, Lima, Peru. Editorial "El Carmen 1991, and Webb et al. (1990). and employees to choose contracts or arrangements that produce the minimum possible contributions. 2.16 The IPSS is basically funded by the contributions of employers and workers, but it has no computerized registration system. It appears that there was a computerized system until 1987 but it was discontinued due to lack of personnel and a bad experience with fraudulent data manipulation. Today, employers calculate and pay their contributions according to their own declaration. At the time when pensions have to be determined, workers show their income and tax records. There is accordingly no cross-check on the extent of possible evasion of contributions. Evaluation of the Pension System 2.17 Men are eligible for retirement benefits at age 60 and women at 55 in the general IPSS program. The pension is calculated on the basis of a formula that takes into account years of contribution, (generally, men must have contributed at least 15 years and women 13), the average income earned during the previous 12 months, and the family composition. - 26 - Table 2.43 DISTRBUTION OF CONTRIUTORS ACCOEDING TO :ffCOM (June 1989) RANGE 1UBER PERCENT Up to 1 R.M.A1 39,623 30.9 1 - 2 R.M.A. 51,864 40.5 2 - 3 R.M.A. 17121S 13.4 3 - 4 R.M.A. 6,487 5.1 4 - .5 R.M.A. 3,345 2. .. 5 - 7 R4M.A. 3,647 2.9 7 - 20 R.A. 4,711 3.7 A mbove 20 R.`.A ,117 0.9 Total 128,046 100.0 Note: R.M.A. stands for 1Remuneraci6n Minxima Asegurable* or miniman salaty over which the conribut.ion 4. calculated, SOURCEs DNPIRAD, as~ quoted in Rueds et al. {op. et..) 2.18 It is easy to calculate the pension for the first year after retirement, but inflation makes that calculation much more difficult in subsequent years. Indexation problems have created great uncertainty concerning the real value of pensions. In 1984, a law established a quarterly indexing rule according to the Consumer Price Index (CPI). This indexation was respected until 1988, when due to lack of funding the IPSS discontinued the CPI indexation and replaced it with arbitrary adjustment rules. As a result, pensions have declined in real value, and most are now clustered around the minimum level. This in turn has created incentives for contributions to decline in similar fashion to near the minimum levels. 2.19 The relevant life expectancy at retirement age is today approximately 16.5 years. An individual who contributes the minimum to the IPSS gets the Lima Minimum Wage at retirement. This individual receives a very important transfer of real income through the system. Assuming 40 years of contributions and a capitalization rate of 3 percent, this individual would have to pay 16 percent of the reported minimum income to finance the pension. The system is, of course, even more profitable for workers who contribute for less than 40 years. Evaluation of the IPSS Health System 2.20 Contributions of workers to the health system are independent of. marital status, family size, and age, being based solely on reported income. - 27 - Since 1985, benefits have been extended to the worker's spouse and to children younger than 18. Table 2.4 shows the evolution of a number of indicators of the structure of the system. The declining ratio of contributors to insured resulted not from a gradual process, but from the decision taken in 1985 to transfer a significant surplus from the operation of the pensions system to extend health insurance coverage. Table 2.4 POPULATION COVERED BY IPSS HEALTH INSURANCE (Ratlo in Percent Terms) YEAR CONTRIBUTORS "FACULTATIVOS" CONTRIBUTORS CONTRIBUTORS TO TO TO TO INSURED CONTRIBUTORS LABOR FORCE POPULATION 1981 93.73 2.19 39.59 12.90 1982 93,76 2.55 41.39 12.87 40.61334 1983 t 934 .2 906 1984 93.79 3.33 4.1813.60 1985 5S.8 3.72 14 13.79 1986 50.21 4.01 42.00 14.00 1987 50.20 4.44 42.43 14.23 1988 50.17 4.81 42.91 14.45 1989 50.75 5.03 44.45 15.05 1990 50.62 4.95 48.89 16.58.' Note EA oe be and rs1 loterst apear a. contri.ito: for the first time. SOURCE: IPSS - Direcci6n de Estadiatica- and Cu&nto, S.A. (1991). 2.21 The share of the population which contributes to the financing of the IPSS health care system has not declined over the last seven years. Moreover, the fastest growing category is that of voluntary contributors ("aportantes facultativos"), who pay directly to the IPSS as self-employed or as independent workers. Voluntary contributors usually pay the minimum 9 percent of the income estimated from their tax returns. 2.22 The incentive to become eligible for health dare services under the IPSS seems to be growing. Three factors have, however, helped to strain the system finances. Firstly, the coverage of the health programs almost doubled in 1985. Secondly, real wages declined, particularly since 1987. Although wages are also an important determinant of health costs, total costs of the insurance system have not declined as much as wages. Thirdly, an increasing fraction of contributors are voluntary. Most of them pay the minimum, which is apparently much less than the marginal cost of the insurance obtained. - 28 - 2.23 The IPSS has also faced a recent drain on resources because many new labor contracts are fixed term and the workers involved are not joining the IPSS health services. Workers in small enterprises seem to be voting more often to stop contributing to the IPSS, using the 9 percent payroll tax to buy private health insurance instead. 2.24 The new IPSS administrators have initiated a deep reform program to improve services, reduce unnecessary costs, and recover unpaid contributions from private and public sector employers. The IPSS has become the leader in the strategy of reducing employment through individually negotiated compensations. Total employment was reduced from 44,000 in 1990 to 31,000 in 1991, leaving the IPSS with only 1,000 more employees than in 1985. Monthly revenues are increasing since the introduction of a new system for managing financial resources in which the IPSS is no longer forced to operate with official banks and can therefore obtain much higher returns for its funds from commercial banks. 2.25 The reforms introduced by the IPSS to improve its financial management seems appropriate, although by no means sufficient. The task ahead seems much harder as the IPSS tries to reestablish its position as a reliable provider of health care services. Currently, the system is much too bureaucratic and centralized. Budgetary allocations are made by region and type of service on a historical basis extrapolating past expenditures without any real assessment of actual demand or existing capacity. Private Pensions and Health Systems Complementary with the IPSS: Recent Reforms 2.26 The basic laws governing pensions and health care do not allow private systems to substitute for the existing Social Security system. As a result, the new decrees have introduced reforms based on the principle of complementarity. The existing payroll obligations are maintained, but the new regulations change the nominal incidence of Social Security taxes. While employers were responsible for contributions equal to 12 percent of wages and workers for 6 percent of wages before the reform, these earmarked taxes have been changed to 2 and 16 percent respectively. The total tax incidence has been maintained while the structure has been changed in order to let the employee allocate a larger volume of resources to complementary schemes. 2.27 These complementary schemes include private pension plans and private health insurance. The worker may now allocate a fraction of his contributions to capitalization accounts for complementary pensions which will begin operations in July 1992. Private administrators of pension funds (AFP) will operate under the regulation and surveillance of a special Superintendency. Workers cannot allocate all their contributions to private pension funds. Similarly, complementary private health care providers (OSS) will operate under the surveillance of another special Superintendency. 2.28 The new law establishes two basic programs. The basic pension program and the basic health care program will be financed with a minimum of 1 percentage point of payroll taxes each, which will be paid by the employer. - 29 - Table 2.5 SUMfMARY OF PAX1ROLt TAXES SYDSIAIO (As of Otober 1991) 1.. IUIONS TAX RATE TAX BSE Health Care MotlyWg 32 6 Prgram! - IPSS OPension Planx - IPSS Monthly Wage q6 Acidnt Insurance Xonthly Mage - 1X-12 PremiYa . IPSS SENAT... ..Proll of tose .1.52 Covered FONIAVI othyHae 1 5% ap to 8 UIT SOURCE: ( TPS and IPSS. This earmarked component is called the "solidarity contribution." As the name suggests, it is intended to serve as a transfer mechanism administered by the IPSS. This scheme is a safeguard to maintain the IPSS. 2.29 Given the Peruvian experience with choice in health care programs, the health care component of the reform is likely to be widely accepted. First, according to the law, IPSS contributors can get private medical attention, while the IPSS must reimburse these expenses according to previously established prices. This system was in place before the reforms, but was not used because prices were not realistic. Second, many employees in the private sector already subscribed health care programs through private institutions. Under the old law, this implied that they had opted out of the IPSS by signing temporary labor contracts which were not covered by Social Security obligations. Allowing this option to become legal is a recognition of the real needs of labor markets. 2.30 Implementing the new system also means defining the role of the IPSS. The law gives workers the option, but not the requirement, of staying with the IPSS. This will force the IPSS to become more efficient in order to be competitive. The new law has not defined the exact percentage of contributions that can be shifted to private health care providers, and the fraction that will remain to finance the IPSS, although minimum levels have indeed been determined. - 30 - Table 2.6 SUNfARY OF TAYROLL TAT E . ..... ... . (As of iNovember 1991) CONT.I.UTION AIMUK TA. .ATM TAX BASE EMPLOYTEE 2WLOAXE Health Care Solidari:Sty Contibti >Monthly Wage a1 (IPSS) Complementary Progrem Monthly Wage (8-s)Z (grim, Ad. ....S.) Pensions ..solidarity, Contribto kotly Yae4%I (IPSS) CmplemeItary Program . outbly Wage (# p (Priv. Ada. or ITSS) Aceiden~ts Inaurante Moeathly Wage - 1±-12% Premim (IP$S) Life Insurance Vrei.um fouthly W.ge * .531. 46%* (Choice of Provider) . . SENATI Pajrol of thoe. 152 Covered FOAV$outbly Wage it $% up to8aUIT Wotes; The ,atB cotespQude,t tth o contribution to IPS$ that will financ baei health care. The rate is tobe established from the aetuarial studies. The rate tp. corresponds to the coutribution to the ,I$$ xuat w1 f4nance a basic pension. The rtie s to be established after actuarial ,tudies. . Worker` 1an choose to pay their .ife nurance Premium to a private insurance company. In thatas, empoyers wil deduct the premima .53% for wihitecollarvorkers, .71% for blue collar workers, and 1.46% ini case of risky activiti~e) from their contribution to ihe ZPS$ :eqired åo 1ve x19km aoiotated vstk accidents on the job. SOURCE:-TPS' and IP$ 2.31 The reform of the pensions system is not designed to resolve its present crisis. The conditions of current pensioners will not change much with the new system. Rather, the new rules are aimed at creating the conditions for the healthy operation of a future pension system, one that will - 31 - benefit the existing work force on retirement. The only possible way to give the new system a fair chance is to assume the losses associated with the old one and move on. Any attempt to continue to tie current contributions to the liabilities of the IPSS would compromise the success of the reform. 2.32 A new law on the Consolidation of Social Benefits also aims to increase efficiency in the area of work accidents and life insurance. While life insurance was offered as part of the Social Security Benefits by the IPSS, the new law opens up the possibility of private competition in the administration of this insurance. 2.33 The changes made through the new law should improve the internalization of insurance costs in each industry, and increase the efficiency with which this mandatory insurance scheme is administered. It is however very difficult to assess its specific effects in each industry. It is likely that the purely private supply of insurance will not be possible due to well known problems of adverse selection. A totally free competitive system will trigger heavy demands for industrial regulation, and the result could very well be less desirable than a certain degree of social insurance. 2.34 Tables 2.5 and 2.6 summarize the nominal impact on labor contracts of the old and new systems of payroll taxes earmarked for different programs. In the new system the two key parameters, a and p, which determine the level IPSS share of revenues in health and in pensions, still await final determination, based on the eventual extent of privatization and the scope of the IPSS. These parameters can be used to make premiums a function of age, family size, and other determinants of the value of benefits to contributors. The scheme can be designed to introduce more transparency while reducing the discriminatory effects of these regulations. In this respect, it may be useful to examine some of the lessons learned from the experience of the Chilean reforms. (See Appendix B). C. Main Recommendations CTS and FONAVI are not needed 2.35 The CTS cannot be considered a proper substitute for unemployment insurance. Analysis of the structure of unemployment in Peru reveals that this insurance is mostly required for inexperienced and low-tenured workers. The proportionality of the CTS to tenure makes it very inadequate as unemployment insurance. 2.36 Independent of its shortcomings as a substitute for unemployment insurance, the CTS must be analyzed as what it really is, a forced savings scheme, intended to change the life-profile of net income available to workers for consumption. In general, if these forced savings are in fact lower than those willingly made by workers, the scheme is irrelevant and can only impose additional administration costs. Only in cases where voluntary savings would be lower than the CTS, is this a binding constraint which produces the desired effects, namely an increase in the fraction of earned income saved. - 32 - 2.37 If the analysis is carried one step further, it can easily be seen that the scheme may not end up increasing the total savings of workers. These savings are the product of the fraction of earnings saved and the share of formal wages in total value added in the economy. The forced savings schemes, as payroll taxes, make formal contracts less likely. If the net result is that the share of formal sector wages declines substantially, the total savings of workers may actually decrease as a result of forced savings in formal employment contracts. 2.38 The recent evidence from employment trends suggests that this could be the case in Peru. The decline in the share of formal wages in total value added, very visible in Peru today, has undoubtedly reduced the total savings of the labor force. To be sure, this decline in formal employment is not determined solely by the CTS, but by other obstacles to formal contracts as well. It is however reasonable to conclude that the total burden of so many forced savings schemes on formal contracts is considerable, and should be reduced. When the value of labor in the formal sector has already deteriorated as much as it has over the past three years, the CTS should be decreased or eliminated altogether. It is not an effective or desirable instrument for resource mobilization in Peru today. 2.39 For the time being, while the system still exists, the recent administrative reforms represent a great improvement. The elimination of delayed payments by employers has reduced uncertainty concerning their liabilities, and has improved the transparency of the system. The accumulated CTS still unpaid can also be paid at will by employers, up to the limit of ten years; thus even this component should not present a threat to the financial stability of enterprises. At the same time, the real wealth accumulated in employee accounts is now visible and can be properly administered within the legal constraints. It is however tempting and therefore very likely, that future administrations will change the rules for allocating these funds in an attempt to earmark them for pressing needs. This is a further argument for dismantling this forced savings scheme as soon as that becomes legally and politically feasible. 2.40 Another tax on formal contracts (FONAVI), earmarked for housing, is an obvious candidate for elimination. Most of the arguments against the CTS apply to any other particular form of forced savings. Moreover, the distortionary effect of FONAVI is higher than that of the CTS, because the earmarking scheme is more specific. The likelihood of misallocating these resources is very high. The implied shadow interest rates that workers pay for housing finance through FONAVI is far greater than the cost of borrowing from any reasonably efficient financial institution operating in the same business. It is clear that housing finance should be a component of an efficient financial system and not the result of specific earmarking of payroll taxes. There are no good reasons for burdening formal labor contracts with this particular form of mandatory savings, one which has the additional undesirable effect of segmenting financial markets. Role of SENATI 2.41 Not all earmarked payroll taxes are, however, undesirable. In some exceptional cases, earmarked payroll taxes may be the best or even the only way to provide services with discounted benefits to workers greater than the - 33 - costs of the taxes they pay. This is particularly true of taxes earmarked for the production of some forms of human capital. In most societies, clever and able individuals, simply born in too poor families, do not get human capital and training that would be extremely productive. This most important genuine market failure can be remedied through mandatory payroll taxes when ordinary budget allocations become unfeasible. This is particularly true in situations of extraordinary fiscal difficulties, and that is why in Peru today payroll taxes for certain programs aimed at increasing the productivity of young workers should be maintained and further developed. The SENATI programs, is one example, although it should be improved to properly internalize its benefits to employers and employees. It is well known that the rapid and sustained productivity gains of the German labor force owe much to the tradition of very general apprenticeship programs. Similar results can be obtained in the formal sector of developing countries (see Jimenez and Kugler 1987). 2.42 The potential for increasing productivity in the formal sector through self-financed training and on-the-job learning is a significant component of any sound strategy for economic recovery in Peru today. Consequently, the recommendation for a strategy aimed at the recovery of labor productivity is to eliminate payroll taxes where market failures are not present, and to use these taxes instead to foster the development of mechanisms which can internalize the benefits of the programs aimed at correcting genuine market failures. The programs to be maintained and improved are those aimed at improving pensions and health insurance on the one hand, and at producing the human capital of young workers that is most complementary with on-the-job training in the formal sector on the other. The first two fall within the area of payroll-financed social security, and the last takes the form of expanded apprenticeship programs. 2.43 Entrepreneurs and their employees should be directly involved in the organization of the apprenticeship programs. Most of the training should be in the form of on-the-job training. It appears that SENATI's formal training today does in fact include a substantial on-the-job component. (See Appendix A). Candidates in these programs must be selected according to rigorous and transparent examinations, with the objective of attaining the highest possible gains in worker productivity in the industries which immediately employ these workers. Policies for the Pensions and Health Systems 2.44 Regarding the recent reforms of the pension funds and the role of the IPSS, the most important issue is the stability of the system. The public has to be assured that its funds will be properly protected. A solid regulatory framework is crucial and its implementation requires a long learning process. Most of the relevant issues in this area refer to the important financial implications of the pensions system. It is of course beyond the scope of this report to examine specific policies to use the pension funds as an instrument to develop capital markets. 2.45 An important issue in the area of reforming the pension system is the need to pay for benefits under the old system. The reform must be restricted to providing a new sound set of new rules and institutions that will benefit the current work force, Existing workers cannot be asked to pay - 34 - for the crisis of the current system, to improve the conditions of current beneficiaries. Any attempt to continue to tie current contributions to IPSS liabilities would compromise the success of the reform. The only way to give the new system a chance is to assume the losses associated with the old one and move on. 2.46 What is at stake is the role of the IPSS in the new system. If the IPSS is to become an AFP, then it is best to start fresh with a separate institution. In fact, any operations of the IPSS within the new system would have to be kept in totally separate accounting books. The IPSS along with all other traditional funds must continue their operations as long as they have beneficiaries to serve. Their received contributions will decline as new entrants to the labor force and some old contributors join other new institutions. The obligations already established will have to be honored independently of the operations of the new system. 2.47 In the recently introduced rules for allocating payroll taxes a proportion to be defined later is reserved for the IPSS. It is not clear why the new system has to include this transfer mechanism. If the objective is to secure a minimum pension, the Chilean system shows that this can be attained without the IPSS collecting a fixed share of worker payments. 2.48 Mandatory insurance for retirement benefits should be reduced to a minimum. For most families, a reduction of payroll taxes and the equivalent reduction of benefits cannot imply a less desirable outcome because the family heads leave substantial wealth to their children in the form of both human capital and bequests. Higher mandatory savings for pension benefits for those families simply substitute for within-family transfers of the same amount. This means that mandatory insurance of pension benefits cannot be simply justified on the grounds that most workers save too little. 2.49 In Peru, payroll taxes are mostly paid by those workers with the highest earnings capacity (i.e., those employed in the formal sector). At least during the next ten years it is clear that retirement benefits will not be received by the poorest workers. Among those, the proportion of families that are able to leave bequests will certainly be much higher than for the labor population at large. As a result mandatory insurance for retirement benefits could be safely maintained at fairly low levels as seems desirable to reduce the tax burden of formal labor contracts and increase the tax base. 2.50 Similarly, it would be difficult to justify the pension system as a vehicle to carry out transfers to poor families. Most poor families do not earn their living in the formal sector. To use payroll taxes in the formal sector to effect transfers to informal workers is therefore not only undesirable but unrealistic in Peru today. This is probably the most important reason why it is necessary to underscore the need to create safety nets to address the most pressing needs of the poorest people. In the short run, this can hardly be seen as an objective of the pension schemes for formal employees. 2.51 Thus for all practical purposes a non-unified system of pension benefits and contributions is desirable. First, it must be realistic, taking into account the fact that most work is today found in the informal sector. Second, uniform payroll taxes in the formal sector must finance only a very - 35 - basic level of retirement benefits. Groups with special needs must be identified, and, for them, forced savings may be devised to finance the higher required benefits. If proper earmarking of taxes paid by these groups is not possible, then these groups are natural targets for safety net schemes financed by ordinary budgetary allocations. 2.52 For health insurance, the introduction of the new mechanisms that would promote efficiency in both the provision of the services and in their financing is difficult, but essential. One key aspect of implementing this as yet undefined health care reform is the definition of the new role of the IPSS. To the extent that individuals are given the option but not the requirement of staying with the IPSS, this institution could maintain its current form and purpose while the new competitive environment would presumably make it more efficient. Improvements in the administration of the IPSS are already visible in anticipation of the changes to come. One important lesson from the Chilean experience is that in order to take advantage of market incentives it is crucial to separate the financial functions from the health provision functions within the public system. 2.53 From the view point of the adequacy of labor market regulations, it is clear that if the reforms of the health system successfully induce competition and efficiency, payroll taxes for health should not be very distortionary. For the ordinary level of earned income in the formal sector (real wages in 1987 for instance), most families will end up demanding more total health services that those guaranteed by the mandatory system if rates are maintained at current or lower levels. If the efficiency of the supply of services guarantees that there are no costs involved in substituting mandatory for voluntary demand for health while it presumably has some benefits, these payroll taxes cannot be viewed as very distortionary. This implies that recovery of employment and productivity in the formal sector would not be hindered by the present level of mandatory health insurance. In other words, the most important matters in this area are related to the efficiency of the supply of health care, rather than genuine labor market issues. PART 2: THE BASIC ISSUES The policy recommendations contained in the first part of the report derive from the analysis of the main issues contained in this second part which includes two chapters. One introductory chapter provides a brief overview of the basic demographic and economic data. Then follows a chapter in three sections each one dealing with one of three main issues. The first section analyzes the issue of labor mobility in terms of labor force participation, changes in the occupational structure and job mobility. The second one takes up the issue of unemployment which is particularly relevant for the design of policies aimed at increasing efficiency in labor markets by improving job matching. The final section deals with the issue of distortions in the mechanisms of wage determination. - 38 - III. DEMOGRAPHIC AND ECONOMIC BACKGROUND A. Demographic Trends Population: Level and Structure 3.1 The total population in Peru grew by more than three million during the 1980's, from about 17.3 million persons in 1980 to close to 21.6 million in 1990. The annual rate of growth was 2.2 percent. Just under half of the population are women. Peru is divided into three geographical regions. The Costa consists of the littoral and foothills of the Andes below 2,000 meters; the Sierra consists of the Andean ranges; and the Selva, to the east of the Sierra, consists of tropical rain forest and mountain slopes east of the Andes below 2,000 meters. Table 3.1 provides a brief description of the population in those regions. Table 3.1 POPULATION AND LAND AREA OF PERU, 1991 Costa Sierra Slva :-t tousael. 8).* Land: Area ( M2 10 37 3 989 094 Source: INEI 3.2 As illustrated in Figure 3.1, the age structure of the population has shifted from the very young (age 0-14) toward the productive work years (age 15-64). In 1980, the youngest age group accounted for 41.8 percent of the population, but by 1990 this had dropped to 37.6 percent. Meanwhile, the population 15 to 64 years of age increased from 54.6 percent to 58.6 percent of the total population over the same time frame. The proportion of older Peruvians (65 years and older) has been increasing slightly from 3.6 percent in 1980 to 4.1 percent in 1990. In the early 1980's, urban areas had a greater concentration of working-age individuals than did rural areas (in 1980, 57.4 percent of the urban population was between the ages of 15 and 64 in urban areas as compared to 49.7 percent in rural areas). However, by 1985 the age structure in rural and urban areas was nearly identical. Migration 3.3 During the 1980's the main population flow in Peru was from rural to urban areas. The urban population grew at a net annualized rate of 3.1 percent over the last decade while rural population growth averaged only 0.5 percent. The urban population constituted 64.2 percent of the population in - 39 - 1980 and 69.8 percent by 1990. This trend is expected to continue, with the urban population reaching over 73 percent of the total population by 1995. Figure 3.1. TOTAL POPULATION DISTRIBUTION BY AGE GROUPS 0.9 C 0.7 0.6 0 0.5 0.4 0.3 - 0.2 - 0. 1 1980 1985 1990 1995 aged 0-14 aged 15-64 aged 65. Source: Authors calculations based on INEI data. 3.4 Between the early 1940's and the end of the 1970's, there was a very substantial population flow from the Sierra to the Costa. However, while still noticeable, the rate of movement slowed considerably during the 1980's. In 1980, population in the Costa and Sierra accounted for 50 percent and 38 percent of the total population respectively, compared to population levels of 53.1 percent and 35.5 percent respectively in 1991. 3.5 In 1985, moves due to the search for employment accounted for slightly more than one-quarter of all migration of persons 15 years of age and older. The search for higher paid jobs was of relatively minor importance, accounting for only 2.2 percent of migration. More than half of all moves were for family-related reasons (for example, working age family members moving with a head of household who was seeking employment). Table 3.2 summarizes the reasons for migration over this period. Moves instigated by employment search are most important for individuals residing in the Selva and least important for those in the Sierra. Also, moves for family-related reasons and marriage contributed significantly more to migration from rural areas than from urban areas. Migration related to study accounted for over 13 percent of moves in urban areas but only 2.5 percent in rural areas. - 40 - Table 3-2 Nation by Reason, Percentage based upon reg,onal totals rows: sum tLQZ .ore dp1y Study Marriage Pamily Othet Income 6et Related Total 2.2 26.1 .2 5.5 51.2 4.7 Urban 2.3 26.9 13.3 4.3 4.5 4.8 Rural 1.9 24.3 .5 8.7 58.1 4.6 Coata 2.7 26.9 12.7 4. 47.9 5,1 Se 1.5 21.7 7.8.. 1 7t $6.9 4.7 : 2 Source: INEI B. Economic Background The Economy Inherited 3.6 During most of the last two decades, Peru has had a legal structure and a policy environment inimical to economic development. A state-led development strategy created a large public sector, an intricate network of controls, and largely inefficient state-owned enterprises-in virtually every sector of the economy. The tax code was complex and full of loopholes, tax administration was ineffective, and tax evasion was widespread. 3.7 These conditions seriously distorted most markets in the economy. Job security in the formal sector was constitutionally protected and strictly enforced, and, together with very high tax rates on labor, helped shrink the formal sector. Land reform in the early 1970's severely restricted the marketability of land and its use as collateral. Trade policy relied on import substitution with very high tariffs and many non-tariff barriers, depriving the country of the benefits of its comparative advantage in international trade. Domestic regulatory policies conferred special privileges on incumbents. The financial sector was heavily regulated and burdened by direct government involvement, which included tight control of interest rates. 3.8 The Administration taking office in 1985 attributed chronic inflation and low growth to Peru's external debt, declaring that debt- servicing was responsible for the devaluation of the domestic currency, the erosion of domestic savings, and curtailment of investment. Consequently, it put a ceiling on foreign debt service at 10 percent of the value of exports, - 41 - and later stopped servicing its debt altogether. The authorities instituted widespread foreign-exchange controls on current and capital accounts, and introduced multiple exchange rates that were changed frequently. The government also increased minimum wages, expanded directed credit and subsidies, reduced the VAT rate, and reduced public sector prices. 3.9 This strategy led to a severe fiscal deficit, financed by an unrestrained increase in the money supply. Between 1985 and 1987, this increase had little effect on prices, due to an increased demand for the domestic currency generated by the Administration's initial credibility among Peruvians. Economic activity expanded by 17 percent, the rate of price increases fell, and real domestic liquidity experienced an unprecedented growth. This "boom," of course, proved unsustainable. After 1987 the market lost confidence, the process of monetization was suddenly reversed, inflation rebounded, and the economy deteriorated sharply. 3.10 In key productive sectors such as petroleum and mining, the breakdown of management and finances led to a decline in production. Peru's per capita GNP declined sharply, while unemployment and poverty increased alarmingly. Relations with international financial institutions, bilateral lenders, and commercial institutions were suspended as large arrears accumulated. The government's only remedy was to reduce public investment to historically low levels, with the greatest relative reduction in health and education. Inflation continued out of control, further eroding tax revenues. In 1989, the price level was 267 times higher than in 1987, output had contracted by 19 percent, and real wages had collapsed to less than half their level in 1985. 3.11 By early 1990, economic activity was extremely depressed and investment even more so. The public sector had virtually stopped maintaining its capital stock, the formal financial sector had shrunk, and real domestic liquidity was down to about one-sixth the level of 1985. Trends in the Sectoral Structure of Value Added and Employment 3.12 According to official projections, the economically active population (EAP)11 in Peru has been increasing at a faster rate than the overall population. Part of the declining trend in the dependency rate, defined as the number of persons per 100 economically active persons, can be explained by demographic shifts. The increase in the proportion of workers aged 15-64 means that a greater portion of the population is of prime working age. However, this demographic shift is responsible for only a portion of this downward trend. The higher levels of female participation in the labor force also had an effect. 3.13 The overall productivity of the economically active population, however, is deteriorating. Even those sectors which have experienced growth 1/ In this section, the EAP is defined for individuals over the age of 15. Persons not economically active include housewives, persons at school or university, retired persons, persons living exclusively off income from capital, nuns and monks, persons physically or mentally unable to work, prisoners, and persons under the age of 17 who are not employed or looking for work. - 42 - in value added over the past decade, have not kept lup vith the growth of the economically active population in that sector. As. a result, the real value added per person has generally declined. 3.14 Sectors which have generated relatively high value added per economically active person include mining, electricity, and financial services. However, these sectors engage only 5.1 percent of the economically active population. Mining and electricity tend to be highly capitalized while the financial sector employs significant human capital. Nonetheless, performance in these sectors declined throughout -the decade, the result of declining capacity utilization, lack of new investments, deteriorating capital stock, and inadequate human resource allocation. 3.15 Agriculture and the service sector, which in large part involve traditional labor intensive technology, have extraordinarily low value added. As a result, labo-r compensation in these sectors is very low. These sectors also engage over 60 percent of the economically active population. 3.16 Agriculture. The agricultural sector absorbs the largest part of the Peruvian labor force. During the 1980's, however, the percentage of the economically active population in agriculture declined from close to 40 percent to 34 percent. This decline mirrors the trend toward urbanization as Table 3.3 indicates. Table 3.3 POPULATION IN RURAL AREAS AND SHARE OF AGRICULTURE IN EAP 1ura-:1pul AP`*in- Ariculture;.] [185 .1 33i:02. 16 3662 Source: INEI 3.17 At the same time, agriculture's share of national value added increased from 10.8 percent in 1980 to 12.6 percent in 1985. While this share fluctuated somewhat during the second half of the decade, the trend was upward, reaching an estimated 15 percent in 1989 and 14.3 percent in 1990. Over the decade, value added per EAP in agriculture was the lowest of all sectors, most likely reflecting the use of traditional technologies. 3.18 These developments mean that labor productivity, defined as the average productivity for all individuals active in the sector, grew at the rate of 1.5 percent on an annualized basis from 1980 to 1985. From 1985 to 1990, productivity declined by 0.3 annually. These averages conceal large yearly fluctuations, which perhaps reflect the vagaries of agricultural production. 3.19 Mining. The mining sector involves a relatively small percentage of the EAP. Over the decade, this share has risen from 2 percent to 2.4 percent. Value added by mining as a percent of total value added declined, from 13.2 percent in 1980 to an estimated 10.1 percent in 1990. Value added per economically active individual was significantly higher than in other sectors for most of the 1980's, but fell by more than half from 4,196 Intis in 1980 to 1,824 Intis in 1990 (1979 prices). - 43 - 3.20 Labor productivity in mining declined during the first half of the decade at an annualized rate of 5.6 percent, accelerating to 10.3 percent from 1985 to 1990. This decline was caused in part by an influx of labor into the mining sector. Other factors that caused declining productivity include a deteriorating capital stock, declining demand for products and resources and deteriorating terms of trade for exports. Total value added in the mining sector clearly declined, while total employment increased. 3.21 Manufacturing. For most of the 1980's, about 10.5 percent of the total EAP was involved in manufacturing. During the early 1980's, this figure was slightly higher. Value added attributable to manufacturing ranged from 21.5 to 24.5 percent of total value added, while value added per person ranges from around 1,100 Intis to 1,340 Intis (1979 prices). However in 1989 this figure dropped to 969 Intis, and again in 1990 to 888 Intis. From Figure 3.2, it is clear that large manufacturing firms have significantly decreased their demand for labor in metropolitan Lima, both in absolute terms and in comparison to service and commercial sectors. 3.22 Electricity. This sector consistently involved 0.3 percent of the EAP and produced 1.1 to 1.2 percent of value added. Value added per person fluctuated from 2,323 Intis in 1980 to 2,076 Intis in 1990 (1979 prices). The consistently high value added reflects the modern technology used in this sector and the high level of capitalization. Despite the deteriorating economy, value added in electricity grew steadily, at the rate of 2 percent annually throughout the decade. Though this is a strong growth record compared with other sectors, it still fails to match the growth of the EAP in this sector (3.2 percent annually), or the general population growth of around 2.2 percent annually. As a result, even in this relatively well- capitalized and modern sector, the overall labor productivity declined during the 1980's. However, as a result of 3.1 percent average annual growth of employment from 1985 to 1990, and 3.3 percent annual growth in real value added, the electricity sector did generate positive average annual growth in labor productivity during the last five years of the decade. This was the only sector to do so. 3.23 Construction. The construction sector is rather small and the portion of EAP in construction declined slightly from 3.9 percent to 3.7 percent over the decade. The trend in value added per person is similar to that in manufacturing although there was an especially sharp decline in 1985 to 676 Intis from 759 Intis in 1984, resulting from the recession. As in manufacturing, value added per person declined significantly in both 1989 and 1990. Fluctuations in value added growth rates in the construction industry are more closely tied to business cycles than in other sectors. 3.24 In the construction sector in particular, it is important to recognize that the data examined here are national statistics that do not adequately capture activity in the informal sector. A significant amount of informal employment exists in the construction industry. - 44 - Figure 3.2: INDEX OF EMPLOYMENT LEVELS IN LIMA (Large Firms with 100 or more Workers) 115 110- 105 - 100 - 95 - 80 - 198 Year 0 anufacturino + Co welal 0 Sevices Source: Ministerio de Trabajo y Promoci6n Social: Direcci6n General de Empleo. 3.25 Commerce. The percentage of total EAP in the commercial sector increased steadily, from 13.1 percent in 1980 to 15.6 percent in 1990. Value added fluctuated between 17.5 percent and 18.7 percent of total value added. The value added per EAP is relatively low, dropping from close to 900 Intis/EAP in the early 1980's to only 636 Intis/person in 1988. Since 1988, this has surely declined further as in most other sectors. 3.26 The EAP in the commercial sector, manufacturing, and services has grown more than 60 percent during the past decade. Labor productivity fell more sharply in commerce (5.9 percent annual decline from 1980-85) than in services (2.9 percent annual decline from 1980-85). 3.27 Transportation and Communications. Throughout the 1980's, the transportation and communication sector has consistently accounted for around 6.7 percent of value added and engaged 4.4 percent of the EAP. Value added remains constant at 800-900 Intis/person. 3.28 Financial Services. Financial services generate 8.5-9 percent of value added, and account for about 2.5 percent of EAP. Value added was about 2000 Intis/person throughout the 1980's. Employment in this sector increased at an average rate of 3.2 percent annually between 1980 and 1985, and at 2.3 percent from 1985 to 1990. Real value added grew at 2.2 percent annually from 1980 to 1988. As a result of the diverging growth rates in employment and value added, labor productivity in this sector decreased at an annualized rate - 45 - of 1.2 percent during the first five years of the 1980's, and at 2.9 percent from 1985 to 1988. 3.29 Services. With the exception of agriculture, value added per person in the services sector is lower than in any other sector in the economy, while the service sector also absorbs a large percentage of the EAP. The level of technology is more likely to be traditional, and the mode of organization informal. Information on the service sector is inaccurate, due to the high degree of informal activity inadequately captured by available data. C. Characteristics of the EAP 3.30 Official estimates show that in 1990, over 70 percent of the labor force resided in urban areas, with over one-third of available labor in metropolitan Lima alone. About 34 percent of the EAP was engaged in agriculture, forestry, hunting and fishing. Though still very important, that share has fallen considerably from 47 percent of EAP in 1970. 3.31 Of the EAP in agriculture, 79 percent were self-employed in 1990. This percentage has been steadily increasing over the last twenty years. Salaried employees and wage workers accounted for 20.2 and 0.9 percent, respectively, of the agricultural labor force. In clear contrast, only 47.5 percent of the non-agricultural EAP were self-employed in 1990, while 30.7 percent were wage workers, and 21.9 percent were salaried employees. Again, the relative importance of self-employment has been increasing. 3.32 Public Administration has increased its share of the EAP steadily over the past twenty years, from 7.1 percent in 1970 to 10.4 percent in 1989. Employment in public enterprises increased from 1.4 percent in 1980 to 2.7 percent in 1984, and declined to 2.4 percent in 1988. In the private sector, self-employment remained fairly constant at around 58 percent of private EAP. Wage and salaried employment were 27.5 percent and 14.6 percent of the private sector EAP in 1989. Overall, more than half of the total EAP were self- employed in 1990. Educational Levels 3.33 Figure 3.3 describes the EAP in 1981 by level of education. Over 70 percent of individuals with no education are active in the agricultural sector. Individuals with primary education are frequent in agriculture (43 percent of those with primary education), but also in the commerce and manufacturing sectors (11 percent and 10 percent). Those with secondary education are mostly concentrated in services, commerce, and manufacturing (38 percent, 17 percent, and 15 percent respectively) while only twelve percent are engaged in agriculture. Active persons with post secondary education, either technical or university level, are heavily concentrated in services (60 percent), while less than 3 percent work in agriculture. About 10 percent are engaged in manufacturing, and approximately the same proportion in commerce. 3.34 Of individuals with no educational attainment, over 60 percent are self-employed. Individuals with a primary level education are also mostly self-employed (53 percent). They are also frequently involved in manual labor - 46 - (26 percent). Secondary-level educated individualsare fairly equally divided between manual labor, self-employment, and employee status. Those.vith post secondary education are primarily employees (over 70 percent). Only about 10 percent are self-employed. Figure 3.3: EAP BY EDUCATIONAL LEVEL, CENSUS OF 1981 Not Specifiled C2.650 UnIversity CO.010 No Education Ci1.0M Technical (4.050 Inicial CO.65M Secondary C20.49) PrIfnery C44.M5 Source: INEI Unemployment 3.35 Unemployment rates in 1985 were much lower in rural areas than in urban centers, and were higher in Lima than in other urban areas. For the country as a whole, unemployment rates increased during the first half of the 1980's, peaking at 11.9 percent in 1985, according to official statistics. Trends in unemployment rates for the decade are available only for Lima. They reveal that unemployment fluctuated, reaching a peak of 9.1 percent in 1983, then declining to 4.8 percent in 1987, before increasing thereafter, reaching levels in 1990 comparable to the highest observed during the first half of the decade. - 47 - IV. MAIN ANALYSIS OF LABOR ISSUES 1. LABOR MOBILITY A. Labor Force Participation and Employment Structure Assessment of Trends using Official Statistics (Peru, 1970-89) 4.1 Table 4.1 shows the structure of the labor force for the whole country, by sector, from 1970 to 1989. This table indicates that the share of public administration in total employment expanded from 7.1 percent in 1970 to 10.2 percent in 1989, with expansion coming very slowly until 1984, and at a much faster rate thereafter. On the other hand, employment in public enterprises also increased during the 1980's, expanding more rapidly than in the public administration, from 1.4 percent in 1980 to 2.7 percent in 1984, remaining stable thereafter. Between 1980 and 1988, employment in public firms increased by 90 percent, in public administration by 57 percent, and total employment by 28 percent. 4.2 Within the private sector, there was a gradual contraction of employment in agriculture. In 1970, agriculture represented almost one half of total employment. By the end of the 1980's, it employed only one third of all Peruvian workers. 4.3 According to these official statistics, the share of non- agricultural self-employment has increased regularly during the two decades, from 20.1 percent of total employment in 1970, to 23.5 percent in 1980, and 25.8 percent in 1989. These figures indicate that in the most recent period, from 1985 to 1989, non-agricultural self-employment grew more than total employment. As discussed below, this trend is opposite to that indicated by survey data for Lima. Gender Trends 4.4 Between 1970 and 1986, labor force participation (LFP) rates for men in Lima declined, while those of women increased, as shown in Table 4.2. The overall decline in male participation rates is consistent with increases in school attendance ratios and earlier retirement from the labor market. Increasing female participation could be expected if Peru followed worldwide trends. 4.5 Despite these general trends, Table 4.2 shows remarkable fluctuations in LFP rates. Men's participation, for instance, reversed its long-term decline and actually rose from 69 percent in 1982 to 74 percent in 1986. Female LFP, on the other hand, was not higher in 1983 than it had been thirteen years earlier. - 48 - Table 4.1 STRUCTRE 0F THE LABOR FoCE, 1970-89 Public Sector Prvate Sector Trotal .Workers Publ i ulc Agcutu~al Sectot No-giutr etr(00 Enterpr. Admin, ______________________ Employee Self Total Employee Self Total 1970 7.1% 12.6% 34.7% 47.3% 25.5% 20,1% 45.6% 4080 1971 7.2Z 12.3% 34,42 46.6% 26,02 20.2% 46.2% 4206 1972. 7.3% 12.0% 34.0% 46.0% 26.4% 20.42 46.8% 4336 1973 7.4% 11.6% 33.6% 45,2% 26.8% 20,6% 47.42 4473 * 19.74 7.5% 11.3% 33.1% 44,4% 27.3% 20,8% 48,1% 4617 1975 3,02 7,8% 11.02 32,7% 43.6% 27.5% 21.4% 48.6% 4765 1976 7.9% 10.7% 32.2% 42.9% 27.9% .21.3% *.49.2%. 4920 * 1977 8,2% 10,3% ,31,8% 42.12% 28.1% 21.6% .49,7 5080 * 1978 7,8% 10.0% .31.4% 41,4%.28,2%. 22.6%.50,8% .5239 1979 8.0% 9.6% 30.82 40.4% 28.3% 23.32 51.6Z 5407 1980 1.4% 8.4% 9.4% 30.4% 39.8% 28.3% 23.5%. 51.8% 5586 1981 1,92 8.1% 9.22 30,32 39.5% 28.82 23,6% .52,4%. 5773 1982 2,0% 8,1% 9.0% 30,2% 39,2% 28,9% 23.8% 52.7% 5960 1983 2.52 8.22 8.92 30.4% 39.4% 28.9% 2.3.6% . 52,4%. 6151 1984 2.7% 8.3% 8.4% 29.1% 37.5% 30.0% 24.2% 54.2% *.6341. 1985 2.6% 9,02 8.22 28,4% 36,6% 29,92 24.52 54,42%. 6531 1986 2.6% 9,4% 7,9% 27.9% 35,8% 30.2% 24.6% 54,8% .6740 1987 2.7% 9.9% 7.72 27.5% 35.22 30.02 24.9% 54.92.. 69.53 1988 2.4% 10.0% 7.5% 27.3% 34.8% 30.1% 25.2% 55.2% 7160 1989 10.2 1 7.42 27.0% 34.42 29.6% 25.8% 55,4% 7379 Note: Figures for the 1981-90 period are estimations from Webb et al. (1990), based~ en th.e National Accounts and data from the INEI. Private sector figures include employment in public enterprises. SOURCE: INEI and Webb at al. (1990). Assessment of Trende with PLS Surveys (Lima: 1985, 1990) Gender Trenda 4.6 The PLS data allow an alternative assessment of LFP trends. Table C.1 in Appendix C shows that in Lima total LFP rates decreased from 64.7 percent inl 1985 to 59.3 percent in 1990. Among men, participation dropped only 3 percentage points, while among women it declined from 52 percent to less than 44 percent. - 49 - Table 4.2 Ad ORC` S1A PERCENT OF POPULATION AGED 15 AND OVER-IN LIMA, BY 00 7 6 7 9 72 2 n..74 e 38 35 40 6 40 MSURCE: Hinistry of Labor, Boletin Mensual de Empleo, No.16, April 1987. 4.7 The sharp decline in the female LFP rates between 1985 and 1990 is as expected. As real wages declined rapidly during the period between the two surveys families responded by withdrawing their women from the labor force replacing them with an increased supply of male workers. This substitution effect has been very strong. Analysis contained in Appendix D confirms with cross-section estimates that the wage effect in the female labor supply function dominates the income effect. Purther Evidence of Women's Withdrawal from the Labor Market between 1985 and 1990 4.8 PLS longitudinal data provide stronger evidence of women's retreat from the labor force during this period. While 56 percent of women between 14 and 65 years old participated in the labor force in 1985, by 1990 the LFP rate for these same women was only 52.9 percent. The decline in female LFP is under-estimated by these figures; this sample had five more years of potential work experience by 1990, and that would by itself tend to increase LFP rates. Table C.2 in Appendix C shows that, between 1985 and 1990, departures from the labor force were twice as frequent as arrivals among women between 25 and 50 years old. The reduction in women's labor supply was observed in all schooling groups, but it was much higher among the less educated women. Change in the Structure of the Labor Force Women's Presence 4.9 The decline in female LFP reduced the ratio of women to men in the labor force from 0.70 in 1985 to 0.59 in 1990, as shown in Table 4.3. This effect was not uniform across sectors. In fact, female LFP declined only in the self-employed sector; in all other sectors, the presence of women increased during the period. Sectoral Structure 4.10 The sectoral composition of total employment changed substantially between 1985 and 1990, as revealed by data in Table 4.4. The share of the public sector in total employment has declined. Moreover, taking into account - 50 - Table 4.3 RATIO F W.OMN TO `E sY SECTOR 1985 1990 Private Secto 0.14 0.59 Self Employed 112 075 Priv.Employees 0.39 ,0.41. House Workers .40* 13.50 Publi Sector O54, 0-60 All0.59 DATA SOURCE! PLS. the decline in total LFP rates, this means that the number of public sector jobs as a proportion of the working age population has declined even more (from 11.5 percent to 9.5 percent). 4.11 Interestingly, within the private sector, the self-employed outnumbered employees in 1985, while the reverse was true in 1990. Moreover, the share of self-employment in total employment declined. Consequently, the expansion of the private sector is determined by the growth of the share of private employees in total employment. 4.12 These sectoral trends do not reflect the most important gender substitution effects revealed in Table 4.3. The decline in the self-employed sector is the outcome of two opposite developments. While there were more self-employed men in 1990 than in 1985, the number of self-employed women declined drastically over that period. Both reflect the strong substitution of men for women in the self-employment sector. 4.13 Similarly, different gender trends account for the decline in the public sector share of total employment from 17.7 to 16.1 percent. While male employment in the public sector fell from 19.6 to 16.0 percent of total male employment, for women the figure increased from 14.9 to 16.2 percent. However, the overall decline in public sector employment was sufficient to make the number of public jobs occupied by women a declining proportion of total- working age population (from 7.7 to 7.1 percent). 4.14 In sum, the most important phenomenon revealed by this analysis is the strong substitution of men for women in the self-employment sector. This results from both a declining female LFP and a shift in the structure of women's employment in favor of the formal sector. This is further evidence on the basic family strategies which determine female labor supply and women's occupational choice. - 51 - Table 4.4 DISTRIBUTION OF THE LABOR FORCE BY SEX AND TYPE OF EMPLOYMENT .1985 1990. MEN WOMEN TOTAL MEN VOMEN TOTAL Private Sector 80.4% 85.1% 82.3% 84.0% 83.8% 83.9% .Self-Employed 33.6% 53.72 41.9 35.5% 45.0% 39.0% Prv.Employees 46.2% Z5.:6 37.72 48.3% 33.4% 42.7Z House Workers 0.6% 5.81 271 0.2Z 5.4Z 2.2% Public Secto t 19.6% 14.9 17.72 16.0% 16.2% 16.1% Total 100% 1001 100% 100% 100% 100% Sample Size 1583 1113 2696 1693 998 2691 DATA SOURCE: PLS6 Growth of the Informal Sector? 4.15 A very important issue is the assessment of trends in the evolution of the formal sector. As shown in Table 4.4, public employment, which is an important component of the formal sector, indeed declined somewhat between 1985 and 1990. However, the contraction of self-employment was even larger. To the extent that fluctuations in self-employment correlate positively with those of informal employment, it is not possible to conclude from this evidence that the informal sector expanded. Of course, the informal sector cannot be exactly identified with self-employment, but it is very unlikely that figures in Table 4.4 could be compatible with growth of the informal sector. 4.16 Figure C.1 in Appendix C shows the decline in the share of self- employment among women from 1985 to 1990, a decline which is approximately equivalent for all age groups. The inverse effect occurred among men as shown in Figure C.2 in Appendix C. As the value of labor in general declined, real earnings in the informal sector increased relative to general wages and salaries. In response, the supply of male self-employed workers increased. This could not, however, have occurred without the withdrawal of women from the labor force, which means that total self-employment trends do not directly reflect the increased relative value of labor in this sector. 4.17 In summary, PLS show that there was a significant withdrawal of women from the labor force between 1985 and 1990 (LFP declined from 52 percent to 44 percent), and a small decline in male LFP (from 79 percent to 76 percent). Because of declining female LFP, the sectoral structure of the labor force changed substantially. Those women who remained in the labor - 52 - market have tended to move into the formal sector *s public or private employees, while the male labor force has tended to replace the female labor force in the self-employment sector. Consequently, self-employment has become more frequent among men. Only if one considers males self-employment as an indicator of informal sector trends, it can be concluded that the informal sector expanded in Lima between 1985 and 1990. B. Job Mobility 4.18 In principle, the existing legal and regulatory framework governing Peru's labor markets could be expected to impose very high barriers to mobility in the formal sector. However, studies of Peruvian labor markets do not provide empirical proof of these rigidities. Previous analyses of PLS data have documented features of Peruvian labor markets which suggest that the regulatory framework has not been as detrimental to mobility as could be expected (see Newman, 1988). This evidence shows that turnover rates in Lima are fairly high, while the search for jobs is very frequent and is not limited to the unemployed. 4.19 Labor mobility can be analyzed using PLS data by estimating the specific effect of fundamental variables suggested by economic theory on the duration of the present job, a variable also referred to as job tenure or simply tenure. This analysis is very relevant for policy design, because it measures the real effect of rules and institutions on labor markets, rather than simply extrapolating presumed effects from a simple reading of legal texts. In consequence, this section provides explicit statistical evidence of the main stylized facts concerning labor mobility and its determinants in Peru. 4.20 Except with constant and very high mobility, tenure will be positively correlated with potential experience (age minus schooling years minus age at first schooling experience). Total absence of mobility would produce perfect correlation of tenure with potential experience. With very frequent mobility, tenure would not increase with experience. Regressing tenure against potential experience is therefore a natural approach to the analysis of job mobility. Main Determinants of Job Mobility 4.21 Table C.4 includes regression results obtained with the 1985 PLS survey data for all Peru, and for selected sub-samples of analytical interest. In all regressions, individuals in the sample have at least 7 years of potential experience in order to allow the process that generates turnover work for a time sufficient for accumulated tenure effects to become observable. These estimates reveal that mobility is determined by the fundamental variables that economic rationality suggests. Tenure is twice as high in rural than in urban areas. Public administration employees show more mobility than rural workers. 4.22 An interesting finding is that mobility is not lower in public administration than in the private sector outside Lima. Moreover, even informal employment in urban areas outside Lima shows mobility that is not very different from that observed in the public administration. Only in Lima, - 53 - both for the private sector as a whole and for thelinformal sector separately, is mobility significantly higher than in other urban areas or in the public administration. In Lima, the informal sector does not show much higher mobility than the private formal sector. 4.23 These basic facts to some extent contradict received wisdom; they certainly do not support the view that legislation and public intervention in labor markets are the main determinants of mobility. If that were the case, job attachment in public administration should be much higher than in other sectors. In fact, though, job attachment in public administration is similar to that in the informal sector in urban areas outside Lima. 4.24 As the regional variable is the most important determinant of mobility, job attachment has to be explained by basic economic forces and not by restrictions imposed by the legal framework. The needs of labor markets themselves and the transaction costs involved in operating those markets seem to determine mobility much more than do distortions stemming from administrative constraints. For instance, the nature of jobs in rural areas means that turnover is much less likely than in any other working environments. In contrast, in Lima, lower costs of moving and more opportunities and expected benefits of moving are reflected in lower experience coefficients in both the formal and informal sectors. The dominance of the regional effect is even more evident in the case of urban areas outside Lima, where job attachment is almost the same across sectors, including the informal sector even though the later is not directly affected by labor market regulations. Education and Job Mobilry 4.25 Education is mainly related to labor mobility through the effect on job attachment of learning through experience, and the complementarity of this learning with education. Unless severe distortions prevent natural responses in the labor markets, worker mobility will likely be reduced with more education. Job attachment increases when the labor force becomes more educated mainly because general education before entry into the labor force makes on-the-job learning easier. Because more educated workers are more efficient in learning through experience they invest more in human capital accumulation while on the job. An important fraction of what is learned through experience is knowledge specific to the job, which would be lost by moving. As a result, more education generates more specific human capital, higher costs of mobility and more job attachment. 4.26 The evidence shown in Table C.4, Appendix C, reveals that in Lima education has a strong positive effect on tenure. However, in rural areas and in the private sector in urban areas other than Lima there is no effect of education on tenure. The effect of education on job tenure is particularly important in the public administration independent of geographical location. 4.27 Education should reduce mobility much more in the public administration than anywhere else if human capital accumulation through on- the-job training in the public sector were more important than in private employment. However, evidence examined below indicates that the contrary is true. In Peru, as in many other countries, workers in public administration tend to have flatter life-cycle wage profiles than workers in the private sector. - 54 - 4.28 A possible explanation for this apparently conflicting evidence could be that the measured effect of education on mobility in the public sector is only a short-run phenomenon determined by recent and unusual patterns of high mobility out of public employment among low skilled workers. Time series data on the skill mix of public administration and on mobility could help to answer this important question, but these data are not available. 4.29 If this pattern were not a purely transient phenomenon, the effect of education on mobility would be underestimated in regressions restricted to the private sector. The true effect would in part operate through the occupational selection of workers into public administration, which demands more skills on average than the private sector. Separate estimates for the private sector, however, would still be relevant because many labor policies are designed mostly for workers not in public administration. 4.30 The strong effect of schooling on tenure means that labor policies must take into account the fact that mobility varies across industries according to the skill mix that each demand. In other words, legislation aimed at affecting labor mobility will have quite different effects in industries which use less educated labor than those which demand more skills. 4.31 In terms of trends, as demand for skills increases, the effect of given costs of labor mobility decreases. This explains why legislation which imposes high costs on labor mobility may not be very harmful in advanced countries, while similar laws create powerful incentives to shift labor to the informal sector in developing countries. It is also not surprising that these laws and regulations, transplanted from developed to developing economies without much knowledge of their economic effects, are sometimes difficult to enforce. 4.32 The fact that the schooling effect in the private sector is observed only in Lima is further evidence that capital-skill complementarity in labor markets tends to increase as economies develop. Job attachment is increased by higher education of the labor force which is in greater demand in the modernized sectors of the economy. Urban markets other than Lima are still apparently dominated by industries that do not have a modern derived demand for these factors. 4.33 Regional differences in the determinants of mobility can be seen even in the estimates that cover only informal employment. This interesting finding strongly suggests that regional differences in mobility are determined by forces that go well beyond the effect of labor market regulations or the impact of trade unions. 4.34 Whatever the interpretation of these results they clearly show that only in Lima is the structure of labor mobility consistent with features typical of modern labor markets, notably the effect of schooling on tenure. In rural areas, the predominance of traditional agriculture is the most likely explanation for the absence of schooling effects on tenure. In urban areas outside Lima, the lack of opportunities to change jobs appears to produce both the absence of education effects on mobility and the important job attachment that the experience coefficients revealed. - 55 - 2. UNEMPLOYMENT A. Unemployment Durini the 1980's: General Trends 5.1 The only available official unemployment statistics in Peru for the 1980's come from Ministerio de Trabajo y Promoci6n Social (MTPS): Direccion General de Empleo. These statistics, shown in Table 4.5, indicate that unemployment increased during the first half of the 1980's, reaching a peak in the mid-1980's. For Lima, the peak was 9.1 percent in 1983. Unemployment rates after 1986, only available for Lima, declined to a minimum of 4.8 percent in 1987. They increased rapidly thereafter, reaching 1984 levels again in 1990. Table 4.5 TOTAL UNEMPLOYMENT RATES: OFFICIAL STATISTICS 14.8 179 104 104": 1W9 18 I- SOURCE: MTPS: Direcei6n General de Empleo. Notes Figures for Lima based on the Encuesta de Niveles de Empleo (ENE). 5.2 The figures in Table 4.5 indicate that unemployment rates in Metropolitan Lima during the mid-1980's were lower than in the rest of the country. This conclusion does not, however, agree with evidence from other sources. According to the PLS data, unemployment rates in Lima were higher than in Peru as a whole in 1985. The PLS surveys were conducted, for the whole country in 1985-86, and for Metropolitan Lima alone in July 1990. Unemployment figures in Table 4.6 are obtained from these surveys. Table 4.6 TRENDS IN UNEMPLOYMENT RATES BY SEX AND GEOGRAPHICAL LOCATION: PLS DATA ear8 1985 1985 1 1990 1 Area Peri RRural Urba L Total 16.1 3.2 7.5 9___9 _ 9.5_ Nfen 6.2 3.5 8.3 8.5 B.0 Wome 16.1 2'9 .4 11.8 12.0 DATA SOURCE: PLS. Note: "Urban" refers to urban areas other than Metropolitan Lima. 5.3 These figures reveal that in 1985 unemployment rates were much lower in rural areas than in urban centers and that they were higher in Lima than in other urban areas. By 1985, female unemployment rates were already - 56 - much higher than those of men in Lima; that differehitial became even larger by 1990. The geographical pattern of unemployment rates in 1985 was qualitatively similar for both sexes. 5.4 The overall unemployment rate in Lima did not change much between 1985 and 1990. Female unemployment increased from 11.8 percent to 12.0 percent. Male unemployment decreased during the same period from 8.5 percent to 8.0 percent. It is worth noticing that female unemployment did not decline despite the reduction in women's LFP rates between 1985 and 1990 in response to the decline in real wages. B. Structure and Trends of Unemployment by Age and Education 5.5 Table 4.7 shows unemployment rates by age category for men and women from the 1985 and 1990 PLS surveys. The level of unemployment is higher for women across all age categories, except for the oldest workers where men have unemployment rates twice as high as women. Table 4.7 AGE PROFILES OF UNEMPLOYMENT RATES BY SEX: 1985-90 Age Croups 14 - 2 2 - 3$ S - 45 46 -55 54 5 ttal 1 Ferns 1985 4.4 2 33 4.9 6.4 Women 10,1 6.1 9 4.9 25 69 they~ 1985 Women 20.4 94438.2 3 4 tt,0 e164 6.3 .4 '49 I ., Women 18.8 10.0 8.9 9.2 4 7 12.1 DATA SOURCE:PLS. 5.6 These figures indicate that unemployment is much higher at early stages of the life-cycle for both men and women. The male unemployment profile by age group is u-shaped with a minimum for the 36-45 age group. The female profile is also at a low level with the same middle aged group, but the profile is not quite u-shaped, because unemployment is lowest for the oldest group. 5.7 Unemployment rates in Metropolitan Lima are higher than in Peru as a whole for both men and women, and this is true for all age categories. Unemployment profiles by age groups in Lima have the same configuration as those for the country as a whole, although unemployment rates are especially high for young workers. Between 1985 and 1990, the unemployment level in Lima - 57 - remained approximately constant as did the structure of unemployment by age group. Table 4.8 SCHOOLING PROFILES OF UNEMPLOYMENT RATES BY SEXs 1985-90 Schooling 0 -4 5 6 -9 10 11+ Total Peru0 1985 4-7 910 49 omn36 5.7 7.4 11,9 1, . Lima, 1985 Woen64 10 4 lt&1. 120 12. Hiea 10.6 8i.4 8.88. 5.581 uoe.1,s1. 13.1 16.2 6 .0. . 131 DATA SOURCE: PLS. 5.8 Table 4.8 shows unemployment by schooling, in 1985 and 1990. Both male and female unemployment profiles are bell-shaped, reaching maximum at a level of 10 years of schooling. This corresponds to the completion of secondary education. Unemployment in Peru as a whole is similar for men and women across all educational groups, except for the most educated workers, where men have much lower unemployment rates than women. The unemployment profile by education in Lima is similar to that for Peru as a whole. Women in Lima have higher unemployment rates than men in every educational category. 5.9 The structure of unemployment by schooling groups changed substantially between 1985 and 1990. Unemployment among workers with low educational levels increased for both men and women, while by 1990 highly educated women had unemployment rates even lower than equally educated men. 5.10 In sum, this evidence from the marginal distribution of unemployment by age and schooling shows that, in 1985 and for both men and women, unemployment was very frequent among young workers and among those with average levels of education. Unemployment was especially high when both circumstances were present. Workers in the 14-25 age group with 10 years of schooling had unemployment rates in Lima of 31.9 and 21.5 percent for women and men respectively. The decline in unemployment with very high levels of schooling was particularly strong in the case of men. Furthermore, unemployment was higher among female workers in Lima. The structure of unemployment in terms of education changed substantially between 1985 and 1990 losing its bell-shape profile for both men and women. - 58 - C. Analysis of the Determinants of Unemployment 5.11 The structure of unemployment is fairly complex. Important features are not captured by conventional measures of unemployment which estimate the frequency of persons in a given group that are unemployed at a given time. For policy analysis it is essential to know whether unemployment hits persistently certain workers while it never hits most of workers or unemployment hits more uniformly all the working population. In both cases it is necessary to know why unemployment has a given structure. It may be just because this is the most efficient response of labor markets or on the contrary a symptom of a situation that asks for corrective policies. 5.12 Let us consider two extreme cases. In one case, eighty percent of workers are never unemployed while twenty percent are randomly unemployed half the time. The conventional unemployment rate would be ten percent while twenty percent of workers would experience unemployment at least once during a year. In another case, unemployment hits randomly all workers with spells of one month length and employment periods of nine months. Conventional unemployment rate would still be ten percent but all workers would have been hit by unemployment during a year. In the first case labor economists say that unemployment is very persistent; it hits always the same 20 percent of workers. In the second case, unemployment shows no persistence but perfect turnover instead. 5.13 The PLS data have information on the occurrence of unemployment both during the last week and during the last twelve months previous to the survey. Persistence can then be measured by the indicator alfa defined in Tables C.5 and C.6, Appendix C. The first column shows conventional unemployment rates. The second column provides the fraction of the population that would be hit by unemployment during one year if persistence were zero. The third column is the actual fraction of workers who experienced unemployment during one year. The fourth column provides the measurement of persistence. These measurements reveal that persistence of unemployment is much higher for men than for women. The effect of age and education on persistence is strong and positive in both tables. 5.14 This evidence reveals that unemployment has a well defined structure that conforms to what efficiency of labor markets imply. As workers are more experienced and educated, and therefore have more human capital, unemployment becomes much less randomly observed, that is to say, more persistent. It is important to realize that these features of efficient labor markets are robust and can be relied upon in the design of policies or institutions to help unemployed workers. As noted in the first chapter of this report, the present CTS system which is sometimes viewed as a substitute for unemployment insurance, is not fit with the real needs of unemployed workers given the structure of unemployment and the determinants of its persistence. - 59 - D. Duration of Unemployment 5.15 In order to analyze the duration of unemployment spells it is useful to construct survival curves. These curves are similar to the ones used in demography. Survivors are those workers who remain unemployed, as a proportion of the cohort that entered unemployment at a given time. Survival probabilities decline with duration according to the probability of exit from unemployment at a given time. 5.16 Table 4.9 shows the frequency distribution of survivors derived from duration of unemployment spells reported by males who are currently unemployed. A total of 168 individuals report unemployment at the time of the survey. Individuals with less than 4 weeks of unemployment length have had little time to find a job. As the length of unemployment increases, the proportion of each cohort who still remains unemployed ( the proportion of survivors) declines. 5.17 Errors in the measurement of duration are evident from these data. Although frequencies decline in most cases, there is a clustering of reported durations at 20 and 24 weeks. This clustering is caused by over-reporting of half a year as duration. Note that because these data were collected by uniform sampling throughout the year, seasonal unemployment cannot explain higher frequencies of longer unemployment spells. 5.18 When the length of unemployment spells approaches one year, they become very frequent. More than one third of the spells are longer than 44 weeks. This results partly from errors in the reported length. However, many spells really are longer than one year, which means that the few members of each cohort that remain unemployed after 48 weeks have very long unemployment spells. Survivors of each cohort are less than 10 percent or around an absolute frequency of less than 3 individuals. The average duration of unemployment for the two survivors would be around 120 weeks if they have to generate the 63 spells reported with length longer than 44 weeks. 5.19 These data indicate a very low probability of unemployment duration longer than one year. More than 90 percent of each cohort of workers who become unemployed find a job or leave the labor force in less than one year. On the other hand, little is known of the nature of unemployment lasting longer than one year. Most likely, these extremely long durations reflect something quite different from unemployment, possibly individuals who permanently report unemployment while they are not in the labor force or while they have informal jobs. Frequencies in Table 4.9 indicate not only a low survival probability after one year, but also suggest that more than half of unemployment spells end within 12 weeks. - 60 - ....... .SEV1A POBIIYTable 4,9 . SURVVALPROABIITYREVEALED BY INCOMPLETE SEL2MLS 95 (1) (2) (3) Correeted :_Eotimated Duration Survivors vivere Snrvivl (weeks) Frequency Frequency Probability 1-4 27 27 0.794 5-8 20 20 0.588 9-12 12 12033 13-16 7 ; ~10 0.294 17-20 . 10 9 0.245 21-24 10 7 0.206 25-28 4 6 0.176 29-32 6 5 0.147 33-36 4 4 9.118 37-40 3 3 0.088 41-44 2 2 0.059 45-48 2 2 0059 49-52 41 2 0.059 Notet Colun (1) shows thLe number of male ieorkera idu> Zemaint unemployed out of the echort of workera who be_ame comployed t the ane time. Column (2) corrects the frequency of survivors for each cohort to elmate cuzste.rng errore, Coluan (3) estimates a probability based on the corrected frequency distribution and as starting cohort of 34 woker, DATA SFURCE:.LS 5.20 While these inferences are obtained with incomplete spells of unemployment, the PLS survey also allows the measurement of duration of complete spells through retrospective information provided by workers employed at the time of the survey. This information can be used to estimate the distribution of complete spells. The analysis of complete spells reveals the same survival function. More than half of male unemployed workers find work before the end of the 12th week, and only 11 percent remain unemployed after 36 weeks. The proportion of workers with unemployment spells shorter than one year is around 97 percent of each cohort entering unemployment. Determinants of Duration 5.21 Table C.7 in Appendix C shows estimates from analyses using Cox proportional hazard models. These estimates reveal that duration of unemployment declines with age until 45, increasing thereafter. Schooling has no significant effect on duration, but unemployment spells, are shorter for workers living in rural areas. Married and cohabiting workers also have shorter unemployment spells while more household work is linked to longer spells. - 61 - 5.22 Although most of these effects fit with what economic rationality suggests, the effect of age on duration is anomalous. Older workers should have longer unemployment spells. More labor market experience makes job separation less likely, but once it happens the higher job specificity of older workers human capital should make a new job match more difficult. Although further research will be necessary to provide a satisfactory explanation for this phenomenon it is important to know that young workers not only are more frequently unemployed but they experience longer unemployment spells. These features of unemployment prove that the CTS is not a good substitute for an unemployment benefits scheme. 3. WAGE DISTORTIONS A. Trends in Wages and Salaries Comparison with Other Latin American Countries 5.23 Wages and salaries in Peru plummeted during the 1980's. Table 4.10 shows that this decline was larger in Peru than in any other Latin American country. In real terms, the 1990 wage level in Lima was approximately 44 percent of its level in 1980. However, wages did not decline at a fast rate during the whole period, following instead trends common to other Latin American countries until 1987. In fact, real wages were about the same in 1987 as they had been seven years earlier. This speaks of a very serious economic stagnation, but not of an important decline in the real value of labor. After 1987, however, wages experienced a dramatic sustained decline that has no parallel in any other of the reported countries, even though some of them have faced very difficult economic circumstances. Trends in Earned Income by Type of Employment 5.24 More detailed trends in the value of labor in different sectors can be analyzed by focusing on the nature of labor contracts. Three main categories require special attention: contracts within the public administration, those in the private formal sector, and informal labor relationships. Table 4.11 shows the evolution of minimum wages, wages and salaries in the private sector, earned income in the public administration, and the remuneration of labor in the informal sector. 5.25 The economic adjustment undertaken by the present administration in August 1990 had important effects on the real value of most economic variables including wages and salaries. This makes it specially necessary to make a careful choice of the particular time periods that are to be used to measure trends in real magnitudes. 5.26 Column 4 in Table 4.11 shows the total percent change of variables measured in real terms from 1980 through June 1990, just before the 1990 policy adjustments were implemented. Column 5 shows the effect of the August 1990 economic adjustment on the measured variables. 5.27 These figures show that between 1980 and June 1990, the income earned by Peruvian workers declined dramatically in all areas except the - 62 - informal sector. In the private sector, which represents approximately 90 percent of total employment, salaries and wages decreased by 61 and 64 percent respectively. This means that the majority of workers in the formal sectors of the economy lost approximately 40 percent of their earning capacity. On the other hand, and according to the special index constructed by Banco Central de Reserva, real income in the informal sector increased by more than Table 4,10 TRENDS IN VAGES IN PERU AND IN OTHER LATIN AMERICAN COUNTRIES 1981-90 Anua1a Average Index~ (1980m100) 1981 1983 1985 1987 1988 1989 1990'; ARGENTINA t 89 102 197 100 96 82 8 Rio de Janeiro 109 113 113 102 10 [07 86 See Paule it 105 94 120 143 152 164 140 'COLOMBIA:'11 10 15 11 1 1 2 CHILE ' 109 97 94 95 101 103 105 NEICO 't 104 81 77 73 73 76 PE102 93.7 101 76 42 44 . URUGUAy 107 85 6. 75 76 7 7 Notes; Preliminary s a act e bes" industry wages' deflated using the CPI of Ri6 de Jan.iro. IKanufacturing industry wages in the State of Sao Paulo, deusted ains tqe Sao P 0ulo ot*O living index. :Worker. gages in the manufacturing industry. / Remunerations of non-agricultutd workers.. Manufacturing indust wages. . Workers ages in the private sector of Metropolitan Lima. '-Index of real pverage ages. Source: Instituto Cu&nto, Aiuste v Economia ain 1985 V9i J 199 half during the same period. The legal minimum wage roughly followed the trend of wages and salaries in the formal sector, declining by more than 75 percent in real terms. Minimum wages constituted thus a smaller fraction of wages in the formal sector in June 1990 than they had been in 1980. Real remuneration in public administration declined even more than minimum wages: the average monthly salary in this sector declined by 80 percent. - 63 - 5.28 After the economic adjustment of August 1990, real earned income declined in all sectors. However, the effect of the adjustment has been uneven across sectors. Tighter control of public expenditures forced the real value of public administration remuneration down in dramatic fashion; by . Table 4.11 TRENDS IN EARN~ED INCOME BT TYPE OF EPLOYMENT FES.80 FEB.80 JUJN.90 JUN.90 JUN90. DE..90 AVG .T. JUN DEC .. PERCENT. PERCE.T . ANNUAL 1980 1990 . 1990 C.NGE C. NGE CNG REAL WAGES *' 9943 3852 3152 -61L3% -18.2% -8.3% ('000 I./ per day) .REAL SALA2RIES "i 482 173 162 -64.1% -6.4% -8.9% (m. I./ per month) .INFORMAL SERVICES 't 85 133 100 56.5% -24.8% 4.2% (Index) M.INIMUM WAGE.137 34 25 -75.2% -26.5% -11.9% (in. I. per month) GOY.REMUNERATIONS * 448 91 28 -79.7% -69.2% -14.7% (.I.:. per month) Notest All wag alaries and indices in constant values of December 1990 Sased on the surve of irms :with 10 or more employees conducted in Lima by the M'linistry of Labor.~ Index of real income of informal service workers, by Bato Central: de Reserva. Figures f 1989 and 1990 have been estimated by Cu&nto, S.A. i Based on the survey of General Government employees conducted by the Ministry of Labor. General Gove~rment compises: Central Government, local governments, and public institutions. The "GOV.REMUNERATIONS" figure in Column 1 corresponds to February 1981. December, it had reached a level barely higher than the minimum wage. Within the formal private sector, real wages declined much more than real salaries. Real income for informal service workers, as measured by this index, appears to have declined at a higher rate than wages in the formal sector. This offers a sharp contrast to the 1980-June 1990 period. 5.29 These trends are measured as annual average rates of growth or decline in Column 6 of Table 4.11. As implied by the evidence in Table 4.10 above, the real value of labor fluctuated during the 1980's, with most of the total decline coming during the past three years. - 64 - Figure 4.1: EVOLUTION OF EARNED INCOMES RELATIVE TO PRIVATE WACES 3.2 3 2.8 2.6 - 2.4 - - 2.2 - o .2 1.6 1.4 91. 2 o 0.8 0.6 0.4 0.2 8ie 19132 I 1984 I9IS 1988 Is,- + Minimuas Wag . o Government A Private Salaries X Informal Services 5.30 These new data show that trends have been quite different in the public administration and in the private formal sector. In the latter, real incomes fluctuated around a slowly declining trend until 1988, and fell dramatically during the last two years of the decade. In fact, decline up to 1988 was small enough to be compensated at the peak of the last fluctuation, so real salaries and wages in 1987 were very close to those in 1980. In contrast, public administration remuneration declined rapidly during the first half of the decade, and did not fully recover during the peak of the last fluctuation. The rapid decline in public salaries resumed after 1988. 5.31 The changes in the structure of remuneration induced by these quite different sectoral trends are illustrated in Figure 4.1, which depicts the evolution of minimum wages, and income in the formal private sector, informal services, and public administration as fractions or multiples of wages in the formal private sector. Note that in the formal private sector itself, salaries decreased more slowly than wages before 1985, but faster after 1985. In 1980 and 1990 the ratio of salaries to wages was about two, and fluctuated by up to 20 percent during the period. 5.32 Incomes earned by Government employees and officials evolved quite differently from wages in the private sector. In 1981, remuneration in Government was almost twice as high as private sector wages, and close to the level of private sector salaries. By 1986, both incomes were at similar levels, which implies that incomes in public administration declined much more - 65 - rapidly than that in the private sector. Since 1986, compensations in public administration have fluctuated peaking at more than 40 percent above private sector wages in 1989, but then falling sharply to a level below private wages in August 1990. 5.33 Earned incomes in the informal sector averaged only half the level of wages in the formal sector in 1980. Since then, this ratio has grown continuously, with growth accelerating during the second half of the 1980's. By the end of the decade, earnings in the informal sector were as high as salaries in the formal sector. The minimum wage was about one half of reference wages in 1980, and is again at that level today. It was somewhat lower than one half during most of the decade. 5.34 In sum, these trends reveal important fluctuations in the value of labor during the 1980's, a pattern which deserves careful analysis in terms of both its determinants and its effects on the efficiency of labor markets. The spectacular decline in the value of labor in the formal sector of the economy, and the fact that real earned incomes in informal activities have simultaneously increased, can hardly be read as evidence of efficiency. B. Determinants of the Value of Labor Conceptual Issues: Earninge Generating Functions 5.35 In order to examine the structure of real wages it is necessary to use an analytical framework. Earnings generating functions, as specified by Mincer (1974), are the main analytical tool used by labor economists to measure the effect of human capital on the value of labor, and to assess the impact of distortions derived from a variety of labor market phenomena such as discrimination, barriers to entry into the professions, union behavior, and public regulations. 5.36 Earnings generating functions provide a rigorous method for identifying the contribution of on-the-job training to human capital accumulation, which in turn makes it possible to measure the effect on the value of labor of investments in formal education and training and that of distortions of any source. 5.37 Earnings of workers evolve throughout the life cycle because of human capital production and its depreciation. Through on-the-job training and other forms of learning, human capital keeps increasing at a relatively fast rate during the early stages of professional life. Learning becomes less profitable when workers become older because the cost of learning increases while the years left to use the human capital before retirement decrease. 5.38 The timing of post-school investments in human capital, along with the depreciation of all skills, should generate earnings profiles with the highest slopes at the early stages of professional life. Even if human capital depreciation rates remained constant, the slopes of earnings profiles depicted as a function of experience should decrease with time, reflecting the lower intensity of learning processes for older workers. After a number of - 66 - years of professional experience, depreciation becomes more important than learning new skills, and human capital decreases as well as observed earnings. 5.39 Early specifications of earnings functions (Becker, 1964, and Mincer, 1962) did not explicitly model the effect of post-school investments in human capital. Rates of return to investments in education through formal schooling were estimated in these models by simply adjusting for age. Since more highly educated workers have less experience at a given age, these models produced estimates of returns to schooling that were biased downward. 5.40 In the early stages of the scientific diffusion of Mincer's model, the idea of costly learning by experience was difficult to accept. It seemed to contradict the general presumption that most on-the-job learning was probably free of costs, both socially and individually. For some time it was not clear that even when professional activity entails learning by doing-- which does not involve either payments to teachers or other expenses--this learning is still paid for by workers in the form of foregone earnings. 5.41 Even if learning by doing is physically inseparable from work, and accordingly takes place as a by-product of professional experience, it will not be cost free. Labor markets place a value on human capital acquired by learning through experience. This learning renders workers more productive. Competition drives down the wages of inexperienced workers, for whom learning by doing is intense. Their lower wages mean that this form of learning becomes costly, in terms of their foregone earnings. As a result, learning which in appearance seems to be costless, is in fact acquired at a cost in the market place just like any other good or service. In an economic sense, learning is never cost free although its cost varies across the working population. The costs of on-the-job training are real. Mincer's approach makes them measurable. Relevance for Policy Design 5.42 Improved knowledge of the impact of labor policies through the analysis of earnings generating functions has led to widespread acceptance of the once exceptional point of view that labor supply responses are of great consequence, even when price distortions induced by social policies in general and labor market interventions in particular appear fairly small. 5.43 This awareness of the potential for great damage from even simple policy interventions was absent when most labor market-regulations were introduced in many Latin American countries. Peru is a good example of a country where public policy was inspired by the idea that supply responses were negligible and that regulation of labor markets could be an instrument for attaining desirable social objectives. The formal sector of the economy was expected to support a heavy burden of fiscal and regulatory constraints without real effects on its share of the labor market. Reality has proved that this idea was extremely naive. Employment in the formal sector declined, and Peru ended up with a dramatic drop in the real value of labor. The supply responses to distortions in wages in the formal sector have produced in Peru substantial movements out of the labor force and from the formal to the informal sector. Better knowledge of the nature of these distortions is a requisite for improved policy design. - 67 - C. Empirical Analysis: Main Results Returns to investments in formal education. 5.44 The most important source of economic growth is accumulation of human capital through formal education. The estimates of earnings generating functions presented in Table C.8 in Appendix C lead to one clear and fundamental policy implication: returns to schooling were in 1985 in Peru high enough to make educational policies the single most important component of any labor market strategy. Estimates indicate an average return of 11 percent. This means that one additional year of schooling generates approximately 11 percent higher lifetime earnings. 5.45 There is no evidence in these data of decreasing returns to schooling. In fact, estimated rates of return are 9.3 percent for workers with less than 10 years of schooling, and 12.6 percent for those with 10 years or more. The complementarity between formal education and on-the-job training that Mincer has stressed is also revealed by this analysis. The effect of experience on wage growth is stronger for the more educated workers. The Value of Labor in Rural Areas 5.46 The complementarity of human capital investments with other forms of investment in disembodied capital is a well-known phenomenon, which has far-reaching policy implications. These complementarity effects are the essence of modern development which, in most cases, does not rely on the exploitation of specific natural resources. The comparative advantages in international trade that most countries exploit originate in the speed with which productivity grows through accelerated human capital investments, complementary with other forms of investment. 5.47 One basic issue is whether education is as complementary with physical investments in agriculture as in other activities. If not, measured rates of return to schooling should be lower in rural than in urban areas. At first sight one might expect important derived demand for education from industries only in urban areas. This view was implied in the development strategies adopted by a number of countries in the past. It was thought that returns to education in rural areas were much too low, for such investments to be justified. 5.48 It is indeed an historical fact that general literacy in many countries emerged hand in hand with the growth of developed industry in urban centers. This evidence was taken as proof that education was not profitable in rural areas. Developing countries were advised to use labor-intensive production methods in agriculture which low wages would make profitable. 5.49 This view proved wrong. Policies designed to promote agrarian reforms based on the idea that land was the main source of revenue for farmers failed repeatedly. Modern growth cannot rely on traditional agriculture; the key to understanding and promoting growth in rural areas is the realization that traditional agriculture has to be transformed (Schultz, 1964). In modernized agriculture, as in other sectors of the economy, human capital is - 68 - complementary with investments in physical capital. Moreover, education is an important determinant of the costs of transforming the agricultural sector. More education in rural areas makes this cost much lower, and thus accelerates the process of technical and economic change. 5.50 The analysis of Peruvian data reveals that rural location has a very strong negative effect on earnings. On average, earnings in rural areas are more than 20 percent lower than in urban areas, even after correcting for education and experience. Estimated rates of return to education, however, are similar in the urban and in the rural sample, as shown in Table C.9 in Appendix C. 5.51 The fact that earnings are much higher for the same level of education and experience in urban areas means that there are sizeable expected gains from migration. Investments in education in rural areas generate returns for the worker similar to those obtained in urban areas even if the workers does not migrate. For migrating workers returns to acquired human capital become much larger. Taking into account that migration is more likely for the most educated workers, these estimates indicate that returns to years of schooling in rural areas are, in fact, much higher than in urban areas. This important conclusion has obvious implications for the allocation of public resources and for the design of a sound strategy to favor economic development in Peru. Moreover, policies that would help to educate the rural population are not only the best for fostering economic development, but are also very effective in increasing the wealth of workers living in the poorest areas of the country. 5.52 While this analysis shows that schooling has higher returns in rural areas, it does not measure the total costs of schooling. Obviously, the high returns to education prove that these costs must be fairly high. Thus, the key policy implication of this analysis is that the cost of investment in education to families in rural areas must be reduced through public intervention, until the educational attainment of children born in rural areas matches their natural endowments. Defining the Informal Sector 5.53 For many years, economists have realized that development in most poor countries involves a complex transformation from a traditional rural society to a modern economy where most workers will have to be educated for non-traditional jobs. This will require learning in formal schools first and substantial post-school investments in human capital later. A certain share of the labor force will be able to remain self-employed, but this share will constitute a much lower proportion of the labor force than in many traditional rural societies. Development necessarily requires the expansion of the formal sector of the economy. 5.54 Recent trends in employment described above show that derived demand for labor in the formal sector has been shrinking in Peru. De Soto (1988) estimates that 48 percent of the labor force and 61 percent of working hours are devoted to informal activities contributing 39 percent of GDP. These trends cannot be seen as a sustainable process compatible with economic development. - 69 - 5.55 The fact that only the informal sector has been resilient in periods of economic difficulty is seen by some observers as evidence that a sound strategy for economic recovery in Peru could rely on helping the increasingly numerous small entrepreneurs. Without denying the role of the formal sector in some particular industries, these alternative heterodox development strategies would try to foster the growth of self-employment in the economy, and would distort the allocation of financial resources in doing so. 5.56 This view is not based on a deep understanding of the role of the formal sector in modern economies. To analyze the role of informal labor a rigorous analytical framework is needed. This framework is provided by a well-known theory proposed by Coase (1937) many years ago. The key idea is that the essential function of formal labor contracts is to save on the sizeable transaction costs involved in operating markets. If these costs were always very low, workers would never be employed by entrepreneurs. Most family heads would be entrepreneurs themselves, producing for markets with the help of family members. In other words, every worker would be self-employed. 5.57 This very basic idea, for which Coase was awarded the Nobel prize in economics, provides the necessary perspective from which to analyze informal labor markets and their role. The theory states that the formal organization of labor in enterprises starts where the market ends. There is a frontier between the world of organized labor inside enterprises which operates under a command system, and the world of free markets supplied by self-employed individuals. This frontier moves across industries and through time according to the relative costs of producing things with each of these two alternative economic organizations. Enterprises compete not only among themselves, but also with the market. 5.58 The operation of the comand system that is the essence of enterprises is very imperfect. It involves the centralization of decisions, a process which in general entails great internal inefficiency. However, the reason why enterprises can compete with the market is that operating markets is also very costly. As a result, there is a natural equilibrium in all economies with a certain fraction of all economic decisions taken by coordination through markets (with more or less political interference) and all other decisions taken within centralized command systems that are the enterprises. 5.59 The fundamental difference between free self-employed producers and workers subject to the enterprise command system did not go unnoticed by social scientists. Economists from Adam Smith to Marx stressed the alienation of human nature involved in work organized through command systems. They understood the fundamental difference between old societies where coordination occurred mainly through markets supplied mostly by self-employed individuals, and modern industrial societies where most individuals work within rigid command systems under the authority of entrepreneurs. The earlier economists did not notice however that enterprises compete with markets because they save on transaction costs. 5.60 The practical value of this theory is commensurate with the time it took economists to use it in their analyses. Understanding that enterprises can be replaced by the market only at a cost makes the task of promoting the informal sector where it has no comparative advantage less attractive than it - 70 - is sometimes thought to be. Distortion of the natural share of the formal sector is costly. When, through subsidies or other artificial methods, the cost of formal labor contracts is reduced, there is too much command system, and society incurs losses by forgoing the efficiency of markets. If, on the other hand, the use of organized labor in enterprises becomes artificially costly through heavy taxation, harmful regulations, and misguided public interventions as has been the case in Peru, the share of the informal sector increases, with equally important losses to society in terms of overly high transaction costs. 5.61 The natural equilibrium of an undistorted economy produces a certain share of employment in the formal sector in each industry. Some industries are naturally organized in the form of fairly big formal enterprises--as in the case of most utilities--while in other cases self- employment predominates, as in certain sectors of modern agriculture. This equilibrium must be maintained in order to foster economic development. Clearly, there are high costs in the artificial impoverishment of self- employed individuals and families through policies that rely on forced industrialization via a heavily subsidized formal sector. Equally, there are significant costs to increasing the fiscal burden and the cost of organizing production through formal labor contracts to a point where many activities shift from the formal to-the informal sector. There are sizeable costs of operating informal markets. 5.62 This theory explains why labor contracts in the formal sector need to be fairly rigid in order to save on transaction costs by inducing easy to monitor behavior and predictable outcomes. Wage contracts which simply monitor performance through one input, time, are the most prevalent form of compensation in the formal sector. Labor contracts cannot be flexible enough to fit the complexities of production processes. Entrepreneurs obviously try to monitor the performance of their employees, but they cannot carry this task to the extreme of disintegrating the firm into teams of self-employed producers exchanging their products. This is precisely what the informal sector does by the direct exchange of products and services in market transactions. The essence of enterprises is to eliminate these costly transactions. Integration within enterprises, both vertical and horizontal, is aimed precisely at saving on transaction costs, replacing transactions in products and services with fairly rigid explicit and implicit labor contracts. 5.63 These fundamental features of the formal sector of the economy explain why labor contracts in the command system that is the firm have to be fairly rigid. The proposition that efficiency of a well organized capitalist system requires flexibility in labor contracts is obviously at odds with the very nature of the capitalistic firm. The theory of the firm due to Coase relies on the recognition that there are two alternative ways of organizing production; through firms and through markets. Both are efficient but not simultaneously. It is simply not possible to have at the same time the freedom of markets and the reduction in transaction costs obtained in firms. Earnings in the formal and informal sectors 5.64 Using these basic concepts it is possible to delineate the formal sector by the nature of labor contracts. All self-employment is considered informal. In addition, workers who are not self-employed will be considered informal if: a) the worker has no written contract that specifies the terms of - 71 - employment; and b) the worker does not receive benefits from the Peruvian Social Security Scheme (Instituto Peruano de la Seguridad Social-IPSS). Signing a contract or participating in the IPSS are indicative of a permanent contract, and hence put the worker in an inflexible regime. 5.65 Estimates in Table C.10 in Appendix C, obtained with data for 1985, reveal that in urban areas self-employed individuals earn more than anybody else, but informal employees earn less than formal workers. In rural areas, however, formal workers earn more than the self-employed. In Lima, earnings of formal workers are 12 percent above those of informal employees. The earnings differential becomes almost twice as big in the sample for all urban employees. 5.66 Between 1985 and 1990, the relative wages of the formal sector experienced a drastic reduction. Estimates in Table C.11 in Appendix C indicate that, duly corrected for human capital endowments, the earnings of the self-employed were 13 percent higher than those of employees ( both formal and informal) in 1985. This earnings differential was already 32 percent in 1990.11 5.67 A very important finding of the analysis is the insignificant returns to investments in education obtained in the earnings generating functions in the informal sector. This reveals that the high complementary of investment in human capital with physical investments is lost in the informal sector. A strategy of development which would rely on an increasing role of the informal sector would be doomed to fail. The recent trend in Peruvian labor markets of increasing earning differentials against formal sector employment is very detrimental to long term growth because it distorts the wage structure discouraging investments in human capital. Male-female waze aaps 5.68 Estimates in Table C.12 in Appendix C reveal that there is no gender wage gap in Peru among employees covered by the Social Security. In contrast, in 1985 among employees without Social Security men earned wages 33 percent higher than women with the same education and measured experience. However, for many women, the measurement of experience which is simply time since the worker finished school, overestimates their true professional experience. This measurement error produces an upward biased estimate of the gender gap. The gender wage gap is also very high in 1985 among the self- employed. 5.69 The absence of gender wage gap among workers covered by the Social Security is very revealing, because it is in the formal sector of the economy where regulations to correct potential gender discrimination in earnings would be an issue. Since unexplained gender related differences in earnings exist only in the informal sector, there are no grounds for undertaking any measures to correct distortions in the formal sector. / Analysis of wage trends with earnings generating functions can only be carried out for Lima because the PLS 1990 survey does not cover other geographical areas. - 72 - 5.70 None of the estimations for 1990 produced statistically significant gender effects in earnings functions. The trends in the gender wage gap revealed by these estimates fit with trends in labor force participation and occupational structure examined earlier in this chapter. Between 1985 and 1990 women withdrew from the labor market and more than proportionately from the informal sector. The higher earning capacity of women who kept their jobs explains the reduction in the gender wage gap. Effect of Unions and Public Employment on Earnings Functions 5.71 The free mobility of workers across sectors and occupations is compatible with differences in wages, as long as the differences in return correspond to the costs of mobility and that of investments in the skills that the jobs require. Earnings generating functions should provide evidence on any departures from the natural equilibrium that free entry would provide. 5.72 Two factors that frequently lead to higher earnings than those predicted by real investments in human capital are the presence of unions and public employment. Unions may be able to restrict entry into certain jobs, or impose special working conditions and compensation schemes which would produce higher than competitive rates of return to skills. Employees in the public sector may also, protected from competition, obtain earnings above those explained by their human capital. Often, the unionization and public employment effects interact, in the sense that unions tend to be more frequent and stronger in public enterprises and among public employees. As a result, the effect of unionization may be different in the public and in the private sectors. 5.73 Table C.13 in Appendix C shows the results of estimating earnings generating functions with variables that identify the effects of unions, and that of public employment in both the administration and in public enterprises in 1985. These estimates indicate that there is a great difference between earnings in the public administration and those in public enterprises although both are higher than earnings in the private sector. The effect of unions on earnings is not statistically significant in the private sector. In contrast workers in public enterprises where unions tend to be strongest earned 30 percent more than workers with the same human capital in the private sector. Comparable workers also earned 13 percent more in the public administration than in the private sector at the time. 5.74 Similar estimates for 1990 would have shown trends in these earnings differentials. Unfortunately, the PLS data for 1990 do not allow the identification of workers employed by public enterprises. Policy recommendations aimed at correcting these wage distortions have been made in the first chapter of this report on the assumption that the distortions measured in 1985 have not been corrected yet. This is most likely given the rapid decline of earnings in the private formal sector during the last three years, but of course estimation of the evolution of the earnings differential of workers in public enterprises should be a high priority in subsequent analysis of labor market issues. REFERENCES AnAlisis Laboral, Lima, PerG. (various issues) Becker, G.S. (1964). Human Capital. Columbia University Press, New York. Canales-Kriljenco, J. (1991). "Social Security in Perfi: Current Situation and Reform". Lima, Peri. Carrillo Calle, M. (1991). Apuntes Criticos sobre el Nuevo R6gimen de Participaci6n de los Trabaladores en las Utilidades, Gesti6n y Propiedad de las Empresas. ADEC-ATC. Lima, Peri. Coase, R. (1937). "The Nature of the Firm". Econ6mica, Num. 4. Cox-Edwards, A., and P. Griffin (1992). "Wage Flexibility and Labor Market Adjustment in Brazil". Draft Internal Report, LA1CO. Dabos, M. and G. Psacharopoulos (1991). "An Analysis of the Sources of Earnings Variation among Brazilian Males." Economics of Education Review, Vol. 10, No. 4, 1991: 359-77. de Soto, H. (1986). El Otro Sendero: La Revoluci6n Informal. Lima, Peri. Gertler, P., and P. Glewwe (1988). "The Willingness to Pay for Education in Developing Countries: Evidence from Rural Peru". LSMS Working Paper No. 54, The World Bank, Washington, D.C. Gill, I. A. (1991). "Is there Sex Discrimination in Perd? Evidence from the 1990 Lima Living Standards Surveys" in Women's Employment and Pay in Latin America. Part II Country Case Studies. G. Psacharopoulos and Z. Tzannatos (eds.). Latin America and the Caribbean Technical Department, Regional Studies Program, Report No. 10, The World Bank, Washington, D.C. Glewe, P. (1987). "The Distribution of Welfare in Peru in 1985-86". LSMS Working Paper No. 42, The World Bank, Washington, D.C. Glewwe, P., and G. Hall (1992). "Poverty and Inequality during Unorthodox Adjustment: the Case of Peri, 1985-90". LSMS Working Paper No. 42, The World Bank, Washington, D.C. Iglesias, A., R. Acufa, and C. Chamorro (1991). Diez Afos de Historia del Sistema de AFP. Antecedentes Estadisticos 1981-90 . Ed. AFP Habitat, Santiago, Chile. Instituto Cuanto, Aluste y Economia Familiar, 1985-1990, June 1991 Instituto Nacional de Estadistica (1988). Encuesta Nacional de HoRares sobre Medici6n de Niveles de Vida, ENNIV (1985-86): An&lisis de Resultados. Direcci6n General de Censos y Encuestas. Lima, Peri. Instituto Nacional de Estadistica (1991). Peri: Compendio Estadistico 1990- 91. Direcci6n T6cnica de Indicadores Econ6micos. Lima, Peri. - 74 - Jacoby, H.G. (1990). "Shadow Wages and Peasant Family Labor Supply: An Econometric Application to the Peruvian Sierra". LSMS Working Paper No. 73, The World Bank, Washington, D.C. Jim6nez, E., and B. Kugler (1987). "The Earning Impact of Training Duration in a Developing Country". Journal of Human Resources, Fall 1987. Khandker, S. R. (1991). "Women's Labor Market Participation and Male-Female Wage Differences in Pero" in Women's Employment and Pay in Latin America. Part II Country Case Studies. G. Psacharopoulos and Z. Tzannatos (eds.). Latin America and the Caribbean Technical Department, Regional Studies Program, Report No. 10, The World Bank, Washington, D.C. King, E.M. (1990). "Does Education Pay in the Labor Market?: the Labor Force Participation, Occupation, and Earnings of Peruvian Women." LSMS Working Paper No. 67, The World Bank, Washington, D.C. Kuznets, S. (1971). Economic growth of nations: Total output and production structure. Cambridge: Harvard University Press. Lago, R. (1991). "The Illusion of Pursuing Redistribution Through Macropolicy: PerG's Heterodox Experience, 1985-90", in The Macroeconomics qf Populism in Latin America. Dornbusch, Rudiger and Edwards (eds.). University of Chicago Press, November 1991. Lexislaci6n de Seguridad Social (1991). Ed. El Carmen, Lima, Pero. Mincer, J. (1962). "Labor Force Participation of Married Women" in Aspects of Labor Economics. H. Gregg Lewis ed. Princeton University Press. Princeton, N.J. Mincer, J. (1974). Schooling Experience and Earnings. Columbia University Press, New York. Mincer, J., and L. Leighton (1982). "Labor Turnover and Youth Unemployment", in The Youth Labor Market Problem. R. Freeman, and D. Wise. University of Chicago Press and NBER, 1982. Mincer, J., (1988) "Education and Unemployment". Report Submitted to the National Center for Education and Employment. Teachers College, Columbia University, New York. Moock, P., P. Musgrove, and M. Stelcner (1990). "Education and Earnings in Pero's Informal Nonfarm Family Enterprises". LSMS Working Paper No. 64, The World Bank, Washington, D.C. Murguia, A., and N. Alcab6s (1988). Sistematizaci6n de la Legislaci6n Laboral Peruana. Centro Interamericano de Administraci6n del Trabajo, Lima, Pero. Newman, J.L. (1988). "Labor Market Activity in Cote d'Ivoire and Pero". LSMS Working Paper No. 36, The World Bank, Washington, D.C. Nueva Ley de Compensaci6n por Tiempo de Servicios (1991). Ed. Olimpico, Lima, Pero. - 75 - Paredes, C., and J. Sacks (1991). Perfis Path to Recovery: A Plan for Economic Stabilization and Growth. Brookings Institute, Washington, D.C. Psacharopoulos, G., and Z. Tzannatos (1991). "Women's Employment and Pay in Latin America." Regional Studies Report No. 2, The World Bank, Washington, D.C. Peacharopoulos, G., and Z. Tzannatos (1992). "Latin American Women's Earnings and Participation in the Labor Force." Policy Research Working Paper.No. 856, The World Bank, Washington, D.C. Rueda, M. and C. Contreras (1989). Report presented to the "Reuni6n Internacional sobre Prestaciones Econ6micas y Pensiones." Lima, PerGi. Shultz, T.W. (1981). Investing in People. University of California Press. (1964). Transforming Traditional Agriculture. Univ. of Chicago Press. Sistema Nacional de Pensiones: Jubilaci6n. Ed. El Carmen, Lima, Perfi. Stelener, M., J. van der Gaag, and W. Vijverberg (1987). "Public-Private Sector Wage Differentials in PerGi, 1985-86." LSMS Working Paper No. 41, The World Bank, Washington, D.C. Stelcner, M., A. Arriagada, and P. Moock (1988). "Wage Determinants and School Attainment among Men in Perfi." LSMS Working Paper No. 38, The World Bank, Washington, D.C. SuArez-Berenguela, R. M. (1987). "Perfi Informal Sector, Labor Markets, and Returns to Education". LSMS Working Paper No. 32, The World Bank, Washington, D.C. TASC #30. 1992. Modificaciones a la Actual Legislaci6n del Sistema de ISAPRE. Ilades, Santiago, Chile. Tuma, Nancy B. 1982. "Effects of Labor Market Structure on Job Shift Patterns," in Longitudinal Analysis of Labor Market Data, Heckman and Singer, eds.,'Cambridge University Press, pages 327-363. van der Gaag, J., and W. Vijverberg (1988). "A Switching Regression Model for Wage Determinants in the Public and Private Sectors of a Developing Country." The Review of Economics and Statistics, Vol. LXX, no. 2, May 1988. pp. 244- 252. van der Gaag, J., M. Stelcner and W. Vijverberg (1988). "Public-Private Wage Comparisons and Moonlighting in Developing Countries: Evidence from Cote d'Ivoire and Perfi." LSMS Working Paper No. 52, The World Bank, Washington, D.C. Webb, R., and G. Fern&ndez Baca (1990). PerG en NOmeros 1990. Cuinto, S.A. Lima, Perfi. APPENDIX A Page 1 of 5 TECHNICAL EDUCATION IN PERU Backxround Technical education has been neglected for many years in Peru. Only a limited number of public secondary schools have technical education programs, and most of them lack equipment and supplies. They are seriously understaffed because many qualified technicians have left for private business as a result of low salaries. In Peru, university education also tends to be more highly regarded than technical education. In 1990, about 504,700 students were enrolled at universities, compared with about 173,000 at technical schools.' However, the available data suggest that (post-secondary) technical education has been effective in .reducing vulnerability to severe welfare loss during the period of economic crisis. According to the PLS in 1985 and 1990, those least vulnerable to declines in consumption levels were those with secondary technical or post-secondary (non-university) training. Those with technical training also began the period with higher consumption levels than average. Moreover, considering the length of school years (three years at post- secondary technical schools compared to seven-eight years which most students actually spend at universities) and the fee level ($100-200 per month compared to $270 per month), investment in technical education appears to have greater returns than that to university education.2 In Peru, there are at least two good-quality post-secondary technical schools in the private sector, the Servicio Nacional de Adiestramiento en Trabajo Industrial (SENATI) and the Instituto Superior Tecnol6gico (TECSUP). They have close ties with private enterprises, particularly in the industrial sectors, specially with small manufacturing companies with good productivity and export potential. Strong partnership with the private sector .ensures that training is given which corresponds to the occupational reality in the productive sector. This section summarizes the activities of SENATI and TECSUP. Servicio Nacional de Adiestramiento en Trabalo Industrial (SENATI) General SENATI was established in December 1961 at the initiative of the business community. Its objectives are to develop human resources in industry, promote technology development, and improve the productivity of private enterprises in industry. SENATI undertakes three types of activities: training, advisory services, and technical services. Major emphasis has been Source: The Ministry of Education, Oficina de Estadistica e InformAtica, quoted from "Perfi en NOmeros 1991" by Cuinto, S.A. 1 Based on interviews with SENATI and TECSUP staff. Costs for private universities and private institutions for technical training. APPENDIX A Page 2 of 5 - 78 - placed on training. Its training courses cover: metal mechanic, graphic arts, carpentry, electronics, automobile mechanics, textiles, and shoemaking. SENATI has a headquarters in Lima and 18 local centers in the Northern, mid- North, Central, Southern, mid-South, and Southeastern parts of Peru. Typical courses consist of basic theoretical training (at the SENATI centers) and practical training (at private enterprises). At present, about 1,700 enterprises accept SENATI trainees for practical training. There are about 750 workers at SENATI, of which 450 are instructors and 300 are administrators. Description of Training Courses: Roughly speaking, there are two types of training courses: (i) youth employment program; and (ii) professional program for adults. Youth Employment Program (Programa de Aprendizale): This program provides youth aged 14-20 years with professional skills required for the industrial sector through two-three years of full-time training. The program starts with theoretical training at the SENATI centers, followed by practical training at both private enterprises (four days per week) and the SENATI centers (one day per week). During the period of practical training, trainees receive 50 percent of the minimum wage, social security, and personal expenditures. At present, about 4,500- 5,000 youth are enrolled at this program. In addition, SENATI runs a short training program for youth aged 16-22 (Programa de Formaci6n Bisica Industrial). This aims at training youth for basic skills in one semester on a full time basis (870 hours), both at the SENATI centers and in private enterprises. Those who have graduated from this basic training program could participate in the more complete Programa de Aprendizaje in the future to develop professional skills. Professional Qualification Program (Proxrama de Calificaci6n Profesional): This program provides adult workers with professional skills through theoretical and practical training. Training is provided on a part-time basis (three hours per day, either in the afternoon or at night). Training consists of various stages based on "SENATI modules," so that individual trainees could start from different stages according to the levels of their skills and knowledge. As a result, the period of training differs by trainee. Seventy eight percent of the trainees are workers in the private enterprises, which cover their training costs. About 8 percent are independent, and they are required to pay 15-20 percent of training costs. In addition, SENATI has a training program for highly specialized skills (Programa de Actualizaci6n y Especializaci6n) on a part-time basis. Costs of the Youth Employment Program: The Youth Employment Program costs about $100 per trainee (monthly), or about $1,000-$1,200 per year. This covers the costs of workers' salaries APPENDIX A Page 3 of 5 - 79 - during the period of practical training (at 50 percent of the minimum wage), social security, and the use of SENATI training facilities. No fees are charged to youths participating in the Programa de Aprendizaje. Selection of the Participants in the Youth Employment Program: Participants in the program are selected competitively. The minimum qualifications include: an appropriate educational background, and guarantees that a sponsoring company will provide the applicant with practical training. (Although completion of secondary education is not a formal requirement, applicants are expected to have an equivalent .level of knowledge.) Once selected, trainees receive one month of pre-apprenticeship training. Only those whose performance is judged reasonable are allowed to participate in the next steps: (i) a basic skill formation course at SENATI (three-five months of theoretical training); and (ii) a specific skill-formation course at both private enterprises and SENATI (18-30 months of practical training). Those who successfully complete the full training cycle are awarded SENATI certificates. Only about 20 percent of the initial participants of the program obtain certificates.3 During the period of practical training (ii), trainees receive salaries (50 percent of the minimum wage) and social security, and are allowed to use training facilities. Relations with the Private Sector: The SENATI training program is based on a strong partnership with the private sector. The Ley de Organizaci6n y Funciones del SENATI (issued as part of Decreto Legislativo 175) requires that private enterprises engaged in "manufacturing activities" pay a levy equivalent to 1.5 percent of wage costs to SENATI.' These private contributions finance SENATI's operations and the cost of installing machinery. Also, private enterprises contributing to SENATI are required to sponsor one trainee per 25 workers, if no enterprises agree to sponsor the trainee voluntarily. At present, about 8,000 enterprises contribute financially to SENATI activities. Follow-Up of SENATI Graduates: A follow-up survey conducted in 1988 for 951 SENATI graduates (representing 12.4 of the total number of graduates) of the Youth Employment Program in Lima during the period of 1969-1986 indicates the following: I According to SENATI data of July 30, 1991, out of the 8,429 persons selected for the pre-apprenticeship training, 7,507 persons receive theoretical training, and 4,467 receive practical training. Only 1,840 persons are awarded certificates. " However, small industries with less than 15 workers are exempted from this requirement. APPENDIX A Page 4 of 5 - 80 - Educational Background: 79.2 percent of the trainees were enrolled at secondary schools, with 35.1 percent having completed secondary education. Most trainees did not have previous exposure to actual technical training, and only 13.6 percent of trainees enrolled at technical schools. Only 1.1 percent received only primary education (including incomplete). This underscores the importance of receiving some measure of secondary education. Job Opportunities: 85.2 percent of the graduates are currently employed, with 3.8 percent unemployed, 83.4 percent have found jobs in manufacturing industry (26.4 percent of the graduates now work at their sponsoring companies), 77.1 percent of graduates consider that their jobs match their specialized training at SENATI. Most work as supervisors or operators of production facilities. It should be also noted that 64.1 percent of the graduates were not the first SENATI employees in the enterprise. This implies that the private sector regards SENATI training as some guarantee of the quality of the workers. TECSUP (Instituto Superior Tecnol6gico) General TECSUP was established in 1984 by a Peruvian business man. It is the only private Instituto Superior Tecnol6gico in Peru. TECSUP's objectives are to develop middle-level technicians whose professional skills meet the needs of private enterprises in industry. Its training courses cover: computers, metal mechanic, electronics, automobile mechanics, chemical and physics. TECSUP has a major center in Lima and is planning to expand its facility in Arequipa. All the courses are offered at TECSUP, and the ratio of theoretical to practical training is about 2:3. There are 55 full-time instructors, and 50 part-time instructors. Description of Training Courses: Like SENATI, TECSUP offers courses for both youths and adults workers. Regular Program (Programa de Formaci6n Regular): This program provides youths age 22 or less with three years of intensive technical training (six semesters, on a full-time basis) at TECSUP's modern workshops and laboratories. After successfully completing the program, the trainees are awarded "technical professional certificates" in industrial specialization, as authorized by the Ministry of Education. At present, about 750 youths are enrolled at this program, with 570 at day-time class and 180 at the recently-created night courses. The trainees are charged fees ($170 per month on average) to cover part of the operational expenses, i.e., textbooks and materials. One of the important characteristics of TECSUP training is the provision of Fondo de Cr6dito Educativo -- education loans operating as revolving fund. The fund was created in 1985, with IDB support ($450,000). Currently, APPENDIX A Page 5 of 5 - 81 - about 40 percent of the trainees receive educational loans, which cover either 50-100 percent of the fees, according to financial need. Terms of the loans include a six-month grace period on graduation from TECSUP and six years of amortization. The monthly payment is set not to exceed 17 percent of the average annual family income. To qualify for the loans, the trainees must demonstrate good performance by every semester. Continued Training Program (Programa de Capacitaci6n Continua): This program aims to upgrade the professional skills of adult workers. The course length ranges from 40 to 60 hours (four-six weeks per course). Training is provided at both TECSUP and private enterprises. At present, about 300 trainees are enrolled at this program. There is no examination for selecting the trainees, but they must be employed by private enterprises because the fees (about $200 per month) are paid by the private enterprises. Costs of the Youth Employment Program: On average, the Regular Course (for youth) costs $170 per month ranging between $100 and $230. Fees are set to cover operational expenses, including textbooks and materials. Transport subsidies are also provided to the trainees. Selection of the Participants in the Youth Employment Program: Applicants must be aged below 23, have completed secondary education, and pass the TECSUP general examination. The examination covers mathematics, history, chemistry, physics, and geography. Selection is very competitive. Every year about 2,000 persons apply for 230 slots. Here, good secondary education is one of the most important requirements. Relations with the Private Sector: TECSUP receives major financing from fees and private contributions. Of the private contributions, Peruvian private enterprises account for 60 percent and external donors for the remaining 40 percent (USAID, GTZ, CIDA, and IDB). External donors typically finance the cost of advisers and the initial installation of machinery. Follow-Up of TECSUP Graduates: Although no comprehensive survey has been undertaken, most of the graduates find jobs closely related to their training. About 90 percent of the graduates are currently employed. APPENDIX B Page 1 of 6 EXPERIENCE WITH THE PRIVATE ADMINISTRATION OF PENSION PLANS IN CHILE, AND ITS RELEVANCE FOR THE PERUVIAN CASE A system of pensions based on the individual capitalization of funds started in Chile for the first time in May of 1981. The new system determines the minimum contribution, and makes the benefit a function of the individual's contributions and the returns to them over the working lifetime. Legislation introduced in 1980 made contributions obligatory for dependent workers who joined the labor force after December 31, 1982, and made it voluntary for independent workers and those who had already contributed to the traditional Pension Funds. The reform responded to a need that had been recognized before, when the pay-as-you-go system faced serious financial difficulties. Several options for reform were considered. These included; administrative reform; change toward a system based on benefits determined by the accumulation of reserves; and change toward a system based on the accumulation of contributions. The basic argument behind the choice of the third alternative was the notion that whenever the benefits are defined independently of contributions, the system becomes too vulnerable to political pressures. In the reformed system, the state plays a fundamental role in regulating and monitoring the system, and in guaranteeing "solidarity in the base" through a minimum payment. Workers have the right to a minimum pension, even if their lifetime contributions to the system result in a smaller benefit. Besides the individual capitalization of funds, there are six other important aspects of the Chilean system: 1. The new system is a substitute for and not a complement of the old distribution system. Since there are important economies of scale in the administration of pension plans, it was decided that all dependent workers (those not self-employed) who joined the labor force after December 31, 1982 would contribute to a Pension Administration Fund (AFP). Independent workers (the self-employed) and those with contributions to the old Pension Funds were not obliged to contribute to the AFPs. On the other hand, the AFP cannot reject any worker's application to join if it follows approved legal procedures. 2. The benefits are limited to pensions for retirement, incapacity and survival. This limited role helps to ensure the efficiency of the system; limits the opportunity for political or sectoral pressures for additional programs that could be discriminatory; and helps to focus public sector spending on the groups that need it the most, especially through the minimum pension program. APPENDIX B Page 2 of 6 - 84 - Eligibility for retirement is established by a minimum age (65 for men, 60 for women) or earlier if the accumulated funds are sufficient relative to taxable income. Those eligible for retirement, incapacity, or survival benefits may dispose of the balance in the individual account, using one of the following methods to withdraw the funds: a) Full transfer of the balance to a life insurance company, purchasing a fixed monthly income and a pension plan in favor of the beneficiaries. This option is only available if the monthly income is above the established minimum pension. b) Programed retirement, which maintains the remaining balance after monthly withdrawals. Every year, the monthly withdrawals are recalculated on the basis of the individual's life expectancy. Any remaining funds are transferable to the legal beneficiaries. 3. The rules are the same for all workers. The benefits are independent of activity or economic sector of employment. They are solely a function of contributions and returns. The basic contribution is equal to 10 percent of disposable income. 4. The system is competitive, and workers are free to choose the administration company and to move as often as they want. The choice is individual, not by enterprise or other collective form. The AFP regulations of APPs ensure free prices and entry into the industry. 5. Separation of Pension Funds and AFPs. The responsibilities and funding services of the AFPs and Pension Funds are separated. The AFPs do not guarantee a return on the funds invested, and may only vary their cost of administration which is charged to the individual depositor in the form of a monthly (or annual) fee and an incorporation fee. The government is responsible for regulating investment decisions. With this objective in mind, the law established precise norms to secure the diversification and transparency of the resources invested." The law also established mechanisms to ensure that the returns for an individual AFP are close to those of the system as a whole. In case of financial failure, the AFP is liquidated and the state guarantees a minimum return (as a function of the system's return) to the AFP's members. 6. Insurance to Finance Incapacity and Survival Benefits (widows and orphans). Since these events can occur before individuals have accumulated enough funds relative to their taxable income, insurance has been found appropriate. Benefits for incapacity or survival are financed by a complementary insurance with financial reserves. 1/For a detailed description of these norms see, Iglesias et.al. (1991). APPENDIX B Page 3 of 6 - 85 - Individuals began signing into the new system in May of 1981, and there were a large number of transfers from the old system. Between 1981 and 1983, about 1.5 million individuals transferred their funds, a figure which had reached about 1.85 million by 1990. The new system has increased coverage, with the fraction of workers covered growing from 58 percent in 1985 to 67 percent in 1990. The fraction of dependent workers covered has increased from 79 percent in 1985 to 92 percent in 1990. Independent workers have a low participation rate. The AFPs were authorized to charge a fee to finance their administration. The most important restriction was that, with a few exceptions, fees had to be the same for all affiliates in a given AFP. The Chilean experience has shown that after fees increased between 1981 and 1983, competition has resulted in a steady decline in the cost to individuals. In 1990, the cost for an "average contributor" was 33% lower in real terms than in December 1983. Commissions have fallen from 5 percent of taxable income in 1985 to about 3.2 percent in 1990. Strict norms regulate the investment of Pension Funds. These resources can only be invested in certain instruments, and there are clear limits to the distribution of investments by type of instrument. The dynamism of the Chilean capital market during the last ten years has forced constant revisions to these norms. Currently, pension funds are the largest institutional investors in Chile's capital market, representing a 26.5 percent of GDP in 1990 (compared to a 0.9 percent in 1981). Between 1981 and 1990 Chilean Pension Funds had an average real return to investment of 13 percent. The real return of individual accounts (after subtracting fees) followed the same trends, varying between 10.4 and 9.2 percent depending on the level of taxable income. Lower rates of return, of the order of 5 percent, result in pensions equivalent to 76 percent of taxable income (with variations by individual cases). These impressive results partly reflect the rapid development of a previously underdeveloped financial market. The energy and effort required to establish a competitive system like that described above may not pay off if the traditional system of distribution retains a significant fraction of individual's contribution. It is not clear why the newly-planned Peruvian system for administering pension funds needs to transfer a fraction of contributions to the IPSS. If the objective is to secure a minimum pension, the Chilean system manages to offer it without collecting a mandatory fraction directly from workers. This point raises questions about the role foreseen for the IPSS within the newly reformed pension system. If the IPSS is to become an AFP, then it is best to start fresh with a separate institution. In fact, any operations of the IPSS within the new system would have to be kept in totally separate accounting systems. The IPSS, along with all the other traditional "funds," must continue their operations as long as they have beneficiaries to serve. Their contributions will decline as new entrants to the labor force are APPENDIX B Page 4 of 6 - 86 - covered by the new law and as some older contributors choose to transfer their pension plans to the new system. In short, the obligations already established will have to be honored independently of the operation of the new system. 6. The Experience with Private Administration of Health Care in Chile, and its Relevance for the Peruvian Case After a series of reforms in the early 1980s, the Chilean system of health care insurance has become one of choice. Dependent workers are required to contribute 7 percent of their taxable incomes to an insurance program. They can choose between the public system, which is administered by the National Health Fund (FONASA), and a private system, made up by a set of Private Health Institutes (ISAPRES). Those that do not contribute, have access to the public system. Around 80 percent of the population is affiliated with FONASA, 15 percent which the ISAPRES, and the remaining 5 percent with other systems including those of the armed forces. Income per capita is the most important determinant of the choice between FONASA and the ISAPRES. The income level of ISAPRE's affiliates is estimated at four times that of the FONASA's. In fact, the ISAPRES receive about 55 percent of total contributions to the health care system. In 1979, the Ministry of Health was drastically reformed under new legislation. One crucial aspect of this reform was the separation of the three functions of this Ministry. The provision of health care is in the hands of the "Sistema Nacional de Servicios de Salud," which combines 27 autonomous entities which offer health care in different regions and respond directly to the Ministry, which retains the normative function. This organizational structure permits more decentralized decisions and increased executive capacity at local levels. The financial function was put in the hands of FONASA in 1980. FONASA is in charge of allocating the financial resources to the decentralized health care institutes. Until 1978, funds were allocated using a simple criteria, "past expenditure." Starting that year, a series of mechanisms have been introduced that attempt to generate incentives which would allow a more efficient allocation of very scarce resources. The public system offers services to the population through two distinct plans. The "free election" mode allows individuals to get attention in private institutions as long as they are registered in the system and are willing to pay a pre-specified fee. Individual "co-payments" correspond to at least 50 percent of the cost. Under the "institutional" mode, services are rendered in public institutions, and individuals make a co-payment equivalent to a relatively small fraction of the cost. In this mode, fees are determined by income level. APPENDIX B Page 5 of 6 - 87 - Between 1981 and 1988, the Ministry of Health decentralized the administration of Primary Health Care Institutions. These are now the responsibility of the municipalities, or have in turn been transferred to private non-profit organizations. According to current legislation, the ISAPRES are insurance companies, and are not required to offer health services. Nevertheless, these institutions have tended to construct their own infrastructure and to offer services at a lower cost. They sign contracts with their beneficiaries, whereby a plan of services and a system of payments and co-payments is chosen. The member contributes 7 percent, plus the additional payment required to cover the value of the contract. The ISAPRES are not allowed to end these contracts unilaterally, unless there is a breach of contract by the member, who can in turn only end the contract after a year. The Chilean system has been called regressive, and the ISAPRES have been accused of discriminating against individuals who show a clear propensity to use the health care services. Attending to these and other problems, a new law is under discussion. One frequently mentioned problem relates to the fact that health insurance programs are difficult to comprehend and compare. Some regulation of the terminology used and its interpretation seems to be needed. A second problem relates to the fact that all individuals, independent of age and other determinants of their health care needs, are required to contribute the same 7 percent of their taxable income, as a minimum, to a health care program. Yet individuals with chronic diseases, who were not previously insured, are not in a position to buy "insurance" but rather they buy "care." Institutions will increase premiums in these cases to cover the costs. The law contemplates the case in which an insured individual becomes chronically ill. The ISAPRE cannot increase the annual premiums in relation to the health status of the individual. Nevertheless, there are no established penalties for violating this norm. In the case of the elderly, who have health care costs four to five times higher than the middle-aged, the ISAPRES either do not accept them or charge very high premiums. Thus, this group has generally chosen to stay with the public system or go back to it after reaching a certain age. This represents a growing financial problem for the public system. The new proposal contemplates the creation of a savings account destined for retirement. These savings would be administered by the APPs, and would correspond to 0.9 percent of taxable income. The remaining 6.1 percent would go to fund the health care plan. The proposal also establishes the practice of returning to members any difference between the cost of the plan and their contributions. The ISAPRE would improve the benefits, put the funds in the saving account, or contribute to the co-payments. The new law would also add precision to many of the concepts associated with health care services. For example, it defines allowed exclusions and the minimum services that institutions are obliged to offer, among other elements. APPENDIX B Page 6 of 6 - 88 - Although it is beyond the scope of this paper to present a complete evaluation of the Chilean reforms, we can draw some conclusions that are more specific to the functioning of the labor market. As a result of the reforms, Chileans as individuals have taken a larger degree of responsibility for their health. Their contribution increased from 4 percent to 7 percent of taxable income. The private system of (the ISAPRES) receives no government subsidies. Users finance a larger fraction of their health care needs. This is evident from the reduction in the public sector contribution to the health care system from 56 percent in 1980 to 39 percent in 1990. This was accomplished with an increase of 3 percentage points in the individual contribution,. and with the use of co-payment systems that are a function of individual incomes. APPENDIX C Page 1 of 15 NOTES, TABLES AND FIGURES FOR CHAPTER 4 Note C.1 Definition of Labor Force Participation A man is considered to be within the labor force if any of the following conditions are fulfilled: a. worked during the week prior to the survey b. looked actively for a job during that week c. had a permanent job even though did not work the week before the survey d. did not work, did not looked for a job, and did not have a permanent job, but: 1. reported to carry out household tasks and was in the age bracket 18-59 2. was on holidays, on strike, waiting for seasonal harvesting, waiting for the answer of an employer, or waiting for the beginning of a new job 3. was sick but worked at some point during the year before the survey 4. reported not to be looking for a job because "there is not work" or "other reasons" but worked during the year before the survey, or did not work but was not enrolled in a school A woman is considered to be within the labor force if any of the following conditions are fulfilled: a. worked during the week prior to the survey b. looked actively for a job during that week c. had a permanent job even though did not work the week before the survey d. did not work, did not looked for a job, and did not have a permanent job, but: 1. was on holidays, on strike, waiting for seasonal harvesting, waiting for the answer of an employer, or waiting for the beginning of a new job 2. was sick but worked at some point during the year before the survey 3. reported not to be looking for a job because "there is not work" or "other reasons" but worked during the year before the survey APPENDIX C Page 2 of 15 - 90 - Figure C.1: WOMEN'S SELF-EMPLOYMENT. LIMA, 1985-90 0.0 0.8 - 0.7 a.e II 0.5 - 0.4 0.3 I 14-20 220 31-40 41-50 si-so Age Cateor Iee 0 1995 + 1990 APPENDIX C Page 3 of 15 - 91 - Figure C.2: MEN'S SELF-EMPLOYHENT. LIMA, 1985-90 0.416 0.45 - 0.44 - 0.43 - 0.42 - 0.41 - 0,4 0.4 9 o 0.27 0.34 0.33- . 0. 21 0.3 0.29 a . 27 0.27 0. 25 4-20 21-30 31-40 41-50 51-60 Age Ctaer ies a 1985 + 190 Data Source: PLS APPENDIX C Page 4 of 15 - 92 - Figure C.3: MALE AGE LFP PROFILES BY GEOGRAPHICAL LOCATION, 1985 0.9 C - 0.8 0.7 o.e- 0.6 0.4 0.3 44-20 24-20 34-40 44-S 54-60 64-70 Age Q0oupe 0 shAr" + Othe wmn o Lim Data Source: PLS APPENDIX C Page 5 of 15 - 93 - Table C.1 LABOR FORCE PARTICIPATION MEN :OMEN TOTAL Lima 1985: LFP Rate 78,9% 52,0% 64.7Z SampleSize"2265 2534 4799 Lima 1990; LVP Rate 76.2% 43.8Z 59.$% Sample Size 2442 2668 5110 .ATA. SOURCE; PLS. Table C.2 WOMEN' S TRNSITIONS TO AND FROM LFP 1990 In Out Total 19859 In 301 118 419 71.7% Out 59 106 165 Tte 360 224 584 61.67 38.4% Note: The sample includes those women older than 24 and younger than 51. DATA SOURCE: PLS. ’ㅑㅑ―-j ㅑ―｀의．&F ^PPENDIXC Page 7 of 15 •95- 號 APPENDIX C Page 8 of 15 96 TAM (4) ALIPA OBSER :Så0ö1 -2: 0.08~ -0'.*99,::- 0.8 0.17 0 ---0, 14 4160 o- ... ;0,33 1. 00,* 2 54 2 - .. .... . Y*~ 0.04. -0487 0. 42'.:, '54 '486 0.04- 4. 49, -:*.45- .. .. . .. . 0.04* .0-89, 0.'037-' 4.18: ... .... . .. 0.05, -56 --Ö.03 . .. 0 338.. 10. 1*1 38 -,4.09. 0 vli: c . ---Individuals - in xhe:läbo- i: :f r * ca- var loviäd-*-'.. dtiiin tbc, pIriQr 7ý'Q' :X-be, -e U L l 2 P r* o- p*` 6r t l 6 n** öf Ihdividuåls'ln Che 1 a b o r* k 6 ci a** v; .6 :`å" x-`.pgrierýced 97ýeåt at durin - - etläd f :6né-,Year prior-to the* ALFA (ULI 1 2,. - URM L) :Piareistönec lndlca , torý::,:..:.:*.:., . ...... . ... APPENDIX C Page 9 of 15 - 97 - DEFINITION OF VARIABLES USED IN COX MODEL FOR DETERMINANTS OF DURATION VARIABLES DESCRIPTION Dependent Variable DUR Duration of the last unemployment spell in weeks Censoring Variable COMP - 1 if unemployment spell is complete, 0 otherwise Education, Age and Training SCH Years of schooling AGE Age of the individual AGE2 Age of the individual squared SCHAGE Product of the variables AGE and SCH (AGE * SCH) TRAINED - 1 if ever received professional training, 0 otherwise Marital Status MARCOH - 1 if married or cohabiting, 0 otherwise Family Variables LHHINC Natural log of monthly total income generated by all members of the family from their main job Geographical Variables RURAL - 1 if living in rural areas of the country, 0 otherwise Other HSWRK Number of hours per week spent at home carrying household cores APPENDIX C Page 10 of 15 - 98 - COX ESTIMATES OF THE DETERMINANTS OF DURATION S 98 SCR -0.0637 -00189 0006 0.0002 90 028 0 002 "0.5 0 0'004 6 3 GE 01334 0 .1442 0 .443 01161 0.1084 00 6.4 6.2 6.3 4.7 4.4 9 AGE2 -0.0016 -.0016 000 6 0.0023 0.00 -5.6 5.6 45.6 45 3 35 cG 0.0014 0 00016 01 0018 -- ? MARC R.-"0,4-24'.*:*.-- -443 '-3 TRAIN D 0457 05 09 Hs R. .66 -.0212 .-3. DUR 8 2.8 2 COF .0708- ..70 0.70 .. .70 SCH 7.7 7.75 :775 7.75 7.75 7 75 AGF 29.2 29.2 29,2 29.2 29.2 9.2 UJRAL . 0.17 - 0.173 0.173 0.173 1LARCOH 0. 77 0.377 0.37 DATA SOURCE: PLS APPENDIX C Page 11 of 15 - 99 - Table C.8 BASIC EARNINGS GEATING JUNCTIONS (1) (2 (3) 4) schoo<ID chool>-10 (32,0) (6.8) (LO0.6) (15.,3) SCH2 .001 (2.2) ... EXP .0428 .042.. .08 00 (14.2) (14.0) (9.6) (9.3) EXP2 -.0004 -.0004 -.0004 -.0006 (9.1) (9.1) (6.5)> 58 Note: Student-t in~ pa'rnthes±u (absolute values), DATA DURCE: PL... Table C.9 GEOGRAPHICAL EFFECTS N EANI::OS All A11 Ruiral Urban Rural Urban (1) (2) (3) (4> (5) (6) SCH .0594 .1064 .1088 .1047 01 067 (4.4) (27,9) (31.9) (25.3) (0.3) (4.1) .0026 -008 4002 (3.6) (44.6) <2.) EXP .041 .0427 . 0240 049...02 .049 (13.9) (14,2) (3.6) (14,7) (3.1) <14.5) EXP2 -.0004 -,0004 -.0001 .-.005 .-.0001 -.0006 RURAL -.2194 -.193 (6.7) (6.0) Note; Student-t in1 parenthesis (absol~ute va.lues), DATA SOURCE: PLS APPENDIX C Page 12 of 15 - 100 - Table C.1 NATURE OF CONTRACTS AND EANINGS .FNCTIONS Rural Lima Urban (1) (2) (3) (4) (5) (6) SC0 .1106 .10 .1149 .113 .1039 10 (12,2) (11.7) (20.6) (20.5) (24.8) (24.3) EXP .0293 .02 .0507 .047 .0518 .047 (2 4.4) <3.5) (11.5) (10.7) (15.6) (14.2) EXP -40002 -.0002 -.0005 -.0005 -.0006 -.0005 (2.4) (1.9) (7.2) (6.9) (10.8) (1.3) FOML .183 .35 -.017 .113 .0428 .212 (2.8) (4.2) (0.4) (2.2> (1.4) (5.4) SELF .25 .217 .25 (3.3) (3,9} 62 Note: Student-t iLn parenthest. (aboolute values), DATA SOUJRCE: PLS APFENDt1C Pag.13 of 15 一工01- 魚 APPENDIXC Page 14 of 15 一102- 斗 APPENDIX C Page 15 of 15 103 läpp- tillät -:K. Må I my ... ké* tý 0 N ENTEkåtä Åm OTXER.`:PUBL 00 - t N tt -M.' jul ý::,k åm lewke mentor I tro n 1 -1 n det s- somt, oms sans t mas Y :*.'..;..,;:ý.,ýý ci-7 `91::-: EW ... :(8.0 0004 tillsnäst Vo, 3, -tvbdättttl.-0.,nemtt 1943:- .149 (3.8) ..285 -i-5Y!IYSI el.Voss flottist, sto.. MötC-SamplåSE urban formal vorkers. qtlldent-t in parenthecia DATA SOURCE: PLS APPENDIX D Page 1 of 5 ANALYSIS OF THE DETERMINANTS OF FEMALE LABOR FORCE PARTICIPATION This Appendix examines the effects of age, schooling, geographical location and other variables on female LFP, using multivariate probit models. This analysis makes it possible to uncover interactions between variables not revealed by simple marginal distribution analysis. Probit estimates in Table D.1 reveal the impact of education, age, marital status, husband's income, number of small children in the family, and geographical location on a woman's likelihood of participating in the labor force. The specification presented in Column 1 reveals that female participation decreases linearly with schooling. This effect was not expected, as higher educational levels are usually linked with higher participation rates. Column 2 offers further evidence about the effect of schooling on the likelihood of female participation by revealing the significance of the quadratic schooling coefficient. This new specification shows, therefore, that education has a negative effect on female participation only for women with less than 9 years of schooling; LFP increases with additional years of education after that level. In order to further examine the non-linear effect of schooling on female participation, the sample has been divided into two groups based on the level of education. Since the variable indicating widow or separated status was not significant in previous specifications, it has been eliminated from probit models in columns 3 and 4. Column 3 shows that for those women with less than 9 years of schooling, participation decreases with schooling. On the other hand, Column 4 shows the positive effect of schooling on participation for women with more than 8 years of schooling. Column 5 shows that if the sample used excludes women with less than five years of schooling, the linear specification produces a positive and significant effect of education on LFP. In other words, only the high participation rate of illiterate women produces a negative schooling coefficient for the whole sample. Participation rates increase with age, at a decreasing rate, until 37 years of age, as shown in Column 1. Female participation declines thereafter. This confirms the results obtained during the preliminary analysis of the marginal distributions, which produced bell shaped profile of participation by age category. These results are robust to changed specifications in Table D.1. Columns 1 to 5 show that married and cohabiting women are less likely to be in the labor force than single women. On the other hand, there is no evidence of different participation rates among widowed, separated, and single women. These results reveal that the reservation wages of married and cohabiting women are higher than those of other women. APPENDIX D Page 2 of 5 - 106 - DEFINITION -OF VARIABLES. USED N THE LABOR FORCE PARTICIPATION PROBIT ANALYSIS VARIABLE DESCRIPTION * ependent Varabl LF ~ i £vithin~ the labor fce, 0 therw±,e EducatlLo and ge. SCH Years o- ,esqlure SCH2 Years of schooling quared AGE Age o. he .d..vdua ÀGE2 Aie of the ndividual squared Marital Statu M A RIED - iar ied 0 eth ei .w. -COH"«ZT. 1 chbtng0oá îa MIDOW: - 'I ifidow,. 0.,,therwise....:-... SEPAR i L earated, Q otherwise MARCOH - 1 If arried or cohabit.ig, othezwis amily Variable. T1fl Dumber ofchildrenLo the head of fiy who are younger than7 LWHg 1atu~ral 1g o nhsaiu'snordh y ±çom i m o jQ eographicx .,b:e -RURAL - ing tra areas o the country, 0 otherwNe LIM .ni In et.opolitan L.m.. ..then e. APPENDIXO Pag.3 ofs 一107• 丰 APPENDIX D Page 4 of 5 - 108 - Estimates in columns 1 and 2 show that female participation in rural areas is higher than in urban areas. Within urban areas, women living in Lima participate less than women in the other urban areas. These results match the preliminary conclusions obtained through marginal distributions analysis. Howeverg multivariate analysis allows a deeper identification of the effect of geographical location, which is not confirmed for all levels of schooling. For women with low educational levels, participation rates in urban areas are lower than in rural areas, and these rates are even lower in Lima. Yet, for women with more than 8 years of schooling, the coefficient of the variable indicating geographical location loses statistical significance, as does the variable indicating residence in Lima. For more educated women, the wage effect measured by the schooling coefficient is the same in any environment, as if a more educated labor force promoted integration of labor markets and increased efficiency. Columns 6, 7, and 8 in Table D.1 show estimates for women who live with their spouses. In general, the effect of education on the likelihood of participating in the labor force is similar to that for the whole sample. Among more educated women, spouses living with their husbands have higher participation rates for additional years of education than the rest of the female population. Participation rates decrease with the presence of children younger than 7 in the household. The greater number of young children living at home increases female reservation wages, and thus inhibits female participation. The higher the husband's income, the smaller the probability of the spouse's labor force participation. Higher husband's income also increases women's reservation wage, again lowering participation. In aum, participation rates in Peru increase with education for most women, except that illiterate women participate more than women with limited education. For women with low educational levels, additional years of schooling are associated with lower participation rates. The age effect on participation is positive until approximately 40 years, and negative thereafter. There is no evidence of retreat from the labor force after marriage. Participation rates are higher in rural areas than in urban oneaq and, within urban areas, rates are lower in Lima. However, the effect of geographical location is only true for women with less than 9 years of schooling. For more educated women, geographical location has no impact on Up. The presence of a spouse or young children in the household also inhibits female participation. Finally, husband's income is negatively related to the spouse's participation in the labor force. A similar percentage increase in the wages of both husband and wife increases the female UP rate, a result which is consistent with observed trends. Between 1985 and APPENDIX D Page 5 of 5 - 109 - 1990, as wages declined, female LFP rates dropped, departing from the long run trend. Female LFP Probit Estimates: Comparison with Previous Research Studies Most of these estimates agree with those obtained by previous researchers using the same data (King, 1989). There are, however, a few findings with important policy implications, which were absent or different in previous analyses of female LFP. The most important of these is the estimate of the schooling coefficient. King's most recent study concludes that years of schooling have a very small and, in fact, insignificant negative effect on LFP. Apparently, even with non-linear specifications, she did not obtain more significant results. The new estimates shown here indicate that the effect of education is very strong and positive for women with more than 8 years of schooling. Even for a sample which only excludes women who went to school less than 5 years, probit estimates of the effect of education are significant and positive. King also found that geographical location had a negative and significant effect on female LFP. In fact, geographical variables are not a determinant of LFP of highly educated women. Finally, with King's specifications, the presence of even very young children in the household appears to have no effect on participation. This result is, however, not robust. The new estimates show a very significant and negative effect of the number of children younger than 7. On the other hand, experimentation with a variable for children in the 7-13 age bracket produced insignificant coefficients. 1/ These results confirm those of other studies of female LFP carried out, using the same data source, by some of the authors of this report and by other researchers. (See Gill, 1991 for estimates and references to previous studies.)