Mexico - Urban development (Vol. 1 of 3) : Text

Report No. 1449-ME Urban Development in Mexico Kin Three Volumes) Volume 1: The Text January 31, 1977 Development Economics Department Latin America anid the Caribbean Region Urban Projects Department IFOR OFFICIAL USE ONLY' Document of the World Bank This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization.   FOR OFFICIAL USE ONLY This is one of two reports derived from the findings of a mission to Mexico in January/February 1974. The mission was led by Douglas Keare (Development Economics Department) and Ian Scott (Latin America and the Caribbean Region) and included Roberto Cuca (Development Economics Department), Sadasumi Hara (Transportation and Urban Projects Department), Y.J. Hwang (Development Economics Department), Anna Sant'Anna (Development Economics Department), Professor John Friedman (Senior Consultant) and Jaime Biderman (Consultant - Researck Assistant). The report has been prepared by Ian Scott with principal assistance by Douglas Keare. Editorial assistance was provided by Linda Lessner. Thi document ha a atricted distribution and may b used by recipients only in the rformanc d their olaIk dutis. Its contents may not otherwis be disclod without World Bank. uthorition.   URBAN DEVELOPMENT IN MEXICO Table of Contents Page No. INTRODUCTION ... . . . . . . . . . . . . . . i-v 1. THE ANTECEDENTS OF THE URBAN SYSTEM . . . . .1 A. Urban Development Before c. 1875 . . . . .1 The Pre-Columbian Period . . . . . . . .1 The Colonial Period . . . . . . . . . . . 2 The Independence Period: 1820/75 . . . . 3 B. Urban Development 1875/1910 ... . . . . . . 6 The Dynamics of Urban Development . . . . 6 The Structure of Urban Development . . . 13 C. Urban Development: 1910-40 . . . . . . . . 15 The Dynamics of Urban Development . . . . 15 The Structure of Urban Development . . . 23 2. THE DYNAMICS OF URBAN DEVELOPMENT: 1940/70 . . 26 A. Introduction . . . . . . . . . . . . . . . 26 B. Transport Development . . . . . . . . . . . 29 The Railways . . . . . . . . . . . . . . 29 The Roads . . . . . . . . . . . . . . . . 31 Air Transport . . . . . . . . . . . . . . 32 Transport 'and Urban Development . . . . . 33 C. Industrial Development . . . . . . . . . . 37 Industrial Development and Urban Development 39 D. Tertiary Sector Development . . . . . . . . 57 3. THE SOCIO-ECONOMIC STRUCTURE OF THE URBAN SYSTEM: 1940/70 . . . . . . . . . . . . . . . 58 A. The Demographic Structure of the Urban System 58 The Demographic Characteristics of the System 58 The Origins of Urban Growth . . . . . . . 60 The Geography of Urban Growth . . . . . . 61  Table of Contents (Cont'd) Page No. B. The Economic Structure of the Urban System: 1940/70 . . . . . . . . . . . . . . . . . . . 62 Employment and Output . . . . . . . . . . . 62 Consumption ... . . . . . . . . . . . . . 77 4. THE SPATIAL STRUCTURE OF THE URBAN SYSTEM . . . 88 A. Area Development and the Structure of the Urban System . . . . . . . . . . . . . . . . 88 Demographic Structure . . . . . . . . . . . 88 Patterns of Area Development . . . . . . . 90 B. Regional Development and the Structure of the Urban System ... . . . . . . . . . . . . . 96 Regional Definitions . . . . . . . . . . . 96 Regional Profiles . . . . . . . . . . . . . 99 ANNEX A: Agriculture as an "Urban" Sector ANNEX B: The Development of the Railway Network Prior to 1910 ANNEX C: Airway Flows ANNEX D: Real and Nominal Distance ANNEX E: Relative Accessibility ANNEX F: Urban Economic Profiles ANNEX G: Public Services - Factor Analysis ANNEX H: The Demographic Components of Urban Growth ANNEX I: The State Development Index ANNEX J: Technical Notes  URBAN DEVELOPMENT IN MEXICO INTRODUCTION i. This report was prepared as a background report to the policy-oriented "Spatial Development in Mexico". In addition to providing a compendium of adjunct information essential to an understanding of "Spatial Development in Mexico", it also consti- tutes a source of considerable historical and socio-economic information and data on Mexico, organized for the use of researchers in the fields of urban and spatial development and related fields of study. ii. "Urban Development in Mexico"' is divided into four chapters. The first provides an historical perspective of the urban system up to 1940; the remaining chapters treat the period: 1940 to 1970. The second chapter covers the dynamics of urban development, focusing on development in the transport, industrial and tertiary sectors. The third describes the socio-economic structure of the urban system. The last discusses the spatial implications of the urban system, concentrating on both area and regional development. Annexes are attached. Background iii. In 1940, Mexico was predominantly a rural nation. Seventy percent of its population lived in communities of fewer than 2,500 inhabitants, depended upon an agricultural economy and existed in a rural setting. Presently, nearly two-thirds of the population live in urban communities and are part of an urban economy. Not only has the past seen extensive change; the future seems to promise a continuation of these wide-sweeping changes. Mexico's population will have grown from 60 million in 1970 to 120-150 million by the end of the 20th century. Three-quarters (90-110 million) of this population will live in towns and cities. iv. Urban development in Mexico reflects the processes of modernization and sectoral change which have occurred since 1940. It also measures the growth of a new category of economic issues which require urgent attention. Some problems arise directly from the increasing share of the national population residing in urban areas and engaging innon-agricultural occupations, the impact of rural-urban migration and the provision of jobs, shelter and urban services for newly arrived migrants. Such problems are urgent in every developing country.- But some countries, such as Mexico, are also seriously affected by other aspects of urban development, which relate to the size and structure of the urban system and in particular to the dominating role of Mexico City.  - ii - v. This report is primarily concerned with the country's largest cities, and specifically with those which had populations of 100,000 or more in 1970. The 37 cities in this category have accounted for a very large proportion of the nation's urban population over a long time span as seen from Table 1.55, which shows the percentage share of the total urban population residing in these 37 cities in 1940, 1950, 1960 and'1970. On the basis of a "threshold" of 2,500 inhabitants, the percentage of the urban population living in the largest cities during the 1970s remained constant at 66 to 68 percent of the total. When the urban popula- tion is defined in terms of a "threshold" of 10,000, 15,000 or 20,000, the percentage share of the urban population living in the largest cities is of course higher, but again relatively stable. In terms of threshold of 10,000 inhabitants, the largest cities of the 1970s accounted for 100 percent of the total urban population in 1940, and for 94 percent of the urban population in 1970. vi. Tables 4.15 and 4.16 show the extent to which employment in sectors other than agriculture was concentrated in 1970's largest cities. 1/ In 1950 and 1970, the 37 cities contributed large shares of total employment in both the secondary and tertiary sectors of the economy. For example, in 1950, their aggregate sector shares were commerce, 57 percent and electricity, 75 percent; while in 1970, 61 percent of construction employees and 71 percent of services employees were' in these cities. The same tables show that, in 1950,1 Mexico City accounted for 28 to 40 percent of total national employment in each branch of the secondary and tertiary sectors and that these shares were in the same general range in 1970. vii. After 1940 urbanization was associated with major changes in levels of consumption. Trends of consumption were linked to the growth of output, investment and exports, but there was no clear relationship between consumption and income distribution. The cities which occupied the upper parts of the urban hierarchy in 1970, accounted for at least a proportionate share of national consumption, as extrapolated from data on disposable family income in the 37 cities. Generally, the economic dynamism of the city was the major factor accounting for income differentials and for relatively higher levels of consumption. viii. Against this background, examination of the non-spatial aspects of the urban system in Chapter 3 provides information for the further analysis of general questions concerning urbanization and urban development in Mexico. It is less than a comprehensive background, because it does not account for cities with fewer than 100,000 people but, as any analysis of urbanization and urban development has to be based on a sample of some kind, this sample is chosen because it captures a very large share of the total urban system. However, the major reason for focusing on Mexico's 1/ See Annex A for a discussion of the role of agriculture as an "urban" sector.  - iii - major cities is to provide background for the discussion of spatial policy issues in Volume II. For this purpose, the country's largest cities constitute the basic spatia:L framework of the national economy. The major features of the geographic distribution of population, production, poverty and prosperity may be explained in terms of the location and spatiaL distribution of the country's largest cities. Furthermore, if we interpret the national space economy in terms of a set of regions, covering the whole of the nation's economic space, the nation's largest cities will be the centers of these regions because they represent the major nodes of the interregional transport network, the major markets, the major concentrations of industrial production, and the major commercial centers. ix. Thus an analysis of the upper parts of the urban hierarchy is a means of approaching two key economic issues: (a) area development policy, and (b) regional policy. The latter essentially refers to the national economy, whereas area development policy focuses on subdivisions of the nation. A more detailed view of the urban system (i.e., one encompassing the lower as well as the upper parts of the urban hierarchy) would be unrealistic, because it would necessitate the manipulation of unwieldy amounts of data. Furthermore, an adequate understanding of the area and regional structures of the economy can be gained from an analysis of its major cities. x. The emphasis on the largest cities of the 1970s is much greater in Chapters 2, 3 and 4 than in Chapter 1, because the latter concerns the general evolution of cities under condi- tions in which there was no urban system, but rather a series of separate and largely unrelated sub-systems. The origins of many of the cities, which are now among the country's largest, figure prominently in the discussion of the pre-industrial antecedents of today's urban system. The Antecedents of the Urban System xi. Although :Large-scale urban growth did not begin in Mexico until the 20th century, Chapter 1 traces the roots of Mexico's urban development from the pre-Colombian and colonial period prior to 1875, through the independence period, to the period between the Revolution and 1940 when consolidation of previous developments and institution building commenced. xii. Tenochtit:Lan, the capital of the Aztec empire, repre- sented the maximum urban achievement of pre-Colombian Mexico, but there were other important settlements as well (Uxmal, Chichen-Itza and Oaxaca). The arrival of the Spaniards in 1519 did not greatly alter the spatial-order of the pre-Colombian  - iv - period, when there was sufficient integration between terri- tories to maintain political control, but very little, if any, economic interaction. The colonial pattern of urban develop- ment in Mexico was similar to that in other countries. Tenochtitlan was rebuilt and renamed Mexico City, retaining its position as the hub of the colonial, as well as in the pre-Colombian, system. Although there was greater emphasis on the exploitation of mineral resources, the distribution of colonial towns coincided with the earlier distribution of settlements. xiii. Urban development during the presidency of Porfirio Diaz, from 1876 to 1910, was characterized by the spatial incidence of the growth of export industries and the associated development of the railway network. In general, urban development, and more specifically the selective growth of certain cities, were largely a function of (a) the spatial incidence of the growth of industries producing for foreign markets; and (b) the associated development of the railway network. To some extent, but to a marked extent only in the case of Mexico City, urban development was also the product of agglomeration economies arising from earlier demographic and economic developments and the growth of internal markets. Radical changes did not characterize urban development during 1910-40, when the growth of cities emphasized the continued expansion of places which were already relatively large by 1910. The pattern of relative continuity progressed, with the development of the transport, industrial and agricultural sectors again providing the lynchpins of the system. The Dynamics of Urban Development: 1940-1970 xiv. Urbanization in Mexico in this period, 1940-1970, continued while the population grew at unprecedented rates; the rate of urbanization was faster during this period than at any time prior to 1940. The dynamics and structure of urban growth were determined by the transport, industrial and tertiary sectors. Chapter 2 describes this process by relating trends in total population and urban population growth. In addition, each mode of transport (rail, road and air) is reviewed. The importance of cities being located near the center of a transport network and the impact of the relative connectivity between major cities and between cities and their nearest neighbors is stressed.  xv. The factors underlying the relationship between urban and industrial development are presented and complemented by a brief account of other aspects of the urban-industrial link: market potential, agglomeration economies and the role of the public sector, especially with regard to taxes, credit, public investment, wages, prices and tariffs. Tertiary activities also played a major role in generating urban growth, in the border cities for the most part. The Socio-Economic Structure of the Urban System, 1940-1970 xvi. This chapter describes the demographic characteristics of the urban syste:m, the origins of urban growth, and the geography of urban growth, concentrating on the set of 37 cities which constituted the upper parts of the urban hierarchy in 1970. The economic structure concentrates on employment and output, showing contrasts between and within ci.ties; and on consumption, contrasting income levels and distribution and other measures of welfare at the national level, and between cities. The Spatial Structure of the Urban System xvii. The spatial implications of the urban system are examined in terms of area development and regional development. The patterns of area development from 1940-70 are related to the level of urbanization in each state and to the spatial distribution of large cities between states. Finally, a set of economic regions defined in terms of state aggregations and built around the spatial framework of the nation's urban system is constructed.   URBAN DEVELOPMENT IN MEXICO 1. THE ANTECEDENTS OF THE URBAN SYSTEM A. Urban Development Before c.1875 The Pre-Columbian Period 1. Although large-scale urban growth did not begin in Mexico until the 20th century, the country has a long urban history beginning with the construction of major pret-Columbian cities in widely separated parts of what is now Mexico and culminating in Tenochtitlan, the capital of the Aztec emperor, Montezuma. 2. The Aztecs were the last of the Nahua tribes to settle in the valley of Mexico and established Tenochtitlan on an island in the middle of a lake (Barkin and King: 1970). Wlen Hernan Cortes arrived in 1519, it was a city of perhaps 300,000 people. 1/ 3. While Tenochtitlan was the maximum urban achievement of pre-Columbian cultures in Mexico, it was by no means the only one. To the east, in the Yucatan peninsula, the Mayas and the Toltecs built several major urban settlements - notably Uxmal and Chichen-Itza - while to the south, in Oaxaca, the Zapotecs and the Aztecs also achieved high levels of urban civilization. 4. There are certain similarities between the spatial order of the pre-Columbian period and that which developed after the arrival of the Spaniards. In pre-Columbian societies the geographic space of what later became the colony of New Spain was divided into several political territories, each controlled by a distinctive culture (Map 2). Within each, there was sufficient integration to maintain political control, but between them, there was little, if any, economic interaction. 1/ Cortes later wrote of it: "When we saw so many cities and villages built in the water and other great towns on dry land... we were amazed and said it was like the enchantments they tell of the legend of Amadis, on account of the great towers and cues (temples) and buildingsrising from the water, and all built of masonry."  -2- The Colonial Period 5. If the pre-Columbian antecedents of Mexican urbaniza- tion differed from those of other parts of Latin America, the colonial history of its urban places was similar to that of other countries. The initial pattern of colonial settlement closely resembled the pre-existing pattern of pre-Columbian settlement. In the Valley of Mexico for example, the Spanish approach was to interpret Aztec society as a mosaic of towns rather than as a system of tribes (Morse: 1971). 6. The early distribution of towns and cities across the viceroyalty of New Spain was largely coincident with the earlier distribution of settlements, although there was greater emphasis on the exploitation of mineral resources. However, the site of what had been Tenochtitlan (rebuilt and renamed Mexico City) was always the center of the colonial system and the seat of govern- ment and ecclesiastical authority. 7. During the next four hundred years, the initial pattern of settlement laid down in the 16th century was strengthened but modified in two ways. First, as colonial settlement increased, many new towns were established. Second, some of them were ephemeral because judgments about the resources on which they were based were wrong. 8. Colonial urban development continued this pattern. Throughout the period, techniques of production and communica- tions were largely unchanged. The Spaniards brought the wheel and the horse, but through c.1800, there were few major technological developments compared with those which came afterwards. Similarly, the values - economic, social, political and ethical - which tied the colonial society of New Spain to the mother peninsula and to its various parts, were essentially unaltered. The spatial continuity of colonial Mexico was there- fore consistent with the continuity of its technology and its values. 9. The preeminent status of Mexico City was related to its role as the political, administrative, and financial center of a major part of Spain's colonial territory. Many other colonial cities, including Guanajuato, Taxco, Pachuca, Saltillo, Zacatecas, San Luis Potosi and Durango were mining centers; some, such as Salamanca, were agricultural centers, Veracruz was the chief Gulf port, and Guadalajara, Merida, Oaxaca and Aguascalientes were administrative, military and commercial outposts. 10. The Mexican colonial city was, above all, an instrument of the colonial economic and social systems. Apart from political and administrative integration colonial towns were in no sense  - 3 - integrated into an urban or spatial system. They were autonomous population clusters serving the limited function of providing of certain services. As such, cities were linked to some extent with the agricultural areas around them, but not, in general, with other towns and cities. Consistent with the colonial purpose, many of them had direct links with Spain. The Independence Period: 1820/75 11. The change in the technological system and the later change in values precipitated the collapse of the colonial order. The change in values was confined to attitudes towards an over- seas government and paralleled by attitudes which affected the rest of Hispanic America at about the same time and which, not long before, had affected the Anglo-Saxon colonies to the north. 12. The absence of other kinds of valuative change and the continued absence, in the early post colonial decades, of any major change in the technological system, implied that for some time after Independence, the parameters of urban development were not really much different from those of the colonial period. For most of the nineteenth century the nature and pattern of urban develop- ment was similar to that of the colonial era. 13. In general the development of specialized urban functions was impeded by the near impossibility of communication between various parts of the country. This, to a large extent, was an inevitable reflection of Mexico's tortuous topography, which separated its populations into numerous, small, isolated, distinct communities. 14. Many of Mexico's major topographic and geologic features are extensions of those of the United States (Map 1). The Rocky Mountains continue across the Rio Grande to become the Sierra Madre Oriental, terminating in the neighborhood of Tampico, midway down the Gulf Coast. The Sierra Nevada of California and the Basin Ranges of Arizona, after merging and almost disappearing in the deserts around Mojave and Yuma, gain magnitude when, south of the border, they become the Sierra Madre Occidental extending midway down the Pacific coast. The plains of southern Texas likewise stretch southward to form the great central basin of the states of Chihuahua and Coahuila in northern Mexico. The average altitude of this basin is between four and six hundred feet while the ribs of the Sierra rise to two and four thousand feet. 15. The great basin contains many irregular and disconnected mountain ranges, bluffs, and isolated ridges, separated by broad valleys and plains. The characteristics of this region are  - 4 - accentuated by the barrier of the Sierra Madre del Oeste which towers above it to the west, by the lesser and more disconnected Sierra Madre del Este to the east, and by the gigantic volcanoes of the south. To the east, there is an extensive and rugged decline which slopes to the coastal plain of the Gulf of Mexico over a mile below. To the south, the Sierra Madre del Este continues to rise until the mountain mass culminates in the volcanic cones of Orizaba. 16. The Sierra Madre del Oeste is lower in the north than in the south, its highest northern summit barely reach- ing six thousand feet. To the south, its hills mass together in an irregular and confused manner, increasing in height and number until, in northern Durango and Sinaloa, the mountains become much higher and more extensive, and culmi- nate, east of Culiacan, in summits exceeding 11,000 feet. As they increase in height, the crests of the Sierra Madre gradually approach the coast until in the neighborhood of San Blas, rugged mountain faces rise above the ocean to altitudes of two to seven thousand feet. South of the Rio Balsas, there is a narrow and precipitous range which reaches altitudes of 10,000 to 12,000 feet and separates the valley of the Balsas from the Pacific Coast. 17. This strongly accidented landscape has had a funda- mental influence on communications and urban development in Mexico. Given the topographical conditions of the interior, nineteenth century Mexico was a country of trails and primitive roads; the major means of transport were Indian porters, pack animals and two wheeled carts. By 1810 there were 7,600 km of roads. Regular stagecoach services come into being in 1849. When the first line was established between Mexico and Puebla, other services were subsequently started between Mexico City, Veracruz, Tepic and Tampico. 18. Powell (1912) records that the principal regional routes in pre-railroad Mexico were (Map 18): (1) Veracruz to Mexico, via Jalapa or Orizaba; (2) Tampico to Mexico, via Pachuca or San Luis Potosi, Guanajuato and Queretaro; (3) Natchitales to Mexico, via San Antonio, Presidio de Rio Grande (or Piedras Negras), or via Laredo, Monterrey, Saltillo and San Luis Potosi; (4) Santa Fe to Mexico, via Paso del Norte (or El Paso), Chihuahua, Durango, Zacatecas, and Guanajuato; (5) Mazatlan to Mexico, via Durango; (6) San Blas to Mexico, via Guadalajara and Queretaro; (7) Acapulco to Mexico, via Chilpancingo and Cuernavaca; and (8) Mexico to Guatemala, via Oaxaca. These routes took advantage of the mountain passes which encircled the central tableland; most of the railroad lines later followed them.  -5- 19. Prior to the coming of the railways, the significance of ports was determined by their location on or near traffic routes from the coast to the interior. Besides Veracruz, which was a center of commercial activity even before it had a railroad connection with Mexico City, the main ports were Bagdad, La Perla, ITampico, Tuxpan, Tecohitla, Alvarado, Prontera, Coatzacoalcos, Carmen, Champoton, Campeche, Progreso, Chetumal on the east coast, and Guaymas, Topolobampo, Altata, Mazatlan, Teacapan, San Blas, Barra de Navidad, Manzanillo, Zihuatanejo, Acapulco, Puerto Angel, Salina Cruz, and Tonala on the west coast. 20. Consequently, towns and cities were relatively unimportant up to the late nineteenth century and served primarily as administrative centers with some trading functions. Most manufacturing was directly tied to primary production but there was, with Independence, a fundamental change in economic organization direct:ly affecting the urban economy. The exploitative, and externally-oriented economic order of the colonial system had replaced the pre-Columbian agricultural society, which had been sufficiently productive and inventive to yield a surplus to support urban civilization. At Indepen- dence, this external orientation was replaced by an internal one. As many of the old mining towns lost their dynamism with the exhaustion of their mineral wealth, the post-Indepen- dence era was marked by the attempted stimulation of industry. In this environment, a number of towns such as Puebla, Queretaro, Orizaba, Guadalajara and Mexico City benefitted from the estab- lishment of textile plants, and the port of Veracruz served as a point of entry for imports of cotton, wool and textile machinery from Europe. 21. In nineteenth century Mexico, the effective units of social and spatial organization were still those of the rural sector - haciendas and ejido villages, both of which were relatively autarkic. The hacienda was particularly self-sufficient and served as a "proto-city", fulfilling many urban functions. There was no effective articulation between the rural and urban sectors. As in the colonial period, the space economy continued to feature the development of separate and essentially autonomous economic enclaves, although Mexico City had already achieved a unique status as an administrative center and was by far the largest population cluster.  - 6 - B. Urban Development 1875/1910 22. The administration of Porfirio Diaz (1876-1910), marked the end of the civil wars and foreign interventions which impeded economic progress for 50 years after Independence; but the economic system of this period, sometimes called the "Porfiriato", had special, and ultimately self-destructive, characteristics. Superficially the records show that the national population increased at an average annual rate of about 1.4 percent - from 9.0 million in 1870, to 15.1 million in 1910 but that the GDP grew faster, at an average annual rate of 2.7 percent. This moderate expansion was facilitated by foreign investment as was the transportation network which was developed in the context of an economy in which the growth of the domestic market was, at least initially, almost incidental. Urban development in the Porfirian period was consistent with the characteristics of the economy. In general, urban development and the selective growth of specific cities in particular, was largely a function of (a) the spatial incidence of the growth of industries producing for foreign markets and (b) the associated development of a railway network. Although to some extent (but to a marked extent only in the case of Mexico City), urban development was also the product of agglomeration economies arising from earlier demographic and economic development and the growth of internal markets. The Dynamics of Urban Development (a) The Development of the Export Sector 23. Although the largest share of foreign investment before the Revolution was in railroads (Table 2.4), another quarter was in the extractive industry and smaller proportions were in manufacturing and public utilities. In the early stages of the Porfirian era, most Mexican capital was con- centrated in agriculture and mining. With the realization that large profits could be made in manufacturing, increasing numbers of Mexican landowners invested in urban areas, acting alone or in conjunction with foreign entrepreneurs. 24. New investments, coupled with the development of the railroads, facilitated dramatic export growth; the value of exports from 1877 through 1911 (Table 2.5) increased more than 600 percent. Expansion was led by an increasingly diversified mining sector, whose output grew at an average annual rate of 7.0 percent between 1900 and 1910. After 1890, agricultural exports expanded rapidly.  -7- 25. There was an intimate relationship between urban development and deve:Lopments in mining, agriculture, manu- facturing and transportation. Different cities in different parts of the country develop on the basis of the stimulus of growth in one or other of these sectors but most often on the basis of a combination of two or more. (i) Mining and Manufacturing 26. The greatest era of expansion for Mexico's mining industry was between 1900 and 1910 during which time metal production almost doubled (Table 5.1). The pattern of mining development established by foreign entrepreneurs in the Porfirian period influenced the subsequent structure of the mining sector. 27. Most of the larger cities of the north and center had important mining functions; minerals included gold, silver, zinc, lead, coal and mercury. There was iron in the northeast, and hydrocarbons near the coast in southern Tamaulipas. The stimulus of mining expansion was exemplified by the rapid growth of the capital cities in the northern states (Table 1.22) which reflected the shift of production from gold and silver to industrial minerals such as iron, coal, lead and copper. 28. In this environment, the old mining centers of the states of Guanajuato and Hidalgo in the central region and of Zacatecas and San Luis Potosi in the north - which had produced large quantities of precious metals over many centuries - gave way to newer centers in Coahuila and Durango where industrial minerals were rapidly gaining importance. In many of the older mining centers, stagnation or decline set in which affected other economic activities. In contrast, mining in the newer centers led to secondary growth, parti- cularly in manufacturing, which provided mining inputs (e.g., dynamite and metal tools) and/or processed part of the output of the mining sector. 29. The best examples of export multipliers based on mineral development occurred in the northern cities, especially in Monterrey (Derossi: 1972). A lead foundry was established there in 1892; and a steel plant, in 1901. Because of the link between primary minerals production and the development of metal and non-metal mineral processing industries, the major  - 8 - concentration of "heavy" industry in Mexico by 1910, was in Monterrey. 1/ The importance of agglomeration economies in Monterrey's development derives from rapid sustained growth which was attributed to a favorable export base and the development of transport facilities for exploiting this base. 30. Foundries and foundry-associated industries were also established in other mining centers between 1890 and 1910, among which the lead and copper plants of Cananea (Sonora), Concepcion del Oro (Zacatecas), Torreon and Chihuahua were particularly important. In each instance of the growth of new subsectors (most of which had an agglomerative effect and were inevitably "urban" activities), the sequence of development corresponded to that of the classical precepts of export-based urbanization. (ii). Agriculture 31. Where urban development was based on the exploita- tion of regional agricultural resources, there was, as in the case of mining, a close link between urban growth and railroad growth. This was a consequence of one of three factors: (a) prior railroad development associated with mining in the same area; (b) prior railroad growth associated 1/ Considerably before the turn of the century, cotton textiles and beer brewing, utilizing locally grown cotton and grains, achieved importance in Nuevo Leon. The beer industry's need for bottles, metal caps, and cardboard boxes stimulated the growth of glass, metal working, and paper industries. The steel industry also got its early start in Monterrey, because of its favor- able location with respect to the necessary raw material inputs of coal and iron ore. Favorable rail connections enhanced Monterrey's access to raw materials and external markets.  - 9 - with mining in other areas as a result of which railroads were built which passed through agriculturally productive regions en route to the U.S. frontier or the coasts;or (c) the growth of railroads specifically built in association with agricultural development. 32. In all cases, urban-agricultural growth was based on the growth of domestic demand or, more commonly, the growth of external demand for agricultural products, usually in the U.S.A. 33. The expansion of agricultural exports occurred concurrently with the introduction of mechanical innovations in agriculture and of a shortage of labor on haciendas. These factors combined to stimulate (a) the growth of hacienda agriculture and (b) the demand for the expropriation of communal lands to expand the supply of labor for hacienda employment. 34. Consequently, the growth of agricultural exports was closely associated with a dramatic deterioration in the living conditions of the rural population, because the means chosen to increase the hacienda labor supply involved the progressive destruction of the long-established tenurial rights of rural communities. The basis for intervention derived from the land reform laws of 1855-57 and 1894, and the colonization laws of 1883. 35. The collective impact of Porfirian intervention in the rural system was such that by 1910 some 90 percent of Indian villages in the central plateau had no common lands. In the country as a whole, 85 percent of communal villages and 90 percent of rural families were landless and fully 50 percent of the rural population was tied to the hacienda system representing a drastic deterioration from the past. The new structure was associated with the rapid expansion of agricultural exports (at an annual rate of 2.5 percent in the same period), but the average annual rate of aggregate agricultural growth was only 0.65 percent. This was barely half the rate of population increase and a direct reflection of what was, in social terms, the most significant result of Porfirian change, i.e.,an average annual decline of 0.5 percent in foodcrop production.  - 10 - 36. The welfare consequences of this trend were disastrous. Income expansion in the leading sectors was captured by the owners of capital, land and subsoil resources and the redistribution of income towards profits, interest and rent (given the distribution of the ownership of capital and natural resources) increasingly accrued to foreigners. Not only did the Porfirian system encourage a progressive deterioration in material conditions, it also increased social and political tension. This was the background to the Revolution of 1910 and to the later growth of export oriented hacienda agriculture, which led to the development of cities as collection points for shipments of agricultural products, the growth of new urban functions related to the development of commercial agriculture and the development of new manufacturing activities based on agricultural inputs. (b) The Growth of Internal Demand 37. Even before 1910, factors besides the growth of the external market had in a few places achieved "independent" importance in urban development. These places were those cities which had developed as nodal points in the evolving railnet, but which were not well located in terms of regional resource endowments, and those which had already achieved sufficient momentum to sustain continued growth, to attract railroad linkages and whose growth was not based on exports. 38. The best example of the first type of city was probably San Luis Potosi, which along with Tampico, Monterrey and Torreon, became a major center of regional growth after the opening of new railroads. Several lines passed through the city which gradually developed into an important nodal point although it offered no significant advantages in terms of natural resources. The development of the railnet itself was thus the source of new industrial and commercial activity and a major source of new employment opportunities in the urban economy. By 1910, San Luis Potosi had the largest engine repair shops and the largest rail equipment plants in the country. 39. The major example of the second kind of "exceptional" urban growth was Mexico City. Although located in a region which produced some mineral and agricultural exports, more significant to urban growth was its role as the national capital and center of political, ecclesiastical and administrative authority and as the major center of commerce and industry.  40. In general, the exploitation of agricultural and mineral resources outweighed urban concentration as a factor in the development of a manufacturing base during the Porfiriato. Clearly agglomeration economies had already become important. As early as 1877, the large market of Mexico City permitted many of the factories located there - which operated on steam power or on animal traction - to compete effectively with factories in other states using cheaper hydraulic power. With the introduction of electricity, which proved more efficient and more readily accessible to all states than previ.ous sources of power, the advantage of location in the capital was increased. The broadening of geographic markets via the expansion of the rail network, for which Mexico City served as a primary hub, further enhanced the status of its manufacturing sector and brought about considerable diversification. Because of these cumulative advantages, the Federal District had already begun to outdistance its rivals in total manufacturing employment by 1910. PeiVafiel (1902) shows that there were 35 types of industry in the Federal District compared with 29 in Jalisco and 24 in Puebla, the next two most highly diversified areas. (c) The Development of th-e Transport Network 41. Irrespective of the origins of demand, urban development during the "Porfiriato" was intimately related to the evolution of the transport and communications system. The growth of. towns and cities was stymied by the absence of effective communications - or stimulated by their presence. 42. We have already observed that one-third of all foreign investment during the "Porfiriato" (which accounted in turn for more than two-thirds of total investment), was devoted to the construction of railroads. Although the development of the railnet was the outcome of numerous independent decisions by private entrepreneurs, it facilitated a massive growth of exports and was a major factor in account- ing for the growth of some cities and the relative decline of others. 43. The railway,era in Mexico began when a line of 16 km was built between Veracruz and Tijeria in 1854. This was extended to Mexico City in 1873. From 1877 to 1892, more than 4,500 km of track were built; by 1905 another 12,000 km had been added and by 1910 the total network extended over 24,000 km. There were few new lines there- after.  - 12 - 44. Railroad development before 1900 was in no way part of an attempt to construct a systematic network (Map 19 and Table 10.1). Railroads were constructed to meet the needs of individual developers to transport mineral and agricultural products outside the country. Most of them therefore connected mining areas with seaports (Tampico, Veracruz, Coatzacoalcos, Campeche and Progreso) to provide easy access to international markets (New York, Havana and Europe) or with inland ports on the U.S. border (Nogales, Ciudad Juarez, Piedras Negras, Nuevo Laredo and El Paso) to provide access to markets in the U.S. 45. The late nineteenth century was Mexico's "Railway Age" 1/ and the importance of railroad development between 1870 and 1910 is reflected in many aspects of the country's subsequent macro-economic and spatial development. Above all, the decisions which were made by independent groups of railroad investors for purposes essentially connected with the short run aim of facilitating commodity exports, had unforeseen and unintended consequences - as will be shown later in this report. Table 10.2 shows that while network expansion initially proceeded at a faster pace than the growth of railroad activity - as reflected in passenger and cargo revenues - by 1901 the growth of traffic corresponded to the growth of the railnet itself. 46. The port sector figured more actively in Mexico's economic development before 1900 than it has done since that period. The two mountain ranges, which run down the length of the country, (Map 1) prevent easy access to the coast from the interior, and the emphasis of the rail net- work was on north-south rather than east-west connections. In 1900, two-thirds of the country's imports by volume were shipped across the U.S. border and about 75 percent of the remainder was brought in to Gulf coast ports, whose 1/ The only railroad built entirely after 1910 was the Baja California line, which was the longest in the country (540 km). It was built to connect Tijuana and Mexicali with Sonora, and was completed in 1947, when it facilitated direct access from the northwest peninsula to Mexico City by inland transportation for the first time.  -13 - relative importance vis-a-vis those of the Pacific reflected Mexico's traditional trade patterns with the United States, Europe and the Eastern seaboard of Central and South America. The land "ports" were, however, less important for export trade. Two-thirds of the total volume of exports in 1900 were shipped by sea.l/ 47. Some towns, such as San Luis Potosi, developed almost entirely on the basis of their role as railroad centers while the ports too depended on transport functions for much of their urban growth. Throughout the country, however, urban growth was predicated on a favorable location vis-a-vis the transport network. The Structure of Urban Development 48. By the early 20th century, a number of cities were beginning to develop specialized functions and, for the first time, the means existed to integrate some of them through an incomplete transport network - although the network was built for purposes other than that of facilitating inter-urban communications. 49. Whereas the sets of largest cities at the beginning and end of the "Porfiriato" were different from one another, the largest cities of 1900 were to remain the largest cities thereafter. The macro-economic and spatial changes which occurred during the Porfirian era had a lasting effect on the structure of the Mexican urban economy. 50. Between the beginning and the end of the "Porfiriato" there were many changes. The 25 largest cities in 1877 are shown in Table 1.20; the 25 largest cities in 1900 are shown in Table 1.21; a comparison between them is shown in Table 1.25. There are striking contrasts. First (Table 1.25(a)) only 14 out of 25 appeared for both years and several of those which were l/ In terms of valure, this amounted to less than 50 percent of the total value of commodity exports, since it consisted for the most part of such products as sugar, sulphur, minerals, cotton, corn and petroleum with low value/weight ratios.  - 14 - on both lists had negative growth rates in the intervening period, suggesting that they were not positively affected by the growth generating forces of intervening period - specifically transport improvements and the development of export oriented primary production. Among the cities which appeared in both 1877 and 1900 and which had positive growth rates, all but two (Salamanca and Puruandiro) were among the 37 cities which in 1970 had more than 100,000 inhabitants. Some of 1970's largest cities moreover were already large in relative terms in 1877 - notably Mexico City and Guadalajara - which then, as now, ranked first and second in the urban hierarchy. 51. The spatial structure of Mexico's 25 largest cities in 1877/1900 is shown in Map 8. All of the cities which were important in both years were in the center of the country, and all of the cities which declined in importance were in a belt stretching west-east from Jalisco to San Luis Potosi. With the exceptions of Pachuca and Orizaba, all of the cities which first achieved prominence in 1877/1900 were in northern Mexico.l/ All of the cities which declined in importance in this period were located in the vicinity of other cities which grew in importance. In the state of Jalisco, for example, there was a heavy concentration of "declining" cities but there was also Guadalajara - the second largest city in Mexico. In the state of Zacatecas, there was the city of Zacatecas and two "declining" cities. In the state of San Luis Potosi, the state capital was a growth center while three other cities declined. All of the growth 1/ The growth of Pachuca in the late 19th century was a consequence of a mining boom in the silver-rich areas around the city. The growth of Orizaba was attributable to its development as a textile town given (a) the availability of water power and (b) its location mid- way between the port of Veracruz,through which cotton for processing was imported, and the major market of Mexico City. Most of the capital investment in Orizaba was of French origin.  - 15 - cities in these states had good railnet connections. None of the declining centers did. On a regional scale, a process of inter-city competition facilitated by the development of the railnet was thus under way and those places which lost out in this process did not recover; whereas most of those which succeeded continued to succeed after the Revolution. C. Urban Development: 1910-40 52. Urban development in 1910-40 was not characterized by the kind of radical changes which occurred during the "Porfiriato" and from many points of view, the period between the Revolution and 1940 was essentially one of consolidation and institution building rather than new trends. To a large extent, therefore, the growth of cities emphasized the continued expansion of places which were already relatively large by 1910. In accounting for this pattern of relative continuity, the development of the transport sector and the development of the industrial and agricultural sectors were again the crucial determinants. The Dynamics of Urban Development (a) The Role of Transport Improvements 53. Empirical evidence from many countries suggests that up to a certain stage in the evolution of an urban system, transport developments are the major keys to urban growth. Thereafter, other factors become relatively more important. The evidence for Mexico suggeststhat by 1L940 this critical juncture had not yet been reached and that transport improvements were still the single most important influence on urban demographic and economic growth; this also applies to the "Porfiriato". In the earlier period, transport development referred almost exclusively to the railroads; after the Revolution, other modes, particularly roads, also began to be important, although the role of railroad development remained pre-eminent. (i) Railroads 54. During the first half of the 20th century, the railroad system was maintained at a roughly constant size (Map 19). Railroad operations declined during the Revolution  - 16 - but expanded thereafter as the availability of services continued to exert a major influence on selective urban growth. As interregional transport improvements became more conducive to large-scale manufacturing, specialized production in a limited number of cities was, to some extent, self-generating. Rate competition and freight volume economies became important agglomerating forces. These attracted new manufacturing establishments to favored cities and stimulated the expansion of existing plants, thereby diminishing the importance of less favored locations not served by the network. By and large, however, relative advantages and disadvantages conferred by railnet accessibility did not alter much after 1910. (ii) Roads 55. During the '"Porfiriato", the development of rail- roads and ports was closely related because the railnet was partly designed to provide access to foreign export markets via the seaports of either coast. Similarly, in the early phases of the development of the road netw.6rk, there was a close relationship between roads and railways, the configuration of the national road network being similar to that of the rail network. The major rail lines, which by 1910 stretched north and south-west from Mexico City, were complemented after 1920 by the construction of new roads. There was, however, no road from Mexico City to the southeast until the late 1950s nor were there major east-west connections in either the north or the south of the country until after 1940, and progress in general, was impressive only by comparison with the period before 1910. 56. Prior to the Revolution there was no such thing as a national highway policy. During the 1920s a road construction strategy began to emerge and the Federal Government established the National Roads Commission to construct and maintain roads and introduced a gas tax system to generate revenues. 57. The basis of the road network strategy was to develop regional roads as links between major urban centers and Mexico City. In the 1920s and 30s road construction was slow because resources were limited, and traffic demand restricted. Only 700 km of all-weather roads, most of them in the vicinity of Mexico City, were in service in 1928. But during the six-year period, 1928-34, the Federal Govern- ment extended the network more than sixfold - to 4,260 km. Less than 30 percent (1,186 km) of the total length was paved, 1,291 km was surface treated, and 1,768 km was gravel (Table 10.8). The first inter-regional road connected the main ports of the Pacific with the Gulf of Mexico through Mexico City. The Mexico City-Acapulco road was completed in 1930 and the Veracruz road was finished three years later. Road  - 17 - connections were also built from other regional cities to the nearest ports, such as those from Monterrey to Nuevo Laredo and Merida to Progreso (Map 20). 58. By 1940, the road network extended almost 10,000 km and connected Mexico City with most of the country's urban centers. A major road had been built from the capital to Nuevo Laredo and other highways had been extended between Mexico City and (a) Guadalajara and Tepic, (b) Aguascalientes and Zacatecas and (c) Oaxaca. Even so, only 33 of the 50 largest cities of 1940 were linked by the national road network. With respect to regional distribution, Map 20 suggests that the northwest and the southeast were at a strong disadvantage relative to other areas in 1940 and that most of the large cities which lacked connections with the national road network, were located in these regions. All but two of the 50 largest cities had railnet connections and neither lacked :both rail and road network linkages. The road network development supported the growth of cities which had already benefitted from railnet development and road transport did not, before 1940, have an indeperndent influence on the urban system but tended to reinforce the effects of the railnet. (iii) Ports 59. Data for cabotage activity in individual ports for 1930 are not available but data on total arrivals and depar- tures (Table 10.18) show that from 1900 to 1929 the Gulf coast handled almost 70 percent of inbound and outbound ship movements and the Pacific coast ports, the other third. 60. In 1940 (see Table 10.18) about a quarter of all cabotage traffic was handled through Veracruz and another quarter through Coatzacoalcos where two-thirds of the total volume consisted of exports, Tampico also being primarily an export port (mostly oil and related products). Passen- ger services were relatively unimportant except in Veracruz. 61. Data for 1940 are available for only two Pacific ports but show that Manzanillo handled less than 5 percent of the nation's cargo traffic; imports made up 80 percent of the traffic while passenger movements were negligible. Mazatlan handled even less, two-thirds of the total being imports. 62. Although they were less significant than railroads or highways in the transport system, the ports provided the  - 18 - major mode of transportation to such important centers as Acapulco, La Paz, Ensenada, Campeche and Progreso (Merida). Moreover, maritime transport continued to be the only means of communication to certain areas, notably the Yucatan Peninsula. (iv) Conclusions 63. Compared with the massive transport improvements of the Porfirian period, the essential characteristics of the transport net from 1910-40 remained unchanged. Highway modes, particularly for long distance movements, were not seriously developed until after 1940. The primary roadnet was very similar to the railnet and did not modify the basic network very much. The rail/road network in 1940 emphasized the relative ease of north-south accessibility and the diffi- culties of east-west movement, reflecting both physical geography and the inertia resulting from the construction of railroads during the "Porfiriato". 64. In 1940, three major network zones were clearly distinguishable. First, in the north-central, north-east and central regions, there was a fairly comprehensive network of north-south and east-west routes, with a strong focus on Mexico City. Second, in the northwest, Baja California and Sonoro remained isolated from the rest of the country by the Sierra Madre Occidental. Third, the south-east, in particular the Yucatan peninsula, was also isolated until 1938, when the Yucatan railroad system was linked with the Mexico Central network. 1/ 65. During the "Porfiriato", railroad development played a major role in selective urban growth and contri- buted to the rise of some cities and to the decline of others. The evidence concerning changes in relative urban accessibility in 1910-40 strongly suggests that cities which lacked railroad connections at the beginning of the period suffered by comparison with those located on railnets. Prior to the Revolution, when most cities were centers of tertiary activity, differential accessibility gave those 1/ The Sonora-Baja California system was not linked until 1942.  - 19 - which were relatively well-off in this respect a decisive "edge" over others. Relative accessibility helps to explain contrasts in urban size and urban growth and also to explain the process of the increasing concentration of the urban population. Therefore, Mexico City (above all) and Guadalajara, Puebla, Monterrey, and such smaller cities as Aguascalientes, were relatively accessible and grew rapidly. 66. Cities not enjoying easy accessibility tended to grow less quickly, and to lose relative population-size status. For example, Oaxaca had 35,000 inhabitants in 1900 and a thriving urban economy. Aguascalientes was about the same size and also had a thriving urban economy. By 1940, Oaxaca had a population of only 29,000, whereas that of Aguascalientes had increased to 104,000. There were, of course, other (for the most part local) factors which influenced the growth of these cities, but accessibility was crucial. 67. The composition of the set of the 25 largest cities in 1910 and 1940 (Table 1.39) shows that all of the cities which were in the 1910 list but not in the later one (Celaya, Colima, Jalapa, San Francisco del Rincon) either (a) lacked railnet connections (San Francisco), (b) were located on relatively unimportant railroads (Oaxaca), or (c) were close to other cities with which they could not successfully compete (e.g. Colima with. respect to Mazatlan and Jalapa with respect to Puebla). 68. Other cities, which by 1910 were easily accessible (such as Monterrey, Torreon, Chihuahua, Tampico, Culiacan and Mazatlan) experienced rapid growth in 1910-40 as economic policies began to affect levels of activity in areas which were fairly distant from the major market.s of the center but were well connected by the railnet. 69. Transportation improvements and the momentum derived from earlier improvements during the "Porfiriato" had an important effect on urban development in 1910-40. As the system improved, cities located at strategic nodes began to exploit the.ir comparative advantages and to become regionally dominant centers. Clearly the urban structure was shaped by the interaction between relative accessibility and other aspects of economic activity.  - 20 - (b) Industrial Development 70. Economic growth during the revolutionary period, particularly during 1915-16, when armed intervention from abroad was designed to frustrate the Revolution, was inevitably halted (Table 2.1). In spite of contradictory data it appears that GDP rose only slightly from 1910 to 1921, (there are no data for intervening years), that manufacturing output probably declined less steeply, and that mining suffered a precipitate collapse. 1/ Only the petroleum sector, operating in heavily armed, foreign-owned, enclaves, appears to have grown in response to the surge of external demand created by the First World War. Because the profits of the petroleum sector were spent outside Mexico, this was not of much benefit to the economy. The major domestic impact of the growth of the petroleum industry was that resistance to the Revolution was partly financed from its profits. 71. Because of the vigorous foreign antipathy to the Revolution and the consequent non-availability of foreign loans, the lack of physical security, and the collapse of the banking system, there was virtually no new industrial investment before 1921. In terms of production and invest- ment, the revolutionary period was characterized by economic stagnation and rapid inflation. 72. Manufacturing expanded during the second half of the 1920s because of the search for higher profits which stimulated the transfer of Mexican capital from the rural to the urban sector. External trade was hampered by strained relations with some of the major industrialized countries. Insofar as Mexico was not unduly dependent on the external sector, the effects of the world economic depression were relatively short-lived. The collapse of many export markets, 1/ According to Solis (1970) the sectoral product fell from 1,537 million pesos to 917 million pesos between 1910 and 1921. Gold output declined from 41,400 kgs to 7,300 kgs, silver output from 2,400 tons to 1,200 tons, and lead output from 124 tons to 5.7 tons during the period.  - 21 - however, induced a recession in mining, petroleum and commercial agriculture and an associated decline in domestic incomes and demand. 73. Recovery began in 1933. Although rapid growth was not sustained, the 1930s were generally associated with expansion. By 1939, the manufacturing sector was set on a course of steady growth and other sectors had more than recovered their pre--depression status. A comparison of real GDP in 1933 and 1939 shows an average annual growth rate of 5.6 percent.. By 1939 the economy in general and the industrial sector in particular were ready to profit from the preparations for war and to begin a period of sustained growth which had few parallels in the developing world. 74. Foreign antipathy to post-Revolutionary Mexico was one factor in accounting for the diminished size of the external sector, but: also important was the reversal of the Porfirian emphasis on external trade, and the drive to develop a domestic market oi sufficient size to absorb the output of domestic industry. This too had a significant impact on the nature of indust:rial and urban development. 75. The consequences of "internalizing" the development process affected the existing urban structure. The large market in the center of the country, focusing on Mexico City, was the key to this process, although secondary markets were also significant. The diminishing importance of external trade weas paralleled by a decline in the importance of coastal cities, such as Veracruz. 76. The improved transport system facilitated better access to internal markets, thereby favoring those cities which had the largest market potentials for the location of new manufacturing activities. A more "open" economic policy permitting imports could have resulted in the slower growth of domestic manufacturing, and thus in the slower economic growth of certain towns and cities, parti- cularly those at the center. This could well have produced relatively less contrast between the urban growth rates of the center and those of the periphery, and a less direct relationship between market potential, urban size growth and urban industrial development.  - 22 - (c) Agricultural Development 77. The impact of the Revolution on the rural sector was lasting and enormous. During the Revolution foraging armies seized cattle and crops. The general mobilization of the countryside meant that agricultural labor was scarce and the production of some staple foodstuffs declined from the already inadequate levels of the late Porfirian period. At first the hacienda system was not destroyed, but it was badly damaged, and this implied the transfer of both labor and capital from the rural to the urban sector. The move- ment of rural labor was a new trend but the movement of capital represented an extension of Profirian procedures. 78. The movement of labor and capital to towns and cities was offset by the strengthening of the ejido system, particularly during the Cardenas administration (1934-40) when a massive program of land reform was implemented. This program was based on Villa's Land Reform Law of 1915 and on Article 127 of the Queretaro Constitution of 1917, which dealt, inter alia with land tenure. In 1930, the rural communities which had been systematically dispossessed during the Porfiriato, held only 13 percent of total crop- land. A decade later their share had risen to 47 percent, 1/ and the ejidos contained almost half the rural population. 2/ Within six years, 10.2 percent of the country's continental area was designated for eventual redistribution. The latifundia system was destroyed twice as quickly as it had been created under Porfirio Diaz. The available evidence is inconclusive, but the process of accelerated land redistribution does not appear to have induced a decline in agricultural output. 1/ Wilkie: 1971, p. 76. 2/ The ejido was effectively recreated by the Revolution, though its antecedents stem from pre-Columbian times. It is a form of collective land tenure based on usufruct.  - 23 - 79. One important effect of the ejido system was to stem the tide of rural-urban migration, if only temporarily. Many who had already migrated to cities returned to the country-side. The development of large-scale commercial agriculture in the northwest in the 1930s also encouraged a movement back to the country-side. During the "Porfiriato", a docile, cheap labor-force had been mobilized to work on haciendas by means of the destruction of the ejidos. With the strengthening of the ejido system in the post-Revolutionary period the trend of rural-urban migration was abated. 80. The development policies of the post-Revolutionary era stressed the needs of the rural and urban sectors. The drive towards industrialization which began in the 1.920s inevitably had an urban image, and the drive towards agri- cultural development, particularly after 1930, was a response to production requirements and a device to distribute wealth and welfare more equitably among the population, which had the not necessarily intentional effect of linking the rural and urban economies. The sectoral thrust of economic policy after 1920 implied urbanization and the growth of rural/urban interdependency. The Structure of Urban Development (a) Urban Population Growth 81. Mexico's population in 1940 was 20 million, as compared to 13.6 million in 1900; this implies an average annual rate of increase of 1.0 percent (Table 1.1). On a decennial basis, the rate of change varied from a growth rate of 1.1 percent in 1900/10 to 1.7 percent in 19:30/40, with a negative rate--associated with the fact that the Revolution claimed about 2.3 million lives--in the revolutionary decade 1910/20. The overall mortality rate from 1900 through 1940 declined as the death rate fell from 34/1000 in 1900 to 23/1000 in 1940. The birth rate from 1900 to 1920 was roughly stable at 45/1000 but had risen to 48/1000 by 1940. 82. The measurement of urban population growth is complicated by conceptual and methodological issues con- cerning the use of urban data. (See Technical Annex 4). The recorded urban population, defined in terms of a threshold of 2,500 inhabitants, increased from 4.3 million in 1910 to 6.9 million in 1940 and thus defined, the level of urbanization increased only slightly between 1910 and 1940. However, alternative "thresholds", (Table 1.3) show that  - 24 - urbanization began to accelerate well before 1940. This is most conspicuously true of population clusters larger than 50,000, but even on the basis of "thresholds" of 10,000 or 20,000, the level of urbanization was clearly greater in 1940 than in 1910. (b) Urban Size Distribution 83. The pattern of urban agglomerations reflected a clear contrast between the relative rates of growth of larger and smaller cities. 84. At the beginning of the 20th century, Mexico had only two cities with more than 100,000 inhabitants, and only seven with more than 50,000. The total number of urban places (on a "threshold" of 2,500) was less than 600. Fewer than 7.0 percent of the total population of the country lived in cities of 50,000 or more (Table 1.4). 85. By 1940, there were four cities with more than 100,000 people and 13 with more than 50,000 (Table 1.5); these contained almost 14 percent of the national population. The average size of cities with more than 100,000 inhabitants increased from 295,000 in 1910 to 500,000 in 1940. 86. These data show that the population was becoming more urbanized and that the population as a whole and the urban population were increasingly concentrated in large cities. Table 1.51 illustrates the increasing importance of Mexico City and the increasing "primacy" of the urban structure. Table 1.51 shows that the size distribution of the largest cities in 1900 was such that the indices of primacy ranged from 41.1 (2 cities) to 0.4 (25 cities), by 1940 these indices had increased to 6.6 (2 cities) and 0.7 (25 cities), resulting in an increasingly large "gap" between Mexico City and other urban centers. (c) Hierarchical Stability 87. The nature of urban growth from 1910/40, implies that the urban hierarchy would have been rather unstable. Although there was a trend towards increased concentration, which meant that many of the same cities became larger  - 25 - over time, a comparison of the membership of the hierarchy in 1910 in terms of the 25 largest places reveals a changing composition of cities. 1/ 88. When assessed in terms of the ten largest cities in each year, the set was relatively more consistent; this was true to an even greater extent in respect of the five largest cities (Tables 1.40 and 1.41). 89. The rank order correlations of the sets of largest cities were low (Table 1.44). Consistent, however, with increasing concentration in the very largest cities, the coefficients of rank-order correlation for the sets of 10 and 5 largest cities in 1910/40, were much higher than those for the set of the 25 largest cities. 90. While most of the largest cities at the beginning of the period (Mexico City, Guadalajara, Monterrey, Puebla and Leon) were also among the largest cities in 1940, some of the relatively smaller cities failed to grow and even some of those which were relatively large failed to maintain their relative importance. There was not,therefore,a close correlation between initial size and subsequent growth among the 25 largest cities of 1910. This was inevitably reflected in the fact that the composition of the 25 largest cities in 1910 had also chan,ed by 1940 (Table 1.39). However, (Table 1.26) the largest places grew faster than the others and Mexico City grew fastest of all. The coefficient of correla- tion between the ranks of the largest cities in 1900, and compound average growth rates for the same cities in 1900/40 was relatively low, whereas the coefficient of correlation between the size of the 25 largest cities in 1940 and their compaind growth rates in 1900/40 was higher. This reflected the dynamism of the- "new" cities which began to grow after the Revolution. 1/ There was, however, a curious pattern of membership in 1920 and 1930 when fewer of the 25 largest places were included in the corresponding sets for these years than in 1940 (Table 1.42). One possible explanation is that. some places grew with unusual speed during and after the Revolution but did not sustain rapid growth rates after 1930; whereas other cities, which had been relatively larger in 1900, grew at more steady rates throughout the period.. Such an interpretation is consistent with the intervention of ephemeral determinants of population growth such as war-induced migration.  - 26 - 2. THE DYNAMICS OF URBAN DEVELOPMENT: 1940-70 A. Introduction 91. A declining mortality rate underlies Mexico's population growth from 20.2 million in 1940 to 50.4 million in 1970, 1/ reflecting a change in the rate of growth from an annual rate of 2.6 percent in 1940/50 to 3.4 percent in 1960/70. 92. In terms of an urban "threshold" of 2,500 (i.e., the population living in communities with at least 2,500 inhabitants), the level of urbanization rose from 35 percent in 1940 to about 58 percent in 1970 (Table 1.7). During this period, Mexico changed from a rural to an urban society. 93. The rate of urbanization (i.e., the rate at which the urban population increased as a percentage of the total population), accelerated rapidly after 1940 but slowed after 1950. From 1940 to 1970, the rate of urbanization was faster than at any time before 1940 (Table 1.6). The rate of national population growth and the rate of urban population growth have gradually converged. The acceleration of the national rate coincided with the deceleration of the urban rate during 1950-70. While urban growth has continued to exceed national growth by a substantial margin (4.5 percent to 3.4 percent in 1960/70), urban population shows a declining share in national population growth (85 percent in 1940/50 and72 percent in 1960/70). 94. The process of urbanization in Mexico after 1940 did not run parallel to the country's pattern of economic growth in terms of annual fluctuations or even ten-year fluctuations. The acceleration of urban development did, nonetheless, coincide with the beginnings of sustained economic growth. Different in kind from the earlier European and American urbanization experiences, urbanization 1/ The death rate fell from 23.2/1,000 in 1940 to 7.8/1,000 in 1970. The birth rate also fell, but more slowly, from 48.1/1,000 in 1940 to 43.1/1,000 in 1970.  Figure 2-1 Growth of Urban Population, GDP and Sectors: 1940-1970 LEFT SCALE '000 MEX. RIGHT SCALE PESOS IN URBAN CONSTANT POPULATION (1960) PRICES (MILLIONS) 400 - 300 - 200 - * 100~~~~~~~~~~~~~~~~~~~~~~~~~~~0 80~~~~~~~~~~~~~~~~~~~~~~~~~~~~00 100 - G9.-' 60~~~~~~~~~~~~~~~0 2C _ _ 0,30 40~~~~~~~~~~~~~~~~~~~1 t _ I ~~~~~~~~I _ 1940 1950 1960 1970 World Bank-15571  - 28 - in Mexico was clearly related both to industrialization and developments in the rural sector, but which took place alongside unprecedented population growth. 95. The first correlate of urbanization and economic growth derives from the increase in income and the shift of consumer preferences towards non-agricultural goods. While demand for foodstuffs becomes increasingly inelastic, demand for manufactured goods and services tends to increase. The demand increases for products which can be most feasibly produced in urban agglomerations where scale economies, transfer cost reductions, intersectoral linkages and a wide array of externalities are uniquely available. The city provides a "natural" environment for innovation and technological progress, whether of original or adaptive nature. Industrialization and urbanization are linked by necessity and logic, both of which are supported by theoretical and empirical explanations. Rapid urbanization in Mexico was a consequence of rapid industrialization after 1940. Certainly there is a close historical link between the two (Figure 2.1). The Mexican experience of urban- industrial growth was in many ways synonymous with economic development as elucidated by Clark (1940) and Kuznets (1958). 96. In the United States and Europe in the nineteenth and earlier twentieth centuries urban-industrial development was characterized by relatively labor-intensive production methods in the early stages; the parallel growth of output and employment in the agricultural sector; and moderate population increase. Urbanization in Mexico in this later period occurred under differing conditions. The economy was unable to absorb the labor force growth arising from rapid population growth because (1) agricultural output did not keep pace with population growth and (2) industrial development occurred within an environment of capital- intensive technology. 97. The determinants and structure of urban growth are conceived in the development of the secondary and tertiary sectors. The conclusions about the relative importance of these factors in determining the course of urban growth differ significantly from those referring to urban development up to 1940.  - 29 - B. Transport Development 98. This section describes the development of the transport system after 1940 and interprets its significance in terms of urban development. We shall examine each mode of transport - rail, road and air - and then analyze the urban implications. The Railways 99. After 1940 the railway system was shortened (from 23,000 km in 1940 to 19,900 km in 1970) while its equipment and operation were significantly improved. Steam locomotives were phased out in favor of diesels, all tracks were! laid to a standard gauge, telecommunication systems were modern- ized and the number of freightcars was increased by more than 10 percent. 100. The most :important change was the unification, in 1960, of almost all railroad companies (except those of the Sonora-Baja California and Southern railroads) into the National Railroad of Mexico. The new entity became responsible for more than 80 percent of the total railnet. Because development in the railroad sector after 1940 did not involve the construction of new lines, there were no significant changes in the spatial distribution of railroad facilities compared with the prior period (Map 4). 101. Freight traffic, which increased at an average rate of 3.3 percent per annum from 1940 to 1960 and reached over 9.5 billion tori-km in 1959, was increasingly confined to bulk (agricultural, mineral and forest) commodities. Passenger traffic increased at 2.3 percent a year and rose to a total of seven billion passengers/km in 1960. 102. The decline in demand for rail freight and passenger service was closely linked to the development of highways and commercial aviation after 1940. The rail system did not clearly define functions in relation to these competing modes of transportation. 103. With respect to freight traffic, the status of the railroads was diminished by highway and pipeline development. A 1951 Eximbank survey reported that in all but one instance where an all-weather'highway had been built parallel to a rail line after 1946, railroad freight traffic had either leveled off or diminished the year the highway was put into service. During the five-year period 1946-50, railroad freight activity grew by only 15 percent - a much lower rate of growth than in the previous five-year period when freight traffic had risen by 25 percent.  - 30 104. After 1960, the rail system began to find a new role. Freight-ton kilometers increased from 13 million in 1961 to 18 million in 1966. By 1960 the railroads were carrying more than 20 million ton kilometers a year, but passenger services increased by only 35 percent - from 30 million passenger kilometers in 1966 to 46 million passenger kilometers in 1971. 105. There are no data on interregional railroad flows before 1970 but the inter-urban origin and destination data for 1970 (Table 10.7) reveal that railroad movements were relatively well balanced throughout the country (unlike road or air movements) and that railroad traffic was not dominated by the major metropolitan centers. In 1970, the entire system carried over 49 million tons of freight of which only 7.7 percent (3.8 million freight tons) terminated in the Federal District. Monterrey received 1.8 million tons and Guadalajara about .9 million tons (less than 2 percent of the national total). 106. The pattern also suggests that by 1970, railroads were mainly used for the shipment of bulky commodities over long distances. Most incoming freight to large urban centers, such as the Federal District,originated in remote industrial towns, or in major ports. 1/ Ten of the 49 cities shown in the matrix generate more than 77 percent of the total traffic (2.9 million tons out of a total 3.8 million tons). 107. There were heavy flows between medium sized industrial cities such a Monclova, Ciudad Obregon, Tampico, Torreon and Toluca, some of which were located close to metropolitan cities and thus served as centers where bulky commodities were broken down before being transported to the major metropolitan market. 1/ The Federal District received the largest amounts of freight from Veracruz (about 800,000 tons), Nuevo Laredo (about 580,000 tons), Monclova (about 350,000 tons), Matamoros (205,000 tons), Ciudad Juarez (135,000 tons), Ciudad Obregon (241,000 tons) and Progreso (175,000 tons).  - 31 - The Roads 108. During the 1940s and 1950s the largest portions of road investment were allocated to new inter-urban roads and to physical improvements (pavement, realignment and safety). The need for improvements in the original trunk system which was built in the early thirties arose from the increased volume of traffic and the improvement of vehicles after 1950. 109. By the 1950s there were more than 20,000 km of all weather roads in Mexico and road transport services reached 92 percent of all urban places. About 15,000 km of the total network was equipped with two paved lanes and 3,000 km was surface-treated. Network development generally emphasized the trends that already existed featuring new north-south roads from Ciudad Juarez to Mexico City and from Nogales to Mexico City and a trunk road between Mexico City and the Guatemala border through Tuxtla Gutierrez thereby connecting the northern and southern borders of the country. Another important achievement was the completion of a new highway between Villahermosa and Yucatan which connected the capital with the southeast. A third development was the Saltillo- Zacatecas route which significantly reduced mileage between these cities and facilitated movement between the three largest metropolitan centers. 110. In the north, an east-west route was deve]oped from Matamoros to Durango, through Reynosa, Monterre!y and Torreon, and was extended to Mazatlan on the Pacific, Coast. In the south, another traverse route across the Isthmus of Tehuantepec was completed in 1960. The three major highways from Mexico City to the northern border, the principal traverse roads and many local branch roads followed the routes of existing railroads,notwithstanding that rational highway policy was supposed to stress building roads to locations which were served inadequately, or not at all, by the railnet. By 1950, consequently, a number of major cities were still not integrated into the national road network. 111. Road mileage grew throughout the sixties. In 1960, 45,000 km of roads were in service. By 1970, 72,000 km of roads were in use (7,000 km were dirt, 20,000 km tarred, and 42,000 km paved). The road density index of the Central region increased from 61.6 km per 1,000 km in 1960 to 100 km per 1,000 km2 in 1970. The index in the north-central and northwest regions for 1970 was five times above the national average. The southern region experienced the largest gain in the density index (49.9 km per .1,000 km2 in 1970 compared to 24.6 km per 1,000 km2 in 1960).  - 32 - 112. Toll roads were introduced in the 1960s in response to the rapid growth of traffic demand in the metropolitan region. By 1970 more than 1,000 km of toll roads had been developed in the Mexico City region. 113. The Government continued its efforts to improve inter-regional highway conditions during the 1960s. In 1966-73 for example, auto-operation time improved by 10 to 25 percent. Major improvements were achieved in corridors to Mexico City (Table 10.14) while transverse movement across the north-central region was significantly improved (Tables 10.14 and 10.15). The time-distance values show that on the Tijuana-Mexico City road, for example, operating time decreased by 6 - 12 km/hr, (a reduction of about five hours on a bus journey of 45 hours). Driving time on the Mexico City-Ciudad Juarez road decreased by more than two hours as did that on the road to Nuevo Laredo, while driving time between Mexico City and Merida was reduced from 28 hours 55 minutes in 1960, to 24 hours in 1970. Air Transport 114. Of great importance to urban integration in a country of the size and physical condition of Mexico is the development of the air transport network which by 1970 covered the entire country. The system was serviced by two trunk carriers (Aeromexico and Mexicana de Aviacion) and several feeder lines. 1/ 115. In the early 1940s the air transport system was entirely maintained by private airlines providing services between Mexico City and other cities. By 1946 five airports had been built by the Federal Government (Mexico City, Guadalajara, Tijuana, Acapulco and Mazatlan) and eleven by private airlines (including those at Merida, Tampico, Monterrey and Nuevo Laredo). However, in order to meet 1/ Most of the feeder lines are under the control of Aeromexico as are the two secondary lines, Aerocarga, S.A. and Servicios Aereos Especiales, S.A. de C.V. (S.A.E.). Aerocarga is destined to become the country's national air cargo carrier whilst S.A.E., which provides secondary scheduled services within Mexico, is intended to become the principal Mexican charter line.  - 33 - the rapid growth of demand and to provide equitable accessibility throughout the country, the Government adopted a civil air transport policy in 1947. Several new airports were built in the early 1950s and a number of major improvement and rehabilitation projects were implemented. By 1951 there were 15 airports equipped with navigation aids, ground support and weather service facilities, and some 365 landing fields. 116. Air passenger kilometers increased almost seven times from 39,000 in 1940 to 308,000 in 1950 and more than doubled (to 625 million) by 1960 (see Table 10.21). The 1960s saw an almost: fourfold additional increase. By 1965, passenger traffic had risen to 2.8 billion p/km. Air cargo traffic also grew rapidly -, from less than half a million tons in 1960 to over three quarters of a million tons in 1965. 117. Since the mid-60s, air transport has become even more important, linking urban centers, tourist resorts and industrial cities. In 1969, more than 4 million passengers were carried and the number of passenger-kilometers rose to about 6.8 billion while the average flight length increased from 570 km in 1965 to 840 km in 1969. Cargo traffic exceeded 1 billion ton/kilometers in 1969 and has since increased at an annual rate of .10 to 15 percent. 118. The regional and interregional air traffic patterns for 1967 and 1973 show that in 1967, Mexico City clearly dominated the national pattern, receiving and generating more than half of all domestic passengers. Consequently, development of other interregional routes was limited in spite of a well developed schedule of inter-city services. By 1973, the dominance of Mexico City had been somewhat reduced, and a number of important interregional routes which did not pass through Mexico City had been developed. Regions outside Central Mexico significantly increeased scheduled services after 1967, especially in the southeast. Transport and Urban Development 119. A simple measure of the relationship between network development and city development was based on a comparative nearest-neighbor analysis of the urban system in 1940 and 1970. 120. This showed that there was, after 1940, relatively little change in the spatial distribution of major cities  - 34 - (Table 10.28). Between 1900 and 1940 there was a marked change in this pattern such that, in terms of the largest cities in the urban system in each year, the average distances between each major city and its nearest neighbor rose from 163 km (1900) to 196 km (1940); the change from 1940 to 1970 was only marginal to 202 km. 121. Given the area of the country (1.9 million sq. km) the hypothetical mean distance separating 25 cities would be 280 km. Thus, in 1900, the actual mean separation distance was equivalent to only 57 percent of the hypothetical distance. By 1940 the percentage had risen to 69.8; by 1970, to 71.9 percent. 122. This suggests that network changes had a much smaller impact on the urban system after 1940, and that, notwithstanding the increased efficiency of existing modes and the intensive development of the air network, the major cities were not brought that much closer to one another than they already had been in 1940. 123. Maps 18-22 show that there has been little development in relative connectivity between transport modes after 1940. The major advantage of a location near the center of the transport network is the obvious one that centrality facilitates access to the rest of the country (see Annex E). This suggests the desirability of locating cities near the center of the transport network as one means of encouraging decentralization. 124. An important indicator of network accessibility is the frequency or efficiency of service and the conditions of routes. This is captured by comparing physical with real 1/ distances, given that accessibility between two points is not necessarily a direct function of physical separation and that interregional connectivity does not depend on the mere existence of a network link. 125. A recent study (CEPAL, 1974), shows coefficients between real and physical distances. Physical distance was given the value of 1 and was used as a base to which time and cost elements were added. A value of 1.20 implies that real distance is 20 percent greater than physical distance, i.e., if physical distance is 100 km the real length is thus 120 km. Real distances were computed between major urban centers along trunk routes or road junctions near large cities. Connectivity between regions was interpreted as connectivity between major urban centers. 1/ See Annex D.  - 35 - 126. Examples of the differences between real and physical distances are shown below: Physical Real Length Length Coefficient 1. Acapulco-Mexico City 419 km 611 km 1.46 San Luis Potosi-Mexico City 417 km 433 km 1.04 2. Oaxaca-Mexico City 534 km 754 km 1.41 Guadalajara-Mexico City 581 km 609 km 1.05 127. In the first example the physical distance between Acapulco and Mexico City is virtually the same. When road condition variables are taken into account, however, the distance from Mexico City to each point is very different. The distance to San Luis Potosi is increased by 16 kin, and that to Acapulco by 192 km. In the second example, although in physical terms, Oaxaca is 47 km closer to Mexico City than Guadalajara, in terms of real distance it is 145 km farther away. 128. Annex E shows that accessibility to the Federal District in terms of physical versus real distance in 1970 was excellent to fair in most cases. The Federal District had high coefficients with points to the north such as Queretaro, San Luis Potosi, Guanajuato, Pachuca and Leon. There was also excellent accessibility between Guadalajara and Mexico City (a coefficient of 1.05). Good accessibility existed between the Federal District and the northwest (an average coefficient of 1.10) and the northeast (an average coefficient of 1.09 except in coastal areas). There was fair accessibility between the Federal District and the Southeast (coefficients of 1.19 to Veracruz, 1.15 to Merida). Accessibility between the Federal District and the southwest was only fair (Acapulco 1.46 and Oaxaca 1.41). 129. Monterrey had good to fair accessibility with most other places. Real distances between Monterrey and such neighboring towns as Saltillo, Torreon and Nuevo Laredo were moderate and there was fair connectivity with points in the northwest such as Ciudad Obregon and Tijuana (coefficients of 1.17 and 1.15 respectively). There was poor connectivity between Monterrey and Guadalajara (a coefficient of 1.28),which implies that conditions are not favorable to the formation of strong inter-city flows between these major metropolitan centers. The coefficients  - 36 - between Monterrey and points south of Guadalajara along the Pacific coast were unrecorded but because road conditions were poor it is assumed that accessibility was low. 130. By comparison with Monterrey and Mexico City, Guadalajara had a lower level of accessibility in terms of real versus physical distance, the coefficients being consistently fair to poor except in points to the east - such as Mexico City which underlies the existence of large traffic flows between the central region and the Southwest. 131. This clearly emphasizes that differential transportation advantages were associated with differential urban growth and particularly with the growth of the three largest cities. (See Annex D for information on other cities). The measure of nodal/network relationships further supports this conclusion. This concerns relative accessibility between major cities and provides a means of assessing the relationship between changes in relative accessibility and relative urban development in 1940-70. The analysis is based on an index of relative accessibility between 25 of the cities which dominated the urban system. The use of 25 rather than 37 cities is explained by the extreme proximity in 1940 and 1970, of the 37 cities (e.g., Mexico City and Toluca in macrogeographic terms were in the "same" location). 132. The cities were selected to ensure a minimum of 100 km between them and it was assumed that each mode operated under normal conditions - which implied different assumptions about average modal speeds and efficiency. Different modal weights were used to reflect qualitative and quantitative contrasts in transport conditions. 133. The results of the analysis presented in Annex E show that the cities with the highest levels of accessibility in 1940 grew more rapidly than others in 1940-70. Thus, within a general environment in which spatial connectivity was greatly improved, the places which improved most in relative as well as absolute terms, were those with the highest degrees of initial transport advantage at the beginning of the period. The analysis also showed that there was a close relationship between relative accessibility and relative urban size and urban-economic growth.  - 37 - C. Industrial Development 134. This section relates the growth and structural characteristics of the industrial sector to the process and structure of urban development after 1940. General Aspects of the Development and Structure of the Industrial Sector 135. Industry accounted 1/ for 26.8 percent of GDP in 1940 and 34.0 percent, in 1970, with substantial contrasts in sectoral and subsectoral performance from one decade to another. In general, the manufacturing, construction and power sectors experienced rapid and sustained growth while the mining sector grew less quickly. Industrial expansion was, over all, faster towards the end of the period than in earlier years. 136. The expansion of industrial employment was slower than the rate of growth of output (4.1 percent versus 7.6 percent) and the rate of growth of the urban labor force. Economic policies after 1940 generally implied that industrial development and diversification were considered prerequisites to the achievement of both rapid economic expansion and the provision of jobs for a growing population. 137. World War II provided fortuitiously advantageous conditions for industrial growth. Government policies stimulated industrialization by emphasizing tariff barriers, import licensing, import substitution and the strategic allocation of public investment. Infrastructural and other developments of the 1930s established a basis for rapid industrialization. By 1940, Mexico was poised for "take-off". 138. Although "new and necessary" industries were clearly defined in successive legislative measures during the 1940s, balance of payments difficulties necessit:ated the introduction of import permits, especially for luxury goods, in 1947. Thereafter, the proportion of total imports 1/ Includes manufacturing, mining, power and construction.  - 38 - requiring licenses increased from 28 percent in 1956 to 65 percent in 1964. Tariffs were introduced resulting in a gradual increase in the proportion of goods subject to tariffs and in the nominal amount of the average tariff. 139. Against this protectionist background, and the growth of demand, industrial output increased by 9.4 percent a year from 1940 to 1945, the major growth sub- sectors being food, textiles, iron and steel, cement, paper and pulp, and chemicals. Postwar growth was slower, the average growth rate from 1940 to the early 1950s being less than 6.0 percent. After 1950, changing subsectoral emphases were associated with rapid progress in the steel, metal products and chemical subsectors, and with the general expansion of producer rather than consumer goods. 140. The increasing importance of producer goods industries is consistent with the pattern described by Hoffman (1958), who showed that producer goods tend to account for an increasing share of manufacturing product as the manufacturing sector develops. This conclusion is also supported by the United Nations (1954) in terms of the European experience and by Chenery (1960) who found that sectoral changes in the composition of manufacturing output were associated with rising per capita incomes and increasing national population size. 141. Differential rates of growth and consequent changes in the subsectoral structure of manufacturing in 1940/60 are illustrated in Table 5.2. This shows that seven subsectors grew at less than the average rate for the sector as a whole and that as a result, their combined share of total manufacturing output fell from more than 75 percent of the total in 1940, to less than half in 1960. Among these subsectors, however, food and textiles still accounted, in 1960, for more than 50 percent of total manufacturing employment and they contributed more than any other subsectors to the total growth of manufacturing, although their growth rates were below the average for the sector as a whole. Apparel, which ranked fourth in terms of its share of employment in 1960, was also in this category, but nevertheless made the sixth largest contribu- tion to the total increase in manufacturing employment. The only absolute decline was in the machinery subsector although this was explained by a decrease in employment in artisan-type machine shops.  - 39 - 142. The leading growth industries in the 40s and 50s were chemicals, prirnary metals, transport equipment and electrical equipment, although almost half of the total growth of output in the transport equipment sector came from auto repair shops and the data for the 1960s show that these trends continued through 1970. 143. Table 5.3 uses 1970 industrial census data to show the percentage distribution of output and employment by subsector, and indicates that consumer goods accounted for more than half the total value of output in 1970, and that foodstuffs and apparel continued to be important in terms of employment, while capital and intermediate goods subsectors continued to be more dynamic. Industrial Development and Urban Development (a) Preamble 144. A comprehensive review of the economic rat:ionale of urban-industrial development after 1940 is beyond the scope of this report. However, some of the general factors underlying the relationship between urban and industrial development - in terms of what may be broadly termed agglomeration economies - have already been mentioned. This section will expand that discussion and complement it with a brief account of other aspects of the urban- industrial link. Market potential, agglomeration economies and the role of government policies will be discussed, with a brief account of some location factors such as the role of enterprise, natural resources and regionally specific government programs. (b) Market Potential 145. There is no absolute proof of the relative importance of the market vis-a-vis other factors as a determinant of urban-industrial growth but it has played a crucial role in influencing the course of industrial- urban growth since 1940. 146. As noted, apart from the development of "new" producer goods subsectors, the continued growth of consumer goods industries was a major feature of industrial growth after 1940. For many such industries, the cost of transport of material inputs and of finished goods pointed to a manu- facturing location near the market. The potential size of the market for manufactured goods was thus a major location factor. This had two aspects: access to population and access to incomes.  - 40 - 147. The interstate population distribution in 1940-70 is shown in Table 1.9 which emphasizes the continuing dominance of the Federal District, Jalisco and Veracruz. At the same time itillustrates the relative increase in the ranks of Baja California Norte and Chihuahua and the relative decline of Queretaro. The heaviest increases were in the northern states and the smallest increases in the south. The contrast was stronger in 1960-70 than in 1950-60, although the relative geographic emphasis of growth was the same in both periods. 148. Maps 8 and 9 reflect the consequences of adjusting state population size to geographic area. In 1940 (Map 8) the rank order of states by absolute size and by density shows that the Federal District was the highest ranking entity. It also shows that the states of central Mexico had a generally higher ranking in density than in size, whereas some of the largest states - in both the north and the south - ranked lower in density than in absolute size. 149. This pattern persisted in 1970 (Map 9), although the density of the southern states had by then notably increased, and the high densities of the central states were also present in the Gulf states to the east (Veracruz), and in those immediately to the west (Jalisco, Guerrero). 150. In terms of the geographic distribution of demand as measured by population potentials, the center was most important. However, in a country at Mexico's level of development it cannot be assumed that the implicit population potential (see Technical Annex 9) embodied in the data on population distribution, coincides with that of potential demand. The latter is essentially a product of population weighed by income, and it is therefore necessary to refer to certain aspects of spatial contrasts in personal income levels and interpersonal income distribution. 151. For certain foodstuffs, purchasing power differentials between rural and urban areas may favor rural families, and to the extent that this is so, the apparent differential between rural and urban incomes shown in Tables 11.2 and 11.3 would be reduced. States with large rural populations are generally unlikely to be major sources of demand for manufactured goods given the low absolute level of rural incomes as well as a skewed pattern of distribution.  - 41 - 152. Except in a few states (in the northwest :in particular), average family incomes in rural areas were low (Table 11.5). There was a marked difference between Baja California Territory where the average rural monthly income was Mex $2,317 and Oaxaca, where the average was only Mex $376. Clearly, the market potentials of the rural population of the northwestern states exceeded those of the less developed, but densely populated, states of the south and center. 153. There were also strong contrasts within the urban sector (Table 11.7). Average monthly incomes ranged from Mex$3,000 in Baja California, the Federal District, San Luis Potosi and Sinaloa, to Mex$925 in Oaxaca. The difference between urban incomes was smaller than that between the highest: and lowest rural incomes. Rural and urban income differentials and the distribution of population between rural and utrban areas account for the pattern of average monthly inc:omes in each state (See Table 11.9). The weight of urbanizat:ion in the Federal District is reflected by its income level, the highest of any entity (Mex$3,133), while rural Oaxaca has the lowest (Mex$537). 154. Average income levels multiplied by population size are a general referent of market potential. Two other important references are the absolute level of income and the distribution of income within each entity. 155. Table 11.10 shows broad contrasts in income levels between six major regions and, like Table 11.9, emphasizes the wealth of the Federal District and the poverty of the south and (in general) the center. In absolute terms Table 11.11, 11.12, 11.13, 11.14, 11.15 and 11.16 show that the absolute monthly income of the Federal District greatly exceeded that of any other region; the central region was next highest followed by the south Pacific (more than half of which came from Jalisco), the north (without particular concentrations), the Gulf of Mexico (two thirds of which came from Veracruz) and the north Pacific (without concentration). 156. The Gini coefficients in Table 13.18 show the distribution of the average level of income in each state. There is a high (.78) coefficient of correlation which suggests that the states in which personal incomes in 1969 were highest were those with the most equitable patterns of distribution. At the extremes, the Federal District had the highest level of income (Mex$628) and the fifth lowest Gini coefficient (.501) while Oaxaca had the lowest level of income (Mex$109) and the highest Gini coefficient (.688).  - 42 - 157. The conjunction of relatively high average incomes and a pattern of relatively good income distribution in the center, the northwest and the north, points to a rational location for market oriented manufacturing activities in these areas. In particular, manufacturing growth in Mexico City was closely related to its huge internal market which gave rise to considerable agglomeration economies in both the District and in contiguous parts of the State of Mexico. By 1960 only two cities - Monterrey and Guadalajara - had surpassed Mexico City's population in 1900 (344,721). This indicates the tremendous advantage accruing to the capital from its market size, an advantage which was also reflected in its share of retail sales (Table 9.3). 158. Interstate contrasts in income potential were coupled with the relative advantages and disadvantages derived from the structure of the transport system (already discussed from a general point of view) to provide a contrasting pattern of market opportunities for different cities. The significance of network accessibility derives from the lack of sufficiently large aggregate populations and incomes to achieve scale thresholds basedexclusively or even primarily on internal demand for other than essential consumer goods. In serving a regional market of a given size, manufacturers in poorly connected cities have longer shipping distance and higher transport costs than those in well connected cities. 159. By 1940, there was already a coincidence between accessibility to markets and network structure. For reasons of resource base development or agglomeration, transport improvements had already conferred a lasting pattern of relative advantage and disadvantage with respect to industrial growth in cities. In 1940, a number of cities with relatively low levels of accessibility in terms of market potential had relatively high connectivity as measured by transport and communications networks. Some of these cities were small in 1940 but managed to grow rapidly in 1940-70. It can be argued that they did so despite limited market potential, but they could not have done so without an advantageous situation of connectivity.  - 43 - (c) Agglomeraition Economies 160. Agglomeration economies derive from the increase in internal and exlernal economies accompanying the spatial concentration of population and economic activity. There are several kinds. The first concern plant size and are internal to the firm. Coefficients for three alternative measures of plant scale in manufacturing - capital invest- ment per plant, employees per plant, and output per plant - show that average plant size was greater in the more developed states (Table 7.14) and in larger cities. The increase in coefficients for emnployees per plant and output per plant between 1940 and 1970 suggests that these scale differences may have become sharper over the thirty-year interval because manufacturing plants in larger cities probably benefitted more from internal economies of scale than others during the period. 161. Manufacturing growth has been fastest in cities where industry was able to make the greatest strides in increasing plant sizes, average plant size (measured by output per plant as a percent of the national average) being generally higher in those states which experienced rapid growth in manufacturing employment and output. During the "Porfiriato", manufacturing growth and scale changes were not, in general, closely associated with differential rates of urban growth. 162. A second class of agglomeration economies are external to the firm. Among them, the supply of labor and capital appear to have been particularly significant. 163. Skilled labor was a more crucial determinant of industrial location than the aggregate supply of labor 1/. This is consistent with a high level of marginal employment in both the urban and rural sectors and with migrat:ory factors that tend to mitigate interregional differences in labor supply. Capital and unskilled labor must, in general, be combined with minimum amounts of skilled labor to determine 1/ Skilled workers are "those who, as a result of theoretical training or considerable experience, are capable of realizing a complex and specialized task and whose services are contracted precisely for that purpose", VII Censo Industrial, 1961, p. SVII.  - 44 - the effective overall supply of labor manufacturing. Thus, a predictably strong relationship exists between the proportion of skilled workers to total workers in manufacturing and the incidence of urban/industrial growth both at the state and the city level. 164. In the integrated economy, which evolved in Mexico after 1940, the local availability of capital is not necessarily an important determinant of industrial location. However, Myrdal (1957) argues that in the early stages of economic development, the expanding scope for external economies in high growth regions will raise profits and income levels which, in turn, will lead to higher rates of savings. Increased savings are then invested, which further expands the scope for external economies. Hirschman (1962) supports the argument that savings, as well as investment, tend to be higher in growth poles experiencing agglomeration economies and both contend that advancing regions tend to secure relatively larger shares of government investment. 165. The interstate distribution of private savings per capita in 1940 and subsequent increase were consistent with high differentials in rates of private savings between more and less developed states and those in which manufacturing development was relatively advanced or retarded (Table 3.6). This pattern continued in later years. In 1970, the pattern of savings emphasized the dominance of the Federal District even more than in 1940; by that time almost half of the total bank deposits were concentrated there despite a marginal decline in dominance since 1960 which might indicate the beginning of a new pattern of savings. 166. Disparities in private credit appear similar to spatial disparities in private savings (Table 3.7). The dominance of the Federal District in private credit was even greater than in savings in that it accounted in 1940 for more than two-thirds of total credit for the whole country. Other areas or relative credit concentration at that time were Nuevo Leon, Coahuila, Sinaloa and Sonora. By 1970, the dominance of the Federal District was even stronger. Nuevo Leon, Chihuahua and Jalisco followed in terms of aggregate credit allocations. 167. There is no available data on the sectoral allocation of either private or official credit prior to 1965. In that year industrial credit accounted for almost half of total private credit, with most of it allocated to commercial activities (Table 3.8). In both respects, the Federal District was strongly dominant. With respect to  - 45 - the smaller volume of total official credit, the sectoral and geographic distribution was roughly similar to that of private credit, although the Federal District had a near monopoly on official credit for industry. The pattern in 1972 (Table 6) was roughly comparable, although the share of the Federal District in private industrial credit had by then somewhat declined. 168. Clearly manufacturing development and urbanization were strongly associated with the accumulation of savings and relatively high levels of investment. Indices of invest- ment in manufacturing and agriculture support the hypothesis that rates of private investment were highest in the most developed states and in the largest and most dynamic cities, suggesting, inter alia, that industrial capital was not generally as horizontally mobile up to 1970 as might have been expected. 169. A third class of agglomeration economies are also external to the firm and comprise a variety of intersectoral linkages. First, the concentration of commercial and service functions in large cities implies that large cities offer the manufacturer greater relative advantages than smaller cities. In particular, the role of Mexico City as the financial and commercial center of the nation was strongly associated with its growth as a manufacturing center. Access to financial and other specialized services was also an important variable in investment decisions. Derossi (1972) points out that access to Government services was particularly crucial, even for entrepreneurs with firms located in Monterrey, and that it is not uncommon to find that large enterprises located in other cities maintain permanent offices in the capital. 1/ 170. Second, good access to coal and ore deposits were very important to the expansion of the basic metals industry in Coahuila and Nuevo Leon, but these raw materials were not located in the State of Mexico and the Federal District. Both states experienced substantial development in this sector, which can be mainly ascribed to the effects of transfer economies arising from relatively low freight rates on raw materials. In all four states, the concentration of basic 1/ In at least one instance (no doubt there are others), a firm which has its only plant in Monterrey has its head office in Mexico City, 600 miles away.  - 46 - metals stimulated growth in the forward-linked fabricated metals, transportation products, and electrical machinery and apparatus industries. In turn, the automobile assembly industry, located in the Federal District and the State of Mexico, gave impetus to the backward-linked rubber tire industry in these states. The resulting large concentration of industrial demand (except in Coahuila) also promoted the production of intermediate goods, such as synthetic fibres, industrial acids, cement blocks, and glassware; while the accompanying concentration of consumer demand, resulting from a large labor force, stimulated the output of pharmaceuticals, processed food, and other consumer goods. 171. Transfer economies derived from the adjacent location of closely-linked industries and from internal economies of scale created by the spatial concentration of large industrial and consumer markets. The early prominence of Mexico City and Monterrey, as population and manufacturing centers, placed them in a particularly advantageous position to attract the new large scale industries which the economy was able to support after 1940. Conversely, where industrial growth was based on access to primary materials in rather small urban areas (Monclova for example) 1/ consumer and other intermediate goods industries (e.g., textiles and chemicals) did not grow rapidly. (d) The Role of the Public Sector 172. Government policy in the evolution of the spatial structure of the industrial sector after 1940 revolves around two issues. First, macro-economic policies, which had an important effect on spatial and urban development but in which there was no formal spatial dimension. Second, policies which had an explicit spatial dimension and which were designated as such, or if not, can be recognized as, at least in part, deliberately spatial policies. 1/ Monclova was not one of 1970s largest cities.  - 47 - (i) The Urban Impact of General Economic Policies for Industrial Development 173. The most important policy in this category (and the only one which will be discussed here) concerns external protection. As noted earlier, Mexico's industrialization after 1940 was closely associated with import controLs and tariff barriers with emphasis on the former. Compared with other Latin American countries, Mexican industry is not unduly protected, but this strategy has affected the spatial structure of the industrial sector. 174. Import substitution, which began in the 1940s, led to certain inefficiencies in industrial development. High production costs were associated with the fragmentation of much of Mexico's industry into small firms, often using backward techniques of production and management. They were also related to inadequate specialization within the manu- facturing sector because the system of import licensing encouraged backward integration with firms producing various parts and components for their own use rather than buying from domestic suppliers. 175. Protection hindered efficiency because firms were assured of high profits in a closed domestic market. Thus, technical progress was slow and the quality of many manu- factured goods, poor. The low quality of industrial, inputs also adversely affected commodities at higher levels of fabrication. Import substitution behind protective barriers led to the establishment of industries (mainly capital- intensive) in which Mexico had little comparative advantage. The mechanics of protection emphasized quantitative controls rather than tariffs, with the benefits going almost entirely to the private rather than the public sector, resulting in a regressive impact on income distribution. 176. Lavell (1972) has analyzed the spatial impact of protection with specific reference to the Law of New and Necessary Industries introduced in 1941 and amended in 1946 and in 1955. Industries covered by the Law included: basic metals, transport, chemical, mechanical, electrical, metal and food products. The law grants total relief from excess profits, stamp, industrial and income taxes and from import duties on raw materials or components needed in production. 177. Exemptions were granted for periods of five, seven or ten years depending on the industry and its relative importance for national development. "New" industries included branches of manufacturing which were not well  - 48 - represented in Mexico and those producing a similar product to one already produced in a national plant, provided that the new company guaranteed a 20 percent saving in the price or the lifespan of the new product. "Necessary" industry was defined as that which was of fundamental importance to the nation's industrial growth. 178. The law had no explicit spatial dimension. As Navarrete (1967) correctly emphasized, the Federal Government granted identical exemptions throughout the country, without reference to spatial, regional or urban development. In practice, however, most of the exempt firms were located in the Valley of Mexico and at least 65 percent of the capital investment and 40 percent of the employment linked with the law was in or around Mexico City. By 1960, 19 of the 32 states had not received a single exemption. By 1965 the number had been reduced to 15. 179. Table 6.6 shows the regional distribution of exemptions in three periods; the heavy concentration in the Central Region and to a lesser extent in the northwest are clearly apparent. Table 6.7 emphasizes the same point, but more strongly, because it can be inferred that 98 of the 167 plants established under the Law in 1959-64 were in or near Mexico City, and 21 were in Monterrey. Table 6.7 shows the sectoral incidence as well as the geographic distribution of the exemptions. 180. The absence of a spatial dimension in the law and in the policy of protection in general, does not therefore imply that it had no spatial effect. The lack of a spatial component is attributed to the government's concern to promote national industrial development. When the law went into effect, some parts of the country were severely disadvantaged compared to others. In terms of infrastructure and accessibil- ity, to have encouraged location in the periphery would have been inefficient because already inefficient producers would have had to contend with cost disadvantages attributable to location. But, if the law had included a spatial objective, the response of entrepreneurs would have been less enthusiastic and Mexico would have experienced less real growth than it did. Nonetheless, external protection and industrial central- ization were closely related phenomena.  - 49 - (ii) The Spatial Impact of Industrial Policies with Explicit Spatial Dimensions 181. Among the policies in this category are (1) fiscal policies - which were developed by the states; (2) credit policies which were developed by the Federal Government; (3) public investment policies; and (4) wage, price and tariff policies. - Fiscal Policies 182. Between 1940 and 1960, almost all of the states introduced tax exemption laws to encourage industrial develop- ment. These laws allowed for exemptions from the state's share in the sales tax, from stamp duty, and from local land taxes and were applicable to both new and expanding industries. Exemptions varied from 10 to 30 years, depending on the state concerned. 183. Romero Kolbeck and Urquidi (1952) concluded that the effects of exemptions on industrial development in the Federal District had served little useful purpose and had resulted in an unnecessary loss of fiscal revenue. They also argued that the objectives of tax exemption were, in the case of the Federal District, dubious because there was already an excessive concentration of industry in the area. They added that if exemptions were to be continued, they should only apply to industries which would "lead to an improvement in the standard of living of the population." Finally, they proposed that if exemptions were to be cancelled for 61 percent of the factories studied, the cancellation of exemptions would lead to a 10 percent loss in profits and for 14.9 percent of them, a 25 percent loss. If passed on to the consumer, this would imply an increase in prices of only 1 or 2 percent a year. Not long after the publication of their study, tax exemptions were abolished in the Federal District. 184. It has been argued that essentially similar exemption laws had a countervailing impact on industrial growth in other parts of Mexico and, to the extent that exemption was used, a loss of fiscal revenue ensued. More- over, on the basis of somewhat scanty empirical evidence, the laws did not produce a positive effect on location decisions. Lavell (1972) reports that very few, if any, of a group of entrepreneurs in the states of Puebla, Guanajuato, Queretaro and Cuernavaca accorded much significance to the exemption laws.  - 50 - - Credit Policies 185. In 1953, the Guarantee and Development Fund for Small and Medium Industry was used as an explicit instrument of industrial location policy. By 1960, the industries covered by the Fund accounted for 57 percent of all industrial plants in Mexico and for 79 percent of industrial employment. By June 1970, the Fund had awarded some 10,000 credits to more than 5,200 firms; the credits represented 32 percent of total capital investment in the companies concerned, which employed a total of 187,000 workers. The spatial strategy of the Fund was to "attempt" to select factories located outside the areas of industrial concentration. Table 6.8 shows the regional distribution of credits granted between 1953 and 1970. It shows that, over time, the relative importance of Jalisco, Mexico and Sinaloa increased, but the average size of credits in the Federal District and the State of Mexico was relatively large throughout the period. 186. Table 6.9 confirms that marginally the share of the Federal District decreased. In 1961-62 it attracted 48 percent of all credits granted, whereas in 1969-70 it attracted only 23 percent. However, the marginal participa- tion of the State of Mexico increased over time. The data indicate that, given the rising shares of Jalisco, Sinaloa, Sonora, Guanjuato and Oaxaca in the numbers of credits granted, the policy was a small step towards the spatial objective of decentralization. 187. The limited flows of credits to provincial areas may have been related to the lack of knowledge about credit facilities outside the central core and a few other regions. Lavell (1972) finds it difficult to believe that during the last 15 years, the economies of Zacatecas, Campeche, Colima, Nayarit and Tabasco could only support a demand for 10 or fewer factories requiring credits. The Fund took specific notice of this problem in trying to publicize its existence to less developed states. - Public Investment 188. The direct relationship between the spatial allocation of public investment and that of industrial growth concerns the allocation of public investment to the industrial sector. Prior to 1959 there are no data on the geographic distribution of public expenditure. From 1959 through 1970 there are detailed data on outlays in each state by sector.  * 51 - The industrial sector accounted for 8 to 15 percent of total public investment in this period, about 80 percent of which was in the hydrocarbons sector. Industrial investments were highly localized and Tamaulipas and Veracruz accounted for some 60 percent of hydrocarbons investments while steel investments were heavily concentrated in Coahuila. 189. The indirect relationship between the spatial structure of industrial growth and that of public investment was more important than the direct relationship, because geographic differences determined the availability of transport facilities, power, water and other goods and services which, in one way or another, had a bearing on the attractiveness, and even the viability, of alternative locations for industry. 190. In the period for which consistent data are avail- able (1959-70) the geographic allocation of public investments, in sectors other than industry and agriculture, shows that the Federal District received the largest share of total outlays throughout the period, although the share has declined somewhat in recent years (Table 3.16). The geographic distribution of infrastructural investments of all kinds (transport, power, water) was found to be closely related to the incidence of urbanization at the interstate level. There was a similarly high degree of correlation between urbanization and public investment in education, the Federal District again receiving a disproportionately high share (with respect to its share of the nation's population) of total outlays in these sectors. The Federal District. also accounted for an even larger share of total investment in administration because of its role as the seat of the Federal Government, although this category of expenditure was less important than others. - Wage, Price and Tariff Policies 191. Since 1946, minimum wages in Mexico have been set by a tripartite commission and appearances might suggest, given that wages were consistently lower outside the largest metropolitan areas ard the northern frontier (see be:Low) that wage differentials encouraged industrial growth in peripheral areas. However, minimum wages were mainly designed to take account of cost of living differentials and there is little evidence that geographic differences reflected anything more than this. It is difficult to find evidence that wage policy was ever used in a conscious effort to promote industrial growth in the periphery.  - 52 - 192. With respect to public sector prices and tariffs, general evidence shows that government policies did not encourage decentralization. In the Federal District, water prices were held down so that consumers paid less than the incremental cost of bringing water to them. Prices of corn, electric power, diesel fuel and public transport were also subsidized in the Federal District from 1940-70. Although some of these policies were to benefit the urban poor, substantial parts of the subsidies benefitted other groups and inevitably reduced the costs of locating economic activity in the Federal District, thereby creating a disincentive to locate elsewhere. 193. As argued earlier, progress in transportation was a positive factor in facilitating decentralization after 1940. However, two issues deserve attention here. First, rail freight rates were structured to favor transport routes between Mexico City and other points over other routes. Second, the basis of rate-making was "value of service" and this tended to increase the costs of transporting bulk commodities. The effect of this pattern of cross subsidiza- tion was to encourage location near markets rather than near raw materials, and thus to favor the core rather than the periphery. Finally, property in the Federal District was relatively under-valued for tax purposes. Other states taxed at higher rates because local taxation was necessary to finance public education, whereas education was financed by the Federal Government in the Federal District. This too reduced the relative costs of locating in the Federal District. (e) Other Determinants 194. Other circumstances--some of general (but minor) relevance, others of local (but great) importance--affected the process of urban-industrial development. A complete review of these factors other than those included in the three broad categories we have so far considered is impossible. But a selective review, focussing on locally relevant factors is not only feasible but important because the evidence suggests that such factors may, at this level, have outweighed the more general conditions we have already discussed. 195. The first such factor concerns the decision-maker. Industrial location decisions are necessarily made by people with varying abilities to make rational choices and who may or may not be well informed when taking the decision. Entrepreneurs (or industrial decision-makers) may vary in willingness to take risks, etc. Contrasts between  - 53 - entrepreneurial groups may have an important bearing- on industrial development. The available data suggest that such contrasts in Mexico are specific to cities rather than to states or regions. 196. An exhaustive review of this topic is beyond the scope of this report. The general significance of enterprise as a location factor can be illustrated by reference to the example of Monterrey. Derossi (1971) has shown that: by 1940 the effect of antecedent industrial development had already established this city as a leading industrial center and that its growth after 1940 was, to a large extent, based on the continued exploitation of its initial advantage vis-a-vis other cities and regions. Even so, the continued ability of Monterrey to compete successfully with other cities after 1940, in spite of its peripheral location relative to the central market, depended upon the ability of its entrepreneurs to identify and develop new opportunities for expansion. There may be a parallel here between the role of en-terprise in Monterrey (and, by extension, Nuevo Leon), and in Medellin (and, by extension, Antioquia), Colombia. 1/ Certainly there is some evidence that the entrepreneurs of Monterrey like those of Medellin have sought new opportunities not only in situ but also in other parts of the country. 197. Another relatively localized determinant of urban- industrial growth after 1940 concerned natural resources. In earlier periods (see Chapter 1) we found that these played a crucial role in industrial development. After 1940, their importance appears to have been local rather than general and can best be illustrated by reference to the experience of the northern states of Chihuahua, Sonora, Tamaulipas, Morelos, and Baja California Norte, where raw material processing industries (most of which produced for export to national or international markets rather than for the local market) provided the main source of manufacturing expansion.. 198. Cotton ginning and weaving utilized part of the large amount of raw cotton grown in these states, while the oil extracted from cotton and other oil seeds was important for industrial inputs. Cattle raising stimulated the meat 1/ See Hagen (1965). San Pedro Sula, Honduras; Santa Cruz, Bolivia;and in a somewhat different way, Sao Paulo, Brazil might also fall into the same cat:egory.  - 54 - packing industry in the city of Chihuahua while the wheat and rice milling industries processed the large quantities of wheat and small quantities of rice produced in the northern states. Beerproduction, based on local grains, achieved prominence in Tijuana and Mexicali and petroleum refining gave a significant impetus to industrial expansion in Tampico. Favorable access to raw materials inputs, and external demand for the products of these industries, stimulated heavy capital investment and the establishment of relatively large-scale plants in the northern cities. The expansion of these industries also seems to have induced secondary growth (via multiplier effects) in the vehicles, tires and tubes, and electrical apparatus industries; in the lime, plaster and stone industries providing construction materials from local mineral inputs; and in some local consumption goods industries, such as beverages and printing. Other growth stemmed from insecticide and fertilizer industries providing inputs for the agricultural sector, which in turn furnished raw materials to some of the major. growth industries. 199. The final example of a "local" factor which had a direct and important bearing on urban-industrial growth after 1940 concerns the impact of a regionally specific government policy: the "assembly industry" or maquila program of the northern (U.S.) border, which was started in the 1960s. 200. The border region stands apart from the rest of the country. Its trade and cultural relations are largely with the U.S.,and the high levels of immigration to the region after 1940 were closely linked to the possibility of obtaining employment in the United States. From the 1940s the development of the "Bracero" program, which facilitated the temporary migration of Mexicans to work in farms across the frontier, provided an important source of employment. By 1964, it employed nearly 200,000 workers. In 1965 the U.S. Government terminated the program, and the Mexican government initiated the Border Industries Program to facilitate sub-contracting by U.S. industrial firms, taking advantage of the availability of inexpensive wage labor in the Mexican border states. 1/ 1/ Articles 806.30 and 807 of the U.S. Tariff Schedule provide that imported products assembled in foreign countries with components made in the U.S. are subject to customs duty only on the basis of value added abroad.  - 55 - 201. The program was launched with multiple objectives. These included: the growth of employment; the generation of foreign exchange; the upgrading of the labor force; widening the domestic market for Mexican producers by injecting income into the region; increasing fiscal revenues; stimulating multiplier activity in non-manufacturing sectors and expand- ing the market for skilled Mexican labor. 202. The legislation introduced in 1965 was such that the so-called "maquila" industries located in the Baja California and Sonora Free Zone were governed by regulations which differed from those applicable to other border areas. They were allowed to import any item included in the Free Zone Import List, whereas plants located in other border areas had to obtain "temporary import" permits for their specific required inputs. The initial temporary import permit for new "macuila" industries in the border areas but outside the Free Zone had to be sanctioned by the Ministry of Industry and Commerce and the Ministry of Finance in Mexico City. Repeater permits were granted by a local committee in the concerned border area. These administrative arrangements worked fairly smoothly, and border area industries of both kinds considered themselves advantageously placed compared with plants in the interior which also required temporary import permits. 203. The major "maquila" industries, whether located in the Free Zone or in other border areas, were subsidiaries of US corporations. The latter provided capital, physical inputs, and often management and guaranteed to purchase the entire output. The terms under which the Government allowed them to be established required them to declare costs and revenues and to pay Federal and State taxes on sales and profits. But these conditions were generally bypassed-- particularly by foreign-owned companies. Many foreign firms did not keep financial data or accounts in their Mexican plants and received inputs of nominal value and transferred their output at or near the domestic cost incurred in Mexico. Hence the "value" of exports declared to government authorities was often an approximation of domestic value added. This position was clearly very different from that of Mexican firms located in the interior, which imported inputs on "temporary" licenses for processing or assembly into export products. Mexican firms in the border areas appeared' to be in a somewhat ambiguous position, depending on the discretion of local officials.  - 56 - 204. There can be no doubt that the program had a dramatic impact, both on the frontier region and on Mexico's manufactured exports. It is, however, unreasonable to claim that it has been fully successful even with respect to its stated objectives. Analysis of the consequences of the growth of the border industry program suggests that it may have produced some undesirable effects. 205. Its quantifiable impact on the local, regional and national economy cannot be determined with great precision, but it is estimated that by the early 1970s, the program had generated some 50,000 jobs, mainly for women in the age groups 15-24. Minimum wages in the border area had long been higher than elsewhere, and rose quickly after 1966. By 1972 they ranged from an average of Mex$21,000 per year in Nogales to Mex$14,400 in Laredo and Matamoros. However, while employers in the interior, and especially in the Federal District, have to incur considerable expenses on fringe benefits, border industries generally pay only the minimum prescribed wage. The difference in total cost per unit of labor is not so great as that shown in published minimum wage schedules. It seems likely that there is widespread use of on-the-job trainees by border plants, a practice which may point to the exploitation of Mexican labor laws. 206. Foreign exchange earnings disaggregated into gross and net exports originating in assembly plants are not available prior to 1970. On the basis of the growth of employment, foreign exchange earnings before 1970 were lower than might be expected. It is estimated that 70 percent or more of the wages earned by Mexican employees is spent in the U.S., representing an enourmous "leakage" and indicating the failure of the program to widen the market for Mexican producers, largely because of the influence of U.S. consumption patterns in the border cities. 207. From a fiscal standpoint, the results have been poor. If the U.S.-owned firms in the border areas were to comply fully with the basic terms of the regulations and to declare the full value of their inputs and outputs, the Government of Mexico would gain in the collection of both sales and income taxes. Parent firms in the U.S. would not suffer from paying Mexican income taxes on their subsidiaries because of the U.S. provision of credit for foreign tax pay- ments. The sales tax payments would, however, represent an additional cost.  - 57 - 208. Finally, it is unlikely that the program has had the intended results with respect to labor force skcills. 209. The most serious "failure" of the border industry program is that it has not reduced the traditional dependency of the region on the U.S.A. There is still a mass:ive "leakage" of resources across the frontier. Although the program has generated more income in the area, much of it is "lost". As a result, the border cities remain enclave economies which are essentially disarticulated from the rest of Mexico. This judgment may seem harsh in view of the apparent success of the program in certain respects. Yet, if spatial development implies the generation of "spread" effects within the national economy, the results of the program through 1970 did not achieve this. D. Tertiary Sector Development 210. While profitability may be regarded as a general test of the relative viability and attractiveness of. alternative locations for tertiary activities, there are certain contrasts with the secondary sector and there are also important differences between one tertiary industry and another. Commerce and urban services are dependent on market potential within the framework of a functional hierarchy of central-place activities. Tourism, however, has very little relationship to demand originating in an urban or even a regional market and is, to a very large extent, related to international demand. 211. There are few cities in which initial tertiary multipliers seem to have played a decisive role. Tro some extent, services, especially government, were crucial sectors in Mexico City, in the sense that its initial size (in 1900) was in many ways a consequence of its historical development as an administrative and political center. But the major instances where tertiary functions played a decisive role in generating urban growth were those of the border cities of Tijuana, Mexicali, Ciudad Juarez, Reynosa, Nuevo Laredo, and Matamoros plus the major tourism centers of Acapulco and Mazatlan on the Pacific Coast. 212. The key to growth of the border cities is the large market potential--not in Mexico, but in the USA. After 1940, the development of new industrial production for the US market became a major cause of expansion, and in the late 1960s the border industrialization program also became import:ant. The development of a cluster of  - 58 - entertainment and recreation industries in Tijuana and in other border cities was a major force in the growth of their populations and economies. 213. With respect to growth based on tourism, the physical attributes of the Pacific Coast were clearly paramount in determining sectoral development in Acapulco and less markedly, in Mazatlan. Both enjoy remarkable natural settings and this was the primary force in generating growth after 1940. Clearly, it required enterprise and enthusiasm to exploit these resources. But growth was essentially based on an initial development impulse from new investment in non-industrial activity and corresponded to an export-base development model. Normally an export-base is associated with the growth of secondary sector activities, but here is a case in which it was linked with tertiary development, the multiplier effects of which led to the subsequent growth of industrial and tertiary activities. 3. THE SOCIO-ECONOMIC STRUCTURE OF THE URBAN SYSTEM, 1940/70 A. The Demographic Structure of the Urban System The Demographic Characteristics of the System 214. The structural development of the urban system is based on a set of cities which constituted the upper level of the urban hierarchy in 1970, as conceived in terms of a population of about 100,000. Thirty-seven cities fall into this category, the largest being Mexico City (8,500,000) and the smallest Pachuca (92,000). The population of each in 1940, 1950, 1960, and 1970 is shown in Table 1.23. Their relative sizes are shown in Table 1.24 and (in size cohort form) in Table 1.27. 215. The population growth rates are shown in Table 1.28. The highest growth rates in 1940/50 were those of Tijuana (11.5 percent), Mexicali (10.9 percent) and Reynosa (10.6 percent). The fastest rates in 1950/60 were those of Tijuana (9.7 percent), Mexicali (8.5 percent) and Hermosillo (8.1 percent). In the final decade of the period (1960/70) when urban growth was generally slower, the fastest growth rate was in Acapulco (10.9 percent).  - 59 - 216. During the period of 1940-70, four cities - Tijuana, Acapulco, Mexicali and Ciudad Juarez - grew at an average annual rate exceeding 7.0 percent; five other cities sustained an average growth rate of more than 6.0 percent; and five more grew at an average annual rate of more than 5.0 percent. There was, in general, no relation- ship between relative population size at the beginning of the period (1940), and subsequent population growth, which implies that initial advantage in population terms was not a crucial determinant of differential urban size growth. 217. The growth pattern in each city shows contrasts compared with the average rate of growth for all 37 cities (Table 1.29). Some cities (e.g., Culiacan, Minatitlan and Jalapa) grew faster than the average rate in one decade and more slowly in another; some cities (e.g., Guadalajara, Tijuana and Mexico City) sustained a faster than average growth rate throughout the 30-year period, and some (e.g., Aguascalientes, Morelia and Orizaba) grew at a less than average rate in each of the three decades after 1940. 218. The trend of increasing urbanization was associated with an increasing concentration of the urban population in the 37 cities which dominated the urban system in 1970 (See Table 1.55). Within this framework there was further concentration in the ten largest and five largest cities and in Mexico City (See Table 1.54). The 37 largest cities accounted for 65.0 percent of the nation's urban population in 1940 and 68.0 percent in 1970; the ten largest cities, for 46.0 percent in 1940 and 50.0 percent in 1960; the five largest cities, for 38.0 percent in 1940 and 43.0 percent in 1960 and Mexico City, for 26.0 percent in 1940 and 30.0 percent in 1960. 219. Up to 1950 increasing urbanization was associated with increasing primacy (i.e., an increasing proportion of the total population of the largest cities was concentrated in Mexico City). In terms of the 25 largest cities there was a continuous increase between 1940 and 1950 in the ratio of the size of Mexico City to the total population of the cities occupying the second through the 25th ranks in the urban hierarchy (i.e., the rank order of cities). After 1950, however, there was a decline in the ratio between the size of Mexico City and the size of the second ranked city (Guadalajara) and also in the ratios between the size of Mexico City and the cities occupying the third through the tenth ranks. These contrasts imply that the maximum degree of primacy in the urban system was reached in the ]940s and that thereafter, the country's major regional centers became relatively more dynamic than Mexico City. As a result of the first implication, the "balance" of the urban system was gradually improved after 1950.  - 60 - 220. We also find that the urban system became increasingly stable over time. Analysis shows (Table 1.43) that the member- ship of the sets of the five, ten, and 25 largest cities in 1940/70 was remarkably consistent, suggesting that once estab- lished in the upper limits of the urban hierarchy, cities generally grew fast enough to remain there. This conclusion is reinforced by rank-order correlation analysis (Table 1.45) which shows that the stability of the order of urban system was very high throughout the period of 1940/70, even in terms of the 25 largest cities. 221. These findings suggest that the urban system became increasingly "mature" after 1950 but this is not supported by the results of a rank-size rule analysis (Tables 1.47 to 1.50). It is frequently assessed that a "mature" urban system is characterized by a harmonic lognormal progression. Analysis shows more deviation from a rank-size structure in 1970 than in 1940 and that the discordance was greater in the ten largest and five largest cities than in the 25 largest cities. The degree of discordance increased from one decade to the next. The structure, however, is severely distorted by the absolute size of large cities, such as Mexico City. The Origins of Urban Growth 222. There were marked differences in the relative importance of natural increase and migration as sources of population growth in the nation's largest cities. The period of fastest urbanization - from 1940 to 1950 - was marked by a massive growth of rural-urban migration in the 1940s, and diminution thereafter, and by the trend of accelerating natural increase after 1950. Thus the growth of large cities in the 1940s was mainly the result of migration. In the 1950s and 1960s, the growth was primarily due to natural increase, although the absolute numbers of migrants continued to rise. 223. In addition to migration and natural increase, a third, and only locally important factor in accounting for urban population growth, was the process of accretion whereby cities expanded, and in so doing, incorporated existing and formerly separate rural communities. This process was significant in the densely populated central region and was particularly important in the case of Mexico City, where urban size growth and the outward extension of the city towards formerly suburban areas was a continuing phenomenon. The relative importance of accretion (sometimes called coalescence) is shown in Tables 1.31 and 1.32.  - 61 - 224. Because data on fertility and mortality are not available at the city or municipal level, we have resorted to estimates derived from state data. The largest cities in 1970 are divided into groups showing estimates of the subcomponents of natural increase (Tables 1.33 and 1.34). While natural increase accounted for an increasing share of urban population cjrowth, the rate of natural increase in the set of 37 cities was lower than in the country as a whole. However, there were variations. Natural increase in the cities of the southwest, for example, was below the average for the set,while the rate in the northern cities was sub- stantially higher.. This implies that urbanization was associated - on a regionally differentiated basis - with a relative decline in the rate of natural increase. Although there is no valid basis for determining whether this was a function of lower fertility rather than higher mortality, it is likely, given the higher income levels and health conditions of the urban sector, that the former explanation is correct. This needs further investigation. 225. While migration was a less important city-forming process in the 1950s and 60s than in the 1940s, it:s absolute volume continued to grow (Table 1.37). Its major impact on urban growth was limited to a relatively small nunber of cities (Tables 1.36 and 1.37), among which Mexico City, Guadalajara, Monterrey, Puebla, Ciudad Juarez, Mexicali, Tijuana, Torreon, San Luis Potosi and Merida were outstanding. These cities accounted for 76 percent of all migratory growth in cities of 15,000 or more in the 1940s, while Mexico City, Guadalajara, Monterrey, Puebla, Ciudad Juarez, Mexicali, Tijuana, Leon, Veracruz, Hermosillo and Chihuahua accounted for 82 percent of all migratory growth in this category of cities in the 1950s. Within this framework, the roles of Mexico City, Guadalajara and Monterrey were outstanding. In the 1940s, Mexico City accounted for 49.0 percent of all migratory growth and the three cities together for 60.0 percent of all migratory growth. For the 1950s, Mexico City accounted for 42.0 percent and the others for 64.0 percent. This is consistent with the earlier conclusion that Mexico City was a relatively less dynamic growth center than other large cities after 1950. The Geography of Urban Growth 226. The geographic diffusion of urban development after 1940 is shown in Maps 16 to 19. These show that large cities developed for the first time outside the central core, that most of them were in the north, and  - 62 - several were on the U.S. frontier, and that by 1970 there were still only two southern cities with more than 150,000 people (Merida and Acapulco). These trends were derived from differential growth rates (Maps 20 to 23) which emphasize the sustained expansion of the northern cities. They also show that by 1970 a few southern cities, notably Coatzacoalcos and Villahermosa, were among the fastest growing in the country; providing a contrast between the static picture of comparative size and the dynamic picture of relative growth rates. B. The Economic Structure of the Urban System, 1940/70 Employment and Output (a) Contrasts Between Cities (i) General Contrasts Between Cities 227. Tables 4.1 (1940) and 4.2 (1970) show the distribution of employment in the secondary and tertiary sectors among 1970s' largest cities. (See Tables 4.4 and 4.5 for percentage distributions which emphasize the highly uneven shares of total employment among the 37 cities, and, in particular, the dominant position of Mexico City). In 1940, Mexico City accounted for 49 percent of non-agricultural employment in the 37 cities; this number increased to 51 percent by 1970. 228. In 1940, Mexico City occupied a dominant status in almost every secondary and tertiary sector, with the exception of hydrocarbons and mining (Table 4.4). There were also rather high levels of concentration elsewhere, with Tampico accounting for 45 percent of employment in hydrocarbons and Pachuca for 39 percent of employment in mining. 229. By 1970 (Table 4.5) the relative importance of Mexico City had increased in all sectors except transport, commerce and construction, where its relative shares declined slightly. Tampico continued to account for a large share (21 percent) of total employment in hydrocarbons, while Reynosa, Coatzacoalcos and Minatitlan each accounted for 11 percent of all employment in this sector. Outside Mexico City, most other sectors were fairly widely dispersed among the 37 cities, although there were "pockets" of concentration in mining (8 percent in Pachuca, 7 percent in Monterrey, 6 percent in Guadalajara); commerce (8 percent  - 63 - in Monterrey, 6 percent in Guadalajara) and services (6 per- cent in Monterrey, 7 percent in Guadalajara). There was, however, a fundamental contrast between concentration in the hydrocarbons and mining sectors and other sectors. 230. The general relationship between city population size and manufacturing is indicated in Table 7.3, which shows implicit location quotients for manufacturing in each city in 1940 and 1970. In general, the larger cities such as Leon, Monterrey, Orizaba, Guadalajara, Puebla, San Luis Potosi and Queretaro had higher quotients of manufacturing employment than the smaller ones. This relationship is further confirmed in terms of city size cohorts in Table 7.4. 231. There was a similar pattern in the tertiary sector, with the distribution of employment in services being generally correlated with city population size, although commerce was disproportionately important in such cities as Tijuana, Ciudad Juarez and Guadalajara. I 232. The distribution of manufacturing, commerce and services was thus generally consistent with that of urban population size. However, irrespective of location quotients, there is an important a priori difference in the relationship between population size and manufacturing, and between population size and commerce and services. Theoretically, the latter sector has an hierarchical structure which provides successive levels of commercial and service functions in different cities. The largest cities may thus be assumed to offer higher "orders" of these tertiary functions than smaller cities, because their larger populations facilitate the achievement of successively higher "thresholds" for different sectors. Manufacturing location is, by comparison, less closely tied to the market although, as we have seen (Chapter 2), market potential has been a very important determinant of Mexico's industrial geography. 233. Finally, the transport and government cent:ers have distribution patterns which are related both to size (so that larger cities have higher location quotients than smaller ones), and to the disproportionate development of these sectors in certain cities. The location of key transport centers is related to network development. Aguascalientes, San Luis Potosi, Torreon and Veracruz were among the country's leading transport nodes. Similarly, the historic importance of state capitals such as Chihuahua, Guaclalajara,  - 64 - Monterrey, Culiacan and Veracruz explains the geographic distribution of cities which were relatively important centers of government and administration. (ii) Contrasts Between Cities in the Secondary Sector 234. The distribution of industrial subsectors among the 37 cities is shown in Tables 7.1 and 7.2. Some were highly localized while others were ubiquitous. In general, food, beverages,furniture and paper approached ubiquity while the extractive industries, tobacco, textiles, leather, oil, basic metals and vehicles were relatively localized. 235. While some subsectors had high degrees of concentra- tion, unrelated to the size of the cities concerned, others (mostly consumer goods industries) were distributed in rough accordance with differences in urban size. There were, how- ever, relatively high location quotients in foodstuffs in the northwestern cities and also in Irapuato - implying a direct link with agricultural development in the surrounding areas. Other high location quotients occurred in the tobacco subsector (Cuernavaca) and the wood and cork subsector (Oaxaca). 236. By comparison, the location quotients in the "heavier" and more modern industries such as chemicals, oil, non-metal mineral processing, basic metals, metal products, machinery, electricals, and vehicles and in the mining sub- sectors were generally higher. Chemicals were heavily con- centrated in Mexico City, oil in Coatzacoalcos and Tampico (Ciudad Madero), non-metal minerals and basic metals in Monterrey, and vehicles in Toluca, Queretaro and Monterrey. 237. Mexico City accounted for 67 percent of total industrial employment among the 37 cities, Monterrey (8 per- cent), and Guadalajara (5 percent). At a subsectoral level, the pattern varied considerably, so that Mexico City accounted for more than its average share in the whole sector (i.e., a location quotient greater than one) in textiles, wooden furniture, paper, printing, linoleum, chemicals, metal goods, machinery, electricals and vehicles. Its ostensibly large share (notwithstanding a low implicit location quotient) of the hydrocarbons sector (30 percent) is attributable to the location of the headquarters of Pemex, the state-owned oil company, rather than to the presence of extraction of refining activities.  - 65 - 238. Table 7.8 shows that most of the relatively smaller cities (among the 37) had fewer subsectors than the larger ones. The full range of subsectors (25) was present only in Mexico City (Table 7.9). Minatitlan - which was the next to smallest city - had only 14. (iii) Contrasts Between Cities in the Tertiary Sector 239. Table 9.4 shows the distribution of commercial activity, measured in terms of employment, in the largest cities of the 1970s. The aggregate picture emphasizes the importance of Mexico City, Guadalajara and Monterrey. 240. Table 9.5 shows implicit subsector location quotients for each branch of the commercial sector. These indicate the relatively disproportionate importance of the capital in sales of machinery and other equipment. They also show that its share of total sales of household and primary goods was in line with its share of total employ- ment. 241. This pattern indicates (1) the relatively high level of activity in food sales in the agricultural centers of the northwest, which is consistent with the relative industrial specialization of these cities; (2) the relatively high level of activity in sales of fuel in the border cities - presumably associated with border trade; and (3) a relatively high level of activity in sales of machinery and tools in Monterrey, suggesting linkages between the industrial and commercial sectors of the urban economy. 242. Other general indicators of commercial activity did not show the same close association which has been identified between urban size and sectoral structure. Table 9.9 shows that some of the relatively smaller cities (Culiacan, Hermosillo, Irapuato, Coatzacoalcos) had high levels of sales per worker in commerce. Conversely, Guadalajara and Mexico City had relatively lower levels of sales, with the lowest levels occurring in Oaxaca and Acapulco. 243. Basic data for the services sector (i.e., SIC subsectors 81 to 89) are shown in Table 9.10. This suggests a pattern of activity which was generally in line with variations in urbean size. Mexico City, Guadalajara, Monterrey, Acapulco, Tijuana, Ciudad Juarez and Merida are also important in absolute terms.  - 66 - 244. Table 9.11 gives the percentage share of each city by service subsector, and points to the large shares of Tijuana in recreation, Acapulco in hotels and Ciudad Juarez in restaurants. There was not, however, a high degree of concentration in education, health or professional services, which were, in general, distributed in close accordance with urban size differentials. The aggregate share of Mexico City, Monterrey and Guadalajara in each subsector was fairly similar to the average for all subsectors. Professional services (74 percent in the 3 cities), personal services (71 percent), medical services (72 per- cent), and education (72 percent), were relatively more concentrated, while the shares of total employment in recreation (52 percent), hotels (46 percent), and restaurants (46 percent) were lower than the combined average share of total employment in services. (b) Contrasts Within Cities (i) General Contrasts Within Cities 245. Tables 4.6 and 4.7 show employment in each sector in each city in 1940 and 1970. In both years the manufacturing sector occupied a dominant position, accounting for an average of 25 percent of total employment in 1940 and an average of 29 percent in 1970. By comparison, the average shares of other sectors in total employment were generally low, with the important exceptions of commerce (23 percent), and services (16 percent) in 1940 and services (31 percent), and commerce (16 percent) in 1970. To the extent that the growth of services was a consequence of the reclassification of certain sectors within the intersectoral structure, this expansion was more apparent than real (Technical Annex 7). It was also somewhat illusory because it disguised the proliferation of marginal activity. 246. Aggregating the hydrocarbons, mining, manufacturing, construction and electricity sectors, and the commerce, transport, services and government sectors as comprising, respectively, the "secondary" and "tertiary" sectors, a minor change between 1940 and 1970 concerned the larger share of the tertiary sectors in total employment in the latter year (53 percent versus 50 percent). More important changes occurred within each of these aggregated sectors. Manufacturing increased its average share of "secondary" employment from less than 50 to 60 percent, while the share of services in "tertiary" employment rose from 30 to 60 percent. The shares of hydrocarbons and mining in the "secondary" sector, and of commerce in the "tertiary" sector, declined.  - 67 - 247. Sector shares varied widely. Among the "secondary" sectors in 1940, Leon (55 percent in manufacturing), Orizaba (51 percent in manufacturing), Minatitlan (42 percent in hydrocarbons), and Pachuca (34 percent in mining), had the most "unbalanced" sectoral structures. Several other cities, including Irapuato, Guadalajara, Monterrey, Oaxaca, Puebla, Queretaro and Culiacan also had higher than average levels of employment in manufacturing, which thus accounted for more than 25 percent of all employment in those cities. 248. Among the "tertiary" sectors there were high concentrations of employment in 1940 in Ciudad Juarez (32 percent in commerce), Tijuana (36 percent in commerce), Merida (33 percent in commerce), Matamoros (32 percent in commerce), Aguascalientes (20 percent in transport), and Cuernavaca and Mexico City (20 percent each in services). 249. In 1970, high "secondary" sector shares occurred in Leon (53 percent. in manufacturing), Monterrey (37 percent in manufacturing), Orizaba (36 percent in.manufacturing), Guadalajara (32 percent in manufacturing), Puebla (32 per- cent in manufacturing) and Saltillo (30 percent in inanu- facturing). Hydrocarbons by this time accounted for no more than 27 percent of total employment in Minatit:Lan. 250. Among the "tertiary" sectors in 1970, services provided between 30 and 35 percent of total employment in Mexico City, Meridai, Jalapa, Veracruz, Nuevo Laredo, Matamoros, Villahermosa, Hermosillo, Ciudad Obregon, Mazatlan, Culiacan, Oaxaca, Cuernavaca, Morelia, Chihuahua and Tijuana and some 40 percent of total employment in Acapulco. 251. Table 4.3 shows that, notwithstanding the increase in the relative share of manufacturing in total employment between 1940 and 1970, the growth rate of manufacturing was significantly lower than the rates of the other "secondary" sectors and was also lower than that of services, which registered the fastest growth rate (8.3 percent) of all sectors during the period. The discrepancy arises from the relatively large share of manufacturing in 1940, so that despite its s:Lower growth, it still increased its share of total employment. Services also began with a relatively large share in 1940, grew rapidly thereafter and, as a result, experienced a larger net increase.  - 68 - 252. Table 4.3 shows the compound growth rate for each sector in each city in 1940-70 and points to some dramatic changes. Among those which stand out are the increases of 30 and 19 percent, respectively, in hydrocarbons employment in Reynosa and Villahermosa, the growth of 14 percent and 13 percent in mining in Queretaro and Matamoros, the increase of 11.7 percent in manufacturing in Tijuana, the increase of 18 percent in construction in Durango and the increase of more than 10 percent in services in Mexicali, Tijuana, Ciudad Juarez, Durango, Acapulco, Monterrey, Ciudad Obregon, Hermosillo, Matamoros, and Reynosa. Also of interest are the negative growth rates in mining (Pachuca, Chihuahua, Torreon, San Luis Potosi), in commerce (Merida), and in government (Jalapa, Minatitlan). Some of these reflect population increase, while others reflect structural shifts in urban economies. 253. Tables 4.8 and 4.9 illustrate the basis on which each urban economy can be designated as "diversified", "semi-diversified", "semi-specialized" or "specialized". 1/ This classification is based on the Index of Specialization. (See Annex 2) 254. Table 4.8 shows that, in 1940 seven cities were "specialized", whereas in 1970 (Table 4.9) there was only one city in this category. Similarly, the number of "semi- specialized" cities fell from seven in 1940, to three in 1970 while the number of "semi-diversified" cities rose from 5 to 12 and the number of "diversified" cities increased from 18 to 20. This suggests a relationship between increas- ing size and decreasing specialization with respect to the aggregate urban economy, in the sense that absolute increases in population size were associated with declining indices of specialization. 255. Tables 14.4 and 14.5 confirm the existence of a general (but weak) relationship between specialization and urban size - in 1940 and 1970. These weak correlations suggest there was no particular link between economic structure (defined in terms of specialization) and population size in either year. However, in 1940 the very largest cities (Mexico City, Guadalajara and Monterrey) had relatively 1/ The numerical values of the index are divided into a four-part scale for this purpose: "diversified" = less than 0.3; "semi-diversified" = 0.3 to 0.4; "semi- specialized" = 0.4 to 0.5 and "specialized" = 0.5 and above.  - 69 - diversified economies while some of the smallest cities (Minatitlan, Coatzacoalcos) had high indices of specializa- tion. This pattern of contrast at the extremes of the size distribution scale also occurred in 1970 (notwithstanding the absence of a close overall relationship between population size and economic specialization). Mexico City was thus the largest and least specialized city and Minatitlan (one of the smallest cities among the 37) the most specialized. Again however, some of the smaller cities (Queretaro, Cuernavaca, Irapuato, Mazatlan, Pachuca, Nuevo Laredo - all of which had populations of less than 200,000 in 1970) had rather low indices of specialization; at the same time, such larger cities as Leon (420,000) and Tampico (276,000) had fairly specialized economies. 256. Similarly, while the correlations between economic specialization in 1970 and population growth in 1940-70 were low, the data in Tables 14.6 and 14.7 point to a general contrast between extremes. For example, Minatitlan had a slow growth rate in 1940-70 and a specialized economy at the beginning of the period (1940), while Mexico City and Hermosillo, which had diversified economies in 1940, were among the fastest growing cities in 1940-70. 257. All of the cities which grew at above the average rate (for all 37 cities) in 1940-70, had much lower indices of specialization in 1970 than in 1940; whereas in some cities which grew most slowly in 1940-70, there was a relatively smaller decline. The data point to a contrast between Mexicali, Tijuana, Cuernavaca, Monterrey and Nuevo Laredo and Minatitlan and Pachuca. But, even in this respect, there were contradictory patterns, inasmuch as in cities such as Acapulco - the second fastest growing city in 1940-70 and the fastest growing of all in 1960-70 - there was an increase in the index. This leads to the conclusion that specialization and diversification were not consistently related to urban size or urban size growth in 1940-70. 258. Urban growth theory maintains that the "city forming" sectors of an urban economy are those which generate "export" surpluses from which employment and income multipliers are derived. "City forming" activities at the level of the urban economy are identified by an index of surplus employment applied to employment data for all sectors in 1940 and 1970 except agriculture (See Tables 4.10 and 4.11).  - 70 - 259. The pattern for 1940 stresses the importance of manufacturing and commerce in the growth of most cities and, less importantly, that of transport and hydrocarbons. By comparison with 1940, 1970 data suggest that commerce had become a more generally important "growth" sector, and that government was also far more important in 1970 than in 1940. Manufacturing, transport and hydrocarbons continued to generate high "surpluses" in many cities. 260. In many respects the relative sector emphases of 1940 shown in Table 4.10 were congruent with the sector emphases revealed by the index of sectoral specialization in Table 4.8, at least in the specialized and semi-specialized economies. The role of transport in Aguascalientes, revealed by the index of surplus employment, reflected the high level of relative specialization in this sector of the urban economy. Similarly, the large "surplus" in hydrocarbons in Tampico, Coatzacoalcos and Minatitlan is congruent with a high degree of specialization in hydrocarbons. A similar congruence occurred with respect to mining in Pachuca and manufacturing in Leon and Orizaba. 261. The index of surplus employment shows not only the absolute but also the relative importance of different sectors. In the more diversified urban economies, it reveals that in 1940, manufacturing was the major "surplus" sector in both Monterrey and Guadalajara and that commerce was the major "surplus" sector in Ciudad Juarez and Merida. For 1970, the relative "surplus" values in the specialized and semi-specialized economies generally coincide with those revealed by the index of specialization (hydrocarbons in Coatzacoalcos and Minatitlan, manufacturing in Leon and services in Acapulco). The values shown in Table 4.9 for the non-specialized economies also indicate the sector(s) dependent upon non-specialized economies for growth. As might be expected, they generally had a larger number of "export" sectors than those which were more specialized. 262. Further analysis of the results of the index of specialization and the index of surplus employment for each of the 37 urban economies in 1940 and 1970 is contained in Annex F. Although urban size growth was not consistently related to sectoral structure, certain trends emerge. Manufacturing was a more "important" sector in larger cities. It was usually a dominant sector in that it was not generally associated with other major export sectors, and it was the major export sector in all of the largest cities in 1970: i.e., Mexico City, Monterrey, Guadalajara,  - 71 - Puebla and Leon. These were not, however, the fastest growing of the 37 cities. Thosewhich grew faster were, without exception, associated with the rapid growth of the tertiary sector. Finally, the cities which, in 1940, had been closely and uniquely identified with mining (Pachuca) or hydrocarbons (Minatitlan, Coatzacoalcos) were, with the important exception of Coatzacoalcos, among the slowest growing urban places in 1940-70. One city (Reynosa) became a "hydrocarbons city" during the period and was one of those with faster than average growth. (ii) Contrasts Within Cities in the Secondary Sector 263. The percentage share of each subsector in each city in relation to total employment in mining and manufacturing is shown in Table 7.5. This emphasizes the extent of the domination of the industrial employment structures of parti- cular cities by specific industries. Among those in which one sector accounted for 50 percent or more of all employ- ment in mining and manufacturing were: Pachuca (metallic mineral mining); Culiacan and Irapuato (food); and Leon (footwear). In addition to these highly "specialized" cities, a large numbe!r had more than one-third of all industrial employment concentrated in one sector. These included (apart from those already mentioned): Morelia, Puebla, Mazatlan, Ciudad Obregon, Hermosillo, Villahermosa and Matamoros (food); Acapulco (beverages); Orizaba (textiles); Coatzacoalcos (chemicals); Nuevo Laredo (electrical) and Toluca (vehicles). 264. Contrasting subsectoral structures and subsectoral "mixes" have an important bearing on the nature of industrial development. Although it is impossible to study the inter- industry linkages - because of data constraints - it is possible to identify certain relationships between different subsectoral "mixes" and other characteristics of industrial development. 265. The industrial "mix" of each city was classified on the basis of the relative importance (in terms of employ- ment) of three broad subsector groups: "extractive" industries (subsectors 11 through 16), "final demand" industries (sub- sectors 20 through 30) and "intermediate demand" industries (subsectors 31 through 38).  - 72 - 266. The emerging pattern (Table 7.10) shows that some cities (Pachuca and Minatitlan) were quite heavily "dependent" on extractive industries but that most cities were primarily dependent on either "final demand" or "intermediate demand" subsectors. 267. The former category included Aguascalientes, Cd. Juarez, Durango, Irapuato, Leon, Acapulco, Guadalajara, Morelia, Cuernavaca, Oaxaca, S.L. Potosi, Culiacan, Mazatlan, Cd. Obregon, Hermosillo, Villahermosa, Reynosa, Orizaba, Jalapa and Merida. Over half of the industrial sectors of these cities included industries whose products were final demand consumer goods: food, beverages, tobacco, textiles, footwear, wood and cork products, furniture, paper, printing and leather. Clearly, there were variations in terms of the relative importance of material costs and distribution costs, and some subsectors (tobacco, textiles and wood products for example) were not located near accessible sources of materials. None of these subsectors required major inter- sectoral inputs, so "linkages" were rather unimportant. 268. By comparison, the subsectors which have been termed "intermediate demand" industries, have relatively complex intersectoral linkages. Here too there are subsectoral variations in assembly and distribution costs, particularly in the basic metals industry where the historical record shows that the rationale for the location of the steel industry varied from one plant to another. All of the other subsectors in this broad category were dependent on inputs from other industries; the cities most dependent on these subsectors were Saltillo, Toluca, Monterrey, Queretaro, N. Laredo, Tampico, Veracruz and Coatzacoalcos. Subdividing further (Table 7.10) and into metal using and other "intermediate demand" industries, Saltillo, Toluca and Queretaro fall into the former category and Coatzacoalcos and Veracruz, into the latter. 269. There is no relationship between the subsectoral "mix" as thus defined and urban size or geographical distribution. The ratio between final demand and inter- mediate demand industries ( Table 7.11) suggests that "final demand" industries were more important in smaller cities. There were however some exceptions, such as Queretaro, Toluca and Coatzacoalcos where the relative strength of intermediate demand industries was consider- able in spite of the small size of the city. All of the large cities, however, (Mexico City, Monterrey, Guadalajara, Puebla, Torreon), had fairly low ratios.  - 73 - 270. There is no general pattern for differential urban growth rates in 1940/70 and different subsector "'mixes" in 1970. The fastest growing cities in 1940/70 were often "specialized" in "final demand" industries, although some (including Monterrey) were relatively specialized in intermediate demand industries. 271. From this, one would expect that cities with large concentrations of employment in one industry would be "specialized" rather than "diversified" and thus dependent on the industry concerned. 272. In order t:o measure the degree of relative industrial diversification, a form of the index of specialization was used (see Technical Annex 2) showing the extent to which a given city resembled or differed from the 37 cities EIs a whole in this regard. 273. Table 7.7 shows the results of this analysis: values range from .:159 for Mexico City with a very diver- sified industrial "profile" to 1.27 for a specialized center such as Minatitlan. The index provides a basis for labelling each city's industrial sector as "diversified", "semi-diversified", "semi-specialized" or "specialized". 274. There was some relationship between the index of industrial diversification and urban size. At the extremes, Mexico City had the lowest index and the largest population while the next lowest index values were those of Guadalajara and Monterrey which were, respectively, the second and third ranking cities. Conversely, some of the smallest cities (Minatitlan, Orizaba, Oaxaca, Pachuca, Irapuato) had. high indices of specialization. There were, however, several exceptions such as Leon, Acapulco, veracruz and Tampico, which were large and specialized, and other exceptions such as Durango, Toluca and Jalapa, which were small and diver- sified. 275. Thus, the overall coefficient of correlation between size and diversity was not very high (Table 14.8). The index was fairly well correlated with the number of subsectors established in each city. This relationship is less tautological than it might seem, because the existence of a subsector in a city is not necessarily a reliable indicator of its contribution to the diversity of the industrial structure. Diversification was generally based on a relatively large number of subsectors,although the extent of diversification varied according to the "mix" of subsectors in a given city. The subsectoral values, of  - 74 - the index of specialization are shown in Table 7.6. These generally correspond to the percentage shares of subsectors within the industrial sector of each city (Table 7.5) and high indices thus occur in: Pachuca (metal mining), Irapuato, Morelia and Culiacan (food); Acapulco (beverages); Oaxaca (wood and cork products); Orizaba (textiles); Veracruz (basic metals); Tijuana and Nuevo Laredo (electrical) and Toluca (vehicles). 276. Several ratios were devised to assess aggregate differences in industrial efficiency at the city level. Several "indicators" of efficiency in tables 7.12 and 7.13 reveal that large average plant sizes, and high levels of capital investment per plant had a generally close relation to productivity per worker (as seen in Mexicali, Saltillo, Torreon, Guadalajara, Toluca, Monterrey, Queretaro, Culiacan, Ciudad Obregon, Hermosillo, Coatzacoalcos, Orizaba, Veracruz, Merida and Mexico City). The subsector mixes varied, from a heavy food industry emphasis in Culiacan, Ciudad Obregon, Hermosillo and Mazatlan, to a capital goods emphasis in Saltillo and Veracruz. While some of these cities (Queretaro, Veracruz, Monterrey, Toluca) have been previously identified with "intermediate demand" subsectors, others (Ciudad Obregon, Coatzacoalcos, Orizaba) have been found to emphasize "final demand" industries. This suggests that the industrial "mix" was not directly associated with relative productivity. 277. Apart from some exceptions, such as Irapuato which had a relatively low level of productivity and a capital- intensive industrial sector, there was also a fairly close relationship between the ratio of wages to value added and variations in productivity. The highest level of aggregate productivity (Ciudad Obregon), coincided with the lowest proportion of wages and salaries in value added (.22), while one of the lowest levels of relative productivity (Reynosa) (.79), coincided with one of the highest (.50) proportions of wages and salaries to value added. The highest proportion of wages and salaries in value added occurred in Pachuca (.61) which also had a very low level of relative productivity (.85). 278. Table 7.13 does not suggest a close relationship between productivity and city size. Among cities in the 100- 200,000 size cohort, productivity ranged between .79 and 4.40 (the highest and lowest of all 37 cities) while the average for all cities in this group was 1.78. In the next largest size cohort (400-800,000), the average was lower. Two cities had values of less than 1.00. Finally, in the largest size cohort, the average was higher (2.75) but there were only two cities in this group.  - 75 - (iii) Contrasts Within Cities in the Tertiary Sector 279. In most cities the largest share of tertiary activity was in commerce, with the largest subsector being household goods (Table 9.6). This accounted, on average, for 48 percent of all commercial employment. There were some deviations, such as the low share (2 percent) of this sector in Matamoros, in association with the unusually high share of primary product sales in that city (23 percent versus an average of 15 percent), and there was a similar pattern in Ciudad Obregon. In general, the distribution of commercial activity was consistent from city to city.. 280. The high correlation (r2 = .97) between urban size and the index of specialization in commerce is to be expected. The largest cities (Mexico City, Monterrey, Guadalajara and Puebla) had the most. diversified commercial sectors, and the smallest cities, the most specialized: Pachuca (.43), Ciudad Obregon (.43), Matamoros (.44), Minatitlan (Table 9.7). This is congruent with the assumption that tertiary functions were generally linked wit:h urban population size and, as a whole, the tertiary sector would be larger and less specialized in larger than in smaller cities. 281. There was another very high coefficient of correlation (r = .99) between urban size growth and specialization. This is moresignificant (Table 14.12) because it strongly suggests that the most specialized cities grew slowly, irrespective of size, and the least specialized cities (also irrespective of size), grew quiclcly. 282. The occurrence of a high overall index in sub- sectoral specialization (Table 9.7) was, in most instances, "explained" by relative specialization in household goods. In Minatitlan, however, a high index in the food subsector was the essential cause of a high overall value. Without subsector data beyond the two digit level, it is impossible to refine the analysis of the relationship between size and population threshol(is with respect to commercial activities. One implicit indicator, however, is the relationship between population size and functional specialization. This is consistent with the theoretical premise that the larger the size of the urban population, the larger the market and the more diverse the demands for different commercial activities. Reciprocally, the larger the market, the greater the probability that a given subsector can survive.  - 76 - 283. Table 9.8 shows the "surplus" activity in each subsector in each city and suggests that the largest volume of "export" activity was generally in the food subsector, although this was not true for Mexico City or Monterrey. While there was not a consistent relationship between size and "export" surpluses, Mexico City had by far the largest surplus. This shows the role of commerce as an "export sector" in the Federal District. The export surplus of Monterrey was also large relative to its size. The surpluses of the frontier cities were even larger, suggesting that international rather than domestic demand has been the major key to commercial development in these cities. 284. The tertiary sectors of the frontier cities - and of certain other cities - also differed from others in that their service industry components and variations are re- flected in implicit location quotients (i.e., subsectoral shares within cities) in Table 9.12. In Acapulco, for example, 64 percent of total employment in services was in the hotel subsector, which also accounted for large shares of total employment in Cuernavaca, Oaxaca, Mazatlan, Villahermosa and Veracruz - all of them tourist centers. In Puebla, 50 percent of employment in services was in the restaurant subsector and this industry also accounted for 21 percent of employment in services in Ciudad Juarez. Recreation accounted for 48 percent of all employment in services in Tijuana, 26 percent in Ciudad Juarez, 22 per- cent in Matamoros, 30 percent in Reynosa and 22 percent in Minatitlan. All except the latter city are located on the US border and serve as major centers of "border tourism." 285. The average distribution between subsectors was rather even, with the highest average share for all 37 cities in personal services (14 percent), and the lowest in medical services (5 percent). Nevertheless, some cities had fairly high levels of relative specialization. Table 9.13 shows a range of specialization from less than 0.2 (Guadalajara, Monterrey, Hermosillo, Tampico) to more than 1.0 (Acapulco). 286. Specialization may occur because of the development of a highly specialized service industry such as tourism, or because a city is insufficiently large or insufficiently developed to support a full range of urban services. Examples of the first are Acapulco, Mazatlan and Cuernavaca; all had large tourism industries and high indices of service sector specialization. Tourism thus accounted for a high level of sectoral specialization in four out of the ten cities with indices of 0.50 or above. Of the other six cities in this  - 77 - category, five were on the US border. In four of them the recreation and entertainment sector was the basis for a high level of specialization and in the other (Mexicali), the high index was associated with the food and restaurant industry, which was closely related to entertainment and recreation. The two remaining cases are Aguascalientes and Minatitlan. 287. As in the commercial sector, a very close relation- ship exists between increasing size and decreasing specializa- tion (Table 14.13). This is consistent with the suggested interpretation of high indices in cities, such as Minatitlan, which were small andl unable to provide more than nominal support for certain services. With regard to the cities which had "legitimate" service specializations, the congruence of size and specialization is explained in terms of their relatively small size. 288. Finally, Table 9.14 shows that the index of specialization provides a reasonably good indicator of the sectors which generated the largest "export surpluses" in each city. Among the outstanding instances of high levels of "export surplus" are those of: recreation in Tijuana; hotels in Acapulco, Cuernavaca, Oaxaca and Mazatlan; educa- tion in Guadalajara and Monterrey and Mexico City; restaurants in Ciudad Juarez and Puebla and professional services in Mexico City. Consumption (a) National Trends 289. Urbanization after 1940 was associated with major changes in the level of distribution of the consumption of goods and services. Trends of consumption (measured in terms of aggregate levels or of incidence) were linked to the growth of output, investment and exports, but there was not necessarily a relationship between the level or incidence of consumption and distribution patterns (either interpersonal or interregional). This applied to "private" goods and services (i.e., those exchanged in the market economy), and "public" goods and services (i.e., those produced and dis- tributed by the public sector which were generally "free" to the consumer). 290. Nationally, the trend of improvements in welfare conditions was roughly parallel with that of urbanization. Insofar as welfare depends on the level of public sector services (a valid assumption with respect to education and health services), this may imply that urbanization facilitated  - 78 - the provision of services by opening up the possibility of scale economies in urban areas which could not be achieved in rural areas. Urbanization coincided with the growth of public sector "social" services and with more efficient distribution. 291. Insofar as economic welfare was a function of conditions in the market economy and of the growth of personal incomes (a valid assumption with respect to food, footwear and non-basic consumption), urbanization was also associated with the growth of personal income (Figure 3.1). 292. The trend of urbanization was at variance with the trend of national income distribution (Figure 3.2). This suggests, a priori, that urbanization was associated with a deteriorating distributive trend which may well have been associated with changing sector shares and differential growth rates in urban as compared with rural sectors of the economy. (b) Contrasts Between Cities (i) Income Levels and Distribution 293. Data to analyze welfare trends in the 37 cities from 1940-70 do not exist, but the level and distribution of income and the level of welfare at the end of the period (1970) can be measured. 294. Data on disposable family income in the 37 cities for 1969-70 show substantial contrasts in absolute levels. The ratio of the highest income (Mex$2,255 in Tijuana) to the lowest (Mex$871 in Oaxaca) was 1:2.5. The Gini co- efficients also show considerable variation - from .62 in Puebla to .45 in Mazatlan. The higher the income level, the more equitable the distributive pattern; i.e., the lower the Gini coefficient (Table 14.19). 295. With respect to income distribution, there was no special relationship between urban size and the Gini coefficient, the coefficient of correlation being only 0.09 (Table 14.20). Nor was there a close relationship between the Gini coefficient and urban size growth (Table 14.18). This implies that fast growing cities generally had high income levels; income was not, in general, well distributed although there was a contrast in this regard between growth rates in 1960/70 and in 1940/70. The correlation of the latter with the Gini coefficient for 1970 suggests that cities which, in the longer run, had grown most quickly, had better income distribution than  Average Monthly Income Against National Urban Population as % Total Population 1950- 70 %. 70 60 - 50 40 - 30 - 20- 10 $900 $1000 $1100 $1200 $1300 $1400 $1500 $1600 $1700 $1800 World Bank-15572  Gini Coefficient Against National Urban Population as % of Total Population 1950-70 70- 60 50 40I 3C _~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~C 30 20 10 O l l l l l l l I .. .50 .51 .52 .53 .54 .55 .56 .57 .58 Gini  - 81 - those which had grown more slowly. Recent rapid growth appears moreover to have had a worsening effect on the distributive structure. 296. There was a clear relationship between contrasts in income level and differences in economic structure in terms of the sectoral structure of employment. In general, the fast-growing industrial cities were better-off than others, and fast-growing cities with primarily commercial economies (such as Tijuana) also had relatively high average income levels. Generally, the economic dynamism of the city was the predominant factor in accounting for income differentials, although at the sub-sector level the industrial cities which were associated with "modern" branches of manufacturing were more dynamic than those with more "traditional" industrial structures. They had consistently higher income levels than other cities - with the exception of those on the U.S. frontier. 297. With respect to the relationship between income and unemployment, the 1970 census shows that only 3.7 percent of the economically active population (defined as all those aged 12 and over - equivalent to 13 million people) was openly un- employed. But only 81 percent of the nominally active population was employed for more than 9 months of the year and a large share were engaged in activities at low levels of productivity, i.e., they were employed but they were only marginally productive. 298. About 45 percent of the labor force in 1970 is estimated (Trejo et al 1973) t:o be marginally employed (includes all activities in which monthly income is less than the minimum wage). Those concerned were mainly "engaged" in agricultural or tertiary activities. 299. Table 14.21 shows varying levels of underemployment (defined as inactivity over a 9-month period). Cities with high levels of underemployment were Mexicali, Oaxaca, Culiacan and Villahermosa. Some of the suburbs of Mexico City also had very high levels, rising to 60 percent in one instance. Cities with low levels of underemployment were Monterrey (16 percent), Tampico (19 percent) and some of the other suburbs of Mexico City (as low as 15 percent). 300. There was no apparent relationship between urban size and underemployment. Implicitly, however, there was a close relationship between urban growth and underemployment in several cities, including Hermosillo, Oaxaca, Villahermosa, Culiacan and Acapulco, but underemployment was also high in less demographically dynamic cities such as Toluca and Ciudad  - 82 - Obregon. Rapid population growth has often been associated with the development of economic activities which were not labor absorptive. A test of this hypothesis derives from the incidence of underemployment in the fast growing cities in fast urbanizing states, in which services were a "leading" urban sector, such as in the states of Baja California, Sonora, Chihuahua, Tamaulipas and Nuevo Leon - all in the north - and to some extent also in Veracruz in the south. Table 14.21 shows the percentage level of underemployment in each of the cities (among the 37) in these states, the identity of the leading sector in each state and city and the rate of urban population growth in both the states and the cities concerned. In all of the cities of the six states, the level of underemployment exceeds the average for the largest cities as a group. The major cities in Baja California - Tijuana and Mexicali - are heavily specialized in tertiary activity. Both had relatively low participation rates - Mexicali in particular. In both cities the minimum wage was relatively high. Mexicali was strongly associated with the service sector, and at the same time had the highest minimum wage of any city in the country and the highest level of underemployment. It was also one of the fastest growing cities in Mexico in 1940-70. The data in Table 14.21 suggest that a similar, though less consistent, association between underemployment, rapid growth and service sector development could be made for a number of other cities. (ii) Other Indicators of Welfare 301. Differences between cities with respect to income and income distribution imply contrasts in living conditions although, theoretically, public expenditures could be used as a countervailing instrument to offset differences in income levels and thereby affect living standards. 302. Four types of indicators were used to examine the pattern of living conditions in the 37 cities: health and nutrition, housing, education, and non-basic goods and services (telephones, electricity and automobile ownership). Table 12.2 presents the basic data. 303. The pattern of non-meat consumption revealed that Durango and San Luis Potosi ranked lowest, while Mexico City,  - 83 - Cuernavaca, Puebla, Veracruz and Villahermosa fared the best. 1/ The infant mortality index showed a range around the central value of Mexico City (1.00) and from 0.32 in Culiacan to 1.82 in Leon, a pattern for which there is no obvious explanation. Although it was generally congruent with the pattern of meat consumption (r2 = .78) it supports the validity of the a priori relationship between nutrition and health. 304. Quantitative indicators of medical services show that among the 37 cities, the large cities do not always have the highest levels of services. Three indicators were examined: number of beds, number of doctors and number of nurses in public hospitals (SSA, 2/ state and municipal and IMSS 3/) per 10,000 people. The high ranking cities (most of the three indicators within the ten best) were: Saltillo, Torreon, Pachuca, San Luis Potosi, Hermosillo, Orizaba, Veracruz, Jalapa, Merida and Mexico City. The lowest levels (most of the three indicators within the seven worst:) were in Mexicali, Tijuana, Ciudad Juarez, Irapuato, Mazat:lan, Matamoros and Minatitlan. 305. Persisent difficulties in the housing sector in Mexico have arisen from continuing pressures on supply in view of the high rate of urban immigration and the i.imited scope of public assistance. 306. A 1962 housing survey showedthat 37 percent of urban housing covered an area of less than 50 m2, and that about 40 percent of the housing stock had only one room. It also showed that. about 6 percent of urban dwellings lccked access to water, 24 percent lacked drainage, and about 10 percent had no electricity. In the 1970 census, a large amount of housing still lacked access to water and drainage and about 10 percent still lacked electricity. 1/ There was an apparent relationship with meat production; most of the cities which ranked low on this index were either in meat-producing areas or had high income levels - Pachuca being the only exception. 2/ Ministry of Health and Social Security. 3/ Social Security Institute.  - 84 - 307. The housing "indicators" were not, in all cases, correlated with each other. The overcrowding index showed that the worst conditions were found in Acapulco (53 percent of all dwellings had only one room), and Villahermosa (where 50 percent of all dwellings had only one room), and the best in San Luis Potosi (18.7 percent of dwellings with one room). The three water supply indices were mutually compatible. They showed that the largest numbers of units without indoor water supply were in Mexicali (33 percent), Matamoros (33 percent), Mexico City (32 percent), and Oaxaca (35 percent). When related to population, the pattern was somewhat different. The percentage of population without access to water was 45 percent in Culiacan and Minatitlan and 53 percent in Villa- hermosa, and dropped to 6.2 percent in Monterrey and 8.2 percent in Mexico City. This implies that in some cities where access to water was poor (as in Mexico City), the aggregate level of available water was more satisfactory. In other cases (such as Tijuana) where the convenience level was relatively high, the aggregate availability of water was poor. There was a third category of cities, of which Oaxaca is an example, with approximately the same "convenience" and "aggregate" levels. 308. Except in cities with no access to water, the share of those without piped water and without indoor water supply was higher in Mexico City than in many smaller cities. The same was true of Monterrey but not of Guadalajara. Conversely, some of the smaller cities (San Luis Potosi, Durango) were relatively better-off in spite of means used to measure condi- tions. 309. There is general, but not conclusive, evidence that some of the fastest growing cities had the poorest housing conditions. This was conspicuously true in the border cities of Mexicali, Ciudad Juarez, Reynosa, Nuevo Laredo and Matamoros, all of which had more than an average share of population without indoor water in 1970. Some of the most demographically dynamic cities, which were also, in most instances, economically dynamic, had some of the poorest housing conditions. Conversely, slow growing and economically "backward" cities, such as Durango, had much better conditions. 310. Three indicators of primary education were examined: (i) number of students per class, (ii) the rate of graduation in the first and sixth grades and (iii) the percentage of the population without schooling. The cities with the smallest number of pupils per class (a proxy of schooling environment) were: Morelia, Mazatlan, Matamoros,  - 85 - Orizaba and Jalapa., Those which had the highest rate of graduation were Torreon, Guadalajara, Cuernavaca, Villahermosa, Tampico and Mexico City. For the third indicator, Chihuahua, Pachuca, Monterrey,, Orizaba and Mexico City had high scores. Generally, the first indicator and the other two do not coincide. The quality of primary education is reflected in the second indicator. The medium-sized cities generally had high to medium scores in other urban services, with the exception of Villahermosa. 311. The cities which had poor ratings on the first indicator (class size) were Irapuato, Monterrey, Nuevo Laredo, Coatzacoalcos and Minatitlan. In the second category (the passing rate of the first and sixth grades), low ratings were found in such cities as Durango, Toluca, and Merida. For the third indicator, the poorest group consisted of Irapuato, Leon, Acapulco and Queretaro. While those which had low scores on the second indicator generally had poor ratings on the other two indicators. In most of those which had poor scores on the third indicator, there was a fairly high level of migratory population growth during 1960/70. The cities with the highest percentage of the population with some secondary education included Monterrey (18 percent) and Mexico City (20 percent), while those with the lowest were Leon (8 percent), Irapuato (7 percent), Villahermosa (9 percent) and Minatitlan (9 percent). 312. On the higher education index, Monterrey (25 per- cent) and Mexico City (33 percent) scored highest, while Leon (8 percent) and Irapuato (6 percent) scored lowest. 313. The relationships between these indicators and urban size emphasized the relatively better-off status of the largest cities. There was no special relationship between education levels and population growth. Nor was there a general relationship to economic structure, although the index of professionally and technically qualified personnel per 1,000 persons appears to be related to (a) the significance of industry in the urban economy (Saltillo, Tampico), or (b) the existence of important administrative/ political functions in the city (Jalapa, Guadalajara, Mexico City), and in some cases to both (Puebla, Veracruz, Monterrey). 314. On the first index of telephone use (telephones per 1,000 population), Acapulco (89), Mexico City (87) and Cuernavaca (82) andi Monterrey (74) had the highest scores. The second and third indices which, respectively, include and exclude Mexico City, show patterns similar to the first as well as to each other.  - 86 - 315. This index is probably a fairly good measure of relative prosperity and/or economic structure. The former explanation seems in order for Mexico City and Monterrey. There is an a priori link with the importance of the tourism sector for Acapulco and Cuernavaca. The third index (exclud- ing Mexico City) shows a general relationship with relative economic change. Thus, Hermosillo, Mazatlan and Chihuahua have high indices while Oaxaca and Leon have low indices. In view of the national effort to improve the telephone network, the increase in the number of telephones during the past seven years has been such that cities which had a small number of telephones in service had generally high rates of increase (Queretaro, Reynosa, Coatzacoalcos and Minatitlan). Acapulco and Tampico, however, had both a high level of telephone service and a high rate of increase. There were no concomitant relationships between telephone installations and urban size or urban growth. 316. The Comision Federal de Electricidad (CFE) and the Compania de Luz y Fuerza del Centro, S.A. (Centro) are public enterprises which provide electric power throughout the nation. 1/ CFE generates power and distributes it everywhere, except in Mexico City and the surrounding area, which is supplied by Centro. All urban areas are covered by the supply network. In general, fast growing cities, with the exception of the largest and of Mexico City, appear to be less well off than others. This is consistent with the high demands of rapid population growth on the capacity of the public sector to keep up with the growing demand for public services. With some exceptions, particularly Mexico City, the public sector has not allocated resources to compensate for the consequences of rapid demographic growth. Nor has the public sector tried systematically to offset income differentials and the welfare consequences of contrast- ing economic structures by providing public sector services in urban areas which are unable to absorb an increasing labor force. 317. The highest level of residential electricity consumption occurred in Mexicali, Tijuana, Ciudad Juarez, Nuevo Laredo, Matamoros, Reynosa and Mexico City. The prominence of the northern border cities is striking. The lowest levels of consumption were in Durango, Toluca, Morelia and Oaxaca. The cities with the highest levels of residential consumption were not in all cases those with the highest percentage of homes with electricity. 318. Finally, the automobile indices show substantial differences. The absolute number of registered automobiles in each city showed a wide variation - from 620,800 in Mexico 1/ There is a small amount of private generating capacity too.  - 87 - City to 1,700 in Mi.natitlan. When related to population, Mexico City (72) had a lower index than Mexicali (82), and Minatitlan (18) had a higher value than several other cities, including Ciudad Juarez and Matamoros. This points to interesting cont:rasts among the border cities. When the value for Mexico City is excluded from the calculation, the dominance of Mexicali was greatly increased (2.22) while Tijuana (1.84), Mexico City (1.95) and Pachuca (1.89) also had high scores. The lowest values occurred in Matamoros (.30), Nuevo Laredo (.35) and Irapuato (.46). (iii) Income and Welfare: Conclusion 319. A factor analysis of socio-economic welfare condi- tions, based on selected indicators, is in Annex G. The highest overall mean score, implying the best socic-economic welfare conditions, was that of Mexico City, followed by Torreon, Veracruz, Tampico, Guadalajara, Chihuahua and Monterrey. The lowest was that of Minatitlan followed by Oaxaca, Culiacan and Villahermosa.- Key factors extracted by means of a principal component analysis indicate that two of the four leading factors showed particularly signi- ficant variations. These referred to (a) "conveniences" and (b)"necessities" in terms of the quality of urban life, the former category being weighted by such indices as electricity supply and telephone services, while the latter emphasized water supply, housing and nutrition indicators. There were strong inter-city contrasts with respect: to both. The border cities scored high in terms of convenierLces but poorly in terms of;necessities. Other northern cit:ies were, in general, well-off with respect to both factors. The cities of the central zone (the belt stretching from Guadalajara to Veracruz and including Mexico City) showed strong contrasts. Most cities in the western portion of this zone, with the exception of Guadalajara, were relatively badly off compared with Mexico City and other surrounding urban areas. Cities to the east of Mexico City, with the notable exceptions of Minatitlan and Villahermosa, also fared comparatively well. By contrast, cities to the south of this central zone were, without exception, less well off than those of either the north or the center with regard to both conveniences and necessities. 320. Though relationships between these contrasts in socio- economic welfare and the economic and demographic characteristics of the largest cities do not show consistent patterns, it is worth mentioning that the welfare index is positively correlated with city population and income level.  - 88 - 4. THE SPATIAL STRUCTURE OF TIIE URBAN SYSTEM 321. The spatial implications of the urban system may be analyzed in terms of the area of economic development and in terms of the regional aspect. 322. The area aspect concerns variations in development across the spatial surface of the national economy. It has certain constraints because it can only be measured in terms of statistical areas for which consistent time series data are available (includes the 32 states of the Federation). 323. The regional aspect is also based statistically on these units because the only means of developing a set of regions is to aggregate states. 1/ The key to regional demarcation is provided by the spatial structure of the urban system. The fundamental precept is that the spatial organization of a systemic economic structure is determined by the relationships between its principal cities, which are the major locations of production and consumption, and the most important central places in the national economy. For the purpose of regionalizing the country, the set of 37 cities is larger than need be, because the spatial structure can be adequately defined in terms of a smaller set of urban places. 324. The objective of this chapter is twofold: (1) to relate the patterns of area development from 1940 to 1970 to the level of urbanization in each state and to the spatial distribution of large cities between states; and (2) to construct a set of economic regions defined in terms of state aggregations and built around the spatial framework of the nation's urban system. A. Area Development and the Structure of the Urban System Demographic Structure 325. The interstate distribution of the urban population in 1940-70 is shown in Table 1.15. Certain states maintained almost exactly the same shares of the nation's urban popula- tion throughout the period. Thus, while the urban population of the Federal District increased from 1.6 million in 1940 to 6.6 million in 1970, and accounted for 23.9 percent of the nation's urban population in the former year, it accounted for 23.5 percent in the latter. Jalisco also retained a roughly constant share of the urban population (8 percent), as did Veracruz. But there were important changes in Baja 1/ It would, in theory, be possible to do so on the basis of municipalities but this would be extremely time consuming.  _ 89 - California (0.6 percent to 2.6 percent), Coahuila (4.0 per- cent to 2.9 percent), Nuevo Leon (3.4 percent to 4.6 percent), Puebla (5.2 percent to 4.1 percent) and Yucatan (3.0 percent to 1.7 percent). Most of the states with very smaLl shares of the urban population at the beginning of the period (Colima, Nayarit, Queretaro, Quintana Roo) also hadi very small shares at the end. 326. The absolute size of the urban population in each state in each census year and the average intercensal compound growth rates are shown in Table 1.16. These data explain shifts in the distributive structure. Baja California, for example, had the i-astest rate of urban growth of any state in the nation during 1940-70, but this growth began from a relatively low base in 1940 implying that its share of urban population was still fairly small in 1970. The relative rate of urbanization in each state and the rank order of the rate of urbanization in each intercensal period are shown in Table 1.17. 327. The states with the largest shares of the urban population in each year were not necessarily the most urban- ized. By 1970 the most urbanized states (classified as those more than 7( percent urbanized) accounted for a much larger share of the total urban population than in earlier years (Table 1.19). 328. The rate of urbanization at the state level was not closely assoc:iated with the share of the urban population at the beginning of each intercensal period, an implication confirmed by correlation analysis (Tables 13.16 and 13.30). Some of the least urbanized states in 1940 were among those which urbanized most rapidly during 1940-70, notably Baja California. Conversely, some of the most urbanized states (in terms of the share of the total urban population and of the level of state urbanization) were among those which urbanized rather :,lowly after 1940. In the Federal District, for example, the absorptive capacity of the state had been reached before 1970. Obviously, as the "level" of urbaniza- tion approaches 100 percent, the rate of urbanization must first slow down and eventually halt. 329. Changes in the relative distribution of the urban population and geographic contrasts in relative rates of urbanization necessarily reflect differential patterns of natural increase and migration (Annex H). Table 1.56 demonstrates that 1970's largest cities accounted for a very large proportion of urban demographic development. There were, howevzer, significant interstate contrasts derived from: (a) different rates of urbanization, and (b) intrastate urban structures.  - 90 - 330. There were several instances of slowly urbanizing states in which cities among the nation's 37 largest in 1970, did not account for a large share of the total urban population. These included Guanajuato, Hidalgo, Michoacan and Oaxaca. In addition, there were eight states or Federal Territories which did not include one of the 37 cities. These were: Baja Cali- fornia Sur, Campeche, Colima, Chiapas, Nayarit, Q. Roo, Tlaxcala and Zacatecas. Generally, there were few large cities in the slowly urbanizing states and even where large cities did occur, they did not account for an important share of the total urban population. 331. Conversely, the fastest urbanizing states had one or more of the nation's largest cities, which accounted for a large share of the urban population of the state concerned. There were some states (such as Yucatan and Puebla) in which the largest city or cities accounted for a large share of the urban population, notwithstanding that the state did not, during this period, undergo rapid urbanization. A large share of the state's urban population could thus be accounted for, either by rapid urban-size-growth or by the presence of a large city, irrespective of the latter's rate of growth. 332. The nation's largest cities generally accounted for all or most of the urbanization in their respective states. In the Federal District 100 percent of the urban population was located in the metropolitan area of Mexico City. Else- where, 75 percent or more of state urbanization was accounted for by the urban hierarchy in Baja California Norte, Jalisco, Mexico State, Nuevo Leon, Tamaulipas, and Queretaro. All were rapidly urbanizing states in 1940-70, and all were highly urbanized by the end of the period. 333. In sum, the geographic distribution of the major cities of the urban hierarchy "accounted for" major contrasts in the geographic distribution of the urban population by 1970, and spatial contrasts in urbanization were largely "accounted for" by the differential growth of the nation's largest cities. Patterns of Area Development (a) Area Development and Urbanization 3a4. Against the background of detailed data presented in Annex I, a composite developed index relates the spatial variations in urbanization.  - 91 - 335. The variables of the index are: (1) state product - per capita; (2) industrial output as a share of state product; (3) industrial employment as a share of total employment; (4) capital investment in agriculture; (5) irrigated areas as a share of the cultivated area; (6) electricity consumption per capita; (7) gasoline consumption per capita; (8) infant mortality per 1,000 live births; (9) sugar consumption per capita; (10) percentage of houses with water; (11) percentage of population with shoes; and (12) literacy. 336. Table 3.14 shows that state development resflected trends in the national economy and that progress was far from even. Twenty-four states maintained the same rank over 30 years. But Baja California Sur, Mexico, Tabasco, Sinaloa and Morelos all improved on their initial ranks while Yucatan, Quintana Roo and Durango lost in relative status. These changes, however, dlo not fundamentally alter the general pattern of relative stability. 337. The Federal District retained the first rank at all times. The same states were ranked 1-9 throughout the period, with the exception of Aguascalientes (in 1950) and Baja California Sur (in 1970). Similarly, Oaxaca remained at the bottom from 1940 through 1970 and the same states (Oaxaca, Chiapas and Guerrero) occupied the last three ranks each year. 338. In geographic terms, these data show that eight of the fourteen most developed states were in the north, and that three of the others were around Mexico City (Distrito Federal, Mexico, Morelos) and one was in the Pacific Southwest (Jalisco). In contrast, almost all of the poorest states were in the south. Most of the states in the "intermediate" category were geographically between these extremes. This suggests a three-tier pattern of development broken only by the fact that part of the central area (around Mexico City) and Jalisco, fell into the highly developed rather than the intermediate category. 339. Thus spatial contrasts in urbanization were consistently associated with spatial development. With respect to 1940 (Table 13.11), the relative index of state development is highly correlated with the level of urbanization (r2 = .75). In 1970 (Table 13.12), the association was even stronger (r2 = .85). This finding is confirmed by the fact that relative development measured on the basis of income level was also strongly related to urbanization (Table 13.13).  - 92 - 340. The fact that the degree of association between relative development and relative urbanization was close in both 1940 and 1970 but closer in the latter year than the former, suggests that a time trend analysis would reveal a close relationship between the trend of state development and the trend of state urbanization. 341. This hypothesis was tested by using the same twelve component index as a dynamic measure of relative development showing trends in the relationships between the development levels of each state over three decades (Table 3.15). 342. These measurements are based on the position of each state relative to the Federal District, the highest ranking state in each year, and the tendency towards increas- ing disequilibrium vis-a-vis the Federal District, in the states of the South, moderate disequiibrium in those of the North and a variable situation in the states of the center. (See Map 28). 343. Table 3.5 shows that the states which, in 1940, were relatively most developed, tended to maintain their status relative to other states. The trend for the whole period indicates that the states which suffered the least amount of change relative to the Federal District were the highest ranking in terms of differential development vis-a-vis the Federal District in 1970, although Baja California Norte stands out as an area which developed more rapidly than any other. Conversely, the lowest ranking states in 1970 were those in which the annual rate of differentiation vis-a-vis the Federal District was greatest - notably Quintana Roo and Zacatecas. However, some of the states which had intermediate levels of development also experienced rapid deterioration in relative status, the outstanding instances being Durango and Yucatan. 344. Thus differentiation increased over time and that differentiation was, in general, accentuated in those states which were, at the beginning of the period (1940), among the poorest. 345. A comparison of state trends in urbanization with state trends in development confirmed that there was a dynamic relationship between relative development and relative urbani- zation. Tests showed that the states (led by Baja California Norte) which urbanized most quickly after 1940 were those which developed most quickly. This corresponds to the pattern which has been characterized by Myrdal (1957) as one of "circular and cumulative causation".  - 93 - 346. Because urbanization and area developmentl was generally related, and urbanization and the urban hierarchy were generally related, a general association is presumed to exist between area development and the urban hierarchy. 347. The validity of this assumption is obvious in the case of the Federal District and in such states as Baja California Norte and Nuevo Leon where large and prosperous cities were located in (and accounted for the prosperity of) relatively prosperous states. 348. There were, however, some instances of disjunction between area development and the development of major cities. First, there were some states where the level of area development was higher than expected, assuming a general correlation between urbanization, the urban hierarchy, and area development; in all such cases, however, there was rapid urbanization in 1940-70. Second, there were some states in which the level of urbanization was low and compatible with the level of relative area development - but in which there was one or more large city. Ge:nerally, large cities develop associations with smaller cities in their hinterlands and with rural areas. Consequently, the state as a whole undergoes economic development. In some states, however, large cities were associated with relatively low levels of area development, which relates to the structure and the development of the rural sector. (b) Urban Development Without Area Development 349. Certain conditions govern the spatial diffusion or "spread" of the benevolent effects of urbanization. For example, a city whose locational rationale is the existence of a mineral resource but which is located in the middle of an arid and unirrigated plain may, for as long as the resource base endures and is in demand, continue to grow. Its development wiLll not however "cause" the development of the hinterland. Generally if the physical, economic and social conditions of an area surrounding a city do not facilitate the diffusion of development, such diffusion will not take place. In Mexico, the areaswhich lack favorable conditions for agricultural development generally are not penetrated by the diffusion of development impulses from the urban sector. In several states, the presence of a large city does not coincide with a high level of overall development at thes state level, and there is a disjunction between relative urban development in terms of urban size growth, and relative area development. The spatial structure of the agricultural sector seems to be the key to explaining such disjunctions.  - 94 - 350. Table 13.15 suggests that, in general, the less urbanized states also had low levels of agricultural develop- ment. Disjunction between urbanization and area development with a correlate in agricultural development is more apparent when urban development is considered in terms of the existence of large cities located in relatively unurbanized and under- developed states. Three cases are outstanding: the states of Mexico, Puebla and Guanajuato. In each of these states there is a large city. In the state of Mexico, there is part of Mexico City; in Puebla, there is the city of Puebla and in Guanajuato, there is the city of Leon. The overall level of urbanization in each of these states in 1970 was relatively low: 23 percent in Mexico, 28 percent in Puebla and 35 percent in Guanajuato. The overall development index was also relatively low: -3.1 in Mexico, -3.3 in Puebla, and -2.5 in Guanajuato. In each, the indicators of agri- cultural development were low. In terms of capital investment per agricultural worker for example Mexico, Puebla and Guanajuato ranked 23rd, 25th and 24th out of 32 states. In terms of labor productivity, they ranked, respectively, 19th, 30th and 32nd. In spite of the presence of large, industrialized urban areas in each state (which ranked, in the order given above, 2nd, 6th and 7th, with respect to rural population density), it seems likely that the condi- tions of the rural sector impeded the diffusion of the benefits of urban industrial growth. It also seems that the conditions of the rural sector may have impeded the diffusion of urbanization per se, inasmuch as in these states a very large share of the total urban population was concentrated in the largest city. (c) Area Development Without Urban Development 351. Spatial contrasts in the structure of agricultural production and income have influenced population movements and particularly outmigration. There has been a close relationship between outmigration and a series of indicators of rural/agricultural conditions. Heavy outmigration from Chiapas, Oaxaca and Guerrero was associated with a low relative level of agricultural development (Table 8.6). 352. At the same time, net migration in other states has been positively correlated with a high level of agri- cultural development. This applied particularly to the northwest (Sonora, Sinaloa, Baja California), and suggests that migratory movements were associated with high levels of agricultural development. In these states, the growth of output was not linked with land reform or with the development of the "traditional" sector but with large-scale irrigated commercial agriculture, which was most thoroughly  - 95 - developed after 1940. Although no relationship between the spatial incidence of net population change and land reform (Table 8.7) has been established, it can be argued that land reform efforts were concentrated in states with the most severe problems of population pressure. Although land reforms failed to resolve these problems, this is nct to criticize the effort that was made. Otherwise, outmnigration (as an index of rural conditions) would almost certainly have been greater. 353. The states with the least favorable structural conditions in agriculture generated a large proportion of the migrants who went to states with better agricultural conditions or who contributed to a large share of urban population increase, thereby closely associating urbanization with the deficiencies of the agrarian structure of t:he states of the center and the south (although migrants from these areas often moved to cities far removed from their places of origin - particularly to those in the north). 354. In-migration to the agriculturally developed states of the north was positively associated with urbaniza- tion in another way. Table 8.8 shows that the states with highly developed agriculture (Baja California, Sinaloa, Sonora) also ranked highly in terms of intercensal urbaniza- tion. These states had high relative levels of general development, although they were relatively unurbanized (as measured by the urbanized percentage of the state population). They were major centers of in-migration and rapid urbanization, as measured by the growth of the urban population and areas of advanced agricultural development. The inter-relationships of these indicators is based on the timing of agricultural development, which seems to have preceded urban development. Therefore, urban development in the northwest was,partly, a consequence of the prior growth of irrigated commercial agriculture implying that urbanization in this region was conceptually associated with the development of an agriculturaL export base. This type of urban growth was a common phenomenon during the Porfiriato, 1/ but does not seem to have occurred in other parts of the country after 1940. 1/ See Chapter 1.  - 96 - 355. The development of new industries from an export base depends on the nature of the base. If development leads to the construction of large-scale infrastructure (particularly in transport), production and distribution costs for many activities may be lowered. Growing internal and external economies of scale will then tend to stimulate export growth, thereby expanding the base. If the external demand for a region's product falters, stagnation will likely ensue. It is less likely to occur if the local economy is greatly diversified. 356. Development of the export base depends on the successful generation of new activities from the original- export oriented activity. This is precisely what happened in the northwest after 1940. Driven by massive public investment in irrigation, the regional economy generated large surpluses of agricultural output for the rest of the country and for national exports. The growth of agriculture also generated new demands for services and, to an increasing degree, was linked with the growth of manufacturing (with special reference to food processing industries). In sum, urbanization and agricultural development in this region were intimately linked. B. Regional Development and the Structure of the Urban System Regional Definitions 357. A second, and entirely different spatial dimension of the national economy concerns its regional structure which is defined in terms of the functional spatial framework of economic interaction and exchange. We are concerned neither with identifying homogeneous socio-economic conditions across geographic areas, nor with programatic development regions. The focus is on how the space-economy works and the organiza- tional roles of the country's major cities in that process. 358. A fundamental premise underlying this analysis of the urban system after 1940 has been that the system was fully articulated, in the sense that the national economy developed along interdependent lines and that the national transport system became sufficiently complete to facilitate macro-geographic interaction. Thus, every part of the nation's territorial space must be related to each other part on the basis of access to one or other of the country's major cities which constitute the framework for national spatial integration.  - 97 - 359. There are many levels of interaction and many levels of integration. We are concerned with macro-geographic space (i.e., that of the nation as a whole), so the relevant level refers to a high order of functional interaction based on the city as a major center of economic activity dominating the economy of the surrounding area. This relationship is clearly observed in the structure of the tertiary sector in major cities (generally, as we have seen in Chapter 2, in proportion to their size), which provide services not only for their own populations and those of immediately proximate areas but also for populations at greater distances. Because of the link between urban size and the level of manufacturing activity, which is generally strongest with respect to consumer goods industries, major cities also serve as manu- facturing and distribution centers for their associated regions. 360. Assuming (a) that all parts of the nation were integrated in the national space economy by 1970 and (b) that integration was achieved through linkages to major cities, there are three complementary ways of examining Mexico's spatial structure. The first focusses only on Mexico City and regards the whole country as a single region. Indeed, for certain purposes it is, because of Mexico City's economic and social. dominance. Earlier chapters have dis- cussed the capital city's demographic, industrial, commercial, cultural and political domination. Tucker (1957) states: "In Mexico, all roads lead to the capital of the country; the Federal District is the chief political and governmental center of the nation. Moreover, it is the economic, educa- tional, social and cultural center of Mexico . . . its influence is all pervasive. Other cities are satellites of greater or lesser magnitude held in their orbits by the central sun." This aptly describes the core-periphery structure discussed in the Preface. Table 4.17 shows that Mexico City's primate role is the single most important feature of the Mexican space economy. (See"Spatial Develop- ment in Mexico" for a discussion of the implications of the role of Mexico City). 361. Mexico's space economy may be seen on other levels. Map 29 shows the whole of the nation's space allocated to one of the three largest metropolitan areas - Mexico City, Guadalajara or Monterrey. This is a realistic pattern because, while there are certain functions (those of the Federal Govern- ment in particular) which are performed exclusively in Mexico City, other relatively high order functions are also per- formed in the second and third largest cities and not else- where. The most striking feature of this pattern is that the largest territorial area is assigned to Monterrey. Although the pattern of population density emphasizes the relative  - 98 - dominance of the zones associated with Guadalajara and Mexico City, particularly the latter, territory alone does not deter- mine the relative size of each zone. What it does imply is that the zone associated with Monterrey is so vast that the possibility of functional interaction with its western parts must be questioned. The contiguity of the northern zone with the US frontier is an impediment to spatial economic interaction, but not an absolute barrier. The economies of the northwestern states (which are associated in this scheme with Monterrey) have developed close functional ties with the U.S.A. (See Chapter 2); consequently, a substantial part of the northern zone "looks" towards cities in the US border states including Los Angeles/Long Beach (1970 population 7.0 million), San Diego (1970 population 1.4 million), Phoenix (1970 population 1.0 million) and El Paso (1970 population 800 thousand). 362. These massive "zones" are not "regions" because they are too large to correspond to anything other than an approximate territorial division among the three largest cities with respect to service functions. 363. The third level of spatial organization, shown in Map 30, represents an approximate urban-regional structure built around major cities. The regions thus defined vary greatly in territorial and demographic size and in socio- economic conditions. They vary, too, with respect to the size and characteristics of their central cities - or regional centers. Six major regions have been delineated within the commonly accepted bounds of accuracy for regional demarcation. For measurement purposes, regional boundaries have been adjusted to coincide with state boundaries. The six regions (some of which are subdivided to make a total of 12 subregions) correspond to zones of general socio- economic integration representing "general purpose" regions. These do not necessarily correspond to development regions, if the latter are defined in terms of areas which can be used for integrated development planning. 364. The major regional capitals associated with these regions are: Mexico City, Guadalajara, Monterrey, Tijuana/ Mexicali, Chihuahua, Culiacan, San Luis Potosi, Queretaro, Puebla, Veracruz, Oaxaca and Merida. As with any such scheme of allocation, there are certain anomalies. 365. The first category of anomalies consists of problems which arise with respect to the regional allocation of states which do not include major cities, and which do not have strongly developed linkages with any of the leading metropolitan  - 99 - areas. These are usually either poor, or small or both, examples being Oaxaca, Chiapas, Hidalgo, Guerrero, Colima, Nayarit and Michoacan. The allocation of these states to their respective regions is a somewhat arbitrary matter. The second problem is that some cities, the main examples being Leon, Torreon and Tampico, are located on state boundaries, and their location is also somewhat arbitrary because, in a regional scheme unconstrained by,the form in which data are available (i.e., did not have to respect state boundaries), both of these cities would be important regional centers. These problems show that no scheme of regional allocation can be perfect when the availability of regional information is an important criterion (i.e., when it is necessary to be able to describe and compare the demographic and socio- economic characteristics of different regions). Yet, the urban-regional-structure on Map 30 is- generally consistent with the basic definitional framework of the urban system and thus represents a reasonably accurate outline of Mexico's spatial economic structure in 1970. 366. Earlier chapters referred to more detailed aspects of this picture. TIhe next section will expand the (description of some of the key features of the regional structure in terms of inter-regional contrasts and intra-regionaL charac- teristics. Table 3.17, which gives demographic, social and economic indicators; pertaining to each region (measured on the basis of states), provides a key reference throughout this section. Regional Profiles (a) The Northwest 367. The four states in this region, Baja California Norte, Baja California Sur, Sinaloa and Sonora, fall into two sub-groups, comprising Baja California Norte and Sur and Sinaloa/Sonora, the former centered on Tijuana/Mexicali, which in this context may be thought of as twin regional capitals, and the latter, on Culiacan. The northwest as a whole is distinguished by its advanced level of agricul- tural development, and relatively high level of socio- economic development, but is neither highly urbanized nor industrialized. Its major cities - Tijuana, Mexicali and Culiacan - have gernerally poor social conditions with high levels of underemployment, skewed patterns of income distribution and inadequate levels of service provision.  - 100 - 368. The northwest as a whole covers about 20 percent of Mexico's land area but contains less than seven percent of its population. Just as the northwest was the country's most rapidly urbanizing area in 1960/70 (all four states having achieved well above the national average rate of population increase), so too, their total populations were growing much faster than that of the nation as a whole. 369. Demographically dynamic, the northwest has become an aggressive and upwardly mobile region in terms of economic performance, with the highest levels of agricultural productiv- ity in the nation, an average per capita income level almost 50 percent above the national average, and a relatively good pattern of income distribution as measured by the Gini coefficient. 370. Whilst these conditions apply to the region as a whole, there are sufficient contrasts between the Tijuana/ Mexicali region comprising Baja California (North and South), on the one hand, and the Culiacan region comprising Sonora and Sinaloa, on the other, to provide major sub-regional identities within the northwest. 371. The Tijuana/Mexicali sub-region stands out as an area of rapid demographic expansion in the 50s and 60s, particularly in its urban areas and in terms of its economic dynamism and high average incomes per capita, notwithstanding the relative skewness of the distributive patterns of its core cities. A large proportion of the region's population is concentrated in Tijuana and Mexicali (67 percent in 1970) and the twin regional capitals account for an even larger proportion of its urban population (85 percent in 1970). The region as a whole is strongly focussed on its core area, (see Table 10.17, which shows that the largest proportion of road transport movements of the six major regional areas in 1970 originated and terminated within this region). Although these data also refer to the Culiacan sub-region, they are strongly influenced by the inward-looking nature of the Baja California economy vis-a-vis the rest of Mexico. Because of its location and economic structure, the Tijuana/Mexicali region has very strong linkages with the economy of Southern California in the U.S.A. Taking account of this orientation, the sub-region is less autarkic than it seems. Distance is, and will remain, a major impediment to fuller integration with the Mexican economy; it still takes 44 hours to travel by road from Mexico City to Tijuana and 41 hours to travel to Mexicali.  - 101 - 372. In sum, the Tijuana/Mexicali sub-region is in many ways detached from the rest of the country. Its thriving economy is a result of the development, since 1940, of intense relations with the even more thriving and highly prosperous economy of Southern California, but it also has many problems which are related to the proximity of the United States. The rapid growth rates of Mexicali and Tijuana (See Chaptier 2) have been strongly affected by migration from other parts of Mexico, and there is no doubt that the principal motivation for migrating to Baja California has been the opportunity to move from there to the U.S.A. While these cities are relatively prosperous, they have not been able to absorb the large migratory influx (see Chapter 3 for a discussion of the low-level of welfare conditions). 373. The Culiacan sub-region is far more closely integrated with the Mexican economy than that of Baja Califor- nia. The development of prosperous agricultural economies and of thriving cities has been closely related to the impetus of the agricultural sector in Sonora and Sinaloa, and the market opportunities, enterprise and technological diffusion from the U.S.A. 374. This sub-region has also been demographically and economically mobile, and its core city of Culiacan has grown at a rate well above the national average. Its secondary cities - Ciudad Obregon, Hermosillo and Mazatlan - have with the exception of t:he last, also been extremely dynamic (Table 1.29). Income per capita is much higher than in the nation as a whole, and income distribution is among the most equit- able in the country, the Gini coefficient in 1970 being 0.50 compared with the national average of 0.58. 375. Accessibility between the Culiacan region and the rest of the country is better than that of the Tijuana/ Mexicali sub-region (See Table 10.33) because of the much shorter distance (exemplified by a driving time of 18 hours from Culiacan to Mexico City vis-a-vis more than 40 to Baja California), and the relatively well developed transport services to the rest of the country. 376. Table 10.7 shows that Culiacan has relatively more intensive linkages with the northwest (Monterrey, Monclova) and the Southwest (Guadalajara, Salamanca) than with Mexico City, although its intra-regional link;s (with Hermosillo, Mazatlan, Nogales and Ciudad Obregon) were relatively stronger.  - 102 - 377. The pattern of traffic to and from Hermosillo shows that Culiacan does not dominate this sub-region to the same extent that Tijuana/Mexicali dominatesBaja California. Hermosillo has strong links with the center (Mexico City and Puebla in particular), the northwest and the southwest. The pattern of telephone traffic (Table 10.27) confirms strong linkages with Mexico City, although these are notably weaker than those of most other large cities. There are some prob- lems of data interpretation because railroad freight movements inevitably emphasize the long journeys and de-emphasize intra- regional movements; the same shortcoming applies to the pattern of air-traffic (Table 10.26). The data do suggest that the Culiacan sub-region is less closely tied to Mexico City than most other regions, but that it is more closely tied to the rest of the country than the Tijuana/Mexicali sub-region. 378. The northwest, a large, loosely integrated, prosperous and generally dynamic area which is not strongly dominated by its largest city, is generally well connected with the rest of the country. Its ties with areas other than the center, however, are weaker than those of most other parts of the periphery. (b) The North 379. Again there are two sub-regions, one including the states of Chihuahua and Durango with a regional center in Chihuahua City, and the other including the states of Zacatecas, Aguascalientes and San Luis Potosi, with a regional center in the city of San Luis. The region, as a whole, has little homogeneity and is most validly dis- cussed in terms of its two sub-regions. 380. The first sub-region is extensive, occupying almost 20 percent of the national territory with less than 5.0 per- cent of its population. Sparsely populated, it is not a dynamic region. Its rate of population growth in the 1960s was lower than that of other parts of the northern zone, as was its rate of income growth. It is relatively well-off compared with most other parts of the country. Consistent with the general finding that income levels and income distribution are inversely related, its Gini coefficient is relatively high. 381. Compared with the center and the south, the Chihuahua region is not heavily urbanized and its urban structure is strongly dominated by Chihuahua and Ciudad Juarez. The latter is considerably larger than Chihuahua  - 103 - but is not a regional capital because of its frontier location. Its economy, as in Tijuana and Mexicali, is closely linked with the United States. Road traffic data indicate that the region as a whole (i.e., including the San Luis sub-region as well as the Chihuahua sub-region) appears to be strongly integrated with the northeast (Table 10.17) but has relatively weak connections with the northwest, the west and the center, and hardly any at all with the south. Rail traffic data in Table 10.7 show linkages between Chihuahua and other parts of the country, while Ciudad Juarez appears to have strong linkages with both the center (Mexico City, D.F.) and Guadalajara. These, however, are related to through traffic from the U.S. rather than to traffic originating in the Ciudad Juarez area. The structure of the strictly inter-regional flows is less distorted than would appear. 382. Linkages with the center are an important part of the total air-traffic pattern. As in the northwest, traffic with Mexico City accounts for the smallest share of regional traffic (Table 10.26). 383. The second sub-region of the northern zone is, by comparison with the first, much smaller (occupying 'Little more than seven percent of the country's land area), more densely populated (particularly in Aguascalientes), less urbanized, and poorer. Agricultural productivity, except in Aguascalientes, is very low. Agriculture provides more than half of all employment in this region, so the level of per capita income is well under the average for northern Mexico as a whole, and is only slightly above that of the southern states. Consistent with our earlier findings, income is poorly distributed, the average Gini coefficient being above 0.60. 384. This sub-region is quite strongly dominated by San Luis which is its largest city, the only other sizable urban center being Aguascalientes. 385. Intra-regional rail traffic between San Luis and other cities shows quite strong linkages with Aguascalientes and Zacatecas, although it also shows that the region has a strong relationship with the ports of Tampico and Matamoros and with the inland cities of Monterrey and Nuevo Laredo, all of which are inithe northeast. This suggests that not- withstanding the relative geographic proximity of the San Luis region to Mexico City and to the states of the center (San Luis is only five hours by road from the capital), it should be regarded as part of northern Mexico, as it is classified here.  - 104 - (c) The Northeast 386. Occupying nearly 15 percent of the nation's terri- tory this region is one of the best integrated in the country. It contains several major cities besides the regional capital of Monterrey, including Nuevo Laredo, Reynosa and Matamoros on the U.S. frontier, the port of Tampico in the southeast corner of the region, and Saltillo and Torreon in the west. Its population amounts to more than eight percent of the national population, but because this is a highly urbanized region, its share of the nation's urban population exceeds 11.0 percent. It is, concomitantly, prosperous. More than 25 percent of its population is employed in manufacturing, as compared with an average of 14 percent in the northern states as a whole. Its prosperity is also linked to unusually high agricultural productivity. Per capita income is con- sequently well above the national average, although the Gini coefficient is comparable to that of the northern states as a whole. This region is both demographically and economically dynamic, having had higher rates of population growth and income growth than most other regions during the 1960s. 387. Rail freight data (Table 10.7) suggest that the northeast is integrated strongly internally and that it also enjoys powerful linkages with all other parts of the country, particularly the central region. This is borne out by air passenger traffic (Table 10.24) and road freight traffic (Table 10.17), although the latter suggests strong connections with the northern region. The pattern of telephone traffic (Table 10.27) underlines the importance of both intra-regional linkages and connections with the center. 388. This is, in sum, a large, prosperous, upwardly mobile and well integrated region enjoying a large internal market, close connections with the U.S.A. (which provides its northern frontier) and strong linkages with the rest of the country, emanating from its unchallenged regional capital in Monterrey. (d) The West 389. A comparison between the west and the northeast is of interest because these regions are dominated by the country's second and third largest cities which differ socio- economically and structurally. 390. Territorially, the western region is not very large. But it contains about 10 percent of the nation's population and, in the 1960s, was one of the most demographically dynamic  - 105 - growth areas in the country. However, despite the presence of almost one and half million people in Guadalajara in 1970, the southwest was less urbanized than might be expected. Its urban population was about the same as the national average, while its rate of urbanization in the 1960s was somewhat slower than that oi- the nation as a whole. Employment in agriculture was about the same as the national rate but agricultural productivity was relatively higher, and its share of manufacturing employment in 1970 was less than proportionate to its share of total employment, suggesting that agriculture was the mainstay of the region's economy. Notwithstanding reLatively good productivity, average incomes per capita were beLow those of the northern and central states, while the distribution pattern was relatively favorable. 391. Whereas the northeast, excluding Monterrey, boasts a number of large metropolitan areas which provide a strong framework for regional integration and growth, the west does not. At a regional level, Guadalajara is a strongly primate city, being more than 10 times as large as Morelia, the only other city of more than 100,000 people in this region. 392. Rail traffic movements suggest strong patterns of inter-regional interaction with the northwest, particularly with the Culiacan sub-region, the Federal Capital region centered on Mexico City, and the Bajio region which (see below) lies between Guadalajara and Mexico City with which it enjoys a high level of relative accessibility (Table 10.30). The pattern of road freight traffic (Table 10.16) also emphasizes the importance of linkages with the Chihuahua sub-region to the north, by the air passenger trafific matrix (Table 10.24) and by the telephone traffic matrix (Table 10.27). (e) The Center 393. Although Mexico City is the functionally dominant center for the whole of the central region - an area which includes the States of Mexico, Morelos, Guanajuato, Queretaro, Hidalgo, Puebla and Tlaxcala, as well as the D.F. - there is sufficient diversity within this region and a sufficient number of functionally effective regional centers, to lead us to identify three sub-regions. The first includes the D.F., Mexico and Morelos, with its center in Mexico City; the second, Guanajuato, Queretaro and Hidalgo with its center in Queretaro and t:he third, Puebla and Tlaxcala, with its center in Puebla.  - 106 - 394. Before examining the characteristics of each of these sub-regions, we shall briefly consider the diversity of the region as a whole and the nature of its relationships with the rest of the country, especially as articulated through Mexico City. 395. The central region covers less than eight percent of the Republic's area but contains no less than 36 percent of its total population, almost 25 percent of which is in the D.F. and the State of Mexico. The presence of Mexico City and of many other large metropolitan areas, including Queretaro, Leon, Irapuato, Pachuca, Toluca, Puebla and Cuernavaca, result in a highly urbanized region, with more than 70 percent of its population urbanized in 1970. Thus it is a densely populated region, with by far the highest density of any part of Mexico. Yet it is also diverse because, outside Mexico City, the states of Queretaro and Guanajuato and Puebla remain largely rural, while the population density of Queretaro and Hidalgo is notably lower than that of the region as a whole. 396. In the recent past, it has been a less dynamic region demographically speaking than most of northern Mexico, with the notable exception of the States of Queretaro, Mexico and Morelos. 397. Despite the largely urban nature of the central region, agriculture remains the source of livelihood for much of its population, accounting for half or more of total employment in Hidalgo, Guanajuato, Puebla and Tlaxcala. Agricultural productivity, particularly in Morelos, is generally high. The average income is above 150 pesos/ha in all parts of the region except Tlaxcala. The manufactur- ing and tertiary activities of the central region are its distinguishing feature, accounting for more than 50 percent of all manufacturing employment in the nation and an even larger share of tertiary employment in 1970. 398. One would therefore expect that income per capita would also stand out in a comparative context as in the Federal District (Chapter 3). But, this is not the case in many other parts of the region; Guanajuato, Hidalgo, Puebla, Queretaro, Morelos and Tlaxcala all have relatively low income levels with relatively skewed patterns of income distribution. It is this, rather than the relatively favorable aggregative indices of its employment structure, which reveals the intense contrasts in socio-economic welfare in the central region. For the most part, these variations arise from the highly unequal pattern of land distribution in the rural sector and the widespread rural  - 107 - poverty which occurs within the framework of high avrerage levels of agricultural productivity. 399. As expected, the central region is strongLy focussed on Mexico City as a regional capital. The patterns of economic interaction revealed by transport flows show a very high level of centralization. 400. Patterns of rail traffic (Table 10.7) are not, in this instance, a good measure of intra-regional structure because of the relatively short distances involved and the excellence of the road transport system. Puebla is only two hours away from Mexico City, Queretaro two and a haLf, Leon less than five, Toluca only one, and Pachuca slightLy more. The road traffic matrix for 1970 (Table 10.17) shows that 26 million out of 30 million movements originating in the central region also terminated there, while the pattern of telephone traffic (10.27) confirms that Puebla and Leon were strongly linked with Mexico City. 401. Traffic data also show that Mexico City and other cities of the central region were strongly linked to other parts of the country, with strong rail linkages to the north- east and northwest and to the port of Veracruz to the east (Table 10.7). The pattern of road traffic distribution points,on the other hand, to strong links with the west (Table 10.16) with respect to inward movements and with both the west and south with respect to outward movements. This pattern is repeated in regard to air traffic (Table 10.24). The northeast also had strong linkages with the center in 1970, with the western region accounting for more than 40 percent of the central region's outgoing traffic. 402. The strong links between the center and various regions, when measured in terms of diffeient transport modes, reflects the traditional role of the core which enjoys power- ful links with all other parts of the country while having a relatively strong network of internal linkages, although the latter may be comparatively weaker than those of the north- east. Were internal linkages stronger, greater homogeneity in socio-economic development might be expected in the center. 403. Turning to the sub-components of the central region, the Mexico City sub-region stands out because of it.s large, urbanized population, its emphasis on manufacturing industry, the high level of personal incomes and the dynamism of its recent demographic and economic growth. This is not only a tightly knit sub-region but also the core of the national economy. Inasmuch as Mexico City accounts for a dcominant  - 108 - share of the sub-regional structure, the latter may be conceived as essentially an expanded version of the Federal Capital area, the national role of which has already been discussed in this Chapter. 404. The Bajio sub-region includes the states of Queretaro, Guanajuato and Hidalgo and four of the nation's 37 largest cities in 1970: Leon, Irapuato, Queretaro and Pachuca. The role of Leon within the regional scheme pro- posed in this chapter is - like that of Torreon and Tampico - somewhat anamalous because it was, in 1970, Mexico's seventh largest city. However, straddling the border of Guanajuato and Jalisco, Leon occupies an uncertain status between the southwest region and the Bajio sub-region. Despite its size it does not have, as seen in Chapter 3, a dynamic recent history. Queretaro is the regional capital, although a strong case can also be made for the existence of a nascent linear metropolis, stretching from Queretaro to Irapuato -and enjoying strong intra-subregional linkages, as well as a close relationship (but one preserving some independence) with the Federal Capital region. 405. Finally, in the eastern portion of the central region, the city of Puebla is the undisputed center of a densely populated and generally poor subregion, characterized by low and skewed incomes, the nature and causes of which have been considered in the first part of this Chapter. (f) The South 406. Compared with the regions considered thus far, the southern states present a picture of almost unrelieved poverty and underemployment. However, unlike the central region where the strength of Mexico City provides a single dominant center, the south divides into three sub-regional components. One includes Veracruz and Tabasco, with a principal center in Veracruz; another includes Oaxaca, Guerrero and Chiapas with a center (albeit a weak one) in Oaxaca City; the last includes Yucatan, Quintana Roo and Campeche with a relatively strong center in Merida. 407. The Veracruz sub-region is markedly better off than the others. It is relatively large, accounting for five percent of the nation's area; relatively populous, containing almost 10 percent of the country's population, and, in the 1960s, was quite dynamic. It is, however, relatively unurbanized in spite of its six cities with populations over 100,000: Veracruz, Coatzacoalcos, Minatitlan, Jalapa, Orizaba and Villahermosa. Most are  - 109 - employed in agriculture. While productivity in this sector is quite high, incomes per capita are low relative to those of the states of the northern and central regions. 408. Railroad linkages within the region are still rather weak, although the Gulf Coast serves as a traffic axis between Villahermosa and Veracruz. There are, however, strong inter-regional ties to the central region with respect to outward movements but connections with the northeast with respect to inward movements are stronger (Table 10.7). Road traffic patterns (Table 10.16) emphasize the importance of the center with respect to outward and inward freight, with a roughly balanced flow in either direction. Air passenger movements (Table 10.24) and telephone traffic (Table 10.27) reflect the same pattern. 409. In sum, the Veracruz region is strongly integrated with the central region and somewhat integrated with the north- east. It is an area which depends on internal linkages for its development. As such, it represents an outstanding example of how the core-periphery relationship operates to the general detriment of the development of the periphery. 410. The Oaxaca sub-region is different in that it is hardly developed at all - not as a client of the core nor otherwise. Geography has been the major obstacle. The mountainous terrain - which characterizes the states of Guerrero, Oaxaca and Chiapas - has militated against economic progress. The area encompassed by this fairly homogenous, but barely integrated region amounts to almost 12 percent of the national territory. Its share of the nation population is about 12 percent, and its average population density is close to the national average. It is little urbanized, lacking large cities except in Acapulco and Oaxaca. Its agriculture is poor and productivity is low. Incomes are the lowest in Mexico and very poorly distributed. Its major inter-regional links are with the Central region and the northeast (Torreon) and with the other sub-regions of the south, via the Isthmus of Tehuantepec. This is an under- developed, poor, badly integrated region, which presents perhaps the greatest challenges to development planning in Mexico. 411. Finally, we come to the Merida sub-region with the city of Merida, twelfth in population rank in 1970, providing an eccentrically located but nonetheless dominant center for the states of Yucatan, Quintana Roo and Campeche. There are no other cities of any size.  - 110 - 412. It is comparatively large, sparsely populated, urbanized and poor. Here,too, geography is a critical factor in explaining why this sub-region has remained under- developed. The problem with the Yucatan is that it is on the way to nowhere. In the nineteenth century it became prosperous in the context of henequen cultivation but as world demand for that crop declined, so did the fortunes of the Yucatan; they have never recovered. This region is still heavily agricultural with moderate productivity; thus its income level, although poorly distributed, is somewhat above other parts of the south. 413. Present linkages are mainly with the center which has replaced to some extent the intensive links which this region once enjoyed with the southern U.S.A. (New Orleans) and with Cuba. Its rail links with the northeast are relatively quite strong. 414. This region is, in several respects, the antithesis of the northwest. Both are located "out on a limb" at the extremities of the nation. The Yucatan borders Central America, while the northeast borders the U.S.A. and this, rather than resource endowment or enterprise, has made a fundamental difference in the level of development of each region.  ANNEX A AGRICULTURE AS AN "URBAN" SECTOR 1. The agricultural sector here is identified with the "rural" economy, but is juxtaposed with the secondary and tertiary sectors which are identified with the "urban" economy. Evidence suggests that, after 1940, agriculture may have been a source of employment for part of the "urban" population as in some of the smaller towns which are classi- fied as "urban" on the basis of the 2,500 "urban threshold." This is also true of most of the 37 cities which, in 1970, had populations exceeding 100,000. Probably the use of municipal census tracts for the definition of urban areas leads to some distortion of the true extent of agricultural employment, because it may include people who live beyond the city periphery. This does not (see Technical Note No. 4) constitue a particularly serious problem because a "rural" population, peripherjal to a city, is affected by the urban economy, and is in a sense "semi-urban." 2. Economically, two conclusions affect the basis of the analysis in this chapter. First, to the extent. that agriculture was, in 1940-70, a "legitimate" economic activity in the 37 urban areas, its importance clearly decreased over time. Thus the time trend of the share of agriculture in the employment structure of each city reflects the moderniza- tion of the urban economy and/or the physical spread of the contiguous, built up, urban area. There is no way to separate these components. 3. There is, however, a substantial difference in the structure of urban employment, depending upon the inclusion or exclusion of agriculture. Most cities had fairly high levels of agricultural employment in 1940 which, in all cases,declined thereafter. Some, notably Irapuato, Mexicali, Culiacan and Hermosillo, had high levels of agricultural employ- ment even in 1970. Because of measurement problems, no special inference can reasonably be drawn from variations in the relative importance of agricultural employment between cities. It apparently accounted for relatively large shares of total employment in cities which were quite dissimilar in terms of size, although none were very large in 1970. Many of these cities were in the northwest, which has already been shown to be a relatively highly developed region, but this was not a consistent pattern.  ANNEX A Page 2 4. The second conclusion is that, given reasonable doubt concerning the legitimacy of agriculture as an "urban" sector, and the a priori assumption that the secondary and tertiary sectors were the cornerstones of the urban economy, primary agriculture is not analyzed in this chapter. This may lead to certain distortions, but this procedure serves better our purpose of focusing directly on those sectors of the urban economy which were most closely associated with the process of urban economic growth and development.  ANNEX B THE DEVELOPMENT OF THE RAILWAY NETWORK PRIOR TO 1910 1. In 1880, the Mexican Central Railroad obtained a concession to build a line from Mexico City to Leon and from Leon to Paso del Norte on the U.S. frontier. Regular services were in operation as far as Chihuahua by the end of 1882, and the entire route (845 km from Mexico via Aguascalientes and Chihuahua to Ciudad Juarez) was completed in March 1884. The Mexico Central Railroad also built the Mexico-Guadalajara line, which was completed as far as Irapuato in 1888 and as far as Guadalajara in 1890. 2. Another north-south line was constructed by the National Railroad Company connecting Mexico City, Tolnea, Alhambra, Celaya, San Luis Potosi, Saltillo, Monterrey and Nuevo Laredo, a distance of almost 1,300 km. It was started in 1880 and had been completed by 1886. Also in the north the Monterrey Railroad Company built lines to Tampico (completed in 1891) and Matamoros (completed in 1905). 3. On the Pacific Coast, the Mexico National Railroad constructed a line from Alhambra through Morelia and Uruapan to Patzcuaro in 18388 and added an extension from Manzanillo to Colima, which was started in 1889 and completed in 1902. The Mexican International Railroad built a line from the north-east end of Eagle Pass to the west of Durango, which was open as far as Monclova by 1884 and was extended to Torreon in 1888, where it connected with the Mexican Central. In 1892, this line was extended as far as Durango, from which point it paralleled the Sierra Madre del Oeste to Tehuantepec, but was never extended to connect with the Pacific line at Mazatlan. 4. The North Pacific line was finished in 1898, when the Sonora railway was leased to the South Pacific Railroad, which obtained access to the port of Guaymas. In 1903, the South Pacific Rai:Lroad acquired control of the Cananea, Yaqui River and Pacific Company which had built a short line from the Sonora-Arizon<a boundary to the mining camp of Cananea under a concession granted in 1900. The Guaymas-Guadalajara line was not openesd until 1921, but had been completed as far as Mazatlan by 1909 and as far as Tepic by 1911. From Orendain, an isolated section had been built north to La Suermoda, Jalisco, leaving a gap of about one hundred miles through extremely difficult terrain. At Orendain, a connec- tion was made with the Guadalajara-San Marcos extension of the old Mexican Central, thus preparing the way for communica- tions between the north western region and Mexico City. In  ANNEX B Page 2 1910, however, the South Pacific Railroad received a concession for an independent line from Guadalajara to Mexico City which was never built, nor was a planned extension from Guadalajara to the Isthmus of Tehuantepec, though concessions were ob- tained from the States of Jalisco, Michoacan and Oaxaca. 5. A line over the Sierra Madre del Oeste connecting the Pacific town of Topolobampowith Ciudad Juarez and reach- ing to Presidio (Chihuahua) took more than 70 years to com- plete. In 1899, the Chihuahua and Pacific Railroad completed a section of this route from Chihuahua to Minaca (near the foot of the Sierra Madre) and in 1905, built a branch from Minaca to Temosachie at the instigation of mining and lumber developers. It was decided to extend this route as far as Guaymas and Topolobampoon the Sonora coast. The Ciudad Juarez-Chihuahua section was opened in 1912, and the La Junta- Creel section in 1910. The Chihuahua-Presidio line, however, was not completed until 1934 and the whole railroad from Topolobampoto Ciudad Juarez was not finished until 1961. 6. In the south, the Interoceanic (Mexican Southern) line was planned in 1880 as a means of connecting Mexico City with the port of Anton Lizardo on the Gulf and with Tehuacan, Oaxaca, Tehuantepec and Puerto Angel, on the Pacific Coast. Construction began in 1890, and the line was completed from Puebla as far as Oaxaca in 1892. Two short branches from Oaxaca were constructed in 1911, but the line was never extended as far as Puerto Angel. 7. A Government survey of a possible route across the Isthmus of Tehuantepec was begun as early as 1824; a concession was granted but actual construction did not start until 1890, and the line was not completed until 1894. Because of the poor quality of the railroad, it was not formally opened to traffic until 1907. 8. The Veracruz and Pacific Railroad was incorporated in 1898 to link the Tehuantepec line with railroads to the north and to provide a link between Cordoba (on the Mexican Railroad) to a junction with the Tehuantepec Railroad at Santa Lucrecia. It also operated a branch to Veracruz which joined the main line at Tierra Blanca. On the Pacific side of the Isthmus, there was a connection with the Pan-American Railroad at San Geronimo which led to the Guatemala Boundary. It was completed in 1909.  ANNEX B Page 3 9. The Southern region was historically isolated from the central region and connection was possible only by coastal shipping. There was no direct inland route to the south from the central and northern areas until wel:L into the 20th century. The Yucatan railway system thus cons:isted of a group of small lines designed to link the sisal p:Lantations with the coast, reflecting a direct interest in resource extraction by the railroad developers. The first narrow gauge line built in 1881 connected Merida to Progreso. This was extended to Zagmal in 1890. The peninsula line, connecting Valladolid and Merida with the port of Campeche, was completed in 1898. The Campeche to Mecatepec l.ine opened in 1908, but was not directly connected with the northern railroad system until 1960.  ANNEX C AIRWAY FLOWS 1. Over 46 percent (or 582,509 out of 1,247,000) of all passengers in 1967 originated in the Central region, and about the same number were received there. Data for the Central region shows that Mexico City clearly dominates, suggesting that almost half of the volume of national air traffic passes through Mexico City. 2. Of the half million passengers to or from Mexico City, over 45 percent (265,185) were to or from the south- west where there were two major airports: Acapulco and Guadalajara. Acapulco accounted for 189,000 out of 272,000 passengers between Mexico City and the southwest, and Guadalajara for the remainder. This implies a strong link between demand for air services and income. Most of the demand in Acapulco was generated by tourists and other high income groups. In part, the southwest accounted for the next largest number of air passengers - about 24 percent of the total or 300,000 passengers. As has been shown, approximately 90 percent of this total flew to Mexico City, fewer than 7 percent to the northwest and only a few thousand to the north- east. Acapulco generated about 64 percent of the total out- going traffic of the southwest, which implies that Guadalajara had a minimal level of interregional activity, except with Mexico City. 1/ 3. The northwest with six major airports has relied heavily on air transport but the total volume of regional traffic amounted to only 8.7 percent of the national total (108,889) in 1967. The pattern of regional traffic was fairly evenly distributed. Over 57 percent of passengers traveled to the central region and 19 percent to the south- west while 16 percent traveled within the region. The north- west covers a large area with many physical barriers which impede surface connections. Airlines have facilitated both inter and intra-regional movement since the 1950s. The largest volume of traffic (35 percent of the regional total) originated in Tijuana, of this, 60 percent or 21,438 passengers terminated in Mexico City. Mazatlan and Hermosillo served the next largest numbers of passengers in the region (23 percent and 18 percent respectively). 1/ Guadalajara, however, has other airline connections carry- ing short distance traffic.  ANNEX C Pacfe 2 4. The northeast generated the third largest: volume of passenger traffic in 1967, accounting for about 11.3 per- cent of the national total. Almost 88 percent traveled to Mexico City, about 4 percent within the region and 3 percent to the southwest. The region had four major airports. Monterrey generated more than 50 percent of the regional total, of which 85 percent (61,860 passengers) went to Mexico City, the next most important destination was Guadalajara. Tampico generated about 30 percent of the regional total (43,000 passengers), 90 percent of which traveled to Mexico City. 5. The southeast has five major airports which generated about five percent of the national total (66,276 passengers) in 1967. Over 77 percent of all traffic was with Mexico City and 23 percent was intra-regional. The Merida-Mexico route produced the heaviest volume (over 78 percent of the total) and the Oaxaca-Mexico and Veracruz- Mexico routes had 30 and 25 percent of total traffic, respectively. 6. Air transport has contributed significantly to a monodirectional flow towards Mexico City (and thus the Federal District and the State of Mexico) for a variety of reasons. Air networks are both capital-intensive and inflexible (in the sense that they connect nodes but have no effect on inter-nodal spaces). Second, they are efficient only over long distances. Third, given these characteristics of the node, air services did not develop intensively until after 1960, but by 1967 had reached most of the major cities. Fourth, Mexico City generated and received over half of all domestic air passengers in 1967. Consequently, there was very little development of other inter-regional traffic. 7. The pattern of inter-regional air traffic in 1973 shows important contrasts with that for 1967. The largest percentage increase of total traffic was in the southeast, which increased its share from five percent of the national total in 1967 to 9.6 percent in 1973. Other changes indicate that 54 percent of all traffic in the northwest was to Mexico City, and over 17 percent of all traffic was bound to the southwest, mostly to Guadalajara. East-west traffic to the northeast developed slowly connecting Tijuana, Ciudad Juarez and Monterrey and accounted for 4 or 5 percent of total domestic traffic in 1973. An important inter-regional axis emerged between La Paz-Mazatlan-Torreon and Monterrey, which were accessible only by air.  ANNEX C Page 3 8. The North-Central region had the smallest overall increase in passenger traffic but there was a significant increase in its traffic with Mexico City. Torreon airport handled about 27 percent of the total air traffic of the North Central region, 74 percent of which was to Mexico City. Ciudad Juarez had the lowest level of air dependency among large cities, although its population was the fourth largest in the country. Air traffic generated by this city was relatively less than in other cities. Air connections between the North Central and regions other than the center were few, most traffic being bound for Mexico City. The frontier cities have been growing rapidly since 1950 but were well integrated only with the Central region. Thus most of the growth incentives for these cities came from outside the country. 9. From 1963-73, the demand for air transportation between cities separated by medium distances grew significantly. In many cases, the use of air transport increased when surface transport was unable to provide internodal services in less than five or six hours. This is illustrated by the volume of air movements emananting from Mexico City. Air passenger movements in 1973 show heavy passenger flows for surface transport between the capital, Oaxaca, Veracruz, Tampico and Guadalajara. 10. During 1967-73, the improvement of the civil air network reflected developments in air fleet and airport facilities. The total number of passengers doubled and regional traffic volumes grew in equal proportions through- out the country. This implies that air transport became an increasingly important mode and that inter-regional traffic began to play an important role in regional integration. In 1967, the regional traffic distribution pattern was largely monodirectional, connecting Mexico City and other regions; but in 1973, the origin-destination pattern emphasized major changes in traffic volumes between other nodes. In 1967, there was no direct traffic between the Northwest and the Southeast. In 1973, one to two percent of the total traffic volumes for these regions concerned southeast-northeast traffic (mainly between Merida and Monterrey) and this was, to a lesser extent, also true of movements between the north- west and southeast. Guadalajara received most of the traffic with 12,000 passengers traveling from Monterrey to Guadalajara in 1973 compared with 2,500 passengers in 1967. The Mexico- Monterrey route (in both directions) reduced its proportional weight after 1967, although the traffic volume in 1973 was 12,000 passengers. The Guadalajara-Mexico route had the second heaviest volume of domestic traffic, 73 percent of  ANNEX C Page 4 movements from Guadalajara (177,000) being bound for Mexico City, compared with 80 percent of total movements (83,000) in 1967. 11. Sixteen percent of the passengers (180,000) originated in Mexico City and went to Guadalajara in 1973, compared with 86,000 (15 percent) in 1967. In the south- east, Merida significantly expanded its activity and by 1973 provided schedules to all parts of Merida including Tijuana, Acapulco, Oaxaca and Monterrey. The deconcentration of air traffic was reflected in the decline of the volume of air traffic in Mexico City from 46.7 percent of the total in 1967 to 44.5 percent in 1973, notwithstanding a national net increment of 106 percent for the period. Traffic volumes in most cities increased at an average rate of 1-5 percent during this period. 12. In spite of the fall of 2.2 percent in the number of domestic passengers in Mexico City from 1967-73, Mexico City still dominated the air transport system in 1973 not only in terms of the volume of air traffic but also in terms of flight frequency. But of a national total of 2,609 weekly arrivals and departures, 742 (28.44 percent) originated or terminated in Mexico City. Although these figures show less concentration than those of origin and destination, the dominance of Mexico City is clear. Flight frequencies were higher for services between Mexico City and other urban centers than between any other cities. Guadalajara. handled 257 arrivals or departures of which 109 (42.41 percent) were to or from Mexico City. Of 140 arrivals and departures at Acapulco, 110 (78.4,1 percent) were to or from Mexico City. A notable exception,was Tijuana which had a more evenly distributed traffic, pattern with 121 weekly arrivals and departures of which only 24 (19.83 percent) were to Mexico City. Of the remaining arrivals and departures (36.36 per- cent) had origins or destinations within the northwest indicating that Tijuana was well integrated in intra- regional traffic. 13. Tentatively we conclude that: (1) air transport was not a popular mode until mid-1960s; (2) remo'ter regions were linked with the economic and social activities of the Central region for the first time in 1973, given the rapid services provided by the air transport system;  ANNEX C Page 5 (3) the mono-connective system developed through the 1960s contributed significantly to urban concentration up to 1967 but had been greatly diversified by 1973; and (4) the spatial integration of the network is likely to accelerate the reverse flow of traffic from larger centers to local centers, more rapidly than any other mode.  ANNEX D REAL AND NOMINAL DISTANCE 1. Accessibility between two points is not necessarily a direct function of physical separation. Inter-regional connectivity is not a matter of the mere existence of a net- work link between regions. Such variables as desicgn quality and physical conditions must also be taken into account in order to measure actual connectivity. An analysis of physical distance vis-a-vis real distance must allow for cost varia- tions, with respect to vehicle operation and time. The relationship between real distance and physical distance facilitates assessment of the spatial advantages of different locations. Real distance does not automatically irnply a given level of traffic. For example, real distance between the northwest and northeast is relatively short, and the present level of inter-regional linkages between the two regions is insignificant. 2. A recent study, CEPAL (1974) shows coefficients between real and physical distances. Physical distance was given the value of 1 and was used as a base to which time and cost elements were added. A coefficient value of 1.00 implies the best (or least cost) transport link. A value of 1.20 implies that the cost (including time cost) of transport over the link is twenty percent higher than under the presently prevailing optimal conditions; thus, if physical distance is 100km, real distance is 120km. 3. Two examples of the differences between real and physical distance are shown below: Physical Real Length Length Coefficient (1) Acapulco-Mexic:o City 419 km 611 km 1.46 San Luis Potosi-Mexico City 417 km 433 km 1.04 (2) Oaxaca-Mexico City 534 km 754 km 1.41 Guadalajara-Mexico City 581 km 609 km 1.05 4. The physical distance between Acapulco and Mexico City and San Luis Potosi and Mexico City is virtually the same. When road condition variables are taken into account, the real distance between Mexico City and each point is different. The distance to San Luis Potosi increases by 16 km, and that to Acapulco by 192: km. Similarly, in physical terms, Oaxaca is 47 km closer to;Mexico City than Guadalajara, but in terms of real distance it is 145 miles further away.  ANNEX D Page 2 5. Accessibility to the Federal District with respect to physical versus real distance in 1970 varied. The Federal District had high coefficients with points to the north, such as Queretaro, San Luis Potosi, Guanajuato, Pachuca and Leon. There was also excellent accessibility between Guadalajara and Mexico City (a coefficient (Kv) of 1.05). Good accessibility exists between the Federal District and the northwest (an average Kv of 1.10) and the northeast (an average Kv of 1.09 except in coastal areas). There was fair accessibility between the Federal District and the Southeast except with respect to Oaxaca (Kv's of 1.19 to Veracruz, 1.15 to Merida and 1.41 to Oaxaca). There was fairly good accessibility between the Federal District and coastal locations in the Southwest, such as Acapulco (a Kv of 1.46). 6. Monterrey had good to fair connectivity with other places. Real distances between Monterrey and such neighboring towns as Saltillo, Torreon and Nuevo Laredo, were moderate and there was fair connectivity with points in the northwest, such as Ciudad Obregon and Tijuana (Kv value of 1.17 and 1.15 respectively. There was poor connectivity between Monterrey and Guadalajara (a Kv of 1.28), which implies that conditions are not favorable to the formation of strong inter-city flows between these major metropolitan centers. The Kv values between Monterrey and points south of Guadalajara along the Pacific coast were unrecorded, but because road conditions were poor, it is assumed that connectivity was low. 7. Compared with Monterrey and Mexico City, Guadalajara had a lower level of accessibility with respect to real versus physical distance, Kv values being consistently fair to poor except with points to the east - such as Mexico City. This is consistent with the large traffic flows between the south- west and the central region. The Kv values for points to the west and along the Pacific coast were surprisingly high - 1.21 in Manzanillo, 1.21 in Culiacan and 1.29 in Puerto Vallarta.  ANNE,X E RELATIVE ACCESSIBILITY 1. A relatively complex measure of modal/network relationships concerns accessibility between major cities. This provides a means of assessing the relationship between changes in relative connectivity from 1940-70 and relative urban development during the period. The analysis was based on an index of relative accessibility between 25 of the cities which, in both 1940 and 1970, dominated the urban system. Twenty-five rather than 37 cities are used because some of the 37 cities were close together (e.g., Mexico City, Toluca) and thus were in the "same" location in macro- geographic terms. 2. The cities were chosen to ensure a minimum of about 100 km between them. It was assumed that each mode operated under normal condit:ions, implying different assumptions about average modal speeds and efficiency. Different modal weights were used to reflect qualitative and quantitative contrasts in transport conditions. 3. The analysis showed that the cities which had the highest levels of connectivity in 1940 grew more rapidly than others during the period. Thus, those places which improved most in relative as well as absolute terms were those which had the highest degrees of initial advantage at the beginning of the period. 4. In 1940, about 53 percent of the 25 largest cities had accessibility indices of 51 points or more. Most of Mexico's major cities were equipped with a minimum level of transportation facilities, but there were only six inter- urban connections (2 percent of the total number) with index values ranging from 60 to 70, indicating that very few cities had regular airline service with a large number of other cities. Mexico City had regular air services only with Guadalajara, Torreon, Tampico, Veracruz and Aguascalientes. Some remote cities, such as Merida and Ciudad Juarez, also had scheduled air services to the capital. At this time the best connected cities had railroad and two- lane highway linkages with other cities. Some 3.7 percent of all cities had index values of 40-50, indicating inter-city connections by all-weather paved roads and either railroad or scheduled air services involving less than three flights a week. According to this index, Mexico City had an accessi- bility index of 40 or more with 20 cities out of the other 24. By comparison, five cities had index values of 60-70, 12  ANNEX E Page 2 cities had an index value of 51 and three cities had index values of 30-40. Torreon, Tampico, San Luis Potosi and Veracruz also had high accessibility values, implying that they were major regional traffic nodes. Large regional centers, such as Monterrey, Merida, Ciudad Juarez and Chihuahua, were relatively isolated. Medium size cities, such as Irapuato, Pachuca and Toluca located close to Mexico City, had relatively high levels of accessibility compared to these large regional centers. 5. By 1970, over 93 percent of the 25 urban locations had accessibility indices of 51 or more. This means that out of 300 inter-urban connections, 279 were equipped with facilities which included a two-lane paved road and rapid train services but lacked scheduled air services. By 1970, regardless of locational distance, all of the 25 major cities were within 24 hours of one another. All cities which had grown at a higher than average rate had accessibility values of more than 71 points. Thus, all of the major regional centers which in 1940 had still been spatially isolated had, by 1970, been fully integrated by all transportation modes. 6. Mexico City had the highest accessibility index in 1970. It was connected with all of the 24 other cities by an index value of 51 points, and was connected to 11 pair cities with accessibility index values of 80-90 points. This implies connectivity via two-lane paved roads, rapid express rail lines and more than 14 scheduled flights a week. Mexico City thus had the most accessible location in the country. Guadalajara and Tijuana were the next most accessible locations. Both were connected with 9 pair locations with index values of 60 or more, while Monterrey and Hermosillo were connected with eight pair cities with index values of 60 or more. 7. Tijuana, Hermosillo, Ciudad Juarez, Culiacan and Chihuahua (all on the frontier) had accessibility index values over 80 points with more than two cities. Two of them were not among the largest 25 cities in 1940 and it is inferred that their growth stems, in part, from the efficient operation of regional transportation networks. Meanwhile, some historically important cities, which were no longer among the 25 largest by 1970, had lost out in inter-urban competition to dynamically growing cities. Thus, Toluca, Queretaro, Mazatlan, Saltillo and Aguascalientes, all of which had been important transport nodes during the railroad era, were less dynamic after 1940. The 1970 accessibility index  ANNEX E Page 3 gives only a two-dimensional spatial integration. It is difficult to evaluate the efficiency of the network in terms of real accessibility, taking account of time-distance or effective price. Thus, the same accessibility value was allocated to every type of road, but there are obviously significant differences in actual operation between different roads having the same nominal classification. In order to improve the index,further research on as yet unavailable data would be needed. 8. Network and modal development played an important role in the development of Mexico's largest and most dynamic cities in 1940-70. By inference, transport improvements were thus coincident with urban-industrial development, although they also coincided with urban growth derived from non-industrial investments.  ANNEX F URBAN ECONOMIC PROFILES 1. Tables 4.8/4.10 (1940) and 4.9/4.11 (1970) show "profiles" of the economy of each city at the beginning and end of the period in terms of (a) the index of specialization, and (b) the index of surplus employment. In 1940, there was a fairly random relationship between the degree of specializa- tion and the identity of the leading "growth" sector. There were, however, some general consistencies. First, in the specialized cities, there was only one major export sector - these were the same sectors which had very high sector values in the index of specialization for 1940. In the semi-specialized and semi-diversified cities, there was usually more than one major export sector. This applied to diversified cities, with one important exception. This was that manufacturing, the major export sector in five diversified cities, was the only important export sector, suggesting that in cities where manu- facturing became a dynamically important activity, it tended to be the most significant source of urban growth. The cities in which this occurred were relatively diversified in terms of the average structure of all 37 cities. 2. In 1970, the specialized and semi-specialized urban economies had only one major export sector; but only in one case (Orizaba) was this manufacturing. As in 1940, manufactur- ing was the dominant growth sector in five of the diversified cities. It did not occur as a growth sector in other semi- specialized or semi-diversified economies, in most of which, construction, government, commerce and services were the ostensibly important growth activities. 3. In 1970, as in 1940, commerce, and also transport and government functions, were major growth sectors in both semi-specialized and semi-diversified economies; the only specialized economy (Minatitlan) was, as already noted, essentially dominated by the hydrocarbons industry. 4. In 1940, irrespective of the degree of specializa- tion of urban economies, manufacturing, followed by commerce and transport, was the most commonly occurring dominant "export" function when assessment is made in terms of the number of times each sector occurs as the single most important activity.  ANNEX F Page 2 5. Compared with 1940, services in 1970 were a major growth sector, particularly taking account of the occasions on which it occurred in association with other sectors. Manufacturing was much less important than in 1940, while commerce and transport also occurred less often as single dominant sectors than 30 years earlier. 6. Almost all of the high income cities in 1970 had dominant service or commercial functions, but Tampico, the city with the highesit income, had a dominant hydrocarbons function. The manufacturing cities, without exception, had relatively low levels of gross output per capita. Of course some cities with service sector economies had rela- tively low levels of imputed gross output. 7. In 1940 there were marked contrasts between cities with different economic functions. The high average size of cities with a dominant service function is explained by including Mexico City in the computation. Ignoring this distortion, manufacturing cities clearly had a larger average size than others. "Transport" cities were also relatively large, whether this sector occurred alone or in association with other sectors. Government, mining, hydrocarbons and commerce, occurred as export sectors in cities of roughly similar but smaller sizes. Construction was an export sector only in the smallest of the 37 cities. 8. Excluding Mexico City, the 1970 results contrast with those of 1940. Manufacturing was still associated with the larger cities but commerce, transport and services, particularly when measured in terms of all occurrences rather than unique occurrences of export sectors, were also associated with cities over 200,000. The other sectors were associated with smaller cities. Among the "transport" cities there were two categories: (1) those which were important nodal centers in the interior of the country occupying strategic locations on the railnet, and (2) the ports. 9. There were also marked contrasts between functional economic identity and relative population growth in 1940-70. The "export identity"' of each city and the population growth of each city indicates that only 10 out of 37 cities retained the same identity. Wlhen defined in terms of partial identity, the ratio was roughly reversed. When transformed to a rank  ANNEX F Page 3 order arrangement of the 16 cities which grew at a faster than average rate, only Monterrey, among the ten top ranking cities, maintained the same absolute identity for 1940 and 1970. Of the six remaining cities, two (Coatzacoalcos and Leon) retained the same sector identity throughout. 10. Further, all of the six fastest growing cities were identified with commerce and/or services in 1970 and had, in every case, added services to their major export sectors after 1940.  ANNEX G PUBLIC SERVICES - FACTOR ANALYSIS The factors examined were: (1) Residentia.l density (1970). This is indicated by the number of persons per room, assuming seven- room units as the largest. (2) Percentage of one-room units in total hous:ing stock (1970). I'his is a proxy variable for the degree of congestion. The numbers of residents in such units are fairly uniform in all 37 cities on the level of 4.3 to 6.0 signifying mostly overcrowded units. (3) Percentage of housing units with electrici-ty (1970). (4) Percentage of housing units with TV (1970). (5) Percentage! of housing units without water supply facilities (1970). (6) Number of students per class in primary schools (1972-73). (7) Passing rate of examination at the completion of the first grade (1972-73). (8) Passing reLte of examination at the completion of sixth grade (1972-73). (9) Percentage of population without formal school education (1970). (10) Operating costs of primary school per student (1972-73). (11) Number of public hospital beds per 10,000 people (1973). (12) Number of doctors in public hospitals per .10,000 people (1973). (13) Number of nurses in public hospitals per 10,000 people (1973). (14) Electricity consumption per residential un:it (1973).  ANNEX G Page 2 (15) Average sales of electricity for residential use (1973). (16) Telephones per 100 people (1973). (17) Rate of increase of telephones (1966 to 1973). 2. The correlation of variables between different sectors reveals the nature of the variables and their inter- relationships. 3. Among the housing sector variables, the first two: residential density (1), and one room (2), showing congestion of the unit are correlated with no water (5), and no education (9). There is also a relatively significant correlation with hospital beds (13). Houses with electricity (3) and no water (5), show a similar pattern of correlation and are correlated with no education (9), doctors (14), nurses (15) and telephones (18). The index of houses with TV (4) shows a somewhat different pattern but there is an inevitable correlation with electricity consumption (16) and telephones (18). The four variables (density, one room, houses with electricity, no water) in the housing sector, reflect the basic service level of this sector, which is related to other basic needs of medical and educational services. 4. Among five education sector variables, the first three variables: number of students in class (6), passing rate 1 to 2 (7), passing rate 6 to next (8) were not mean- ingfully related to other sectors. The no education variable (9) was slightly related with housing unit density (1), housing with electricity (3), no water (5) and telephones (18). Educational costs (10) were related to electricity consumption. No education and education costs were neither related to other education variables nor to each other. This implied (i) the quality of primary education was not always related to the physical service level, and (ii) the physical level indicated by class size and operation costs was also independent from the basic service level of other sectors. The no education variable (9), which is not a precise indicator of the present level of education, reflected the general trend of public services in each place. 5. There was, generally, a meaningful correlation with the housing sector variable, the basic needs variables and telephones. Thus, these variables and the housing variables, reflected the most fundamental necessities of social services.  ANNEX G Page 3 6. Electricity variables (16, 17) have a slight relation with housing with TV (4). 7. The telephone variable (18) was correlated with housing with electricity (3), housing with TV (4), no water (5), no education (9), population (11), and doctors (12). These and the housing with TV variable (4), and electricity consumption (16), are indicators of the conveniences of urban life. 8. Some variables indicate levels of fundamental necessities for urban services. These include housing congestion, no water, no education and medical services. There is a group of variables which refers to the comforts of urban life. They include housing with TV, electricity consumption and telephones. Education variables except for no education (9) are more or less independent. 9. In order 1:o examine urban service levels in aggregate, the score of each city for each variable was ranked (Table 12.3). After each score was ranked, the means and coefficients of variations of ranks for each city with respect to 17 variables were calculated. These results were transformed into ranks (Table 12.4). 10. The highest mean (Col. 1) was that of Mexico City - followed by Torreon, Veracruz, Tampico, Guadalajara, Chihuahua and Monterrey. The lowest was Minatitlan - followed, by Oaxaca, Culiacan and Villahermosa. 11. Next, to show the variety of the ranks of a city in each variable, coefficients of variation were examined (Col. 2). Here the smallest relative variation was in Minatitlan and the largest was in Mexico City. The rankings of mean and relative deviations were generally opposite. The sharpest contrast was between Mexico City (the top in mean, the bottom in deviation) and Minatitlan (the bottom in mean but the top in deviation). This implies that the higher the service level, the more difficult to maintain a high standard in every sector of urban services. This seems plausible in large cities which suffer from congestion as a result of rapid immigration, resulting in lower scores in some service level indicators. 12. Further, to indicate the degree of concentration of high level services in a particular locality, simple indices are shown. These are the five best and the five worst (Table 12.5).  ANNEX G Page 4 13. The largest number of the best five are concentrated in Mexico City (8) and the worst, in Minatitlan (8), Villa- hermosa (7), Oaxaca (6), Tijuana (5) and Culiacan. This shows a clear concentration of high service levels in the larger urban centers and low service levels in medium sized cities outside the influence of the capital region. 14. The 17 variables were processed by principal component analysis based on the correlation matrix. The raw factors were rotated 7 times to obtain the largest meaningful loadings for each factor. Four extracted factors seemed meaningful. The loadings are shown in Table 12.6. 15. The nature of the factors extracted was much the same as anticipated. For convenience these factors might be termed: Factor 1 Basic necessity Factor 2 Convenience Factor 3 Educational quality Factor 4 Growth 16. Factor 1 has heavy loadings on residential unit density (1), one room (2), housing with electricity (3), housing with TV (4), no water (5), no education (9), hospital beds (13), doctors (14), nurses (15) and telephones (18). These are variables which indicate fundamental urban services plus some elements of convenience. This factor explains 27.5 of total variance, the largest single factor variance. 17. Factor 2 has significant loadings on housing with TV (4), education operational costs (10) and extremely high loadings on electricity consumption (17) and electricity sales (18). This factor is obviously heavily inclined to electricity and has some bearing on the convenience of urban life. This factor explains 16.9% of total variance. 18. Factor 3 has high loadings on the two educational variables, passing rates of 1st and 6th graders (7 and 8) and some loading on the population variable. This mainly reflects the quality of primary education. This factor explains 12.5% of total variance. 19. Factor 4 shows loading on one room (2), class size (6), and telephone increase (19). This factor has characteristics related to growth. The variance explained is 12.5 percent of the total.  ANNEX G Page 5 20. The factor scores were ranked according to the following categories. The results are shown in TablLe 12.7. A' 1.50 .A 0.50 1.49 B -0.49 1.49 C -0.50 1.49 C' -1.49 21. With respect to service levels, there were strong contrasts. The border cities scored well on "convenience" services but low oni "necessity" services, while other northern cities were relatively well-off in both respects. The cities of the center varied greatly in terms of service levels. Excluding the largest cities (which had good service levels), there was a fairly close relationship between population growth and services. Most of the cities in the west-center had relatively low rates of growth and relatively low levels of income (with generally poor distribution) and low service levels. The place,s closest to Mexico City, however, had variable service levels (although most were high) but had similar patterns of income and population growth. Cities to the east of the capital were generally better-off, with the notable exception of Minatitlan and Villahermosa. 22. Finally, there were no consistent patterns in the Pacific Coast cities in the south or north although service levels were generally low. 23. The regional patterns illustrated in Maps 33 to 36 are reasonably consistent. But there are some anomalies, including Minatitlan and Villahermosa in the east, and Hermosillo and Oaxaca in the west. Both Villahermosa and Oaxaca had very low income levels, while Hermosillo had among the highest income! levels in the country. Minatitlan thus stands out as the most "deviant" cast, the explanation for which may well refer to the highly specialized nature of its economy.  ANNEX H THE DEMOGRAPHIC COMPONENTS OF URBAN GROWTH 1. The female population in the 15-44 age cohort for 1940 and 1960 includes the most fertile years of the female life-span; differences between states in this respect may therefore indicate reproductive potential. The contrasts between states in 1940 were such that the range was between 51.9 percent in the Federal District and 41.3 percent in Tabasco. In 1960, by which time the average level was lower (41.9 percent cf. 45.7 percent in 1940), the range extended from a maximum of 44.9 percent (again in the Federal District), to a minimum of 38.9 percent (in Tlaxcala). 2. The highest birth rates in 1940 occurred in Coahuila (61.4/1000) and Guanajuato (55.5/1000) and the lowest in the Federal District (33.6/1000) and Tamaulipas (35.4 percent). This suggests that factors other than the age structure were important in determining differential fertility. By 1960 the pattern had changed quite a lot. The highest birth rate occurred in Zacatecas (54.4/1000) and the lowest in Quintant Roo (32.5/1000). Again, however, there was not a close relationship between the birth rate and the female age structure. 3. A recent study of mortality differentials (Colegio de Mexico: 1970), allocates each of the states to one of eight groups and three indicators of mortality - the gross mortality ratio, life expectancy at birth and infant mortality. Two groups of states (I and II) scored higher than the others on two out of three indicators (the exception being infant mortality). Group VIII, comprising states in the south, had the lowest rank on each of the indicators. Group VII (con- taining the Federal District and the State of Mexico) did not, in 1940 and 1950, score very highly, but its rank greatly improved between 1950 and 1960. 4. These data refer to each state as a whole not to urban fertility or mortality within each state. Research findings from earlier censuses suggest that in Mexico there may be fewer contrasts between rural and urban fertility than in other Latin American countries. It appears that there was no consistent pattern in 1960 between crude birth rates and relative urbanization. Thus, Chiapas, Oaxaca, Zacatecas and Tabasco were among the least urbanized states. But with the exception of Zacatecas, birth rates in these states were not particularly high. It is difficult to infer the extent to which urban population increase was a con- sequence of differential fertility. Nor is it possible, on the basis of available data, to reach any firm conclusion about the relationship between inter-state variations in  ANNEX H Page 2 mortality, and differential urbanization. The differential role of natural increase as a component of urban population growth is an important topic for future research. 5. There is a much closer pattern of association between relative urbanization and relative migration. It is reasonable to state that variations in natural increase were intimately related to inter-state migratory moviements. 6. There were wide disparities between states with large numbers of migrants and those with relatively few migrants. In 1950 and 1960, over 60 percent of the population of Baja California were migrants; by 1970, the importance of migrants had greatly diminished. In many states, 95 percent or more of the total population was state-born during the period. 7. There was a close relation between the relative importance of migrants in state populations and interstate contrasts in the rate of urbanization. Thus, the most rapidly urbanized states in 1940-50 and 1950-60 were those in which the migrant population at the end of each inter- censal period was of greatest importance. In 1950, Baja California had the largest relative (interstate) migrant population and the fastest rate of urbanization. The Federal District, Colima, Tamaulipas and Morelos also had large migrant populations and were among the states which had urbanized most rapidly in 1940-50. With respect to such apparent anomalies as Guerrero, Tlaxcala and Chiapas, all of which urbanized rapidly in 1940-50, but in which inter- state migrants did not represent a large share of the state populations, it is inferred that intra-state migration was a key to rapid urban population growth. 1/ 8. In 1950-60, a similar pattern prevailed, but there were fewer anomalies and the relationship between urbanization and the size of the migrant population was stronger and re- mained strong in 1960-70. 1/ This inference is supported by research carried out by the Colegio de Mexico.  ANNEX H Page 3 9. The states which had positive net migratory flows were those in which migration played the most important role in population growth, and there was a strong association between net migration and the level of urbanization. 10. Comparing the relative level of urbanization in 1960 with net migration in 1950-60, the highly urbanized states of the Federal District, Baja California, Colima and Nuevo Leon, were areas of net inmigration. Conversely, the states with the lowest levels of urbanization in 1960 (Chiapas, Guerrero, Hidalgo, Oaxaca and Tabasco - all of which were less than 25 percent urbanized), had the highest levels of net outmigration in 1950-60. This pattern also occurred in 1960-70. The most urbanized states in 1970 (those more than 70 percent urbanized) were Baja California, Coahuila, the Federal District, Morelos and Nuevo Leon and these were also the states with the highest levels of net migration in the preceding decade. 11. Migration was, however, a less important component of urbanization in 1950-60 and 1960-70 than 1940-50. For example, fewer states (17 as compared with 21) experienced net outmigration in 1960-70 as compared with 1950-60. In 1950-60 there were 1.7 million interstate migratory move- ments while the total population increased by 9.0 million - a ratio of 1:5. In 1960-70 the corresponding ratio was 1:7, reflecting a total of 1.9 million interstate migratory movements and a total population increase of 13.3 million.  ANNEX I THE STATE DEVELOPMENT INDEX 1. Comparisons of economic development based on single indicators, such as state product, have shortcomings at the intranational as well as at the international level. The choice of a twelve component index based on a variety of social and economic indicators, overcomes some of the defects of a single component measure. This index illustrates differ- ences and trends in agriculture, industry, labor productivity, the size and structure of the labor force, and a variety of social welfare indicators, including the consumption of gasoline, electric power and sugar. Finally, it incorporates differences in housing conditions, health and education. 2. Although the index is an improvement, it too has defects. Being largely based on the percentage of a given population fallinq into (or out of) a particular category, as the percentage nears one hundred, differences are obscured. Nevertheless, this index, first developed by Unikel and Victoria (1970), is the best so far available for Mexico. 3. The construction, on the basis of this index, of a dynamic index of relative disequilibrium, is a means of assessing contrasting patterns of change (from 1940 to 1970) in each part of Mexico relative to the Federal District.  ANNEX J: TECHNICAL NOTES Technical Note Title No. 1 The Rank-Size Rule No. 2 The Index of Specialization No. 3 The Index of Surplus Employees No. 4 The Definition of Urban Places No. 5 Population Growth Analysis No. 6 The Location Quotient No. 7 Census Adjustments Relating to Employment in the Services Sector No. 8 Population and Income Potential No. 9 The Beta Coefficient as an Index of Connectivity No. 10 The Konig Number an Index of Centrality  TECHNICAL NOTE NO. 1 The Rank-Size Rule Auerback (1913) first described regular relationships between the size of towns and their rank. The rank size rule is given by the formula: Pn = P1 (n) where Pn is the population of the nth town in a series 1-2-3. . . n, in which all towns in a region are arranc.ed in a descending or(der by population and P1 is the population of the largest or primate town or city. The 5th largest town woulc have a population one-fifth that of the largest town. Stewart (1958) has shown that the rank-size rule is an empirical finding rather than a theoretical notion. Regularity howe-ver is greater in the lower than the upper "limbs" of u;:b.n hierarchical systems.  TECHNICAL NOTE NO. 2 The Index of Specialization This is given by the formula e. e. et or Ei Ei t Et Et where e. = local industry or sector employment 1 et = local total employment E. = industry or sector employment in 1 nation or set of places or areas E = all employment in nation or set of places or areas  TECHNICAL NOTE NO. 3 The Index of Surplus Employees This is given by the formula S = el E et E E t where S represents the absolute number of "surplus" workers or employees in industry or sector (i). The upper limits of the index approaches e. asymptotically. The lower limit is et The index is useful as a general measure of Et E the basic quality of an industry or sector in an urban economy. It is used here as a general indicator of "export" activity.  TECHNICAL NOTE NO. 4 The Definition of Urban Places 1. Many students of Mexican urban development have argued that the census definition of urban population - 2,500 - is unrealistic. Favoring a definition which regards urban places as those with populations of 15,000 or 20,000, some have argued (Unikel and Torres: 1970) (Fox: 1968) that part of the so-called urban population captured by the census is in fact rural - in economic and cultural terms. They have also argued that the administrative map of Mexico reflects past rather than present reality. There are, for example, 571 municipios in Oaxaca but only 52 in Nuevo Leon, because Oaxaca had a large indigenous population in the sixteenth century. Generally, the south of the country has more municipios than the north. Consequently, urbanization has been more easily achieved in the south than the north, not- withstanding that the south is, in general, much less developed. 2. These are valid arguments which the mission respects. But in deciding to work with an urban definition correspond- ing to that of the census (2,500) several factors were weighed and,on balance, the counter arguments were more powerful. 3. First, there is the fundamental but unsettled question of the difference between "rural" and "urban". If the difference were economic, urban areas would, by definition, have no agricultural activities. If the difference were cultural, urban areas would have a different way of life. If the difference were a simple matter of size there is no reason that a nuclear population of 2,500 people cannot be an urban population. On the other hand (Davis and Golden: 1953), it is argued that an urban nucleus becomes more urban the larger it is. 4. These factors have led Unikel and Torres to propose a definition of urban places in Mexico which has a lower cut-off point of 20,000 - i.e., smaller populations are not considered urban at all. The "cities" thus defined have very little agricultural activity and are almost entirely concerned with the properly "urban" functions of the secondary and tertiary sectors. 5. The use of the urban municipio or urban census tract as a statistical base has certain conceptual advantages. If one is interested in urban change it is possible to measure - as in this report - the process of urban change within the census tract over time by analyzing the changing structure of  economic activity - as revealed by employment data from one year to another. There is moreover no certainty that agri- culture may not be an urban activity in the sense that some of the economically active populations of large cities may well tend to work outside the city. Moreover, from a cultural standpoint there is less justification in I.atin America - Mexico is no exception - for arguing that urban ways of life are markedly different from rural ways of life, that many migrants - to large cities in particular -- recreate rural ways of life in new urban environments (rurs in urbis) and that they may continue to live in a "rural" environment long after they have joined the urban population. 13ut, the nominally rural populations of areas proximate to cities are associated with the city; they may even work there regularly or occasionally and will probably visit more often than those who live beyond its immediate sphere of influence. 6. In short, any definition of what is "urban" and what is "rural" is arbitrary. While the mission found the census definition unsatisfactory on some grounds, i- also found reasons for accepting it. 7. The second question is methodological rather than conceptual. Unikel and Torres have for 1940, 1950 and 1960, published data adjusting the urban census tracts to "cities", but their information concerns only urban population size and the economically active population. This report refers to many other chara-cteristics of urban population from popula- tion census data and other materials, notably those of the industrial, commerce and service censuses, fiscal data, savings and loans data and many series of current statistics - all of which are available on a municipal base only. As a practical matter, the conversion of these data into a base something like Unikel's "cities" would have been impossible even if it had been unambigously desirable on conceptual ground - which (for reasons already cited) was not the case. 8. Third, there was the question of the particular focus of this report. The concern was with the past, present and future of the country's major cities. We were therefore dealing with a set of urban places which in 1970 had more than 100,000 people. Our interest in the general process of urbanization was limited to the analysis of the framework of urban size growth - in the country as a whole and on the regional level - as a background to the analysis of the development of the largest cities.  9. Finally, what difference does it make? The best answer is that at the national level the ratio of the urban population defined by the census to the urban population as defined by Unikel falls from 1:1.20 in 1940 to 1:1.11 in 1970. 10. These differences are not insignificant. The use of one or the other definition does not change the conclusions about urban development nor about other processes (including industrialization and economic growth). Nor does it change the trends of the urban development process or produce a different rank order of cities in terms of any indicator of urban development. In terms of results, it does not make much difference, but it does make more difference in some places than in others. For example, the urban population of Oaxaca in 1960 was 78,639 according to the municipal census tract and 75,196 according to Unikel, and the population of Mexicali (the worst case of distortion) was 281,333 according to the census and 179,539 according to Unikel. 11. No "correct" or fully satisfactory answer exists. That which has been given is believed to be defensible on conceptual grounds, unassailable on methodological grounds, reasonable in terms of the purpose for which it has been used and pragmatically acceptable since it makes little difference one way or the other. 12. Most of the analysis in this report therefore refers to the municipal census tract. However, where it is felt useful to introduce a different definition (most often the 20,000 plus definition) this is done and the reasons are explained. What is not done is to compare one definition directly with another, although this has been done in at least one recent paper (Unikel and Garza 1971).  TECHNICAL NOTE NO. 5 Population Growth Analysis The 37 c:Lties were divided in six regions because (a) vital statistics and migration consist of estimates which are subject to serious error, and grouping of municipalities therefore decreases the error factor, and (b) population characteristics affect entire regions rather than individual cities or municipalities. Three factors were taken into account (1) whether a state had experienced net inmigration or net outmigration during the previous three decades, (2) whether it had experienced a relatively high or a relatively low rate of natural increase during the last three decades, and (3) whether two or more states had common borders. (1) Baja California, Colima, Morelos, Nuevo Leon, and Sonora had net immigration and high rates of natural growth during all three decades; (2) Aguascalientes, Campeche, Coahuila, Durango, Queretaro, S.L.P., Tlaxcala and Zacatecas had high rates of natural growth but were net outmigration states during the last three decades; (3) The Federal District had net immigration and a low rate of natural growth during the! three decades; (4) Chiapas, Oaxaca, Puebla, and Veracruz had low rates of natural growth and net out- migration during the last three decades. The other 14 states fell into different groups in different decades,, so it was decided to base the classification on the situation in the last decade 1960-70. Thus: (1) High natural growth - net immigration: Baja Ca:lifornia, Baja California Territory, Sinaloa, andl'Tabasco;  -2- (2) High natural growth - net outmigration; Chihuahua, Nayarit, Yucatan, Guanajuato, Guerrero, Michoacan, and Jalisco; (3) Low natural growth - net immigration: Mexico, Quintana Roo, and Tamaulipas; (4) Low natural growth - net outmigration: Hidalgo. The following groups thus emerged: (1) Baja California, Baja California Territory, Sonora and Sinaloa, all of which had a high rate of natural growth and net immigration in the last decade; (2) Chiapas, Oaxaca, Puebla, Veracruz, and Hidalgo, all of which had a low rate of natural growth and net outmigration in the last decade; (3) Aguascalientes, Coahuila, Durango, Queretaro, S.L.P., Zacatecas, Chihuahua, Nayarit, Guanajuato, Guerrero, Michoacan, and Jalisco, all of which had a high rate of natural growth and net outmigration during the last decade; (4) Mexico and Mexico City with net inmigration and low rate of natural increase during the last decade. This left out the states of Tabasco, Morelos, Nuevo Leon, Campeche, Tlaxcala, Yucatan, Quintana Roo, Tamaulipas and Colima. Because of the number of states in group three it was decided to divide that group by taking Guanajuato, Guerrero, Michoacan, and Jalisco away from it because of their geographic proximity and somewhat similar characteristics in the three decades. Guanajuato, Michoacan, Guerrero, had low natural growth and outmigration in 1950-1960 while in the other two decades these states had high natural growth rates and outmigration.  - 3 - The states of Tabasco and Yucatan were added to group 2 because of their geographic location. For t:he same reason Morelos was added to group 4. Nuevo Leon and Tamaulipas were joined intc) a single group because of their physical proximity. In review the regions are as follows: Region I: Mexico City, Mexico and Morelos 3 Region II: Jalisco, Guerrero, Michoacan, Guanajuato 4 Region III: Nuevo Leon, Tamaulipas 2 Region IV: Puebla, Yucatan, Veracruz, Tabasco, Oaxaca, and Hidalgo 6 Region V: Coahuila, Chihuahua, San Luis Potosi, Aguascalientes, Durango, Coahuila, and Queretaro, 7 Region VI: Baja California, Sonoro and Sinaloa 3 25 The other seven states: Nayarit, Zacatecas, Campeche, Tlaxcala, Chiapas, and the Territories of Baja California and Quintana Roo were not assigned to any region because they did not have cities of 100,000 or more population in 1970 and were not important for this analysis.  TECHNICAL NOTE NO. 6 The Location Quotient This was devised by Sargent Florence (1948). It measures the degree to which a particular industry or sector is concentrated in an area or place relative to the extent to which all activity is concentrated there - measured on the basis of employment data. It is given by the formula: Numbers employed in industry or sector in place or area of total in nation or set of places Numbers employed in all industries or sectors in place or area as share of national total.  TECHNICAL NOTE NO. 7 Census Adjustments Relating to Employment in the Services Sector The major contrast between the 1960 and 1L970 censuses in this regard refers to the transfer of certain activities previously included in the industrial sector to the services sector. The main category affected is work- shops and similar small establishments. The effect is to inflate the size of the services sector from one year to the other.  TECHNICAL NOTE NO. 8 Population and Income Potential This concept has been used only implicitly in this report. Formally, it refers to the "dynamical radius" of a population (Stewart and Warntz, 1958), defined by the expression: (pd2 ) } /P where p is the population of a small area, d is the distance from the mean center and P is the total population. The formula can be adjusted to apply to market potential, taking account of income differentials.  TECHNICAL NOTE NO. 9 The Beta Coefficient as an Index Connectivi-ty This derives from graph theory. In a given net- work, connectivity rises as the number of "edges" increase, thus improving connectivity between the "vertices". That is: B1 no. of edges no. of vertices' ="edges" o = "vertex"; For example: A B B = 0.86 B = 1.00 C D 0 o _ B = 1.14 B =1.28