gregory k. ingram and alan carroll the spatial structure...

19
World Bank Reprint Series: Number 211 Gregory K. Ingram and Alan Carroll The Spatial Structure of Latin American Cities Reprinted with permission from Journal of Urban Economics, vol. 9, no. 2 (March 1981), pp. 257-73. Copyright by Academic Press. Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized

Upload: others

Post on 26-Jul-2020

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

World Bank Reprint Series: Number 211

Gregory K. Ingram and Alan Carroll

The Spatial Structureof Latin American Cities

Reprinted with permission from Journal of Urban Economics, vol. 9, no. 2 (March 1981),pp. 257-73. Copyright by Academic Press.

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Page 2: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

JOURNAL OF URBAN ECONOMICS 9, 257-273 (1981)

SYMPOSIUM ON URBANIZATION AND DEVELOPMENT

The Spatial Structure of Latin American Cities

GREGORY K. INGRAM AND ALAN CARROLL'

The World Bank, Washington, D.C. 20433

Received August 25, 1979; revised December 31, 1979

Using published census data, metropolitan area population and employmentstatistics are constructed for several large Latin American cities in 1950, 1960, and1970, and compared to similar statistics from selected North American cities. TheLatin cities are experiencing decentralization of population and some decentraliza-tion of employment. Overall population density patterns of large Latin citiesresemble those in older North American cities; newer North American cities havelower densities and are much more decentralized than Latin cities. High-statusgroups are somewhat concentrated in the central cities of Latin American metro-politan areas, but their concentrations there are declining.

The twentieth century has witnessed a profound increase in the world'spopulation and an equally remarkable increase in the proportion of theworld's population living in urban areas. Much of the rapid urbanizationhas occurred in less developed countries where it has brought a host ofproblems as govemments have sought to provide urban infrastructure,control development, and redirect growth from larger to smaller settle-ments. Many urban analysts question the relevance that urban economictheory has for the analysis of the problems of cities in less developedcountries because most empirical work underlying this theory employsdata from developed countries. This paper investigates the generality ofempirical results by analyzing the spatial structure of selected LatinAmerican cities and comparing it with that of selected North Americancities.

URBANIZATION AND LARGE CITIES IN LATIN AMERICA

Table 1 indicates that between 1920 and 1970, the total population ofNorth America nearly doubled while its urbanized population increased

'Support for the work reported in this paper has been provided by the City Study researchproject (RPO 671-47) funded by the World Bank. The views and conclusions reported hereare those of the authors and not of the World Bank or its affiliated organizations. The authorsthank Richard Davis and Yoon Joo Lee for research assistance and members of the CityStudy research staff at the World Bank and at Corporacion Centro Regional de Poblacion inBogota for comments on the work presented here, wlth particular appreciation to RakeshMohan and Jose Fernando Pineda. Ingram was principally responsible for the analysis andCarroll, for the data assembly.

2570094-1 190/81/020257-17$02.00/0Copyright ( 1981 by AcademiC Press, Inc.

All rights of reproduction in any form reserved.

Page 3: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

258 INGRAM AND CARROLL

TABLE ITotal and Urbanized Population in North and Latin America

Year

Region' 1920 1930 1940 1950 1960 1970 1980b

Total population (Millions)

Latin America 90 110 130 162 213 284 374North America 115 135 145 166 199 226 249

Urban population (Millions)

Latin America 20 30 40 67 105 161 237North America 60 75 85 106 139 168 196

Urban population as percentage of total population

Latin America 22 28 31 41 49 57 63North America 52 56 59 64 70 74 79

Source. From [7]; urban population figures are defined by each country.aNorth Amnerica includes, the United States and Canada; Latin America includes all

countries south of the United States.bU.N. projections.

slightly less than three-fold. During the same period, however, the totalpopulation of Latin America more than tripled while its urbanized popula-tion grew eight times as large as its 1920 base. By 1980 Latin America willbe as urbanized as North America was in 1950, and its urban populationwill be absolutely larger than North America's. However, Table 1 alsosuggests that the percent of Latin America's population living in the urbanareas is now growing less rapidly than before, and that the rate ofurbanization is beginning to slow.

The study of urban spatial structure reported here focuses on large LatinAmerican cities-those with 1970 metropolitan area populations over onemillion or with 1970 central city populations of 600,000 or more. The 24cities that meet this criterion are listed in Table 2. For purposes ofcomparison, in 1970 North America had 36 metropolitan areas (34 in theUnited States and 2 in Canada) with a million or more inhabitants.

Having defined the universe of cities that we hope to analyze, the nextstep is to obtain comparable data on the spatial distribution of population,employment, and other activities in (approximately) 1950, 1960, and 1970.Although several demographic censuses are available for virtually all LatinAmerican countries, the availability of data at the metropolitan andsub-metropolitan area varies widely. We have been able to obtain data fora central core and peripheral ring for 11 of the 24 cities; only 8 of the citiesprovide data disaggregated at that level for two decades.2 Since tabulations2See the Appendix for data sources used.

Page 4: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 259

TABLE 2

Latin American Cities with Central City Populationsover 600 Thousand in 1970

Country City Population date Population (000)

City proper Metro area

Argentina Buenos Aires 1970 2,972 8,189Bolivia La paZa 1973 605 -

Brazil Belo Horizonte 1970 1,235 1,606Porto Alegrea 1970 870 -

Recife 1970 1,061 1,793Rio de Janeiro 1970 4,252 7,082Salvadora 1970 998 -

Sao Paulo 1970 5,979 8,195Chile Santiago 1970 1,118 2,820Colombia Barranquillaa 1972 682 694

Bogota 1973 2,855 -

Cali 1973 898 -

Medelina 1972 1,092 1,208Cuba Havana b 1972 2,346 -

Dominican Rep. Santo Domingoa 1970 671 -

Ecuador Guayaquil' 1972 861 -

Guatemala Guatemala Citya 1970 731 -

Mexico Guadalajara 1970 1,199 1,455Mexico City 1970 2,903 8,657Monterrey 1970 858 1,213

Peru Lima 1972 1,448 3,302Uruguay Montevideob 1963 1,159 -

Venezuela Caracas 1970 1,035 2,199Maracaibo' 1970 690 -

'From [6].bFrom [10].

Note: Other figures are from country censuses; see the Appendix.

for the economic censuses are typically disaggregated only to the statelevel, examining the intra-metropolitan distribution of employment ispossible for even fewer cities. And in these few cases, moreover, it is likelythat small enterprises are not well covered, so that economic census datamay only reflect the experience of large establishments. Although problemsof coverage and comparability are more serious than those encounteredwith the U.S. Census, Latin American census data are likely reliableenough to reflect broad trends of growth and change in metropolitanspatial structure.

METROPOLITAN AND CITY POPULATION GROWTH

Table 3 presents estimates of populations and annual average metro-politan population growth rates during the two most recent decades for 13

Page 5: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

TABLE 3Populations and Growth Rates-Selected Latin and North American Metropolitan Areas

Latin American Population (000) Annual growth rate Population (000) Annual growth rate North AmericanCity 1950 1960 1970 50-60 60-70 1950 1960 1970 50-60 60-70 City

MexicoCity 3180 5246 8657 5.1 5•1 9556 10695 11572 1.1 0.8 NewYork ZSao Paulo 2708 4818 8195 5.9 5.5 4152 6039 7032 3.8 1.5 Los Angeles 0Buenos Aires 4723 6739 8189 2.8 2.0 5178 6221 6979 1.9 1.2 ChicagoRio de Janeiroa 3298 5012 7082 4.3 3.5 3671 4343 4818 1.7 1.0 PhiladelphiaLimab - 1846 3302 - 5.4 1508 2077 2861 3.3 3.3 Washington, D.C. >Bogotac 715 1697 2855 6.9 5.9 2414 2595 2754 0.7 0.6 Boston tVSantiagod 1509 2170 2820 4.6 2.7 936 1418 1985 4.2 3.4 Houston QCaracas' 724 1388 2199 6.1 4.7 557 1033 1358 6.4 2.8 San DiegoRecife 819 1240 1793 4.2 3.8 495 935 1268 6.6 3.1 Miami 3Belo Horizonte 475 888 1606 6.5 6.1 612 929 1228 4.3 2.8 Denver rGuadalajara 440 851 1455 6.8 5.5 291 642 1065 8.2 5.2 San JoseMonterrey 376 708 1213 6.5 5.5 332 664 968 7.2 3.8 PhoenixCalic 284 638 898 6.4 3.9

a1947,1960,1970.b19 6 1 ,1972.C1951,1964,1973.dl9 5 2,1960,1970.'1950,1961,1971.Source: For Latin American cities, see the Appendix; data for North American cities are from [8, Table&32] based on 1970SMSA definitions.

Page 6: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 261

large Latin American cities with available data and for 12 large single-centered U.S. metropolitan areas selected for their range of location, size,and growth rates. Of the 13 Latin American cities, 11 had metropolitanareas defined in 1970. The data in Table 3 for these cities were obtainedfor 1950 and 1960 by aggregating populations in the central city andsurrounding municipalities for those earlier years using the 1970 metro-politan area definitions.3 Data for Bogota and Cali, the two Latin Ameri-can cities lacking metropolitan area definitions, are for populations withinthe cities' urban perimeters.

The average growth rates for the metropolitan areas in Table 3 exhibitseveral consistent patterns. First, the decennial average population growthrates in each metropolitan area declined from the fifties to the sixtiesexcept in Mexico City and Washington, D.C., where the growth rates wereunchanged. For the 13 Latin American cities the average growth rate fellfrom 5.5% in the fifties to 4.6% in the sixties, a decline that was less markedthan that of the 12 North American cities whose average growth rate fellfrom 4.1 to 2.5% in the same two periods. In both areas these declinesreflect a reduction in birth rates and a decline in the overall rate ofurbanization, although the average growth rates are still much higher inLatin America than in North America. Second, Table 3 suggests that smallmetropolitan areas tend to grow more rapidly than large metropolitanareas. For example, if large metropolitan areas are defined as those with1970 populations over 2.5 million and small metropolitan areas as the rest,we find that average growth rates for large Latin American areas were4.9% in the fifties and 4.3% in the sixties while average growth rates forsmall Latin American areas were 6.1% in the fifties and 4.9% in the sixties.For North American areas the respective average growth rates are 2.1% inthe fifties and 1.4% in the sixties for large areas, and 6.2% in the fifties and3.5% in the sixties for small areas. Although the Latin American averagegrowth rates are generally higher than those in North America, the lowgrowth rates of large North American metropolitan areas differentiate theNorth American pattern of urban growth most sharply from that found inLatin America. The high growth rates of large Latin metropolitan areas,typified by Mexico City and Sao Paulo, are of major concern to analysts ofLatin American urban development.

Another major difference between North American and Latin Americanurban growth is that urbanization in Latin America is occurring at muchlower real income levels. Gross national product per capita in 1975 was$7100 (in current dollars) for North America and $1000 for Latin America[11]. Based on growth rates of per capita product reported by Kuznets,4

3The metropolitan areas are defined in the Appendix.4Kuznets [3, p. 64], estimates that per capita product in the U.S. grew at 17.2 per decade

from 1839 to 1960.

Page 7: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

262 INGRAM AND CARROLL

North America would have had a per capita GNP of $1000 (1975 dollars)roughly 120 years ago, or in the 1850's. Since no cities in North Americathen had populations over one million, it is obvious that per capita incomeand urbanization have no simple causal relationship over time.

INTRAMETROPOLITAN POPULATION DISTRIBUTION

The pattern of population growth and the distribution of populationwithin cities is often used to characterize urban spatial structure. Table 4presents sumnmary information over a 20-year period for the central andperipheral areas of the 10 Latin American cities with available data andthe 12 North American cities from Table 3. It is striking that both thecentral and peripheral densities vary by an order of magnitude across thecities on both continents. Central densities are very high but are stabilizingin Mexico City and Buenos Aires; in the other 8 Latin American cities,central densities are increasing. For all 10 Latin American cities the

peripheral densities are rising, and only in Belo Horizonte is the peripheraldensity rising less rapidly than the central density. For 9 of the 10 Latincities, therefore, the periphery's population share is rising. Five of theNorth American cities have declining or stable central city densities.Although a declining central density is not strictly a concomitant ofstagnation (Washington, D.C., for example, has a declining central densityyet is one of the fast growing U.S. metropolitan areas), increased centraldensities seem always to be associated with high population growth. For all12 North American cities peripheral densities are risirig; in all but the twosmallest (San Jose and Phoenix), peripheral densities are consistentlyincreasing faster than central city densities.

In comparing the data for Latin and North American cities in Table 4, itis apparent that the central densities of cities in the two continents coversimilar ranges. Most Latin American central cities (with the possibleexception of Mexico City) do not appear to be significantly more densethan older central cities in the U.S. The peripheral densities of LatinAmerican cities are similar to those found in North American cities. Thenewer, rapidly growing cities in the southwest of the United States havecentral densities that are lower than those of older U.S. cities and lowerthan those of most Latin American cities. Only Monterrey has a centraldensity as low as that found in the rapidly growing south-western U.S.cities.

DENSITY FUNsTCTION COMPARISONS

Observations based on the data in Table 4 must be made carefullybecause they are based upon comparisons of arbitrarily specified centraland peripheral areas. Examination of central and peripheral land areas inTable 4 reveals vast differences in the absolute and relative magnitudes of

Page 8: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

TABLE 4Central and Peripheral Areas and Plopulation Densities

Latin American Center or Area Popn. density (Popn./Km2 ) Popn. density (Popn./Km 2) Area Center or North Americancity periphery (Kn 2

) 1950 1960 1970 1950 1960 1970 ( periphery city

Mexico City C 138 16,225 20,558 21,074 10,157 10,015 10,161 777 C New YorkP 2192 432 1,101 2,675 350 612 773 4,758 P

Sao Paulo C 1493" 1,380 2,287 4,005 1,675 2,130 2,390 1,326 C Los AngelesP 6458 79 172 343 210 350 420 9,213 P

Buenos Aires C 200 14,952 14,872 14,897 6,275 6,140 5,825 578 C ChicagoP 3860 473 1,025 1,418 170 295 400 9,054 P

Rio de Janeiro C 1171 2,030 2,824 3,631 6,202 5,995 5,835 334 C PhiladelphiaP 5293 174 322 535 180 264 325 8,868 P

Deogotab C 304 2,352 5,582 9,391 5,077 4,835 4,790 158 C Washington, D.C. ZP - - - - 119 221 355 5,936 P

Recife C 209' 3,594 3,815 5,075 6,735 5,860 5,387 119 C BostonP 1992 148 223 367 662 780 867 2,437 P

Belo Horizonte C 335 1,053 2,070 3,686 580 913 1,198 1,028 C HoustonP 3335 37 58 111 22 31 49 15,250 P Z

Guadalajara C 188 2,204 3,940 6,383 606 1,038 1,257 552 C San DiegoP 1164 52 95 220 21 44 63 10,484 P

Monterrey C 451 752 1,332 1,901 2,833 3,315 3,805 88 C MiamiP 1292 28 83 275 47 124 179 5,201 P

Calib C 85 3,341 7,506 10,565 1,690 2,008 2,092 246 C DenverP - - - - 21 47 77 9,233 P

318 681 1,480 300 C San Jose64 143 202 3,067 P

166 684 906 642 C Phoenix10 10 17 23,069 P

al,622 in 1950 and 1960.bFor 1951, 1964, and 1973.C146 in 1950.

Source: For Latin American cities, see Bibliography; data for North American cities is from (8), Table 32 (based on 1970 SMSA definitions).

Page 9: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

264 INGRAM AND CARROLL

the center and periphery of the 22 metropolitan areas. It is possible,however, to reduce the effects of such arbitrary definitions by estimating apopulation density function for each metropolitan area. Using a techniquedescribed by White [9], with the data at hand and assuming that popula-tion densities decline exponentially with distance from the center, we canestimate the intercept, D, and gradient, b, of the density function

D(x) = Debx

where D(x) is the population density at distance x from the center.Parameters D and b can then be used to characterize the distribution ofpopulation within metropolitanr areas and to provide further insights abouturban spatial structure.

Table 5 displays estimates of density function parameters for the 10Latin American cities that have sufficient data for their calculation, andfor the 12 North Americatn described above. In both the Latin and NorthAmerican cities, the density gradient, b, generally declines over time. Onlyin Guadalajara, Belo Horizonte, Houston, San Jose, and Phoenix does the

gradient increase, and in each case this happens in only one of the twodecades shown. Moreover, these cities all have 1970 populations less than2.5 million and high population growth rates. The density gradients are

steeper for the Latin American than the North American cities: 16 of the28 estimated gradients exceed 0.2 in Latin America while only 4 of the 36do so in North America. Most of this difference in gradients is accountedfor by the smaller (1970 population less then 2.5 million) Latin cities,however. For example, the average 1970 gradient of 0.25 for the smallLatin cities is over twice the 1970 average gradient of 0.12 for the smallNorth American cities, while the average 1970 gradient of 0.12 for largeLatin cities is similar to the 0.105 average for large North American cities

in 1970. These averages also suggest that on both continents smaller citiestend to have steeper gradients than large cities.

The change in intercept density, D, over time is less regular than thatobserved for the density gradient. The intercept density has increasedduring the two decades shown for 6 of the 10 Latin American cities and for6 of the 12 North American cities. Intercept increses are typically accom-panied by high population growth rates, but high growth rates are notsufficient for intercept increases. Mexico City is the paramount example ofrapid growth accompanied by a declining intercept density. Ain NorthAmerican cities, increases in the intercept density are likely to occur in

cities that also have low intercept densities, say below 7500, but thispattern is less clear in Latin American cities where intercept densitiescontinue to increase beyond values of 20,000. In terms of absolute magni-tude, intercept densities are significantly higher in Latin America than in

Page 10: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 265

North America. The average 1970 intercept density of 24,000 in LatinAmerica was more than twice the 10,000 observed in North America.However, in North America there is a strong tendency for interceptdensities to be higher in large cities than in smaller ones whereas thispattern is less apparent for the Latin American cities. In 1970, for example,the North American average intercept densities were 15,700 for large citiesand 4500 for small cities while for Latin American cities the averageintercept densities for the two-size categories were 26,400 and 22,300,respectively.

When we combine the comparison of intercept densities with that madefor density gradients, two major points emerge. First, there is a surprisingdegree of similarity between the density function parameters of large Latinand large North American cities. This similarity is enchanced when wecompare the large Latin cities to only the five older northeastern cities(New York, Chicago, Philadelphia, Washington, Boston) in Table 5. The1970 average intercept and gradient, 17,700 and 0.116, for these five citiesare similar to the 1970 averages, 24,600 and 0.12, respectively, for the largeLatin American cities. Second, it appears that density function parametersfollow different patterns in Latin and North America: as size varies, NorthAmerican cities tend to have fairly constant density gradients and varyingintercept densities, while Latin American cities tend to have fairly constantintercept densities and varying density gradients. Coupling this patternwith the similarity of parameters for large cities suggests that sinall LatinAmerican cities have larger intercept densities and steeper gradients, andtherefore are much more centralized, than small North American cities.

Although the density function comparisons have been based on city sizecategories, the data for North American cities in Table 5 suggest that cityage and transportation technology are also important determinants ofdensity function parameter values. Older North American cities weredeveloped during a period when transit was the dominant transport mode,and the newer North American cities are developing when autos are thedominant transport mode. The older cities are therefore more centralizedand have higher densities than the newer cities. Although mode split dataare not widely available, the scattered evidence we have suggests thatmotorized transit is the dominant mode in large Latin American cities,accounting for roughly seven-tenths of work trip travel. The balance iscomprised of auto travel, taxis, and walking. Walking typically accountsfor 5 to 10% of work trip travel in Latin cities, so its modal share is thesame order of magnitude in Latin as in North American cities. The degreeof motorized travel is, therefore, similar in Latin and North Americancities. This similarity plus the historical dependence on transit in olderNorth American cities probably explains the similarity of density functionparameters in large Latin American cities and the older U.S. cities.

Page 11: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

TABLE S

Population Density Gradients'

Latin American North Americancity 4 b Parameter' 1950 1960 1970 1950 1960 1970 Parameter' b city

Mexico City 1.0 D 69,000 62,000 44,000 62,000 45,000 40,000 D 0.6 New Yorkb 0.37 0.27 0.17 0.16 0.13 0.11 b

S.,o Paulo 1.0 D 8,400 12,000 18,000 4,800 5,300 5,800 D 0.6 Los Angelesb 0.14 0.13 0.12 0.06 0.06 0.05 b

Buenos Aires 0.6 D 54,000 37,000 33,000 27,000 20,000 16,000 D 0.5 Chicagob 0.21 0.14 0.12 0.13 0.10 0.08 b

Rio de Janeiro 0.5 D 8,700 10,000 11,000 20,000 16,000 14,000 D 1.0 Philadelphiab 0.09 0.08 0.07 0.18 0.15 0.13 b

Bogota 0.5 D - 37,000 26,000 15,000 11,000 9,000 D 1.0 Washington, D.C.b - 0.25 0.12 0.25 0.18 0.14 b

Recife 0.6 D 13,000 14,000 17,000 14,000 11,000 9,300 D 0.7 Bostonb 0.25 0.21 0.19 0.16 0.13 0.12 b

Belo Horizonte 1.0 1) 5.000 11,000 19,000 2,100 3,500 4,200 D 1.0 Houston ob 0.26 0.28 0.27 0.12 0.13 0.12 b

Guadalajara 1.0 D 14,000 28,000 39,000 2,100 3,200 3,600 D 0.5 San Diegob 0.45 0.46 0.41 0.11 0.10 0.09 b

Monterrey 1.0 D 6,200 8,500 7,400 8,000 6,800 7,200 D 0.5 Miami r

b 0.32 0.27 0.19 0.23 0.15 0.13 bCali 0.5 D - 43,000 29,000 6,800 6,000 5,100 D 1.0 Denver

b - 0.41 0.21 0.27 0.20 0.16 b620 1,300 3,500 D 0.5 San Jose

0.08 0.07 0.10 b350 2,700 3,100 D 1.0 Phoenix

0.08 0.16 0.14 b

'Calculated using the technique described in [9].bProportion of circle that can be developed.cParameters from density = De bx, where x is distance from center in kilometers; D is central density in persons/km2 .

Page 12: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 267

Although the population density characteristics found in large LatinAmerican cities are similar to those of older North American cities, it isimportant to note that living conditions in Latin American cities are verydifferent. It is likely, for example, that Latin American cities will havelower ratios of capital to land, in terms of both housing and public serviceinfrastructure, than otherwise similar North American cities. Accordingly,similarities in population density patterns between North and LatinAmerican cities do not have much significance as indicators of welfarelevels.

INTRAMETROPOLITAN DISTRIBUTION OFSOCIO-ECONOMIC GROUPS AND JOBS

Because questions about household income are infrequent in LatinAmerican censuses, it is not possible to investigate directly tne in-trametropolitan distribution of income groups. Instead we must examinethe distribution of household characteristics, such as occupation, that arethought to be correlated with income. In Table 6, the center's share of theeconomically active population is compared to its share of up to fouroccupational groups in four cities. Assuming that professional/technicaland office/sales worker ca,gories proxy high socio-economic status, thedata suggest that high status groups are somewhat concentrated in centralcities. At the same time, it is apparent that the central city shares of thesetwo occupational groups are declining over time in the four cities with dataavailable. Auto ownership is another reasonable proxy for high incomes inLatin American cities. The high shares of auto ownership in central citiesalso suggest a concentration of high status groups.5 This concentration islikely attributable to the greater availability and higher quality of utilitiesand public services in central cities.

The summary of available data about central city shares of employmentover time in Table 6 indicate that jobs are decentralizing somewhat,although Recife and Belo Horizonte are notable exceptions.6 Manufactur-ing employment seems to be less concentrated in (he central city thanemployment in either commerce ori services. In most cases, however, itappears that even manufacturing employment is more centralized than theeconomically active population.7 The extensive decentralization of the

5The center-periphery comparison may conceal decentralization of high status groupsoccuring within narrow radial zones of cities, as in the north of Bogota, Colombia. Forreports on the movement of affluent groups outward in specific directions, see [1, 5].

61n Bogota, Colombia both firms and jobs in the manufacturing sector have movedoutward from the center during the 1970-1975 period [4].

7 This conclusion remains tentative due to the census coverage problems mentioned earlier,particularly with respect to small firms.

Page 13: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

268 INGRAM AND CARROLL

TABLE 6Central City Shares of Population and Employment

Percent of Items Located in CenterBy residential location By job location

Econ. active Prof. Office Blue AutosYear popn. tech. sales Service collar owned Mfg. Commerce Service

Mexico City 50 73 86 81 70a - - - -60 59 72 66 49a - - - -70 39 52 47 43 28

SaoPaulo 50 85 - - - - 84 9560 - - - - - 80 92 9070 75 - - - 82 71 - -Buenos Aires 50 65 - - - - 67 83b60 46 62 55 57 - 49' 66b,C70 37 56 48 44 54 -

Rio de Janeiro 50 76 - - - 77 88 -60 - - - - 75 86 8270 56 - - 79 76 - -Recife 50 68 - - - 59 86 -60 - - - - 63 85 8670 63 - - 82 63 - -Belo Horizonte 50 77 - - - 56 91 -60 - - - - 47 92 8870 81 - - 90 53 - -Guadalajara 50 87 95 94 91 -- -60 88 96 94 90g - 92' -70 84 91 89 86 84 87

Monterrey 50 91 94 95 92a - - - -60 86 89 89 850 - - -70 66 78 76 71 70

'Service and blue collar combined.bCommerce and service combined.CFor 1964.dFor 1956.'For 1965.Source: Various population and economic censuses; see the Appendix.

economically active population in Mexico City and Buenos Aires is sirmiarto that of North American cities and is apparently the major reason whythe density functions of these two cities are so similar to North Americancities.

CONCLUDING OBSERVATIONSLatin American cities are experiencing extremely high rates of popula-

tion growth at low levels of income and across all city size categories.Small cities seem to be growing more rapidly than large cities in Latin

Page 14: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 269

America, and the growth rate of urbanization in Latin A.nerica is begin-ning to moderate. While the level of urbanization is roughly 30 yearsbehind that of North America, average per capita product in LatinAmerica approximates that of the United States 120 years ago. High citypopulation growth rates also occur in North America, but not typically inthe largest cities.

A general similatity exists between the spatial structure of large LatinAmerican cities and large North American cities that experienced theirpeak periods of growth in the late nineteenth or early twentieth centuries.However, any similarity between large Latin American and large, oldNorth American cities does not imply that the welfare levels of theirresidents are similar. The spatial structure of smaller Latin American citiesdiffers from that of similar sized, rapidly growing North American citiesby being much more centralized.

There is ample evidence of suburbanization in nearly all of the LatinAmerican cities with available data, as measured by a decrease in thecentral city's share of population and employment, as well as by a decreasein the population density gradient. Various indices of socio-economicstatus suggest that higher income households are somewhat concentratedin central cities, although the center's share has fallen over time.

Although this paper has not identified all of the determinants of urbanspatial structure in Latin America, several contributing factors suggestthemselves. Low incomes are likely to keep housing consumption at lowlevels and to contribute to high densities. Since intra-city travel in LatinAmerican cities is primarily by transit, especially bus, high timne and moneytravel costs are also likely to encourage high densities. The great dif-ferences in density function parameters between new North Americancities and smaller Latin American cities is likely due to the North Ameri-can dependence on autos, but the causal significance of this and otherfactors remains to be tested. Since the highest density cities consideredhave varying population growth rates, the data presented here suggest thathigh growth rates by themselves are not strong determinants of densitypatterns.

APPENDIX

A. Data Sources

Argentina

Population: IV Censo General de la Nacion 1947, Tomo I and II; CensoNacional de Poblacion 1960, Tomo II; Censo Nacional de Vivienda 1960,Tomo III; Censo Nacional de Poblacion, Faniilias, y Vivienda 1970; Vols.Resultados Obtenidos por Muestra; Resultados Provisionales.

Page 15: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

270 INGRAM AND CARROLL

Economy: IV Censo Industrial 1947: Tomo II, Censo Industrial yComerical. Censo Nacional Economico 1964: Vols. Industria Manufac-turera, Comercio, Prestacion del Servicios.

Brazil

Population: Censos Demograficos 1950; Censos Demograficos 1960;Censos Demograficos 1970; Censos Domicilios 1970.

Economy: Censos Industriais 1950; Censos Comercial e dos Servicos1950; Censos Industriais 1960; Censos Comercial e dos Servicos 1960;Censo Comercial 1970; Censo dos Servicos 1970.

Chile

Population: XII Censo Nacional de Poblacion y Vivienda 1952; XIIICenso Demografico 1960, including vol. "Entidades de Poblacion"; CensoNacional de Poblacion y Vivienda 1970, including vol. "Entidades dePoblacion;" Vols. 6 and 7, Santiago.

Economy: IV Censo Nacional de Manufacturas 1968, Tomo III.

Colombia

Population: Censo de Poblacion 1951; Censo de Edificios y Viviendas,1951; Decimotercero Censo Nacional de Poblacion, 1964; II Censo Na-cional de Edificios y Viviendas, 1964; Anuario Municipal de Estadistica,Bogota, 1952; Anuario Estadistico, Distrito Especial de Bogota, 1964;Anuario Estaditico de Bogota, D.E. 1972; Anuario Estadistico de Cali,various years.

Economy: Bogota Urban Development Study Phase II, "EmploymentLocation and Decentralization Technical Appendix," Bogota; September1973.

Mexico

Population: Septimo Censo General de la Poblacion 1950; Vols. DistritoFederal, Estado Nuevo Leon, Estado Jalisco, Estado de Mexico. OctavoCenso General de Poblacion 1960; Vols. Distrito Federal, Estado NuevoLeon, Estado Jalisco, Estado de Mexico. Noveno Censo General dePoblacion 1970; Resumen Nacional Direccion General de Estadistica yDireccion General de Programacion y Estudios Economicos, "EncuestaNacional de Hogares 1976, Area Metropolitana de la Ciudad de Mexico."

Peru

Population: Censo Nacional 1940; Censo Nacional de Poblacion, includ-ing vol. "Centros Pobladas"; Primer Censo Nacional de Vivienda 1961;VII Censo Nacional de Poblacion 1972; II Censo Nacional de Vivienda1972; Anuario Estadistico de Peru 1970-71.

Page 16: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 271

Economy: Primer Censo Economico 1963, Censo de Manufacturas yCensos de Comercio y Servicios.

Venezuela

Population: Octavo Censo General de Poblacion 1950; Noveno CensoGeneral de Poblacion 1961; Vol. 6, Metro Caracas; X Censo General dePoblacion y Vivienda 1971; Nomenclados de Centros Poblados 1971.

Economy: Tercer Censo Economico 1963, Vol. Comercio, Manufactura.

B. Jurisdictiorns Comprising Centers and Peripheries(All demarcations are official; year of demarcation in parentheses)

Belo Horizonte (1970)

Center: Municipio Belo Horizonte.Periphery: Municipios Betim, Caete, Contagem, Ibirit6, Lagoa Santa,

Nova Lima, Pedro Leopoldo, Rapcsos, Ribenao dos Neves, Rio Acima,Sabara, Santa Lucia, Vespasiano.

Buenos Aires (1970)

Center: Jurisdiction within "Gran Buenos Aires" called Capital Federal.Periphery: Partidos of Gran Buenos Aires: Alte Brown, Avellaneda, E.

Echeverria, F. Varela, Gral. San Martin, Gral. Sarmiento, Isla SanFernando, Isla Tigre, La Matanza, Lanus, Lomas de Zamora, Merlo,Moreno, Moron, Quilmes, San Fernando, San Isidro, Tigre, Tres deFebrero, V. Lopez.

Caracas (1971)

Center: Departamento Libertador of Distrito Federal.Periphery: Parroquia Carayaca in Departamento Vargas of Distrito

Federal; plus Distrito Sucre and Municipios San Antonio, Carrizal, andCecilio Acosta in Distrito Guaycapuro of State of Miranda.

Guadalajara (1970)

Center: Municipio Guadalajara in State of Jalisco.Periphery: Municipios Tlaquepaque and Zapopan in State of Jalisco.

Lima (1972)

Center: Distritios Barranco, del Cercado, Chorillos, La Victoria, Lince,Magdalena del Mar, Miraflores, Rimac, Pueblo Libre, San Isidro, SanMiguel, and Santiago del Surco in Province of Lima.

Periphery: Entire remainder of Province of Lima plus entire Province ofCallao.

Page 17: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

272 INGRAM AND CARROLL

Mexico City (1976)

Center: Jurisdiction within Distrito Federal called Ciudad de Mexico.Periphery: Entire remainder of Distrito Federal plus Municipios Atiza-

pan, Chimalhuacan, Coacalco, Cuautitlan, Ecatepec, Huixquilucan, LaPaz, Naucalpan, Netzahualcoyotl, Tlalnepantla, and Tultitlan in State ofMexico.

Monterrey (1970)

Center: Municipio Monterrey in State of Nuevo Leon.Periphery: Municipios Garza Garcia, Guadalupe, Santa Catalina, and

San Nicolas de los Garza in State of Nuevo Leon.

Recife (1970)

Center: Municipio Recife.Periphery: Municipios Cabo, Igarassu, Itamaraca, Jaboatao, Moreno,

Olinda, Paulista, and Sao Lourenso da Mata.

Rio de Janeiro (1970)

Center: City of Rio de Janeiro, equivalent to Distrito Federal or State ofGuanabava.

Periphery: Municipios Duque de Caxias, Itaborai, Itaguai, Mage,Mangaratiba, Marica, Nil6polis, Niteroi, Nova Iguacu, Paracambi, Petr6p-olis, Sao Gongalo, and Saio JMao de Meriti.

Santiago (1970)

Center: Comunas La Reina, Providencia, Quinta Normal, San Miguel,and Santiago in Department of Santiago.

Periphery: Comunas Barrancas, Conchali, La Cisterna, La Granja, LaFlorida, Las Condes, Maipu, Nufioa, Puente Alto, Quilicura, Renca, andSan Bernardo in Departments of Aguirre Cerda, Puente Alto, and Santiago.

Sao Paulo (1970)

Center: Municipio Sao Paulo.Periphery: Municipios Aruja, Barueri, Biritiba-Mirim, Caieras, Cajamar,

Carapicuiba, Cotia, Diadema, Embu, Embu-Guacu, Ferraz de Vasconce-los, Franciso Morato, Franco da Rocha, Guararema, Guarulhos, Itape-cerica da Serra, Itapeui, Itaquaque Cetuba, Jandira, Juquitiba, Mairipora,Mava, Mogi das Cruces, Osasco, Pirapora du Bom Jesus, Poa, RiberaoPires, Rio Grande da Serra, Sales6polis, Santa Isabel, Santana de Parnaiba,Santo Andre, Sao Bernardo do Campo, Sao Caetano do Sul, Suzano,Taboao da Serra.

Page 18: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

LATIN AMERICAN CITIES 273

REFERENCES1. P. Amato, "An Analysis of the Changing Patterns of Elite Residential Areas in Bogota,

Colombia," Ph.D. Dissertation, Cornell University, (June 1968).2. J. Heilbrun, "Urban Economics and Public Policy," St. Martin's, New York, 1974.3. S. Kuznets, "Modern Economic Growth," Yale Univ. Press, New Haven, Conn. 1966.4. K. S. Lee, Intra-urban location of manufacturing employment in Colombia (paper

presented at AEA meetings, Chicago, August, 1978), J. Urban Econ., 9, 222-241(1981).

5. A. Portes and J. Walton, "Urban Latin America: The Political Condition from Aboveand Below," Univ. of Texas Press, Austin, 1976.

6. U.N. Demographic Yearbook 1973, N.Y. Department of Economic and Social Affairs,Statistical Office, 1974.

7. U.N. World Housing Survey, 1974, N.Y. Department of Economic and Social Affairs,Statistical Office, 1974.

8. U.S. Sunmary, Final Report PC(l)-AI, U.S. Census Bureau, 1970 Census of Population.9. L. J. White, How good are two point estimates of urban density gradients and central

densities?, J. Ujrban Econ., 4, No. 3 (July, 1977).10. J. W. Wilkie, Ed., "Statistical Abstract of Latin America," UCLA Latin American Center

Publications, Los Angeles, 1976.11. "World Bank Atlas," World Bank, Washington, D.C., 1977.

Page 19: Gregory K. Ingram and Alan Carroll The Spatial Structure ...documents.worldbank.org/curated/en/... · Recife 1970 1,061 1,793 Rio de Janeiro 1970 4,252 7,082 Salvadora 1970 998 -Sao

World BankHeadquarters:1818 H Street, N.W. UWashington, D.C. 20433, U.S.A.

Telephone: (202) 477-1234Telex: RCA 248423 WORLDBK

WUI 64145 WORLDBANKCable address: INTBAFRAD

WASHINGTONDC

European Office:66, avenue d'1ena75116 Paris, France

Telephone: 723.54.21Telex: 842-620628

Tokyo Office:Kokusai Building1-1, Marunouchi 3-chomeChiyoda-ku, Tokyo 100, Japan

Telephone: 214-5001Telex: 781-26838

The full range of World Bank publications, both free and for sale, isdescribed in the World Bank Catalog of Publications, and of the continuingresearch program of the World Bank, in World Bank Research Program:Abstracts of Current Studies. The most recent edition of each is availablewithout charge from:

PUBLICATIONS UNITTHE WORLD BANK1818 H STREET, N.W.IN'ASHINGTON, D.C. 20433U.S.A. ISSN 0253-2131