National Academies Press: OpenBook
« Previous: 4. Urban Population Dynamics: Models, Measures, and Forecasts
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 155
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 156
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 157
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 158
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 159
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 160
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 161
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 162
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 163
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 164
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 165
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 166
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 167
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 168
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 169
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 170
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 171
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 172
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 173
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 174
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 175
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 176
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 177
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 178
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 179
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 180
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 181
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 182
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 183
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 184
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 185
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 186
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 187
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 188
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 189
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 190
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 191
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 192
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 193
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 194
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 195
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 196
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 197
Suggested Citation:"5. Diversity and Inequality." National Research Council. 2003. Cities Transformed: Demographic Change and Its Implications in the Developing World. Washington, DC: The National Academies Press. doi: 10.17226/10693.
×
Page 198

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Diversity and Inequality Diversity is among the defining features of city life. Seen from one perspective, diversity is a manifestation of the concept of the city as lottery, a social arena where risks and rewards are on display. It is evidence of mobility and possibility. But from another perspective, diversity is experienced as inequity, a reminder of immobility and possibilities frustrated. This chapter explores several of the di- mensions of urban socioeconomic diversity and inequality. Particular attention is paid to the circumstances of the urban poor. On close inspection, the housing and living conditions of the urban poor prove to be more varied than might have been thought, and it is not easy to reduce in- dicators of urban housing quality to estimates of the population living in slums. Even the term "slum" tends to be avoided in careful research on urban housing, although it can be employed as a convenient shorthand. The tone adopted in sci- entific studies resembles that of the United Nations Centre for Human Settlements (UNCHS) (1996:205~: "How simplistic and often inaccurate it is to assume that mostlow-incomegroupsElive~in'slums'or'slumsandsquattersettlements'." On the question of changes in the percentages of urban dwellers in slums, UNCHS (1996) does not find sufficient evidence to draw conclusions, although it does concede that the total numbers living in such settlements are large and probably have been rising. Not until 1990, when UNCHS began its Housing Indicators Programme, was a sustained effort made on a large scale to bring order and coher- ence to empirical measures of urban housing, enabling cross-country comparisons in a few key dimensions. This study (described in UNCHS [1996:1961; see also Malpezzi [19991) laid the groundwork for a more nuanced understanding of urban living conditions and drew attention to the variety of housing markets in which the urban poor participate. We approach urban poverty with the understanding that it has many facets that need to be considered. Housing is of interest, as are levels of income and con- sumption. Other aspects also warrant attention. When poverty is conceptualized 155

156 CITIES TRANSFORMED as having multiple dimensions, the focus of analyses extends from individuals and households to groups, encompassing measures of the economic and politi- cal power held by groups and the qualities and capacities of local and national governments. As will be seen, a recognition of poverty's multiple dimensions suggests broader roles for programs and policies than could be deduced from a consideration of income and consumption alone. As we seek to understand how diversity and inequality manifest themselves spatially, we initially examine city maps of socioeconomic indicators. It is dis- appointing that the analysis cannot then proceed systematically to the level of city neighborhoods and districts. As mentioned in Chapter 2, the conceptual and empirical tools that have been applied to the cities of rich countries await further application to the cities of poor countries. We are confident that the neighborhood data exist, though at present they are inaccessible. To be sure, the literature on poor countries presents many fine-grained portraits of selected city neighborhoods. The micro studies give vivid and compelling ac- counts of the absolute poverty and serious deprivation that can be found in some city neighborhoods, but such studies provide an uncertain basis for wider gener- alization. Household surveys fielded at the national level can offer such a gen- eralized overview in what follows, we rely heavily on the Demographic and Health Surveys (DHS) but they usually lack the sample sizes needed to detect socioeconomic differentiation at the neighborhood level or even at the level of cities. However, household surveys will generally support aspatial analyses of urban populations taken as a whole, and will often allow the urban population to be subdivided by city size class or separated into other broad categories. In the first section of the chapter, then, we can only briefly examine the spatial aspects of inequality and diversity within cities. The remainder of the chapter is necessarily less spatial in orientation. Conditions among urban populations are compared with those among rural populations; within the urban populations, we explore how socioeconomic conditions vary across cities of differing population size and by measures of relative poverty. Applying this aspatial approach, the chapter's second section examines school- ing. For adults, schooling is an important determinant of socioeconomic well- being and demographic behavior. We show the distributions of adult education in urban and rural areas and pay special attention to the educational diversity that marks urban areas. The section following considers the distinctive features of urban poverty, describing recent research that leads to a multidimensional per- spective on disadvantage. We then explore one of these dimensions in detail, examining how access to basic public services differs between rural and urban populations, and differs within urban populations along the lines of city size and relative poverty. Next, we critically assess current estimates of urban poverty in developing countries, arguing that national and international statistics are likely to have understated its prevalence. This assessment is followed by a brief discussion of the risk and vulnerability faced by the urban poor.

DIVERSITY AND INEQUALITY 157 The chapter ends with a consideration of children's lives. How well do chil- dren fare in the cities of poor countries? If adult schooling represents the socio- economic diversity of the current generation, children's schooling represents the potential for differences to emerge in the next generation. Schooling is the result of social investments of several kinds those made by parents, communities, and the state. As will be seen, these investments give urban children a decided advan- tage in terms of school enrollment, on average, as compared with their rural coun- terparts. However, the urban averages conceal substantial variation. When other aspects of children's lives in the city are examined we review what is known about street children a fuller picture emerges of diversity and inequality in ur- ban children's lives. A SPATIAL PERSPECTIVE In the cities of Africa, Asia, and Latin America, the spatial expressions of poverty and affluence are often as vivid as they are in Chicago, Los Angeles, and New York. Figure 5-1 shows the concentration of the affluent in Santiago, Chile, an urban area with some 4.7 million residents within the bounds of the city proper. Here the elites are clustered in the northeastern sections of the city. (Wealthier areas are depicted in darker shading.) The ways in which rich and poor are spa- tially arrayed vary greatly across cities (compare Figure 5-1 with the complex pat- tern seen in Mexico City in Figure 2-1 in Chapter 2), but in one form or another, many cities exhibit spatial evidence of exclusion and segregation. As we have mentioned, however, not all aspects of exclusion express themselves spatially, and researchers are beginning to explore the nonspatial forms. The complexities can be appreciated in a recent study of Buenos Aires (Torres, 2001) that examines the increase in this city's "gated" communities the protected enclaves of upper-income groups. Figure 5-2 depicts the changes seen in Buenos Aires during the l990s, when many such enclaves sprang up. (The white squares represent the locations of gated communities in 1990-1991, and the black dots represent the new communities of this type that emerged by 2001.) As can be seen in Figure 5-3, a number of these enclaves lie adjacent to the neigh- borhoods of the poor. (In this figure, the gated communities are shown as white dots, and the darker areas indicate where the poor live.) In locating near the poor, the rich gain easy access to a pool of cheap labor, persons who can be hired as se- curity guards, gardeners, and maids. The rich separate themselves from the poor not so much by putting them at a distance (though there is some of that), but by fortifying the borders of their enclaves and restricting the terms on which the poor are allowed to come into contact with them. This is a strategy of "proximity and high walls" (a phrase taken from Caldeira, 1996~. If transport costs allow some of the poor to make longer journeys to work, the rich are permitted the additional luxury of distance. As can be seen in Figure 5-3, most of the gated communities

158 CITIES TRANSFORMED · . :: ~ (pi . . . ~ .f,:.:fi x.),,,1i , . ,. S,.,:,.:~e,2. .2.,.- .,,,',,,j':., $f,ff.if.ff6-i: ,.(',2 ,i<,2,:\ ~s,.",$"", -, ,'.2.>2,/~./2.2.~/2, ~,<~,.,2,.,.2.',nj2~ )' i ''' ''a /2'~ ('2 ' ~ ~"1': .~ v . . ;.," .d .. ' '\ ~ ; 'K'i: ~,,,''' ' ' ; :'~ i, ~ ~ it r C / ~ :' "'': <:,': :::.', :'-:: '::~i :i:: .'-:X: ~ .'~ 'i~' ,'-. '-'.: " -2 -'.,.i'.,i~ :-':>2 $.gff;< ~.,': -: :::::::,.,::::, :'-':..if.~',f:.'i. ~ ~$ : r ;i ,'i""2 ""2 ', "' Percentage of the group in each zone ~ .... ti ,i,> Upto2% [:; ~>' 2-5% .~.~ 5-1 0% 10-21 % 42-1 00% . no information ........ ~The elite represents 1a6% of the heads of households ~ :: ::~::\ ::i . $ ~ :2:2: ': ':2 '':''::':2::''2':,i :;~f ~f"<''.:.: '' ., , i j.2 ' . '': : ,, ' ,,-:$ .,.'$ .~ . ~ :.', .-:< ,~ ;; ~ , ~ ,~- .... .~. ~ ~,,,,,,,, ,, ,i,,,i,,, .~.; . ~fff~.,h . ~ j' ::::,:, >: $ .:::: :' ': .: ": : ' . ~': :-': . ~ .'.~": .'N ~ ., i . - j: ..... , $ ::: :: ,:,, ; 1 ~ j; ., i' .':':'.':'.'. '.':'. .'."'" ' '''. .\ i i.'P .: , :, . , .. ,f T~', ,2 '. ;, :,: .'.: :':' :' i '<~' '' ''~';21.~ '2 i ~ 2. i., s., i,~ 2/,~ ~,:~ '', ":2 :: :': :.:.: ': :.:.~:.,~, 5!; . ~,.ji~.,. .' ..:. . .'.' s" " ; 2 i 2>,~, 2 2 i,, " ~2."'2"'.2.2.,"~.~l . ff5~ ": " ": ~ ''5 ~ :': ''-C i ''..~''", "'''''''<~. :'"..:-'" -'": 3 :, :~::.::::,: ::j,:::::: :.:: .:: :.:.::: ::. : s .',.::.:.:.:.,).::::: :::::::.':::::: ::': -:: :::::, :::::$::::,:,:,::::::.~.:::::.:.:> .'':: '.': .':s::::~::: :Y f..2:'2...-,;," 2, 2''';.':.'.,.5 2 ' .'', O2'.-.''2 ''';' ''' 5': '' '' '' ' ::'>s''.f' ~:"':" ':' "' ' . 'i~ ~, ,,, ,,,,, sic.. ~i ~,,,,,,, $, ',',, >, - - . ,:3i ,,:,: ,j,~ , i~.- .-, -~ ,,,,.,,,,-., .:::,,,,,:i.:,{.'j',f',ff~:::;:,::: . :5:'~,::: 5:::,:~'.,j2:.:"".::',.i::',::::::,:: :-;' ':: .: :' >-i'-f S~if-~f ifffff~-~- ~ ''''' i:,.,2,.,.i$"."~ .,,',,,,.,i.'5,~"."." $. '"'"''""'"'-.''''''"" $'""''"'''" "'":'"'''" ""'''"' ' ' '' - ''""'-" ''~'' $,,,, i X2'22' .2''''i,',,,:.2.2,'.". "'~.':.2..."..:2." ;,',, . " '': ':: ""` '~'' '''"" " i.'.2} f $. . ."i2."2 .2 S '~ "> . ~ .~,.,2 ~'~,., 's' ,",' :' j i: -: : :' $ .:.::' .. , ,, ., ~ . '~'. .' :. ' '1.,. .'.' , .. FIGURE 5-1 Spatial concentration of the elites of Santiago, Chile. Wealthier neighborhoods are depicted in darker shading. SOURCE: Sabatini and Arenas (2000~. Of Buenos Aires are situated far from poor neighborhoods but close to the main transport routes. A recent analysis of poverty and inequality in Argentina demonstrates that there are great differences in the quality of basic infrastructure among the neigh- borhoods of greater Buenos Aires. Inequality as measured by differences in pro- vision of basic services (water, sanitation, and housing) is three times greater than inequality measured in terms of education and health status. These striking intraurban differentials illustrate the insights that can be derived from a focus on neighborhoods (Cohen, 2002~. Another spatial dimension to be considered is the situation of small relative to large cities. As noted earlier, there are a number of reasons to believe that economic conditions in small cities are often worse than those in large cities,

DIVERSITY AND INEQUALITY ... .. . . . · is. :.~..~.. . ~ - The Great Buenos-Aires :~/~: Agglomeration in 1991 i .- - ~. ~ ~ . ~ : .. a> a- ~ ~~ ~) :~: ': it: Y' K -,, - :: :'N : <~ <O :~ ~'1~.~,= ~ > / ,. :'~: }./ :~::: ·:: .,::-- ~Y' 159 FIGURE 5-2 Increase in the number of gated communities in Buenos Aires in the l990s. SOURCE: Torres (2001~. . . Q:: -. ., ~ T . ..v ~ ~ ~ ~ ~ ,. . .. ~~ ~ a: ~~ ~ ~ If: High : ^ ~ ~` ~ ~ A.: ' ~ ~ ~ . ~ Medium '[ (.~ ~ .. / Low FIGURE 5-3 A number of gated communities lie adjacent to poor neighborhoods in Buenos Aires. SOURCE: Torres (2001~.

160 60 - 40 - c~ _, (in o Q 20 - o O- CITIES TRANSFORMED · Abidjan C1 Other Cities . ~ - T 1 1984 1987 1990 1993 1996 1999 Year FIGURE 5-4 Poverty in Abidjan compared with that in the secondary cities of Cote d'Ivoire, 1985-1995. SOURCE: Grimm, Guenard, and Mesplee-Somps (2002~. at least on average. Some evidence to this effect is available for Cote d'Ivoire. For the period 1985-1998, Figure 5-4 shows the levels and trends in poverty in Abidjan relative to those in the country's secondary cities. (Point estimates are given with their confidence bands.) As can be seen, the proportion of residents estimated to be living on less than US$2 per day has consistently been lower in Abidjan than in the smaller cities. Through the mid-199Os, macroeconomic deterioration drove up the poverty rates of Abidjan along with those of secondary cities, but never erased Abidjan's advantage. We report evidence of such smaller- city disadvantages throughout this chapter. HUMAN CAPITAL: SCHOOLING As a principal measure of human capital, adult educational attainment is of funda- mental importance to incomes and socioeconomic standing. Figure 5-5 depicts the distributions of schooling in urban and rural areas, and Table 5-1 provides further region-specific detail.) It is not surprising to see that the average level of schooling iThe figure and table present summaries of estimates from 61 surveys fielded by the DHS pro- gram in 44 developing countries between 1985 and 1999 (see Table C-1 in Appendix C for a list of

DIVERSI~AND INEQUALITY 40 - 30 - 20- 10— O- 161 Rural Urban l...................... l...................... l...................... l...................... l...................... l...................... l...................... l....................... l....................... l....................... l....................... l....................... None ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... ,...................... - ............ ....................... ....................... ....................... ....................... ....................... ....................... ....................... - ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... ....................... 1"""""""""""" 1'''''''''''2 1'''''''''''2 1'''''''''''2 1'''''''''''2 1'''''''''''2 1'''''''''''2 Primary Secondary Incomplete Primary Incomplete Secondary Higher ........... ....................... l ....................... l ....................... l ....................... l 1 l 1 _ Level of Completed Schooling FIGURE 5-5 Completed schooling for adults, rural and urban areas. See footnote 1. is higher in urban than in rural areas, with higher percentages of urban adults hav- ing secondary schooling or more, and lower percentages having no schooling. But to focus on the extremes of these educational distributions is to overlook another important feature: the urban distribution is less concentrated than the rural, with appreciable percentages of urban adults appearing in each of the educational cat- egories. Here is evidence of an urban advantage higher levels of schooling on average coupled with evidence of greater urban diversity. Urban/rural differences in education stem from many causes. Among these, we would single out migration because it is known to be selective of those with higher levels of schooling. A portion of the urban/rural difference might well be due to the outmigration of better-educated, formerly rural residents.2 As will be seen later in this chapter, however, urban children have strikingly higher levels of school enrollment than their rural counterparts. Hence, the urban/rural differences shown above must also reflect long-standing differences in educational invest- ments between the countryside and the city. countries). The estimates refer to both women and men. The numbers shown are based on survey- specific estimates, which are weighted so that an estimate from a country with two surveys receives a weight that is one-half that of a country with only one survey. Estimates from countries with three or four surveys are similarly downweighted. This is our practice throughout the report. Note, however, that the number of surveys and countries varies depending on what is being analyzed. Educational attainment is measured in the DHS household modules, but before the recent rounds of the DHS pro- gram, not all surveys included education questions for all household members. Although fertility data are available for 90 surveys, adult educational data are available for far fewer surveys. 2 Unfortunately, we cannot isolate the migration component. DHS surveys generally do not iden- tify the former residences of adults in the household, other than the woman selected for the main interview.

162 TABLE 5- 1 Adult Educational Attainment, Rural and Urban Areas CITIES TRANSFORMED DHS Percentage of Adults with Surveys Some Complete Some Complete in Regiona None Primary Primary Secondary Secondary Higher North Rural 59.5 14.0 4.9 11.5 7.7 2.4 Africa Urban 28.0 13.6 9.9 22.9 15.0 10.7 Sub- Rural 51.5 27.6 9.7 9.5 1.2 0.5 Saharan Urban 27.7 22.5 13.3 27.2 5.9 3.8 Africa Southeast Rural 11.0 26.7 25.6 20.5 12.0 5.7 Asia Urban 3.9 11.9 17.2 27.0 26.6 17.8 South, Rural 33.2 8.4 9.9 17.2 26.3 5.0 Central, Urban 17.3 6.6 8.5 22.8 27.0 17.8 West Asia Latin Rural 23.8 41.3 15.4 13.8 4.0 1.7 America Urban 7.9 21.2 13.9 28.5 15.1 13.3 TOTAL Rural 41.0 26.3 11.4 12.4 7.1 2.0 Urban 20.7 18.5 12.6 26.4 13.0 9.4 a Number of countries surveyed: North Africa, 2; sub-Saharan Africa, 23; Southeast Asia, 2; South, Central, and West Asia, 8; Latin America, 9; all regions, 44. Why does educational diversity matter? In Chapter 2 we referred to the eco- nomic and social theories that draw out its implications. The collective social- ization theory of Coleman (1988) and Wilson (1987) posits that educated adults may wield beneficial influence in poor neighborhoods; the economic theories of Jacobs (1969), Rauch (1993), and Moretti (2000) suggest that economic interac- tions between the better and less educated may generate positive externalities in city labor markets and firms. What is central to these theories, but missing from Figure 5-5, is the aspect of interaction. It is one thing to note that cities contain a diversity of educational levels, but quite another to say that adults with differ- ent levels of education commonly interact, whether in their social or economic relations. Spatial segregation and exclusion no doubt inhibit interaction, but indi- viduals may interact in their workplaces and other settings across a metropolitan region. (Measuring such interactions is admittedly difficult.) Still, the mere pres- ence of educational diversity in cities raises the possibility of beneficial spillovers in city neighborhoods and labor markets. Comparing educational attainment across city size classes (see Figure 5-6), we find that adults living in large cities (especially in those of 1 million or more popu- lation) tend to have acquired more schooling than those in small cities and towns. This is easily seen in the percentages having secondary and higher schooling.

DIVERSI~AND INEQUALITY 40 - 30 - 20- 10 - o 163 Under 100,000 100,000-500,000 ~ 500,000-1 million 1~ 1 1-5 million I I Over 5 million ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ ........ . 1 ..... ·:-:-:-: _ .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .... .:.:.:.: :::::::: . 11 1 .......... . L .......... 1"""" ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ ..... ........ 1 ma.. _ ........ ........ i ........ ........ . ........ ........ . ........ ........ . ........ ........ . ........ ......... ........ ..... ........ ........ . T _ . . ~ , - ~ ............ ........ 1 ............ ........ 1 ....... ........ 1 ......... ........ 1 ......... ........ 1 ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ......... ,~ . .,.,.,.,. . ~ _ _ None Primary Secondary Incomplete Primary Incomplete Secondary Higher Level of Completed Schooling FIGURE 5-6 Completed schooling for adults, rural and urban areas by city size. Again, however, there is evidence of substantial diversity in schooling levels, even in the larger cities. For the 16 countries having two or more DHS surveys, we are able to examine changes over time in the educational attainment of adults.3 Over the period be- tween surveys about 5 years on average there was a decline in the percentage of adults having no schooling or at most incomplete primary schooling. In 11 of the 16 countries, the percentage of adults with no schooling fell in both rural and urban areas, with the rural decline generally being larger than the urban. In 7 countries, the percentage of adults with incomplete primary education also de- clined in both rural and urban areas, but here the decline tended to be larger in urban areas. Examining the other end of the educational distribution, we find that the percentage of adults with higher schooling increased in 9 countries, with the urban increase again being greater than the rural on average, and much the same was true at the level of completed secondary schooling. In summary, where the trends in adult schooling can be examined, the data show that the relative situation of rural and urban populations remains much as it is portrayed in Figure 5-5: an urban advantage on average, together with greater urban diversity. 3 The countries are Bangladesh, Bolivia, Cameroon, Colombia, the Dominican Republic, Egypt, Ghana, Guatemala, Indonesia (where three time points can be compared), Kenya, Madagascar, Niger, Peru, the Philippines, Tanzania, and Zambia.

164 CITIES TRANSFORMED URBAN WELL-BEING: CONCEPTS AND MEASURES The path from educational attainment to living standards and well-being is com- plex and indirect, involving, among other things, many mediating institutions and policies. Although education data are helpful in understanding socioeconomic di- versity, they inevitably tell something less than the full story. This section takes up the problem of how to define urban living standards, treating the issues in broad, conceptual terms, but commenting in passing on measurement concerns to be discussed at length later. A glance at the literature shows that the conceptualization of urban poverty by researchers is increasingly diverging from the methods used by governments to measure it. Recent research has taken to describing poverty in multidimensional terms, even as the official poverty measures continue to be expressed in highly simplified, unidimensional terms. Poverty measures typically take the form of income-based or consumption-based poverty lines, with the thresholds defined mainly in terms of food consumption. Such restricted definitions testify to the difficulties involved in measuring poverty's multiple dimensions, but they can also distort understanding of its causes and may unnecessarily narrow the scope for intervention on the part of local and international agencies. Box 5.1 lists some of the key dimensions of urban poverty identified in the lit- erature.4 Although we emphasize its implications for urban poverty, this research owes a great deal to studies of rural poverty (see especially Chambers, 1983, 1995; Beck, 1994~. The roles of income and assets are central and well recognized in the new research, but other dimensions of poverty are also singled out for considera- tion. A leading example is a lack of voice within political systems that keeps the concerns of the poor from being heard; another example is the poor's inadequate security and lack of protection from violence, theft, and fraud. It is the multiplicity of deprivations and the connections among them that char- acterize the circumstances of many urban poor. As Navarro (2001) notes, a low- income family with only one income earner living in an illegal settlement on a flood plain cannot really be said to have three distinct problems (low income, 4This box draws on a typology developed by Baulch (1996) to describe rural poverty, modified to reflect the dimensions of poverty that are common in urban areas. It is meant only to illustrate dif- ferent aspects of urban poverty; others could certainly be added. The multiple aspects of poverty are described by Moser (1993, 1996, 1998), Moser, Herbert, and Makonnen (1993), Amis (1995), Wrat- ten (1995), Rakodi (1995), Satterthwaite (1996b), and Anzorena, Bolnick, Boonyabancha, Cabannes, Hardoy, Hasan, Levy, Mitlin, et al. (1998), among others. Perhaps the best-known of the multidimensional indices is the Human Poverty Index (HPI), devel- oped by the United Nations Development Programme and introduced in the 1997 Human Development Report (United Nations Development Program, 1997b). This index, generally applied to national pop- ulations but sometimes to population subgroups, takes three factors of poverty into account: vulnera- bility to death before age 40; the illiteracy rate, and the standard of living, this last being measured on the basis of access to health services and safe water and the percentage of children under age 5 who are malnourished.

DIVERSITY AND INEQUALITY 165 BOX 5.1 The Multiple Dimensions of Urban Poverty Income and consumption Poverty is conventionally defined in terms of incomes that are inadequate to permit the purchase of necessities, including food and safe water in sufficient quantity. Because incomes can be transitory and are difficult to measure, levels of consumption are often used as indicators of the longer-term component of income. Assets The nature of household assets also bears on the longer-term aspects of poverty and the degree to which households are shielded from risk. A household's assets may be inadequate, unstable, difficult to convert to monetized form, or subject to economic, weather-related, or political risks; access to credit may be restricted or loans available only at high rates of interest. For many of the urban poor, significant proportions of income go to repay debts (see, e.g., Amis and Kumar, 2000~. Time costs Conventional poverty lines do not directly incorporate the time needed for low-income households to travel to work or undertake other essential tasks. Such households often try to reduce their monetary expenditures on travel by walking or enduring long commutes (Moser, 19964. Time costs also affect the net value of some goods and services. Shelter Shelter may be of poor quality, overcrowded, or insecure. Public infrastructure Inadequate provision of public infrastructure (piped water, sanita- tion, drainage, roads, and the like) can increase health burdens, as well as the time and money costs of employment. Other basic services There can be inadequate provision of such basic services as health care, emergency services, law enforcement, schools, day care, vocational training, and communication. Safety nets There may be no social safety net to secure consumption, access to shelter, and health care when incomes fall. Protection of rights The rights of poor groups may be inadequately protected, there being a lack of effective laws and regulations regarding civil and political rights, occupa- tional health and safety, pollution control, environmental health, violence and crime, discrimination, and exploitation. Political voice The poor's lack of voice and their powerlessness within political and bureaucratic systems may leave them with little likelihood of receiving entitlements and little prospect that organizing and making demands on the public sector will produce a fair response. The lack of voice also refers to an absence of means to ensure accountability from public, private, and nongovernmental agencies. insecure tenure, and exposure to environmental risk) its difficulties are manifes- tations of a fundamental deprivation. Of course, if the multiple dimensions of poverty were always so closely related, a single poverty measure might suffice to identify poor households. Why, then, is it necessary to elaborate on each of the multiple dimensions?

166 CITIES TRANSFORMED Mitlin and Satterthwaite (2001) develop the rationale. First, a number of case studies show how the deprivations associated with low income can be eased with- out increasing income as such. Several routes have been explored: increasing asset bases, improving basic infrastructure and services, and fostering political changes that give low-income groups a means of negotiating for greater govern- mental support or less harassment (see Patel and Mitlin, 2001; Baumann, Bolnick, and Mitlin, 2001~. Second, governments and nongovernmental organizations (NGOs) may have few direct means of increasing the incomes of the poor, but may be better equipped to address other dimensions of poverty. Recent experience with micro finance in cities (and with funds that support public works and employment) suggests a greater scope for raising urban incomes than many had supposed. In most cities, however, the prospects for effective direct intervention will doubtless remain lim- ited (see Benjamin, 2000; Amis and Kumar, 2000; Etemadi, 2000~. Third, an effective intervention in the nonincome dimensions of poverty some- times has the potential to increase real incomes. Better-quality housing and improved public services can enhance income-earning opportunities for home en- terprises, allowing rooms to be rented out and small enterprises to be maintained. The provision of piped water can reduce a household's water bill, which in turn may permit consumption of more food; see Cairncross (1990) on the mechanisms. Improved infrastructure and services protect health and reduce fatigue (as when water piped into the home replaces a long trek to a standpipe) and in this way may increase net income (as when reduced illness and injury mean less time taken from work and lower medical costs). Fourth, even increased incomes may not enable households to escape poverty. In many cities, the provision of infrastructure and services is so limited and the capacity to expand them so weak, that families with incomes above the poverty line can find it difficult to locate housing with adequate services and protections against risk (Hardoy, Mitlin, and Satterthwaite, 2001~. These families may suffer from most of the deprivations of poverty without being officially classified as poor by the standard income criteria.5 The deprivations identified in Box 5.1 are not problems of the income-poor alone; those with adequate income can also lack political voice and suffer from inadequate protection. Fifth, a nuanced understanding of urban poverty helps policy makers expand the set of agencies charged with roles in poverty reduction. Most municipal agen- cies do not see their principal responsibilities as lying in the area of poverty alle- viation when poverty is defined in terms of income alone. But if the definition is expanded, as in Box 5.1, then public transport authorities, hospitals, occupational health and safety agencies, and water companies may come to understand that the 5See McDade and Adair (2001) for an empirical assessment of the overlap in three dimensions of poverty in Cebu, Philippines levels of consumption, access to infrastructure, and access to health services.

DIVERSITY AND INEQUALITY 167 services for which they are responsible can affect poverty. When poverty is de- scribed in expanded terms, the poor are given a new vocabulary with which they can assert claims to state resources. Links to the state and to external private groups can be critical in bringing resources to poor communities (Anzorena, Bolnick, Boonyabancha, Cabannes, Hardoy, Hasan, Levy, Mitlin, et al., 1998; Mitlin and Satterthwaite, 2001~. In Chapter 2, this point is discussed in connection with the bridging role of local social capital, and illustrated by the experience of the Mumbai-based alliance of S PARC, the National Slum Dwellers Federation, and Mahila Milan (recall Box 2.2~. As that case shows, the quality of the relationship between the poor and external organizations depends on organizational transparency and the degree to which outside organizations are accountable to the poor. A number of case studies demonstrate how much the urban poor can achieve when they have good relation- ships with local and external organizations (Baumann, Bolnick, and Mitlin, 2001; Patel and Mitlin, 2001~. A broad conception of poverty also draws out its implications for women. Because women take responsibility for care of children and the sick, in addi- tion to routine household management, they often bear much of the burden of the time costs described above. Poor women can be exposed to greater health or social risks than poor men (Moser, 1996, 1987, 1993; Beall and Levy, 1994; Lee- SmithandTrujillo,1992;Crewe,1995;Sapir,l990;SongsoreandMcGranahan, 1998~. For instance, women are often responsible for the disposal of their house- holds' human wastes when provision for sanitation is inadequate, and this can elevate the risks to their own health. ACCESS TO PUBLIC SERVICES Most countries gather data on the provision of infrastructure and public services, although few include access to services in their official definitions of poverty. The official reports on service provision make their way into the international databases, where they are duly recorded as national, urban, and rural percent- ages. It is unclear, however, whether these reports are reliable. What counts as "adequate provision" is often left vague in the official accounts. For example, are households that take water from a standpipe that works for 2 hours a day as adequately served as households whose water is piped to the home (Hardoy, Mitlin, and Satterthwaite, 2001~? Even in countries with sophisticated data col- lection abilities, such as Argentina, local authorities often know surprisingly little about the quality of service provision in their jurisdictions (Navarro, 2001~. In an effort to improve the situation, some international agencies are beginning their own programs of data gathering and monitoring. Box 5.2 describes the Cities Data Book project under way at the Asian Development Bank. In what follows, we rely on the data on access to services gathered by the DHS in the course of its household interviews. Although these data are not all

168 CITIES TRANSFORMED BOX 5.2 The Asian Development Bank's Cities Data Book The Cities Data Book project for the Asian and Pacific Region was organized by the Asian Development Bank in 1998. The objectives of the project are: · To build the capacity of local governments to collect and use urban indicators to improve coverage and operational performance. · To develop criteria and methods for the measurement and evaluation of urban service delivery. · To enhance interaction and information exchange among local governments. The project, which involves some 17 local governments from 15 countries, lends assistance to local governments in developing methods for collecting and analyzing data. The core questionnaire covers such social indicators as population, land use, income and employ- ment, poverty, health, and education, as well as sectoral areas of service provision such as shelter, transport, communication, energy, water supply, and sanitation. Data on local gov- ernments cover the structure of revenues and expenditures, measures of satisfaction with services, and indicators of the roles of public corporations and privatized urban services companies. Over the course of the project, quantitative and qualitative analyses will ad- dress city organizations, relationships with higher levels of government, major issues in service provision, and the effectiveness of service delivery. The Cities Data Book also includes theme papers that explore the implications of the local government data for urban development strategies, capacity building for efficient data collection and analysis, best practices in urban monitoring and data collection, and the role of local data in promoting efficient and sustainable urban management. that could be desired, they have the important advantage of being recorded at the household level. There is some variation from survey to survey in how access to services is measured, but the DHS has imposed a coding scheme that keeps this variation within reasonable bounds. By choosing to define broad categories of access see Box 5.3 for the scheme adopted by the panel we have organized the data in a way that should permit valid comparisons across countries and over time, at the cost of omitting some of the country-specific detail. Urban and Rural Access to Services Table 5-2 summarizes the data on services for the rural and urban sectors as a whole. The urban advantage can be seen at a glance. There is an enormous gap between urban and rural areas in the provision of piped drinking water: the percentages in urban areas range from nearly 5 times those of rural areas in the case of sub-Saharan Africa to a ratio of about 2 to 1 in the case of North Africa.6 Access to drinking water through standpipes and other neighborhood sources is less unevenly distributed, with the rural and urban percentages being roughly equal in sub-Saharan Africa and Southeast Asia, but higher in rural areas in the 6The only countries surveyed in North Africa are Egypt and Morocco.

DIVERSITY AND INEQUALITY :: 169 owes a great deal to studies of mral poverty (see especially Chambers, 1983, 1995; Beck, 1994~. The roles of income and assets are central and well recognized in the new research, but other dimensions of poverty are also singled out for considera- tion. A leading example is a lack of voice within political systems that keeps the concerns of the poor from being heard; another example is the poor's inadequate security and lack of protection from violence, theft, and fraud. It is the multiplicity of deprivations and the connections among them that char- acterize the circumstances of many urban poor. As Navarro (2001) notes, a low- income family with only one income earner living in an illegal settlement on a flood plain cannot really be said to have three distinct problems (low income, insecure tenure, and exposure to environmental risk~its difficulties are manifes- tations of a fundamental deprivation. Of course, if the multiple dimensions of poverty were always so closely related, a single poverty measure might suffice to identify poor households. Why, then, is it necessary to elaborate on each of the multiple dimensions? Mitlin and Satterthwaite (2001 ) develop the rationale. First, a number of case studies show how the deprivations associated with low income can be eased with- out increasing income as such. Several routes have been explored: increasing asset bases, improving basic infrastructure and services, and fostering political changes that give low-income groups a means of negotiating for greater govern- mental support or less harassment (see Patel and Mitlin, 2001; Baumann, Bolnick, and Mitlin, 2001~. Second, governments and nongovernmental organizations (NGOs) may have few direct means of increasing the incomes of the poor, but may be better equipped to address other dimensions of poverty. Recent experience with micro finance in cities (and with funds that support public works and employment) suggests a greater scope for raising urban incomes than many had supposed. In most cities, however, the prospects for effective direct intervention will doubtless remain lim- ited (see Benjamin, 2000; Amis and Kumar, 2000; Etemadi, 2000~. Third, an effective intervention in the nonincome dimensions of poverty some- times has the potential to increase real incomes. Better-quality housing and im- proved public services can enhance income-earning opportunities for home enter- prises, allowing rooms to be rented out and small enterprises to be maintained. The provision of piped water can reduce a household's water bill, which in turn may permit consumption of more food; see Cairncross (1990) on the mechanisms. Improved infrastructure and services protect health and reduce fatigue (as when water piped into the home replaces a long trek to a standpipe) and in this way may increase net income (as when reduced illness and injury mean less time taken from work and lower medical costs). bility to death before age 40; the illiteracy rate, and the standard of living, this last being measured on the basis of access to health services and safe water and the percentage of children under age 5 who are malnourished. 5

170 CITIES TRANSFORMED BOX 5.3 Measures of Access to Basic Public Services in the Demographic and Health Surveys Drinking water For the panel's analyses, access to drinking water is coded in three cate- gories: (1) water delivered through in-residence pipes or an in-residence well, the latter accounting for a small proportion of cases; (2) water available not in the home but in the neighborhood, usually via standpipes or neighborhood wells, but includ- ing cases in which water is purchased from vendors and tanker trucks; and (3) water drawn from other sources, such as from open streams and rainwater. These access measures are meant to represent the household's main water source, and only one response per household is recorded in a DHS interview. No data are available from DHS surveys on water costs, the variability of water supply, or perceived levels of contamination. In addition, many countries have fielded a question on the time required to fetch water and return. This indicator is presumably sensitive to congestion (waiting) costs, as well as to the distance of the household from its main water source. Waste disposal The household's access to a method of waste disposal is measured by whether the household (1) has access to a flush toilet; (2) uses a pit toilet or latrine; or (3) uses some other method of disposal, such as "throw-aways" or deposits in the bush. As with access to water, each surveyed household responds in terms of its principal method of waste disposal, and multiple methods are not recorded. Some DHS surveys include additional questions on the privacy of access to flush toilets, pit toilets, and latrines, which could show how many households must share such facilities. Unfortunately, not enough surveys include these questions to justify exploring them in the panel's analyses. No information is gathered on the frequency with which facilities other than flush toilets are cleaned, or on the time costs of access in the case of shared facilities. Electricity Access to electricity is measured crudely through a "yes-no" question that omits the dimensions of cost, reliability, and adequacy of power. other regions. The drawing of water from other sources is seen mainly in the countryside. These urban/rural differences in types of access to water are reflected in the time costs of access as well, as shown in the "time to water" column of Table 5-2. Even when attention is limited to households lacking in-home access to water, as is the case here, rural households need substantially more time than urban to fetch water and return, with Latin America being the exception. Where waste disposal is concerned, a marked urban advantage is seen in access to flush toilets, an advantage that is due mainly, no doubt, to differences in the provision of piped water. The use of pit toilets and latrines is quite common in the cities of sub-Saharan Africa, where about two-thirds of urban households dispose of their wastes in this manner. A significant fraction of households also use pit toilets and latrines in South, Central, and West Asia and in Latin America.

DIVERSITY AND INEQUALITY 171 What does it matter that a pit toilet is used rather than a flush toilet? In rural areas, a well-constructed and -maintained pit latrine can provide the same health advantages as a flush toilet if water is available for washing after defecation. The use of other methods of waste disposal also merits comment. In small, uncrowded rural villages, defecation in the bush or in the open can pose little risk of contact with feces or contamination of water sources. In crowded urban areas, however, poorly maintained pit toilets, open defecation, and "wrap and throw" may present substantial health risks. Table 5-2 shows that significant percentages of urban households use such methods of waste disposal in sub-Saharan Africa; in South, West, and Central Asia; and even in Latin America. According to the estimates in Table 5-2, some 90 percent or more of urban households have access to electricity except in the case of sub-Saharan Africa, where only 42 percent of households are electrified. Rural households in sub- Saharan Africa are rarely supplied with electricity, although 41 to 55 percent of rural households have access to it in other regions. The last column of Table 5-2 presents a composite indicator of disadvantage that summarizes lack of access to public services. This measure identifies house- holds that lack three such services: in-home access to drinking water, a flush toilet, and electricity. The households that are least well served are the rural households of sub-Saharan Africa, almost 89 percent of which lack all three services. The best-served households are those in urban North Africa, only 2 percent of which lack these services. Can such composite measures of disadvantage be used to define residence in an urban slum? The measure just described makes no reference to housing as such; neither does it address the spatial concentration of disadvantage. But with these limitations noted, how might it perform? In sub-Saharan Africa, the composite measure would place some 44 percent of the urban population in slums. In North Africa, Southeast Asia, and Latin America, however, it would suggest that only 5 percent of the urban population resides in slums. Many researchers, we suspect, would reject 5 percent as being far too low a number. But if a services-based definition is rejected, what properly defines a slum? Is crowded and dilapidated housing sufficient to capture the essence of the concept? In considering trends in service delivery, which can be examined for only 17 pairs of DHS surveys, we find some evidence of improvement outside sub-Saharan Africa. For provision of electricity, the average increase recorded between DHS surveys (over a span averaging 5 years) is 5.9 percentage points in rural and 6.3 points in urban areas. The gains recorded in piped water and access to flush toilets are smaller, on the order of 2 percentage points in rural areas and less than 1 percentage point in urban areas. Little progress is seen for sub-Saharan Africa, as might be expected given its chronic economic malaise. Outside this region, however, the urban/rural gap is maintained as the levels of provision rise. These analyses reveal striking urban advantages in almost all of the service indicators. In most of the countries surveyed, the average urban resident enjoys

172 CITIES TRANSFORMED much better access to services than the average rural resident. This is hardly sur- prising. The presence of middle- and upper-income groups in cities undoubtedly drives up the urban averages, insofar as access to services improves with ability to pay. Another urban advantage lies in the significant economies of scale, scope, and proximity that (at least in principle) reduce the costs per user of some public services. Of course, government investment priorities have often favored cities, whether as a result of political bias or because of the lower unit costs of service . . provlslon. Differences by City Size Within the urban population, are large and small cities equally well served? As discussed in Chapters 1 and 3, much of the urban population of developing coun- tries is found in smaller cities, which are also likely to accommodate much of the developing world's future population growth. If current levels of service delivery in small cities and towns are any indication, large programs of capital investment are likely to be needed. Table 5-3 shows why. Smaller towns and cities particularly those under 100,000 in population are significantly underserved by comparison with larger cities.7 The association between city size and the proportion served is not en- tirely uniform irregularities are apparent in the region-specific estimates but the general picture is one of poorer services in smaller cities. Differentials such as these do raise a question about the substantive meaning of city size. Population size is but a crude proxy for more fundamental city char- acteristics on which data are lacking. It is probable that city size and the quality of public-sector management are correlated only loosely, if at all. Larger cities have bigger municipal staffs and greater total resources to deploy than small cities, but their staff and resources can be defeated by the greater scale of the management tasks they face. Across the range of cities examined by the United Nations Popu- lation Division, it is the smaller cities that tend to have higher population growth rates (see Chapters 1 and 3), and this may subject their generally thinner man- agement teams to greater stress. Many other factors are doubtless involved in the relationship between city size and public-sector performance. Note, however, that even the smallest cities those with fewer than 100,000 residents exhibit levels of service provision that are well above those seen in rural areas (compare Tables 5-2 and 5-3~. Settlements of this size may not be thought to have a strong claim to "urbanness," and of course it is in this size range that countries differ greatly in how they distinguish urban from rural places (see Chapter 4~. Referring back to Figure 5-6, we find that a gap of this sort also characterizes adult educational attainment, with small towns and cities having higher levels of schooling than rural areas. Such gaps reappear in many contexts 7 The contrasts in the table are statistically significant less often in sub-Saharan Africa than in other regions.

DIVERSITY AND INEQUALITY TABLE 5-3 Percentages of Urban Households with Access to Public Services, by City Population Size 173 City Size DHS Surveys in Under 100,000 to 500,000 to 1 to Over Region 100,000 500,000 1 million 5 million 5 million Piped or Well Water on Premises North Africa 79.6 89.0 91.2 82.6 94.8 Sub-Saharan Africaa 35.4 45.1 42.3 55.1 Southeast Asia 36.1 50.2 39.2 56.1 53.2 South, Central, West Asia 64.0 72.4 67.0 74.1 61.0 Latin America 66.0 73.8 79.8 64.9 90.1 TOTAL 49.2 60.3 60.6 65.5 69.5 Minutes Needed to Fetch Water Outside the Home North Africa 23.2 20.2 20.9 22.8 22.6 Sub-Saharan Africa 19.1 14.1 19.7 13.5 Southeast Asia 8.6 9.1 8.4 7.2 11.4 South, Central, West Asia 26.1 15.0 13.1 14.2 20.3 Latin America 17.9 19.4 18.0 19.4 20.8 TOTAL 20.0 15.3 17.0 15.9 18.2 Access to Flush Toilet North Africa 88.3 95.4 95.4 93.1 98.5 Sub-Saharan Africa 18.0 20.6 25.4 30.5 Southeast Asia 78.1 82.5 89.1 84.8 South, Central, West Asia 36.8 56.0 71.5 75.9 82.1 Latin America 42.6 57.7 61.6 65.3 82.3 TOTAL 32.2 44.1 55.0 59.4 84.5 Access to Pit Toilet or Latrine North Africa 6.9 3.5 1.7 4.2 1.0 Sub-Saharan Africa 65.7 70.6 72.6 63.5 Southeast Asia 14.4 10.3 9.2 13.9 South, Central, West Asia 44.4 25.9 17.2 19.2 13.5 Latin America 44.1 29.9 30.1 28.6 14.5 TOTAL 52.3 44.3 39.1 35.2 12.2 Availability of Electricity North Africa 90.9 94.4 93.8 88.7 99.1 Sub-Saharan Africa 33.8 46.7 52.0 65.5 Southeast Asia 83.7 90.4 85.6 97.2 98.8 South, Central, West Asia 81.0 89.1 93.3 94.3 91.6 Latin America 84.0 89.9 97.4 98.0 99.1 TOTAL 57.1 70.4 77.1 85.0 96.2 (continued)

174 TABLE 5-3 (continued) CITIES TRANSFORMED DHS Surveys in Region City Size Under 100,000 to 500,000 to 1 to Over 100,000 500,000 1 million 5 million 5 million Household Lacks Piped Water, Flush Toilet, and Electricity 3.7 1.2 2.9 5.6 0.2 50.1 41.1 34.5 21.7 8.1 14.1 10.1 31.4 North Africa Sub-Saharan Africa Southeast Asia South, Central, West Asia Latin America TOTAL 5.7 8.5 5.7 22.1 5.7 4.6 1.6 15.4 0.2 4.2 1.5 0.4 9.8 2.3 a The DHS survey for Nigeria (1990) did not include questions on access to services in the household section of the questionnaire; hence, Lagos does not appear in this table. throughout this report. Evidently, for reasons that are difficult to pin down, small urban places are somehow quite distinct from rural villages. Although we can- not assign any precise meaning to small or large city sizes, we draw attention throughout the report to the differences associated with size and urge that the fac- tors behind these differences be investigated. Services and the Poor To probe further into intraurban inequalities in service access, we make use of a relative urban poverty measure based on the data available in the DHS. These sur- veys do not collect information on income or consumption expenditures, but they do collect data on a number of proxies for living standards. The proxies which concern ownership of consumer durables and housing quality can be distilled into an index of relative urban poverty. Appendix E explains the essence of the method, and Hewett and Montgomery (2001) give a full account. The approach yields a single relative poverty measure that defines the lowest quartile of urban households to be poor by comparison with other urban households. By "poor," we mean that these households own fewer consumer durables and have a lower quality of housing than do the other urban households in the same DHS survey. In the terms used by McDade and Adair (2001), ours is a "relative affluence" indicator. Table 5-4 compares levels of access to services among three groups: rural households, urban households that are relatively poor by our definition, and urban nonpoor households.8 The remarkable aspect of this table is the near uniformity of the results. For several of the key service access measures piped water, access The table summarizes the results of a multivariate analysis, using probits, in which the explanatory variables for the urban subsample include measures of city population size and urban relative poverty. The results were converted to predicted values and then summarized with the aid of the sampling weights so they can be compared with the rural weighted means.

175 Cot ~ ,~ ·> ~ i .~ o Em ·E-°4 ~ ~- I ~ Do Do o ·_. ~ .~ o o Cot .b V ·= ~ .= O ~ ~ oc) ~ aN ~ 00 ID O oo ~ ~ O ~ ~ O O ~ oo ~ ~ ID ~ ~ O ~ ID ~ O . oo kD ~ oO ~ ~ c`] ~ ~ O kD O ~ ~ ~ ~ oo kD ~ ~ oo kD O ~ ~ ~ O O oo ~ o ~ ~ ~ ~ O ~ kD O ~ ~ O O w) ~ ~ ~ ~ w) o w) kD ~ w) w) ~D ~ ~D ~ ~ ~ O ~ ~ ~ ~ kD ~ kD ~ ~ ~ ~ =\ ~ ~ ~ ~ c~] ~0 0 0 ~D ~ ~D ~ ~r, kD oo ~ O ~ w) ~ O ~ ~ ~ O w) ~ ~ ~o ~ ~ ~ O kD ~ kD O - 0 ~ ~ ~ aN ~ ~ O O ~ ~ O ~ ~ ~ oo O ~ ~ oo O —4 kD kD ~ —4 ') ~ ~ '~= '~D O ~ ~ ~ o w) ~ ~ 0 ~ ~ ~ o ~ ~ ~ 0 ~ oo oo 0 O ~ ~ ~ ~ ~ ~ ~ Cr) ~ w) ~ ~ O ~ ~ ~ kD kD =\ ~ kD ~ ~ ~ O O kD O ~ ~ ~ °O °O ~o ~ °O ~ O O ~ ~ ~ ~ ~ O ~ ~ ~ ~ ~ ~ o Cr) O c`] ~ kD O O oo oo t04 ~ kD oo t04 ~ ~ ~ ~ kD ~ ~ ~ ~ ~ ~ —4 ~ ~ O ~ O ~ ~ ~ O ~ ~ O O ~ kD ~ ~o kD ~ ~ O O ~ ~ O > ~ Cr) w) `~D ~ tr) w) w) ~ '> w) ~ ~ aN ~ ~ ~ O w O O oo ~ oO ~ kD ~ kD O oO ~ ~ O ~ ~ ~ ~ ~ - 4 ~ ~ ~ ~ O O ~ kD ~ O oo ~ ~ O oo oo O ~ ~ oo ~ O oo ~ ~ O kD ~ Cr) ~ ~ aN ~ ~ ~ ~ ~ ~ oo O Cr) w) ~ ~ ~- kD Q O ~ O ~ O ~ O ~ O ~ O O ~ ~ O ~ ~ O ~ ~ ~ ~ ~ ~ ~ O O O C~ O O C~ O O C~ O O C~ O O C~ o o o~ ~ ~ o~ ~ ~ o~ ~ ~ o ~ ~ ~ o ~ ~ ~ o ~ ~ ~ o ~ ~ ~ o ~ ~ ~ o ~ ~ ~ m ~=. .~ ¢ cd c~ ¢ '`a ~ cd ¢ ~ ~ v ¢ .= ~ o ~ o o ~ o ~Q ~Q v: ~ E~ ~ .~ t) o ~ o ~= =o c~ ~ c~ ~ c~ ~ cd s~ 00 =, c~ ,=) c~ o O ~ =0 ~ .o .= ~3 s~ ~ o C~ ~ C~ C~ Ct (D 6 o =~ 0° .= ~ ~ =e o 8 Ct ~ C~ C~ =~m C~ ~ ~ ,~o.= ~o o ~ ~t (D C~ .= Q ' c~ O ~ .= ~ c~

176 CITIES TRANSFORMED to a flush toilet, and electricity we find that the urban poor are in a distinctly inferior position compared with other urban residents, but in a decidedly better position than the average rural household. The situation is more complex when we consider access to drinking water in the neighborhood, a case in which poor urban households can be on a par with rural households (as in Latin America) or in which higher proportions of rural households can have access (as in South, Cen- tral, and West Asia). The countries surveyed in sub-Saharan Africa show higher levels of access to water in the neighborhood among the urban poor than in rural areas, and taking the two water sources together gives the urban poor a decided advantage. (Note that the time costs of access to water for those households lack- ing in-home connections are similar for the urban poor and nonpoor.) A similar regional pattern is evident in the use of pit toilets and latrines. The differences in the access of relatively poor urban households and rural households are almost always statistically significant.9 In the case of piped and in-home water, all of the estimates shown in Table 5-4 indicate that the urban poor are significantly more likely to have access than rural households. Much the same is true for the likelihood of having a flush toilet and electricity.~° In the cases of water in the neighborhood and access to either pit toilets or latrines, however, the urban poor are not always significantly advantaged relative to rural households.) Are Migrants Poor and Underserved? We now ask whether city households containing recent migrants more precisely, recent migrants who are women of reproductive age differ from other house- holds in their access to services. Surprisingly, an empirical analysis uncovers few important differences. The details can be found in Appendix E, Tables E-1 and E-2. Households with a migrant woman in residence are only slightly more likely than other households to be defined as "poor" by our relative definition, with the differences ranging from 1 to 5 percentage points depending on the region. Likewise, in comparing access to services between migrant and nonmigranthouse- holds, we uncover only the slightest of differences. The migrant-nonmigrant gap in the likelihood of access typically amounts to no more than 2 to 3 percentage 9We compute standard errors for this test using the delta method. The frequency of significant results is due in part to the relatively large sample sizes for rural areas and the usually generous sample size of the urban sector, one-quarter of which is "poor" according to our relative definition of poverty. i°In each of these cases, only one DHS survey gave an insignificant result. iiIn some 28 of the 59 DHS surveys, the urban poor were significantly less likely to have access to water in the neighborhood than were rural households, whereas in 31 surveys, the urban poor were significantly more likely to have such access. A similar split emerged in access to pit toilets and latrines, with poor urban households being significantly less likely to have access than rural households in 20 of the 56 surveys, but significantly more likely in 34 surveys. i2Recall that by definition, one-quarter of urban households are relatively poor. The relationship is statistically significant in 19 surveys out of the total of 78. A significant relationship emerges in 25 surveys, but 6 of these suggest that migrant households are less likely to be poor.

DIVERSITY AND INEQUALITY 177 points, and it rarely attains statistical significance. Even when the rural-origin mi- grants are separated out they are about one-third of all recent migrants the differences in service access are estimated to be small. This is not at all what we had expected to find. Is the conventional view of migrants so wrong? Can it really be that recent migrants are hardly distinguishable from urban natives and longer-term residents in these socioeconomic dimensions? Is the notion that migrants are found mainly in ill-served urban slums simply incorrect? Or is the problem one of inadequate measurement and an overly restricted sample? After all, the DHS data on migrants pertain only to women of reproductive age, and these women cannot represent the situation of all migrants. We suspect that an analysis of China's unregistered migrants, known as the "floating population," would uncover very large migrant disadvantages (see Chap- ter 8), and doubtless such disadvantages can be found elsewhere as well. Data from the DHS, from which male migrants are generally excluded, simply do not speak to these situations. For the case of women of reproductive age, however, the DHS data cannot be so easily dismissed. Although they are far from being definitive, these data suggest that the conventional views of migrant disadvantage may need to be reconfirmed. Discussion The DHS household-level data on service provision are broadly consistent with the findings of city-level reviews of water and sanitation in African, Asian, and Latin American cities (e.g., Hardoy, Mitlin, and Satterthwaite, 2001~. Studies of individual cities confirm what was seen in the DHS data that members of many poor urban households must defecate outside or resort to unsanitary "wrap and throw" methods. The DHS results for sub-Saharan Africa are also consistent with other evidence. Whereas case studies of large cities in Latin America usually show considerable improvements over time in the provision of water and sanita- tion, this trend is less evident or absent in sub-Saharan Africa (Hardoy, Mitlin, and Satterthwaite, 2001~. For example, a review of how access to water changed in 10 urban sites in East Africa between 1967 and 1997 found that for those lack- ing in-house connections, the average queueing time for water rose from 28 min- utes per day in 1967 to 92 minutes in 1997 (Thompson, Porras, Wood, Tumwine, Mujwahuzi, Katui-Katua, and Johnstone, 2000~. Such increases in waiting time appear to be due to the increased congestion and competition for scarce public resources that often accompany city population growth. Even households with piped supplies experienced deterioration in the quantity and reliability of their water. Micro studies of individual neighborhoods see Boxes 5.4 and 5.5 and Table 5-5 for examples can penetrate more deeply than the DHS surveys into some aspects of services. Perhaps the most important finding to emerge from

178 CITIES TRANSFORMED BOX 5.4 Examples of Intracity Differentials in Water Supply Accra (Ghana) In high-income residential areas with water piped to the home and wa- ter closets for sanitation, daily water consumption is likely to be well in excess of the recommended figure of 200 liters per person. By contrast, in slum neighbor- hoods such as Nima-Maamobi and Ashiaman where buying water from vendors is common, daily consumption is only about 60 liters per person (Songsore, 19924. Dar es Salaam (Tanzania) A 1997 study of domestic water use in four sites, all with piped supplies, found large differences in water use and reliability. The average per capita water use for households interviewed in Oyster Bay (a high-income area) was 164 liters a day. Water use was much less in two lower-income areas: in Changombe, it was 44 liters a day and in Temeke, 64 liters. Some 70 percent of the households interviewed in Oyster Bay received a 24-hour supply, as compared with just 10 percent of households in Temeke and 11 percent in Changombe. The unreliability of the piped water supplies in Changombe and Temeke meant that more than 60 percent of the interviewed households with piped supplies used vendors as their primary source, despite the higher costs (Thompson, Porras, Wood, Tumwine, Mujwahuzi, Katui-Katua, and Johnstone, 20004. Guayaquil (Ecuador) In 1990, average daily consumption ranged from 307 liters per per- son in the well-to-do parts of the city to less than 25 liters for the poor supplied by private water sellers (Swyngedouw, 1995~. Monterrey (Mexico) The proportion of houses with running water varied from 49 to 93 percent among the eight municipalities that made up the metropolitan area in 1990; the proportion with adequate drainage varied from 35 to 96 percent (Garza, 19964. neighborhood studies is the high money cost of water. Evidently it is not uncom- mon for low-income households to spend 10-20 percent of their cash income on water; some case studies show even higher percentages (Cairncross, 1990~. The regularity of water supply also deserves attention: many studies have documented intermittent flows and long outages of supply. One study of water provision in 50 cities in Asia and the Pacific during the mid-199Os (Asian Development Bank, 1997) showed that the utilities often supply water for only a few hours each day33 Very few neighborhoods of Mombasa enjoy a continuous supply of water (the average is only 3 hours a day), and some have seen no water in their pipes for several years (Rakodi, Gatabaki-Kamau, and Devas, 2000~. This dimension of water supply could be examined much more thoroughly in standard demographic surveys.~4 Survey questions could also be fashioned to elicit information about the adequacy and regularity of electricity supply. i3This study reports an average for Karachi of 4 hours per day, and similarly for Madras/Chennai (4 hours), Mumbai (5 hours), Bandung (6 hours), Kathmandu (6 hours), and Faisalabad (7 hours). Understandably, the DHS surveys do not make inquiries about water quality and contamination, which probably could not be recorded reliably through survey questions.

DIVERSITY AND INEQUALITY 179 BOX 5.5 Citywide Study of Environment and Health Among Households in Port Elizabeth, South Africa A 1996-1997 study of a representative sample of 1,000 households in Port Elizabeth (popu- lation 800,000) explores environmental and health conditions at the household level across the city. As in the case of the DHS analyses, consumer durables were used as proxies for living standards. Stark contrasts between the rich and poor emerged in both environmental conditions and access to services, as can be seen in the following table for the case of water supply: Socioeconomic Status Water Source Low Lower-middle Middle Upper-middle High Total Piped inside 11 house 27 57 Piped to yard 36 Communal 51 tap Othera Number of 186 210 207 Households SOURCES: Potgieter, Venter, Thomas, Seager, McGranahan, and Kjellen (19994; Thomas, Seager, Viljoen, Potgieter, Rossouw, Tokota, McGranahan, and Kjellen (1999~. a Includes water from, e.g., a neighbor or river. 37 36 2 0 82 98 35 9 o 14 4 1 O O 227 193 56 24 20 o 1023 Although the wealthier areas of Port Elizabeth are highly homogeneous in measures of income and services, the poorer areas exhibit a surprising amount of internal variation. Here the poor often have relatively rich neighbors despite the high average level of poverty in the neighborhood. The variation evident in these poor neighborhoods suggests that ser- vices might be funded in part through cross-subsidization. Little information has been collected by the DHS surveys on the number of people who share toilets, the quality of toilet maintenance, and the monetary or time costs of access.~5 In some cities, pay-as-you-go public latrines consume a not-insignificant percentage of the cash incomes of poor households (for Kumasi, Ghana, see Devas and Korboe, 2000; for Bhilwara, India, see Ghosh, Ahmad, and Maitra, 1994~. With all the aspects of reliability and adequacy taken into consid- eration, it is likely that the simple indicators used by the DHS tend to overstate the advantages of urban residence with regard to securing access to public services. The degree of overstatement is not known, but the issue clearly warrants further study. i5The DHS has formulated standardized questions on the privacy of access, but few countries have made use of these questions as yet.

180 CITIES TRANSFORMED TABLE 5-5 Household Environment Indicators in Accra, by Affluence of Neighborhood Indicator No water at source of residence Share toilets with more than 10 households No home garbage collection Main cooking fuel wood or charcoal Flies infesting kitchen Number of households Percentage of Sample Poor Middle Affluent 55 60 94 85 91 790 14 17 77 2 55 44 30 56 18 160 50 SOURCE: Benneh, Songsore, Nabila, Amuzu, Tutu, Yangyuoru, and McGranahan (1993~. MEASURING ABSOLUTE POVERTY IN CITIES There is broad agreement on the principles underlying official poverty lines: the lines should be determined by the monetary income required for an individual or household to meet basic consumption needs in a specified neighborhood. In practice, however, few developing-country governments adjust income-based poverty lines to take full account of locational differences in the prices of food and nonfood essentials. The amount of adjustment needed is a matter of debate in high-income and low-income countries alike. In the United States, for instance, there is considerable uncertainty about whether and how to implement location- specific adjustments for differences in the costs of living (see Box 5.6~. Adjusting for housing costs alone would significantly shift the U.S. poverty profile, with the likely effect of raising estimates of poverty in metropolitan areas. Much the same could be expected from locational cost-of-living adjustments in poor countries. Locational Price Differences and Nonfood Needs There is reason to believe that the official poverty lines adopted by developing countries can be seriously deficient in two aspects the treatment of locational price differences and allowances for nonfood essentials. Income-based poverty lines are usually determined according to estimates of the cost of an adequate i6Most governments use the household as the unit of analysis and do not examine intrahouse- hold differentials in access to income. Although undoubtedly important (Levin, Maxwell, Ammar- Klemesu, et al., 1999), intrahousehold differences in consumption and control over resources are very difficult to measure. Many governments also ignore differences in household size and number of de- pendents, although there is some evidence suggesting that adjustments for size can make a significant difference to poverty estimates (Grewe and Becker, 2001).

DIVERSITY AND INEQUALITY 181 BOX 5.6 Adjusting Poverty Rates for Geographic Differences in Prices: The U.S. Experience At present, the official poverty measure used in the United States makes no adjustment for geographic differences in the cost of living. Although strong recommendations have been made to adjust for locational differences (Citro and Michael, 1995), there is disagreement about how best to implement the recommendations given the data available. According to Short (2001), the least contentious issue is that of housing costs, which are thought to be the factor accounting for the greatest portion of locational differences in the costs of living. Building on the methods outlined in Citro and Michael (1995) and Short, Garner, John- son, and Doyle (1999) and linking these to hedonic methods to adjust for variation in housing quality, Short (2001) shows that locational adjustments bring little change to national-level estimates of poverty, but have important effects at the state level. Poverty rates tend to fall in states with relatively low housing costs (e.g., Alabama and Arkansas) and rise in states where housing costs are higher (e.g., New York and California). The change in poverty by metropolitan and nonmetropolitan area is difficult to summarize (laws regarding confidentiality and disclosure of census data prevent detailed inspection of the data at these levels), but the analysis suggests that estimated poverty rates would tend to rise in large metropolitan areas if housing cost adjustments were made. diet, with a further allowance, typically on the order of 15 to 30 percent, for all nonfood expenditures. If the prices of food and other necessities range widely and systematically across locations, there is a potential for systematic misestimation of poverty. The size of the nonfood allowance and the means by which this allowance is calculated also need attention. It has not escaped notice that poverty lines ought to be adjusted for locational differences in prices, and a number of governments have established different poverty lines for urban and rural areas (e.g., the Philippines). For several rea- sons, however, adequate adjustments for locational differences are difficult to put in practice. First, prices can vary widely within locations. Second, households make decisions about where to live and what to consume that cause difficulties in formulating criteria appropriate to any given location. In making such deci- sions, households may accept high prices for some goods (or deficient services) in return for lower prices on others (or better services). Without knowledge of the full range of circumstances in each neighborhood or district, it is difficult to specify the level of income required to reach a "nonpoor" threshold of sub- jective well-being. Local and national authorities have very limited knowledge of the particulars and must do what they can to establish simple, administratively feasible poverty lines. Inevitably some families will be misclassified; the question is whether the errors are tolerable. There is evidence of considerable variation within cities in the prices of wa- ter and food. In some accounts, the cost per liter of good-quality water is said to vary by as much as a factor of 10 depending on location (Hardoy, Mitlin, and Satterthwaite, 20014. As shown earlier in this chapter, middle- and upper-income

182 CITIES TRANSFORMED households generally enjoy better access to piped water, and neighborhood studies show they can pay much less per liter than lower-income groups that rely on wa- ter vendors. Poorer households also tend to pay more for food than higher-income groups (Ruel, Haddad, and Garrett, 1999~. The poor depend to a greater extent on food bought in the street because they lack a means of preparing it in bulk at home and face time constraints in their travels to work. To calculate the costs of food in a manner consistent with the eating habits of the poor would require de- tailed household-level quantity and expenditure data, which are seldom available (Hentschel and Lanjouw,1996~. The unit costs of nonfood essentials may also be higher for the poor. Rents and quality of housing vary both within and across cities (UNCHS, 1996; Malpezzi, 1999~. In many cities, the poor keep their expenses down by building their own homes on the urban periphery, but doing so may entail high transport costs in both time and monetary terms (Barter, l999~. Other households choose to live in rental housing on more centrally located sites, accepting the high rents because of easier access to jobs and lower transport costs. In poor countries the adjustments made to account for nonfood needs tend to be ad hoc and are seldom grounded in careful assessments of the prices faced by the poor (Jonsson and Satterthwaite, 2000a). Housing is the most important nonfood item.~7 A 1989 survey of poor households in six South Korean cities found that 64 percent of their incomes were spent on nonfood items; this propor- tion had increased over the previous 25 years, mainly because of increases in the real costs of housing (United Nations Development Program, 1998~. Other stud- ies confirm that significant shares of income are spent on housing (see Barbosa, Cabannes, and Moraes, 1997; Richmond, 1997; UNCHS, 1993; United Nations Development Program, 1998; Rakodi,1995~. In addition, many poor households face high costs for transport (Urban Resource Centre Karachi, 2001; Malawi, National Statistics Office, 1994; Barter, 1999~; school fees end related expenses (Kanji, 1995; Bigsten and Mugerwa,1992~; health and child care; energy (Islam, Huda, Narayan, and Rana, 1997; Ghosh, Ahmad, and Maitra,1994~; and other ex- penses, such as payments to community-based organizations, bribes to the police, and payment of fines. Is an allowance equivalent to 15-30 percent of minimum food expenditures really sufficient to account for all such needs? When official income-based poverty lines were calculated in the United States in the 1960s, it was thought reasonable to set the poverty threshold at three times the cost of a minimum food basket, rather than at 1.15 to 1.3 times that cost as is current practice in many developing countries. The U.S. approach recognized the many nonfood expenditures that households need to make to avoid poverty i7 there are difficulties in defining the income needed for the acquisition of any durable good, which requires the computation of service flows, imputation of rental values for owner-occupied housing, and the like (e.g., Mozambique Ministry of Planning and Finance, Eduardo Mondlane University, and International Food Policy Research Institute, 1998).

DIVERSITY AND INEQUALITY 183 (Citro and Michael, 1995~. Some researchers have suggested that the appropriate adjustments may be even greater in the cities of poor countries (see Beck, 1994~. Comparisons of Urban and Rural Poverty It is an encouraging sign that an increasing number of governments and interna- tional agencies have adopted separate urban and rural poverty lines this is evi- dence of a growing recognition of the importance of location in the costs of living (Tabatabai and Fouad, 1993; Jonsson and Satterthwaite, 2000a). In the views of many researchers, the urban adjustments give insufficient attention to the prices faced by the urban poor. But not all considerations indicate that urban poverty is understated relative to rural poverty. Some nonfood essentials (e.g., clothes and medicines) can be much more expensive in rural areas, and much of the depriva- tion faced by the rural poor is measured less by prices than by the unavailability of goods and services. If levels of urban poverty are understated for the reasons outlined above, this certainly does not imply that they are understated relative to rural. There is by no means a consensus on the adjustments needed to put urban and rural poverty on a comparable basis. In examining urban and rural price levels in India, Deaton and Tarozzi (2000) find evidence that prices are about 15 percent higher in urban areas, but this is far short of the 41 percent estimate adopted for India's official poverty lines on the recommendation of its Expert Group Report of 1993 (Government of India, 1993~. The lower estimate derived by Deaton and Tarozzi which, incidentally, is roughly equal to the official adjustments made before that Expert Group report excludes housing and transportation, which ac- count for at least a third, and possibly more, of urban household budgets. Even with allowance for the omitted items, Deaton and Tarozzi maintain that the official poverty lines overstate urban costs of living and thus the extent of urban poverty relative to rural. Evidently there is ample room for controversy and dispute, even in countries with sophisticated statistical systems. What, then, can be said about comparisons of urban and rural poverty? The World Bank has developed country-specific estimates of the number of poor peo- ple in cities and rural areas for all of the developing regions. These estimates are based on an international poverty line that is set at (approximately) US$1 per per- son per day. The World Bank's estimates have been highly influential in public debates on development and development assistance; they are in many respects beneficial in bringing attention to the poverty levels of developing countries. Whether the methods used to derive these estimates give a correct assessment of urban poverty levels is another matter. The World Bank (2000a: Figure 11) presents an estimate that some 495 million urban poor lived in developing coun- tries in 2000. This figure implies that of every 4 urban residents, 3 are not poor. In some countries, according to these estimates, the percentage of city residents who are poor is very low. For instance, by the US$1 per day standard, fewer than

184 CITIES TRANSFORMED 2 percent of China's city residents are judged to be poor (World Bank, 2000a: 236~. Is proper account being taken of urban price differences and the costs of housing and other nonfood essentials? We stress this point because estimates such as these have considerable influence in public debates about development and international aid. As China and other poor countries become more urban, the lim- itations of urban poverty estimates cannot be left to delicately worded footnotes and rueful caveats. Urbanization underscores the need for rigorous justification of the basis for urban poverty estimates and clear statements of the limits and uncertainties that surround such estimates. RISK AND VULNERABILITY Because they must live in the narrow margin above subsistence, the urban poor face many difficulties in managing risks in coping with rising prices, falling in- comes, and other personal and societal shocks. Low incomes constrain savings and limit the accumulation of assets on which the poor can draw. The limited asset bases of the poor leave them vulnerable to sudden economic shocks, as well as to longer-term crises such as serious illness or injury (Moser, 1996~. It is usually the poorer groups in urban areas that lose their homes and assets to floods, landslides, and earthquakes (Hardoy, Mitlin, and Satterthwaite, 1992; International Federa- tion of Red Cross and Red Crescent Societies, 1998; Sanderson,2000~. Social networks can provide some important resources for the poor in times of stress (Cox and Jimenez,1998~. Such networks can facilitate transfers of income or services when a network member is in distress, whether from a shortfall in income, illness, or some other shock. Social networks functioning in this way can be likened to informal insurance systems. Although a substantial literature now exists on risk-sharing networks in rural villages (e.g., Townsend, 1994), there has been comparatively little investigation of risk-sharing arrangements in urban communities. Cox and Jimenez note that rural and urban settings differ in two aspects that affect informal insurance mech- anisms. In cities it can be more difficult for network members to monitor the causes of income shortfalls, allowing a negligent or lazy member to make claims for assistance. In rural areas, by contrast, the major causes of income shocks (such as droughts) can be readily verified. Such moral hazard problems could make it difficult to maintain the mutual trust needed to sustain urban informal networks of insurance. Effective monitoring in urban areas might require that the network members reside in the same neighborhood or be linked in other ways (such as through family ties) that maintain trust. But urban areas also present some advan- tages for insurance networks. The greater diversity of city economies and occu- pations in comparison with those of rural areas could allow urban social networks to include more members whose incomes are negatively correlated. In rural areas, by contrast, common price and weather shocks could depress the incomes of all network members simultaneously, leaving little possibility for beneficial transfers.

DIVERSITY AND INEQUALITY 185 In their study, set in a low-income section of Cartagena, Colombia, Cox and Jimenez (1998) find substantial evidence of transfers in urban social networks. Members who had experienced recent spells of unemployment were more likely to have been given assistance by their networks; those with higher incomes were less likely to have been assisted.~9 The diversity of network composition also made a difference: a household whose networks contained other households bet- ter off than itself was more likely to have received assistance. Most of the net- work members some 60 percent lived in Cartagena, and the poorer households tended to have larger networks, perhaps because they needed to spread their social safety nets wider. In addition to relying on social networks, low-income households make use of many other day-to-day coping strategies to survive such stresses without irre- versible damage to their productive capacities. Methods of coping with shocks and stresses include supplying more labor (taking second jobs, working longer hours, entering children and other household members into the labor market); re- ducing consumption; changing diets; selling household assets; and, in the extreme, resorting to prostitution or theft (Dinye, 1995; Moser, 1996~. Box 5.7 describes some of the coping strategies adopted by low-income households in Dar es Salaam (CARE/Tanzania, 1998~. The Dar es Salaam study notes that before seeking outside assistance, urban households "self-adjust" to the extent that they can. It appears that households will restrict food consumption before they borrow money or take food on credit. Those with higher incomes take their children out of school less often and do not cut back on medical expenses to the same extent as lower-income groups. The study also notes that about half of city households regularly send remittances to family members, thereby maintaining extended family relationships that can be drawn upon in times of crisis; some 10 percent report receiving regular remittances from other households. By using means such as these, poor households can protect themselves against some external shocks, but not all households can adjust to the same extent, and the assets on which the poor rely can be rapidly depleted. In addition, many short-term survival strategies have negative longer-term consequences as when all available assets are sold, children are withdrawn from school, and women take up dead-end jobs offering low pay for long hours (Moser, 1996~. Other risky and potentially damaging strategies include compromising the quality of medical care and reducing food consumption. i8 according to the World Bank (2000b), about half of these Cartagena households received a trans- fer in cash or in kind in the month before the survey. For the poorest quintile of households, transfers accounted for 40 percent of household income for male-headed households and 52 percent for female- headed households. i9A related study of the urban and rural Philippines, that of Cox, Hansen, and Jimenez (1999), shows that informal transfers are responsive to the income levels of the poorest households. For urban Filipino households, total transfers received decline with household income among households in the lowest income quartile, but are constant with respect to income for higher-income households.

186 CITIES TRANSFORMED BOX 5.7 Shocks and Coping Strategies: Dar es Salaam A study of shocks and coping strategies in Dar es Salaam, drawn from interviews with 298 households in six wards, gives insight into the coping strategies used by low-income households. In the year leading up to the interview, many households had suffered a severe shock, whether from having a member lose a main income-generating activity, from a major expense such as medical bills, or from necessary festival or ceremony contributions. More than 1 in 10 households had been evicted during the year. The table below (adapted from CARE/Tanzania, 1998: Table 13) shows the coping strategies used. Coping Strategy Used Percent Using Seek more sources of income Pull children out of school Reduce quality of medical care Return to home village Obtain food from rural areas Consume less-preferred or less-expensive foods Borrow money Purchase on credit Reduce the number of meals per day Reduce food portions Limit adult food to give more to children Borrow food Send children to neighbors to eat Skip eating for complete days 24 24 52 25 37 78 39 36 52 56 41 21 16 In cities, environmental hazards represent another dimension of the risk faced by poor households. Vulnerability to environmental risk depends on many fac- tors: income and assets, which determine access to good-quality housing and reduce exposure to some locational hazards; the availability of health care and emergency response capabilities in the community, which can mitigate the health consequences of injury and acute disease; occupational roles that increase expo- sure to risk, such as picking through garbage or disposing of excrete; and other coping mechanisms in a household's repertoire that come into play when a hazard has caused illness or injury knowing what to do, whom to visit, and how to reconfigure survival strategies (Chambers, 1989; Corbett, 1989; Fryer, 1989~. The greater exposure of the poor to environmental hazards may threaten in par- ticular infants, young children, and the elderly, as well as pregnant women.20 Sus- ceptibility to environmentalhazards is determined by various factors: (1) for many 20According to the World Health Organization (1992: 121), "Every stage of the multi-step process of reproduction can be disrupted by external environmental agents and this may lead to increased risk of abortion, birth defects, fetal growth retardation and perinatal death."

DIVERSITY AND INEQUALITY 187 biological pathogens, weak body defenses (some related to age and nutritional status, others to acquired immunity, such as through vaccines); (2) for physical hazards, limited mobility, strength, and balance, as in the cases of young children, the elderly, and the physically disabled; and (3) for exposure to chemicals, age and the state of health at the time of exposure. Micronutrient deficiencies are likely to exacerbate the effects of air pollutants (Romieu and Hernandez, 1999~. Asth- matics and the elderly with chronic respiratory disease appear to be particularly susceptible to certain air pollutants. There are few data available on individual exposure to environmental hazards in developing-country urban environments. Most of the data are aggregate in na- ture, taking the form of areal averages. But some micro-level implications can be pieced together from the aggregate clues. According to the World Health Organi- zation (1992), low-income families in many cities use lower-cost fuels and cook on less-efficient portable stoves or open fires. These practices put poorer fami- lies at risk from indoor and localized air pollution. Air pollution levels in low- income neighborhoods may also be elevated because of spillover effects within the neighborhood, as when many neighborhood households use polluting fuels. The correlations of poverty with other forms of outdoor air pollution, however, may not be especially strong. Exposure may be heightened when the poor live adjacent to quarries, cement factories, and other air-polluting industries. (There is also a tendency for polluting industries, waste dumps, and waste management fa- cilities to concentrate in low-income neighborhoods, a tendency that is evident in high-income nations as well, as the literature on environmental racism has helped document [Wing, Grant, Green, and Stewart, 19961.) The burning of garbage can be a significant source of air pollution in neighborhoods lacking regular garbage collection (Surjadi, 1993~. But the correlations are less clear when an entire city suffers from air pollution or when winds help disperse pollution. Floods and landslides are risks that affect the poor disproportionately because housing and land markets price low-income groups out of safe, well-located, and well-serviced sites. High percentages of the urban poor live in housing that is vulnerable to fire, made of inflammable materials such as wood and cardboard. The risk of accidental fires is much increased when these households cook on open fires or use portable stoves and when kerosene lamps or candles are used for light. As do other urban residents, poor city families also face risks stemming from macroeconomic forces and political restructuring. These issues are discussed in Chapter 8, but it may be useful here to note the spatial distribution of their effects. Macro shocks and political disruptions can have very different impacts in large than in smaller secondary cities. Box 5.8 gives an account of the turbulence sur- rounding the breakup of the former Soviet Union, the repercussions of which were felt differently in Bishkek, the capital of Kyrgyzstan, and the secondary cities of that new country. In the capital, the disruption appears to have been sharp but short; the secondary cities, however, appear to be in a sustained funk.

188 CITIES TRANSFORMED BOX 5.8 The Economic Transition in Kyrgyzstan: Bishkek and the Secondary Cities Compared A number of the countries once part of the Soviet Union have experienced large declines in living standards. In Kyrgyzstan a series of surveys allows the determinants of earnings to be studied over the periods of chaos, stagnation, and nascent recovery (Anderson and Becker, 2001~. The study shows that Bishkek, the national capital, is apparently more resilient than the secondary cities. Kyrgyzstan weathered a loss in gross domestic product of 50-60 percent following the collapse of the Soviet Union, and per capita income today is only about US $300 at official exchange rates. For the nation as a whole, poverty rates peaked in 1996. Since then Bishkek has experienced declining levels of poverty. The situation in the secondary cities, however, is mixed: they have a lower incidence of poverty than rural areas (in partic- ular, lower levels of extreme poverty), but there is no clear time trend and apparently little prospect for long-term improvement. Bishkek's resilience is evidently due to its skilled workforce, the favorable composition of its industries, and the fact that it is Kyrgyzstan's entry point for international trade and foreign aid. CHILDREN'S LIVES Children's lives reflect many of the aspects of inequality and diversity discussed in this chapter. This section begins by focusing on children's schooling, which is a measure of the human capital that governments and families invest in children. (Child health and survival are examined in Chapter 7.) Earlier in this chapter, we presented evidence of a decided urban advantage in adult schooling. Here we ask whether urban residence is also associated with higher levels of investment in children's education. We investigate whether children in relatively poor urban families are at a disadvantage by comparison with other urban children and ex- amine their position relative to rural children. We then explore what is known about a subpopulation of urban children who suffer from serious deprivation- street children. We show that urban children as a group are advantaged and that poor urban children retain some of that advantage by comparison with rural chil- dren. However, street children are burdened not only by poverty, but also by the special social and health risks to which urban life exposes them. School Enrollments in Urban Areas The tables and figures that follow show enrollment proportions for two age groups of children: those aged 9-10, who are of an age to be enrolled in primary school, and those aged 15-16, whose enrollment rates are likely to reflect a mix of the late primary, middle, and early secondary school levels.2i For both age groups, Table 5-6 shows strikingly large differences in enrollment between urban and rural children. On average, the difference is on the order of 2iThese analyses are based on 62 DHS surveys from 45 countries; see Table C-l in Appendix C.

DIVERSITY AND INEQUALITY TABLE 5-6 Percentages of Children Enrolled at Ages 9-10 and 15-16, Rural and Urban Areas DHS Surveys in Region North Africa Sub-Saharan Africa Southeast Asia South, Central, West Asiaa Latin America Proportions of Children Enrolled Number of Aged 9-10 Aged 15-16 Countries Rural Urban Rural Urban 66.6 57.3 75.0 67.0 76.7 2 23 2 8 9 61.7 91.7 57.1 78.6 93.1 97.0 79.4 90.2 84.8 93.4 33.8 39.3 57.2 51.6 48.6 TOTAL 44 68.9 85.3 44.0 64.3 a India (1992) collected enrollment data only for children aged 6-14. 189 16 percentage points for the 9-10 age group and reaches nearly 20 percentage points for those aged 15-16. In Latin America, only 49 percent of rural children are enrolled at the latter ages, whereas almost 77 percent of urban children are still in school at those ages. Focusing on variation in enrollment rates by city population size, we find smaller differences than those seen in the urban/rural comparisons. Table 5-7 presents the city size findings. As one might expect, enrollment rates in the small- est cities are somewhat lower than in the larger cities, but the differences are on TABLE 5-7 Enrollment Percentages for Urban Children, by City Population Size DHS Surveys in Region City Population Size Under 100,000 to 500,000 to 1 to Over 100,000 500,000 1 million 5 million 5 million Enrollment at Ages 9-10 North Africa 90.5 93.2 93.4 91.7 93.8 Sub-Saharan Africa 75.3 79.6 85.9 83.7 95.7 Southeast Asia 97.4 96.1 96.3 98.7 97.5 South, Central, West Asia 86.1 90.5 90.7 91.6 87.5 Latin America 92.5 92.8 95.0 95.1 97.2 TOTAL 82.6 86.6 90.6 90.4 92.9 Enrollment at Ages 15-16 North Africa 64.5 69.3 67.2 63.2 77.9 Sub-Saharan Africa 56.4 57.9 60.6 55.2 70.7 Southeast Asia 70.4 78.2 74.1 73.9 79.2 South, Central, West Asia 61.2 68.6 68.2 66.4 52.7 Latin America 74.0 76.9 79.0 80.3 85.6 TOTAL 61.9 66.0 68.6 66.7 70.7 NOTE: See note to Table 5-6.

190 CITIES TRANSFORMED TABLE 5-8 Predicted Enrollment for Children Aged 15-16 by Residence and, for Urban Areas, by Relative Poverty DHS Surveys All Urban Urban in Region Rural Poor Nonpoor North Africa 33.4 47.4 72.4 Sub-Saharan Africa 39.0 42.9 61.5 Southeast Asia 58.8 63.0 80.3 South, Central, West Asia 43.2 44.2 67.0 Latin America 50.7 66.2 82.1 TOTAL 42.7 49.0 67.9 NATE: See note to Table 5-6. Estimates based on probit models with controls for child's age and sex in rural areas and for age, sex, and city size in urban areas. Some surveys employed in Table 5-6 could not be used in analyses of poverty, causing the rural estimates shown above to differ slightly from those of the earlier table. the order of 9 percentage points for younger children and 6 percentage points for older children. In sub-Saharan Africa, however, the enrollment differences are much larger nearly 20 and 13 points for the younger and older groups of chil- dren, respectively.22 Table 5-8 explores the implications of urban poverty for levels of school enrollment, presenting a comparison of urban poor children, their urban peers from nonpoor households, and rural children. We find that poor urban children are more likely to be enrolled than rural children, but less likely to be enrolled than nonpoor urban children.23 As was the case with public service provision, the urban poor generally occupy the middle position. Note, however, that in sub- Saharan Africa and both regions of Asia, the margin of difference between the urban poor and rural children is very thin. To judge by enrollment rates, poor ur- ban children are not receiving much more in the way of human capital investment than children in the countryside. This finding is dismaying in view of its impli- cations for urban inequality in the future. Table 5-9 provides further evidence on the situation of poor children, with attention to differences by sex. Although there are important differences in enrollment between girls and boys, these differences are not as large as those associated with poverty, which apply with nearly equal force to both sexes.24 22These differences by city size are usually statistically significant they are significant in 46 of 58 surveys for enrollment at ages 9-10 and in 50 of 58 surveys for those aged 15-16. 23 The contrast between urban poor children and urban nonpoor children is almost always statisti- cally significant, being so in 53 of the 57 DHS surveys in which the test could be performed. 24Statistical significance tests were applied to the poverty and sex variables in an effort to discrimi- nate among the four categories shown in the table. The results indicate a strong main effect for poverty in all regions, with poor children less likely to enroll. Enrollment rates for poor boys are significantly different from those for nonpoor boys in 52 of 63 surveys; rates for poor girls are statistically different

DIVERSITY AND INEQUALITY TABLE 5-9 Predicted Enrollment for Urban Children Aged 15-16 by Relative Poverty and Child's Sex 191 Poor Poor Nonpoor Nonpoor DHS Surveys Urban Urban Urban Urban in Region Girls Boys Girls Boys North Africa 43.3 51.2 69.7 75.1 Sub-Saharan Africa 37.7 48.2 55.1 68.4 Southeast Asia 64.2 62.1 78.1 82.8 South, Central, West Asia 38.4 49.2 62.2 71.8 Latin America 65.8 66.7 81.2 83.2 TOTAL 45.0 52.9 63.3 72.9 NOTE: See note to Table 5-6. Estimates based on probit models with controls for city size, the child's age, household poverty status, and their interaction in urban areas. In summary, there can be little doubt of an urban advantage in children's school enrollment. The source of this advantage is less clear. Family background must have a strong influence; we saw earlier that the average level of adult school- ing is higher in urban areas, and this is surely associated with higher levels of children's schooling. In addition, urban residence can be associated with higher enrollment because urban areas are better equipped with schools than rural ar- eas. At the primary school level, the urban/rural differences in the availability of schools are far less important than they once were, although substantial dif- ferences remain in many countries. However, at the middle and secondary school levels, and certainly where tertiary schooling is concerned, access to schools is decidedly greater in the urban areas of many, and perhaps most countries. As dis- cussed in Chapter 2, the presence of more-educated adults in cities may present urban parents with more role models exemplifying the implications of human cap- ital investments, as well as more opportunities for social learning. These urban social interactions could well enhance the perceived returns to children's school- ing. Better-educated adults are not distributed uniformly across the urban space, however; they are less likely to be encountered in city neighborhoods of concen- trated disadvantage. Street Children Street children must represent an extreme in terms of urban disadvantage. A1- though they are a highly visible presence in many developing-country cities, little is known of the total numbers of such children, their characteristics and origins, and the long-term consequences of their life on the street. Researchers agree that from those for nonpoor boys in 54 of 63 surveys. The enrollment difference by sex is more prominent among the nonpoor than among the poor. Among nonpoor boys and girls, the male-female enrollment difference is significant in 43 of 63 surveys. Among the poor, however, it is more difficult to distinguish enrollment rates for poor girls from those for poor boys; the contrast is significant in 26 of 63 surveys.

192 CITIES TRANSFORMED two characteristics of street children must be taken into account where they sleep and the amount of family contact they have but have not yet reached a consen- sus on matters of sampling and measurement. This lack of agreement on methods hinders generalization and leaves room for substantial imprecision in determining who is and who is not a street child. UNICEF distinguishes among three types of children: children of the street, who live and sleep on the street; children on the street, who spend most of their time on the street with little adult supervision but sleep at home; and a broader category of which these two are subsets, denoted children at high risk, consisting of those who live in absolute poverty with little adult supervision (tusk, 1989~. This three-part classification has been widely adopted by researchers and policy makers and helps clarify important differences among children. Unfortunately, the implementation of these concepts in research protocols has been inconsistent. Further methodological difficulties arise from the transient and elusive nature of street children's lives their circulation among shelters, homes, and the streets- which makes them very difficult to study. The result is that researchers often adopt different definitions of street children and employ different sampling strategies. Research on street children remains largely descriptive, based on small, lo- calized studies of doubtful generalizability. Few studies make use of sampling techniques that permit statistical tests or introduce controls for confounding fac- tors. Likewise, few studies make use of comparison groups, either comparing street children from different neighborhoods or comparing them with non-street children. The lack of well-specified comparison groups makes it difficult to situ- ate street children in a larger context and to separate the risks and consequences of life on the street from the more general disadvantages of poverty (for exceptions, see Gross, Landried, and Herman, 1996; Panter-Brick, Todd, and Baker, 1996~. Researchers agree that the number of street children is all but impossible to assess with rigor. Obstacles are presented by definitional problems, the mo- bility of the population, the lack of reliable data, widespread use of purposive sampling techniques, and the fact that many street children elude detection or give inaccurate information when interviewed (Aptekar,1994~. Nevertheless, the available evidence suggests that the number of street children is large and may well be growing. An often-cited 1989 estimate by UNICEF gives a global total of 100 million street children (Barker and Knaul, l991~. This total includes children of the street, as well as those who work on the street. An estimated 40 million of these children live in Latin America, some 25 to 30 million in Asia, and 10 million in Africa (Barker and Knaul, l991~. For children living and sleeping on the street, the estimated total is about 10 million. Demographic profile Over half of the street children in most studies report sleeping at home, which in the UNICEF classification defines them to be children "on" the street. Among those who sleep away from home, most give the street as their residence, though

DIVERSITY AND INEQUALITY 193 others report staying in shelters or with friends. Even among children who do not sleep at home, many report that they maintain some contact with their families, often remitting a portion of their earnings to parents or other family members. The spells of time children spend on the street vary greatly in length, but it appears that most street children, particularly those "of" the street, spend several years in this situation. The majority of street children are boys. Girls also work and live on the street, but when children of the street are considered, the great majority are boys. This imbalance is probably due to gender differences in the socialization of chil- dren, different opportunities for work, the greater vulnerability of girls to physical and sexual assault, and the better protection they sometimes receive (Aptekar and Ciano-Federoff, 1 999; Martins and Ebrahim, 1 995 ). Most street children are in their early adolescence, with the modal age being about 13 years. Children of the street have a mean age of 15-16; they are slightly older on average than children who are on the street. Studies often define the upper age limit of street children to be 18 years; when they pass this threshold, street children are redefined as working or homeless adults. Just over half of street children either live with or come from intact birth fami- lies; the next most common family background includes single-mother families and families containing partners of the child's parents (i.e., stepparents or the equivalent). Contrary to what might be supposed, relatively few street children are orphans, with the percentage of orphans being higher among children of the street than among those on the street. Most street children have attended school at some point, although few have gone so far as to complete primary school, and of course very few remain in school. It is difficult to assess their literacy. Despite a lack of schooling, many street children are numerate, and many are sufficiently entrepreneurial to have acquired critical economic and survival skills. Life on the street and its consequences There can be no doubt that life is harsh and brutalizing for children who live and work on the street. They are exposed to pollution, disease, harassment, abuse, and violence. They generally earn meager wages and are often forced into exploitative or dangerous occupations. They face a constant challenge in finding food, toilet and bathing facilities, and a protected place to sleep. Because of their poverty and exposed living conditions, street children also tend to have many health afflic- tions. Infectious diseases, particularly respiratory illnesses and skin conditions, are pervasive in this population (Mejia-Soto, 1998; Senanayake, Ranasinghe, and Balasuriya, 1998; Wright, Kaminsky, and Wittig, 1993~. Moreover, street chil- dren lack access to health care; some are suspicious of the health care system and actively resist treatment (Reddy,1992~. Street children also engage in activities that put them at risk for other health problems. Alcohol and drug use is widespread in this population; many of the

194 CITIES TRANSFORMED children use inhalants (mostly glue) and tobacco (Campos, Raffaelli, and Ude, 1994; Forster, Tannhauser, and Barros, 1996; Wittig, Wright, and Kaminsky,1997~. Street children also have sexual intercourse at younger ages (whether willingly or as the consequence of coercion), and they tend to engage in riskier sexual be- haviors (e.g., prostitution, multiple partners, infrequent condom use) than those of non-street children of similar ages (Eisenstein and de Aquino, 1992; Swart- Kruger and Richter, 1997~. These behaviors are much more prevalent in children who are of the street. Only a handful of studies have compared street children with other poor and nonpoor children of similar ages. Although the nutritional status of street children can be a problem, they are often found to be healthier than poor children who are not on the street (Gross, Landried, and Herman, 1996; Panter-Brick, Todd, and Baker, 1996~. When they can, street children make use of their varied repertoire of survival and entrepreneurial skills, and their own social networks, to meet basic needs. Although many street children are at great risk for a variety of health problems, at least one study finds that street children in Chandigarh, a medium- sized city in India, learn a variety of useful life skills and coping mechanisms from their work and play in the streets. Their developmental maturity in terms of survival skills is high, but other measures of physical and mental health give cause for concern, and of course the future opportunities for street children are likely to be severely limited (Verma, l999~. Origins and causes Life on the street is the outcome of many factors, some of these being society- wide and others the result of family and personal circumstances. Given the thin research base, researchers can do no more at present than to speculate about the most important causes. There is agreement that many societal factors increasing urbanization and migration, economic recession, civil unrest, conflict, famine, high levels of HIV/AIDS contribute, intensifying poverty and disrupting house- hold structure (Aptekar, 1994; Barker and Knaul, 1991; Martins and Ebrahim, 1995; Rizzini, 1998~. Likewise, researchers agree that there are many factors involved at the family level poverty, family conflict and dissolution, and both physical and sexual abuse. Street children themselves most often cite poverty or the need to find work as their motivation for leaving home, although family dis- ruptions and conflicts, abandonment by or death of parents, and a desire for "street life" also come into play. Many researchers describe street life as the product of combinations of factors (Aptekar and Ciano-Federoff, 1999; Matchinda, 1999), but little is known of the critical stress points and thresholds. Interventions Policy makers, human rights activists, and NGOs are paying growing attention to street children. In governmental policies and programs for children and families,

DIVERSITY AND INEQUALITY 195 street children are increasingly identified as a group with special needs (tusk, 1989; Agrawal, 1999~. At one time, institutionalization and rehabilitation efforts were proposed as the main interventions, but emphasis is now being given to pro- grams focused on basic needs, skill development and training, counseling, and related services. Many current programs engage volunteers or "street educators" who enter the communities of street children to provide them with assistance and services. The change in intervention strategies reflects a shift in the conceptualization of street children. If formerly they were viewed as a "deviant population," they are now being seen as an "at-risk" population. Multisector and multilevel interven- tions, particularly at the community level, are being promoted, and the emphasis is moving, albeit gradually, to prevention efforts aimed at combating poverty and strengthening families. Programs are beginning to target the aspects of street life that are particularly risky, including HIV/AIDS prevention, drug and alcohol in- tervention, and prevention of violence (Crane and Carswell, 1992; World Health Organization, 1997~. Some researchers argue that, when considered in relation to their numbers, street children may now be receiving disproportionate attention compared with other disadvantaged children (MacArthur, 1993~. The fact that the health of street children is often no worse than that of other poor children is compelling testi- mony to the many disadvantages of poverty from which all poor children suffer. Yet street children are undoubtedly seriously disadvantaged, and their elusiveness means that many of them are neither offered nor receive the services they are due. CONCLUSIONS AND RECOMMENDATIONS If within the compass of urban life one finds both street children and the gated communities of the rich, inequality and diversity must be among its defining features. This chapter has explored intraurban and interurban differences in social and economic characteristics. By training an urban lens on human capital invest- ment, poverty and well-being, access to basic services, risk and vulnerability, and the lives of children, we have sought a better understanding of urban/rural differ- ences and intraurban diversity. Our main findings can be summarized in broad strokes: in the dimensions analyzed here, urban residents are better off on aver- age than rural residents; residents of smaller cities are generally disadvantaged by comparison with those of larger cities, although advantaged by comparison with rural villagers; and the urban poor suffer from deprivations that can sometimes leave them no better off than rural residents, but generally situate them between rural residents and the urban nonpoor. In contemplating the rich array of concepts and methods being applied to the study of socioeconomic diversity in the cities of rich countries (Chapter 2), we are struck by the promise they hold for understanding the cities of poor countries. The concepts of neighborhood and social capital are well recognized in the literature

196 CITIES TRANSFORMED on poor countries and are illustrated by many vivid urban examples. Less has been seen of social network analysis, and yet this perspective might fruitfully be applied to the study of job search in cities, to health-seeking behavior, and even to the survival strategies of street children. Although to date the analysis of neighborhoods and social capital has been conducted mainly through powerful and evocative case studies, we see no reason why statistical tools could not also be brought to bear once neighborhood data become available. Conclusions Access to public services The provision of basic services is much better in cities than in rural areas, but smaller cities are less well served than larger ones. The urban poor have signif- icantly less access to basic services than other urban residents. Analyses of the DHS surveys reveal that in almost every country surveyed, the average urban res- ident enjoys better access to basic public services piped drinking water, flush toilets, and electricity than the average rural resident. This finding is not sur- prising given differences in abilities to pay for services, government investment priorities, and (possibly) lower urban unit costs of service provision. Within and between urban areas, however, there are substantial differences in levels of basic service provision. Smaller urban areas especially those under 100,000 population are underserved by comparison with larger cities in all re- gions. With regard to intraurban differences, the urban poor are significantly ill served relative to other urban residents. The effects of urban poverty are strongly corroborated by microstudies of city slums and city-level reviews of service pro- vision. In the DHS survey data, which generally record migrant status only for women of reproductive age, there is surprisingly little evidence that households with recent migrants are disadvantaged in terms of service access. Human capital Urban educational levels are higher on average than rural levels, but the urban educational spectrum is also more diverse. Educational levels are higher on av- erage in larger than in smaller cities, but there is substantial diversity in cities of all sizes. It is to be expected that cities will have higher average levels of ed- ucational attainment than rural areas. But cities also exhibit greater educational diversity. Residents of larger cities (particularly cities of 1 million or more pop- ulation) have higher average levels of schooling compared with their counterparts in smaller urban centers. Important theories suggest that educational diversity can have beneficial social and economic effects, although these theories have not yet been tested in the cities of developing countries. Higher percentages of urban than rural children are enrolled in school, and enrollments are somewhat higher in larger than in smaller cities. In addition, urban poor children are much less likely than other urban children to be enrolled.

DIVERSITY AND INEQUALITY 197 There is a decided urban advantage in children's school enrollment, which is likely attributable to differences in family background and access to schools. Although smaller cities have lower enrollment rates than the largest cities, these differences are not especially great except in sub-Saharan Africa. Poor urban children are also much less likely to be enrolled in school than other urban children, and in some regions (notably in sub-Saharan Africa), they are hardly more likely to be enrolled than children in the countryside. Poverty and well-being Urban poverty is being conceptualized in terms of multiple dimensions, many of which are not summarized by income and assets. Considering the income dimen- sion alone, the poverty lines currently being used in many developing countries appear to need substantial upward revision. The research literature increasingly points to a variety of dimensions of urban poverty, and yet the official methods used to measure poverty continue to be simplistic and one-dimensional. Other di- mensions should be considered, including shelter, access to public infrastructure and other basic services, safety nets, protection of rights, time costs, and politi- cal voice. Failure to recognize the multiple dimensions of poverty can skew un- derstanding of its causes and needlessly narrow the scope of poverty alleviation efforts. In addition, the official methods used to establish income-based poverty lines often fail to account adequately for locational differences in prices and the high proportion of income that many of the urban poor must spend on nonfood es- sentials, especially housing. Urban poverty may well be underestimated in many countries because of these methodological deficiencies. We cannot say, however, that urban poverty is underestimated relative to rural poverty, because formidable methodological and empirical problems prevent direct comparisons. Because they lack significant financial assets and are dependent on cash in- comes, the urban poor are left vulnerable to risks associated with economic shocks, political and social crises, and environmental hazards and disasters. Although low-income households are resilient and employ many coping mechanisms to ad- just, some are harder hit than others and less able to adapt. The loss of assets, homes, and primary income earners that comes in the wake of disasters and shocks can cause poor families to supply more labor, reduce or change consumption pat- terns, borrow, sell household assets, or even resort to prostitution or theft. Poor households are likely to send children to work, cut back on medical care, or restrict food consumption before turning to others for credit or charity. The consequences can be dire; many street children cite poverty and the need to earn money for their families as their main reason for living on the street. Recommendations The analysis presented in this chapter indicates several directions for policy and research in the areas of service delivery, data collection, and a research agenda

198 CITIES TRANSFORMED on intraurban and interurban differences in social and economic well-being in low-income countries. Service delivery In the area of service delivery, we single out several elements of an agenda for basic public services, including water supply, sanitation, electricity, and education: . Reach the urban poor · Improve services in smaller cities · Increase the school enrollment rates of urban poor children; · Create or strengthen social safety nets Data collection Where data collection is concerned, there is an urgent need for collating of available data on socioeconomic conditions within cities, with a particular focus on city neighborhoods and subdistricts. New data collec- tion efforts should also be encouraged, particularly data on access to ser- vices, income and assets, the multiple dimensions of poverty, and education that are comparable among and within cities. Surveys such as the DHS can make an important contribution, especially if measures of the reliability and adequacy of basic services (water supply, electricity, sanitation) can be enriched and made sensitive to urban circumstances. Community-level sur- veys also have a useful role to play. It is critical that these data be collected, and also that they be disseminated widely to policy makers and program managers at the national and local levels. It is disappointing that national statistical offices, which appear to be in the best position to supply spatially disaggregated data on their populations, have seldom done so in the past. The published data from national sources are notably weak in spatial terms. The importance of urban areas must be brought to the attention of the national statistical agencies, and they must bring their resources to bear by supplying adequate local data. Countries need to increase the availability of disaggregated social and economic data by social class, gender, age, and local area to inform policy makers and planners at the local level. These intraurban data-gathering mechanisms are needed to better understand the socioeconomic aspects of spatially con- centrated disadvantage, as well as the extent and nature of urban "slums," for which there is no current agreement on a generally accepted definition. Research Research is also needed on how to define and measure the multiple dimensions of poverty, how to identify vulnerable groups (with attention to migrants), and how to create linkages from poor urban communities to their governments and external sources of funds.

Next: 6. Fertility and Reproductive Health »
Cities Transformed: Demographic Change and Its Implications in the Developing World Get This Book
×
Buy Hardback | $80.00 Buy Ebook | $64.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Virtually all of the growth in the world’s population for the foreseeable future will take place in the cities and towns of the developing world. Over the next twenty years, most developing countries will for the first time become more urban than rural. The benefits from urbanization cannot be overlooked, but the speed and sheer scale of this transformation present many challenges. A new cast of policy makers is emerging to take up the many responsibilities of urban governance—as many national governments decentralize and devolve their functions, programs in poverty, health, education, and public services are increasingly being deposited in the hands of untested municipal and regional governments. Demographers have been surprisingly slow to devote attention to the implications of the urban transformation.

Drawing from a wide variety of data sources, many of them previously inaccessible, Cities Transformed explores the implications of various urban contexts for marriage, fertility, health, schooling, and children’s lives. It should be of interest to all involved in city-level research, policy, planning, and investment decisions.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!