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.
Why Location Matters This chapter explores the ways in which urban environments can influence demo- graphic behavior. Although we refer to empirical findings, our main purpose is to provide a map of concepts, with emphasis on the features that give the urban socioeconomic landscape its distinctive character. The discussion begins at the micro level, examining how neighborhoods might affect individual and family de- mographic decisions. It then moves to successively higher levels of aggregation, surveying the linkages that cross neighborhoods, taking in the broader urban econ- omy, exploring connections among cities, and finally examining the structures of government that are overlaid upon this varied terrain. The themes developed here have international extensions, but this chapter remains within national boundaries. In closing, the concepts that have entered the discussion are reviewed, and a few of the major features that distinguish urban from rural landscapes are identified. Although the discussion in this chapter traverses a great range of contexts, it is guided by a mere handful of concepts: diversity, proximity, externality, network, and centrality. As they are assembled in different configurations, these concepts present a series of urban frameworks through which demographic behavior can be viewed. Urban frames do not always offer novel views of demographic phenom- ena, but they often provide a fresh perspective. At the center of our argument is a proposition so unexceptionable as to be banal: individuals and families demographic decision makers are located or embedded in social contexts that determine what information is available to them and influence the decisions based upon that information. Health, fertility, and human capital investment decisions are influenced by multiple social contexts; so, too, are decisions about job search, migration, and labor force participation. The panel regards these demographic decisions as being inherently multilevel in nature, and in considering the individual-to-group links, emphasizes the roles of urban social interaction, feedback, and diffusion. 29
30 CITIES TRANSFORMED Had demographers but taken their cue from the Taichung experiment, we might now be in a position to summarize 30 years of intraurban multilevel re- search. That randomized intervention its elements were described by Freedman and Takeshita (1969: 109 48) began in 1963, when Taichung was a smallish Taiwanese city of some 325,000 inhabitants. The experimental design exploited the city's neighborhood structure, allocating 2,400 small neighborhoods, or fin, to treatment and control groups. The treatment took the form of provision of in- formation about the intrauterine device (IUD), which was then a relatively new method of family planning. A key question in the analysis was whether such fam- ily planning information "spilled across" the boundaries of the treatment lin to benefit women in adjacent control lin. Remarkably strong evidence of such informational spillovers emerged, and women's social networks were identified as the main mechanism by which news of the IUD was conveyed from the treatment women to those who had not heard of this method firsthand. A number of women outside Taichung proper were found to have learned of the IUD through their social network ties to city residents. These network contacts evidently had the effect of amplifying the program ef- forts undertaken in the treatment fin, acting as "social multipliers" through which information could be diffused across space and socioeconomic strata. The Taichung case introduces several themes that figure prominently in this report. Informational externalities are one such theme. Neighborhood effects are another, as are the conceptual distinctions between neighborhood and social net- work. The Taichung experiment revealed urban/rural linkages that were stronger than expected, thus calling into question the sociological meaning of the city boundaries. It also showed how programs that must operate in specific neighbor- hoods small fin, in this instance can exert influence beyond those local spaces. Had research only continued along these lines, a new "Chicago School" tradi- tion of detailed spatial and social analyses of developing-country cities might well have developed, with demographic behavior as one focus. But as it happened, the urban themes of the Taichung experiment were left dangling, and the demographic literature took up entirely different lines of inquiry. By the late 1970s and 1980s, this literature had come to be dominated by individualistic models of behavior. In- terest in contextual effects did not disappear, of course, and multilevel methods of analysis continue to be refined) In developing countries, however, such methods have seldom been matched to data on urban neighborhoods and local contexts. A continuing difficulty to which we return in Chapter 5 is that in these countries, census data are rarely processed at the level of local areal units. Lacking census data, researchers interested in multilevel approaches have often found themselves iSee, among others, Rosenzweig and Schultz (1982), Casterline (1985a,b), Tsui (1985), Entwisle, Casterline, and Sayed (1989), Entwisle, Rindfuss, Guilkey, Chamratrithirong, Curran, and Sawangdee (1996), Bilsborrow and Anker (1993), Pebley, Goldman, and Rodriguez (1996), Sastry (1996), Degraff, Bilsborrow, and Guilkey (1997), Axinn, Barber, and Ghimire (1997), Axinn and Yabiku (2001), and Mroz, Bollen, Speizer, and Mancini (1999).
WHY LOCATION MATTERS 31 restricted to rural sites, where sample surveys can be used to inventory the local environment. For these reasons, the demographic literature on developing-country cities has never achieved the sophistication and keen appreciation of context that is seen in studies of Chicago (Sampson and Raudenbush, 1999), Oakland (Fischer, 1982), or Glasgow (Garner and Raudenbush, 1991~. As we take up the issue of why location matters in Africa, Asia, and Latin America, we are therefore uncomfort- ably dependent on concepts and findings developed with other settings in mind. We shall, nevertheless, borrow wholesale from these developed-country studies, hoping to bring conceptual parallels and partial analogies to light. PLACES, NETWORKS, NEIGHBORHOODS Urban neighborhoods can be viewed as spatial units that might or might not have significant effects on demographic behavior. To clarify what form neigh- borhood effects could take, we must first separate the concept of place from that of community. As used here, place is a spatial concept, whereas community is a social concept, having to do with individual and group identities, senses of belong- ing, and the presumption of mutual interests and shared values.2 A neighborhood might be defined as a type of community composed of spatially proximate indi- viduals. When the discussion focuses on social capital, social learning, and other mechanisms through which neighborhood effects can be expressed, this commu- nity aspect of neighborhoods comes to the fore. But there may well be effects at- tributable to the local social-spatial environment that are due to the very lack of place-based community ties. In settings where local residents mistrust one another and recognize few shared interests and common values, they may see elements of social and physical risk in the local environment and behave accordingly. Hence, the concept of neighborhood effects encompasses two rather different influences on individual behavior those stemming from local social ties and those due to their absence. The identification of neighborhood with community has gone in and out of fashion in urban sociological research. In the early literature, it was argued that cities had once been home to coherent, functional "natural neighborhoods," or "ur- ban villages," which were ethnic communities akin to rural villages. The forces of modernization were said to have swept away many of these local, place-based social relationships and dispersed their functions among a variety of urban institu- tions. Wirth (1938: 20-21) famously depicted the process as entailing a "substitu- tion of secondary for primary contacts, the weakening of bonds of kinship, and the declining social significance of the family, the disappearance of the neighborhood, 2 Different disciplines attach quite different meanings to the term "place." To geographers (e.g., Committee on Identifying Data Needs for Place-Based Decision Making, 2002: 55), place is an "en- semble concept" that encompasses both spatial and social elements and processes. See also Harvey (1973), Gregory and Urry (1985), Wolch and Dear (1989), and Golledge and Stimson (1997).
32 CITIES TRANSFORMED and the undermining of the traditional basis of social solidarity." In the urban way of life or so it then appeared place-based social ties were replaced by aspatial social relationships or by no relationships at all. When empirical methods were brought to bear on such views, they were found to be simplistic and even misleading. As Chaskin (1994: 12) notes in reference to the bygone era of functional, natural urban neighborhoods, "There is no evidence that such a golden age ever existed." Sociologists began to examine broader con- ceptions of community and to conceive of individuals as participating in several sorts of communities at once, some of which are spatially grounded and others not. Social networks including both personal social networks and those formed through formal and informal associations began to be recognized as a key link- ing mechanism. Through such networks, any given person might have ties to spa- tially proximate and spatially distant partners. The idea of "communities without propinquity" (a phrase due to Webber, 1963) emerged, and sociologists began to think of individuals as being attached to each of their communities on a voluntary and contingent basis (Chaskin,1994~. Empirical studies in the United States showed that urban residence is not nec- essarily associated with weakened personal ties and attenuated senses of commu- nity; rather, it affects the types of ties and communities in which people participate. Sampson and Morenoff (2000: 374) summarize the empirical record in this way: . . . contrary to the popular belief that metropolitan life has led inex- orably to the decline of personal ties, sociological research has shown that while urbanites may be exposed to more unconventionality and diversity, they retain a set of personal support networks just like their suburban and rural counterparts (see e.g., Fischer 1982~. Fischer (1982: 264) puts it succinctly: "Urbanism does not seem to weaken com- munity, but it does seem to help sustain a plurality of communities." If the social networks of urban residents contain sufficient links to other spa- tially proximate individuals, a basis exists for thinking of geographic neighbor- hoods as communities. Some social network researchers question whether mod- ern urban networks are indeed this localized.3 As Wellman and Leighton (1979: 365-67) argue in a memorable passage: To sociologists, unlike geographers, spatial distributions are not in- herently important variables, but assume importance only as they af- fect such social structural questions as the formation of interpersonal networks and the flow of resources through such networks ... the 3Social network researchers maintain that their methods provide excellent tools for assessing the relative frequency and strength of place-based ties. For instance, Wellman and Leighton (1979: 365- 67) write: "By leaving the matter of spatial distributions initially open, [the network perspective] makes it equally as possible to discover an 'urban village' (Gans, 1962) as it is to discover a 'commu- nity without propinquity' (Webber, 1963).... With this approach we are then better able to assess the position of neighborhood ties within the context of overall structures of social relationships."
WHY LOCATION MATTERS identification of a neighborhood as a container for communal ties assumes the a priori organizing power of space. This is spatial determinism. 33 The empirical record confirms that U.S. urban residents are connected to a variety of networks extending outside their neighborhoods. But it also shows that they continue to value their neighborhood ties, making use of them for day-to-day so- cializing and drawing upon them for social support. Evidently, some strong and intimate ties not all, to be sure remain grounded in neighborhoods.4 If neighborhoods are subsets of networks, how are their spatial boundaries to be delineated? Residents of poor U.S. communities often disagree as to the dimen- sions of the neighborhood even residents of the same household can see things differently (Furstenberg and Hughes, 1997: 34~. The perspectives of adults and youth can diverge, as can the views of younger and older children. For adoles- cents, "neighborhoods of sociability" may be at least as important as those de- fined according to residence (Burton, Price-Spratlen, and Spencer, 1997: 135~. In a study of Los Angeles, Sastry, Pebley, and Zonta (2002) find that better-educated respondents conceive of their neighborhoods in broader spatial terms than do those who are less educated. Recent immigrants to Los Angeles hold spatially con- stricted views of their neighborhoods. Hence, even where local space is acknowl- edged to matter, the perceived perimeters of that space can be expected to vary (Committee on Identifying Data Needs for Place-Based Decision Making, 2002~. This social construction of neighborhood is not a merely a subjective matter it may well have an influence on the use of public services and, through ser- vices, on demographic behavior. For instance, neighborhood residents may view a nearby health or family planning clinic as being inaccessible if it happens to be situated beyond a socially defined neighborhood boundary. Or, where privacy is deemed essential as it is often thought to be for adolescents such services may deliberately be sought outside the bounds of the neighborhood. Is there any reason to question the social significance of neighborhood in the cities of poor countries? To the extent that city residents face higher costs of transport and information exchange than their counterparts in rich countries, local social space would be expected to assume greater importance. For instance, in a study of two poor neighborhoods in Santiago, Espinoza (1999) finds that nearly three-quarters of residents' personal network partners live within walking 4 acknowledging this, Wellman and Leighton (1979: 385) nevertheless express skepticism about the relative importance of the local ties: "Neighborhood relationships persist but only as specialized components of the overall primary networks.... if we broaden our field of view to include other primary relations, then the apparent neighborhood solidarities may now be seen as clusters in the rather sparse, loosely bounded structures of urbanites' total networks." Geographers (see Committee on Identifying Data Needs for Place-Based Decision Making, 2002) argue that people can inhabit multiple "places" at the same time, some of these at local spatial scales, such as neighborhoods, and others at larger scales. The spatially distant connections found in social networks serve to define some of the larger-scale places.
34 CITIES TRANSFORMED distance. For these poor Chileans, distance is a constraining factor because of the high cost of transport in relation to their resources. For the United States, Fischer (1982: 251) shows that the social networks of the poorly educated are more spa- tially concentrated than the networks of the better educated. Also, the social net- works of women are believed to be more localized than those of men, containing higher proportions of neighbors and kin (see Moore, 1990, and Stoloff, Glanville, and Bienenstock, 1999, as well as the extensive references cited therein). The panel is not aware of comparable social network research in the cities of low- income countries, but would expect that in these settings, too, the networks of women and the poor would tend to be disproportionately local. Because so much of family demographic behavior depends on the information and resources held by women, differences in the composition of their personal networks can have important demographic implications. Many activities undertaken by governments and nongovernmental organiza- tions (NGOs) are spatially organized and will surely remain so. These activi- ties confer additional social meaning upon local space. Public services water supply, electricity, sanitation are of course spatially grounded. However hazy the boundaries of neighborhoods in the eyes of local residents, city planners are obliged to delineate them to arrange for the delivery of such services (White, 1987: 1-6~. Politics and political access likewise have their territorial aspects. For these reasons, neighborhoods are often taken to be the natural units for program interventions, whether on the part of NGOs, governments, or interna- tional agencies. The popularity of neighborhood-based programs testifies to a widespread belief in the organizing potential of spatial proximity. Chaskin (1994: 32-44) identifies four strands of thinking that have tended to orient program inter- ventions to neighborhoods: the increasing value being placed on the concepts of local empowerment, control, and responsiveness; the need to operate pilot projects and experimental interventions on a manageable scale; the recognition that effec- tive interventions must often have a comprehensive character, which is difficult to achieve without some spatial concentration of effort; and the expectation that important target populations are themselves spatially concentrated. The grow- ing influence of these ideas is evident in many developing countries, and is being expressed in the decentralization of government functions and the Revolution of governmental authority. The twin processes of decentralization and devolution- discussed later in this chapter and in full detail in Chapter 9 are causing govern- ments to be resealed into spatial units that can approximate clusters of neighbor- hoods. All this suggests that urban neighborhoods are likely to retain a good deal of social meaning in low-income countries. Neighborhoods and Demographic Behavior: Theory This section explores the implications of neighborhood contexts for two types of demographic behavior: first, investments in "child quality," which have to do with
WHY LOCATION MATTERS 35 children's schooling and the adoption of time-intensive modes of child care on the part of parents; and second, household knowledge of health care and the nature of health outcomes. For well over a century, specialists in public health have understood that the spatial proximity and intermixing of diverse city populations must amplify the risks of contagious disease. The resources shared within neighborhoods aneigh- borhood well in the famous health puzzle solved by John Snow (1855) present opportunities for social interaction that can have profound epidemiological con- sequences. An appreciation for the neighborhood aspects of child quality invest- ments, however, is comparatively recent. Some of the theories reviewed below stress the acquisition of information about the labor market and the economic returns to children's schooling; others address the time and opportunity costs of child rearing; and still others consider the location of services and institutions. To the panel's knowledge, none of these theories has been properly tested in low-income countries, but we comment on the features that would appear most salient.5 Social learning via social networks Theories of social learning draw attention to the information that is exchanged through peer groups and personal social networks. Such individual-to-group link- ages are a prominent feature of models in many social science disciplines, includ- ing anthropology, sociology, economics, cognitive psychology, and the communi- cation sciences. The common thread is this: In situations of flux and uncertainty, when new choices and strategies are being debated, people naturally look to their reference groups and role models to understand the benefits, costs, and uncertain- ties of these new choices. In the demographic realm, social learning can prompt a rethinking of broad family strategies, or it can be narrowly focused on new tools and behavioral options.6 Social learning about education provides one example of the transformation of family strategies. If left to their own perceptual devices, adolescents and their parents would probably have only the haziest sense of the economic returns to schooling. In diverse urban settings, however, they can gain a keener appreci- ation of these returns by observing local adult role models and reference groups (Wilson, 1987; Borjas, 1995~. Professional and middle-class adults exemplify dis- tinctive life-course strategies; in so doing, they help others understand the implica- tions of educational investments and decision points. If social learning enhances 5This review closely follows the presentation of Gephart (1997: 6-7), who draws in turn from Jencks and Mayer (1990). 6A large literature in sociology, geography, and economics focuses on diffusion and social learning processes see Montgomery and Casterline (1993, 1996) and National Research Council (2001) for extensive reviews with attention to demographic implications. Although the behavioral mechanisms suggested are often plausible, this literature presents few rigorous empirical tests of their significance. We revisit this point later in the chapter.
36 CITIES TRANSFORMED the perceived returns to schooling (it can also raise concerns about educational costs and risks), urban families can be led to make deeper investments in their children's schooling and to forego traditional strategies of high fertility. Social learning about the tools of fertility control modern contraceptives- was central to the success of the Taichung family planning experiment discussed above (Palmore and Freedman, 1969~. The learning mechanisms uncovered in that early experiment are being revisited in models of social networks and the diffusion of modern contraceptive use in sub-Saharan Africa. Montgomery and Casterline (1993, 1996) develop the theory with reference to contraceptive use, and strong confirmation is found in empirical estimates made by Behrman, Kohler, and Watkins (2001) for Kenya and Casterline, Montgomery, Agyeman, Aglobitse, and Kiros (2001) for Ghana. These findings based on longitudinal designs with repeated measures of networks show how contraceptive use (and knowledge of AIDS) can be spread within localized social networks, diffusing as if by force of example. Although these empirical studies have focused on diffusion in either rural (Behrman, Kohler, and Watkins, 2001) or rural and periurban (Casterline, Montgomery, Agyeman, Aglobitse, and Kiros, 2001) networks, there is no reason to expect the effects to be limited to these contexts. Indeed, as the Taichung exper- iment strongly suggests, the socioeconomic diversity of cities and the heterogene- ity of the information that circulates within them probably enhance the prospects for informational spillovers. But surprisingly little is known about urban/rural differences in the composition and function of social networks. Beggs, Haines, and Hurlbert (1996) find that in the United States, rural social network ties are "stronger" in the sense of involving greater intimacy, more frequent contact, and longer duration; that rural networks are more homogeneous (especially in terms of religion) and smaller than urban networks; and that they are more dominated by kin. We know of no comparisons of this sort for developing countries.7 Clustering, common resources, and contagion In describing social learning, some researchers refer to the "contagiousness" of ideas and social examples, and make use of mathematical models drawn from epi- demiology to trace out the implications (e.g., Rosero-Bixby and Casterline,1993~. If models of social contagion are still relatively new and untested, models of bio- logical contagion are by now well established. In epidemiology it is understood that, with other things being equal, the risks of disease transmission among spa- tially proximate urban populations must be higher than the risks facing dispersed rural populations. Clustering obviously affects the likelihood of person-to-person 7 Mitchell (1969) offers many clues as to the nature of urban social networks in southern Africa in the 1950s, but the studies collected here are on such a small scale that they might almost be regarded as anecdotal. More recent evidence on urban social networks in Africa can be found in Tostensen, Tvedten, and Vaa (2001).
WHY LOCATION MATTERS 37 contagion, whether due to airborne or sexually transmitted disease. Also, because concentrated urban populations share certain resources (notably water), the san- itary practices of one group can generate externalities that affect the health of another. As a result, urban populations start from a position of health disadvantage relative to rural populations. Where this disadvantage is erased or reversed, one looks for the cause in several areas: in the investments and regulation undertaken by governments to improve water supply and sanitation, in the provision of both public- and private-sector health services, and in the economic factors that supply urban residents with the means to purchase better health care in private markets (see, among others, Preston and van de Walle, 1978; Ewbank and Preston, 1990; Preston and Haines, 1991; van Poppel and van der Heijden, 1997~. As discussed below and in Chapter 7, neighborhood health externalities are not wholly a local affair they also reflect actions taken by the public and private institutions that transcend neighborhoods. Collective and institutional socialization Theories of collective socialization also highlight the linkages from individuals to groups, but these theories emphasize a form of group influence that is dis- tinct from social learning and contagion. The mechanisms of collective social- ization are powerfully described by Wilson (1987) and Coleman (1988), whose work generated a resurgence of interest in neighborhood effects in the cities of the United States. Their research focuses attention on the role of adults residing in the neighborhood, who can supplement the child-rearing efforts of parents by acting as extraparental sources of authority and social control. Directly and by exam- ple, these neighborhood adults can teach the young the boundaries of acceptable behavior. What if city neighborhoods lack such trusted adults? As will be seen in Chapter 6, parents in developing-country cities often complain of the need for extra vigilance in child rearing, describing this as one of the costs of family life that is decidedly higher in cities. This is, perhaps, the other side of the notion that "it takes a village to raise a child": in city neighborhoods where neighbors are distrusted, the burden of child rearing can fall heavily on parents' own shoul- ders. Furstenberg (1993) describes how in poor U.S. neighborhoods, apprehen- sive parents take pains to isolate their children from the surrounding population and thereby shield them from social risk. Girls in such high-risk settings may be especially closely supervised (Furstenberg and Hughes, 1997: 28~. If simi- lar views prevail in the cities of poor countries we strongly suspect that they do the costs entailed in close parental supervision could strongly discourage high fertility. Institutional socialization refers to the nonresident adults who are figures of influence in a neighborhood because they hold positions in local schools, clinics,
38 CITIES TRANSFORMED police departments, or other institutions. In these roles, adults can affect the young both directly and indirectly. If schools in poor neighborhoods are staffed by inferior teachers, the direct effects may be seen in delayed child develop- ment and discouraged human capital investment. Indirect influence can be ex- erted when teachers or health clinicians require the young to adhere to strict standards of comportment. In such ways, nonresident adults can wield influ- ence much like that of the resident adults envisioned in collective socialization theory. Although institutional socialization theory has generally focused on micro- level outcomes for children and adolescents, it is closely linked to the allocation of societal resources to neighborhoods (White, 2001~. Political and economic processes that deliver good schools and high-quality health clinics to some neigh- borhoods while leaving others ill served can affect the young by altering the depth and form of institutional socialization. Social comparisons and subculture conflict The focus of social comparison theory is on the perception of relative deprivation and the possibility that when young people judge their own situation to be rela- tively unfavorable, the reaction may be either to redouble efforts to improve or to abandon these efforts and drop out of the competition. The theory is described by van den Eeden and Huttner (1982: 42-6) as one of comparative reference groups. The specifically urban aspect is the relative ease with which diverse reference groups can be observed in cities as a result of the spatial proximity and socioeco- nomic heterogeneity of urban residents. When frustrated by blocked opportunities, the young may respond by forming subcultures of resistance. Their individual motivations may be the product of social comparisons, as sketched just above, but the emphasis in cultural conflict theory is on how such motivations are voiced and reinforced by groups (Jencks and Mayer, 1990: 116~. Chapter 9 examines urban gangs and violence from this perspective; see Durlauf (1999) on how disaffected urban groups may constitute a "perverse" form of local social capital. The types of social comparison addressed in these theories have not been ex- plored in much demographic research. A rather different form of social com- parison, however, has attracted a modest amount of attention. We refer to the consumption possibilities exhibited in the behavior of upper-income groups and displayed in advertisements and television soap operas. In rural areas, the social chasm between high-income families and the bulk of the population may be wide enough to render the consumption habits of the rich irrelevant to most residents. In economically diverse cities, however, a greater range of social groups may find some modern consumer durables affordable, and as these items are taken up by middle-class households, they may come to be seen as potentially within the reach of the upwardly striving poor. In this form of social comparison, individuals are
WHY LOCATION MATTERS 39 influenced by the consumption patterns exhibited in their reference groups.8 As Freedman (1979) argues, aspirations for modern consumer goods can exert a pow- erful influence on fertility decisions, particularly when the costs of consumption are understood to compete with the costs of childrearing. Services and the physical environment A prominent theme in demographic research is that the services available in local neighborhoods can either complement or act as substitutes for individual and fam- ily resources.9 In the area of health, for instance, it is thought that mothers who are educated are equipped with information of direct relevance to health care, and that education also helps mothers process the information provided by the media and government, and supplies them with the social confidence needed to seek care for themselves and their children (Caldwell, 1979~. Local health services might therefore complement and enhance the positive effects of maternal education, and in this way could increase the health differentials associated with education. But it is also possible that well-functioning services could supply information to women of low education that they might not have been able to acquire by other means. If so, local services could act as substitutes for maternal education, and the presence of services in the community might then reduce the health differentials associated with education. Examining child mortality rates in Brazil, Sastry (1996) conducts an unusually thorough examination of substitution and complementarily between community measures, on the one hand, and mothers' education, on the other. He finds evi- dence of substitution between mothers' education and community sanitation and water supply in Northeast Brazil. In this region, community infrastructure appears to be more beneficial for the survival of children of less-educated mothers than for that of children of the better educated. But as Sastry notes, even for Brazil the lit- erature offers mixed results, and there is reason to think that the substitution and complementarily effects must be highly context-specific, depending on policies, prices, and levels of development. The theories outlined above are concerned with social interactions and or- dering, but the ways in which local space is physically ordered may also have demographic implications. Highways, waterways, and other physical features of neighborhoods can establish barriers, corridors, and niches that shape social inter- actions and distribute risks across space. As Sampson and Morenoff (2000: 379) observe for the United States, ... the ecological placement of bars, liquor stores, strip-mall shop- ping outlets, subway stops, and unsupervised nlav spaces nlav a Thor economic explorations of consumption reference groups, see Alessie and Kapteyn (1991) and Kapteyn, van de Geer, van de Stadt, and Wansbeek (1997). 9Sastry (1996) provides an excellent review of the salient concepts and literature, touching on earlier studies by Rosenzweig and Schultz (1982), Thomas, Strauss, and Henriques (1991), and others.
40 CITIES TRANSFORMED direct role in the distribution of high-risk situations . . . not only do mixed-use neighborhoods offer greater opportunities for expropria- tive crime, they offer increased opportunity for children to congregate outside the home in places conducive to peer-group influence. Sampson and Raudenbush (1999: 622) point out that such physical barriers limit the scope of beneficial adult surveillance and can inhibit children's social interac- tions, blocking possibilities for collective socialization and weakening the capac- ity of residents to organize for common goals. In the literature on neighborhood effects and crime, the so-called broken win- dows theory stresses the dynamic implications of disordered physical environ- ments. According to this theory, the physical state of a neighborhood testifies to its social cohesion and likely vulnerability. Visual evidence of disorder can sug- gest that residents are indifferent to their neighborhood or incapable of defending it, all but inviting the attention of potential criminals. Such visual cues can also discourage outsiders from investing economic resources in a neighborhood and can cause residents to disengage from community affairs (Sampson and Rauden- bush,1999: 604~. Social Capital The preceding discussion has emphasized the role of social networks in neighbor- hood effects, but certain features of these effects are perhaps more readily seen from a social capital perspective. The distinctions between social networks and social capital are imprecise (Lin, 1999) but may be understood as follows. The term "network" is often employed when the focus is on mechanisms of infor- mation exchange and diffusion dynamics, or when specific linkages to resources are being highlighted. The phrase "social capital" is used when networks and local associations are being described as structures that might support collective action, enforce norms, generate expectations of reciprocity, or foster feelings of mutual trust (Kawachi, Kennedy, and Glass, 1999; Putnam, 2000; Sampson and Morenoff,2000~. Because it places emphasis on the more durable features of net- works and assigns prominent roles to associations and institutions, social capital is often invoked in discussions of civil society and governance. Health and social capital In the field of health, social capital effects have drawn considerable attention, especially in relation to mental health. Networks and groups are thought to play two roles: they can offer general economic and emotional support on a sustained basis, and can make specific resources available in times of severe health stress (Kawachi and Berkman, 2001; Lin, Ye, and Ensel, 1999~. As James, Schulz, and van Olphen (2001) note, the spatial proximity of network members may be something of a requirement if they are to provide one another with day-to-day
WHY LOCATION MATTERS 41 BOX 2.1 Social Capital and Mortality Crisis in Russia Social capital is often defined in terms of features of social organization the density of civic associations, levels of interpersonal trust, and norms of reciprocity that act as re- sources for individuals and facilitate collective action. Using household survey data from 40 regions of Russia, Kennedy, Kawachi, and Brainerd (1998) find empirical links between life expectancy at birth and several indicators of social capital mistrust of government, crime, quality of work relations, and civic engagement in politics. In the impoverished Russian institutional environment, people may be forced to rely on family, friends, and other informal sources of support to meet their health needs. Until formal institutions can be rebuilt, those who lack informal means of support may be especially vulnerable. assistance. The point would appear to apply with special force to the cities of low- income countries, where the costs of transport and communication are relatively high. In the empirical analyses of Kawachi, Kennedy, and Glass (1999), two mea- sures of local social capital civic trust and participation in voluntary associations are shown to enhance individuals' assessments of their health. In addition to material resources, these associations can bring order, coherence, and meaning to the daily lives of their members (James, Schulz, and van Olphen, 2001~. Box 2.1 describes empirical findings on the supportive functions of local social capital in Russia. The Russian context is one of difficult transition to a market economy a transition that has been accompanied by a breakdown of formal health care insti- tutions, a widespread decline in life expectancy, and an increased dependence on informal mechanisms of support. Social capital and community dynamics From the viewpoint of social network theorists, social capital is formed by an accretion of individual investments. When individuals establish a new social net- work tie or choose to strengthen an existing one, they contribute to a mass of social capital. If social capital is constructed when individuals thus invest, it likewise depreciates when they choose to disinvest: as individuals neglect their neighbor- hood ties and withdraw from local associations, they weaken the foundations of local capital (Astone, Nathanson, Schoen, and Kim, 1999~. In describing these processes, social capital theorists tend to stress investments that take the form of participation in local voluntary associations and formal institutions (Narayan and Pritchett, 1999~. They sometimes reconceptualize properties of social networks so as to highlight the links to associations. The differences in emphasis can be seen by reconsidering the theory of collec- tive socialization. Social capital is thought to be strengthened by the "intergen- erational closure" of individual social networks (Coleman, 1988; Gephart, 1997; Sampson, 2002~. Network closure occurs when parents come to know the par- ents of their children's friends; this personal link encourages adult monitoring and
42 CITIES TRANSFORMED supervision. But closure of personal networks is greatly facilitated by the pres- ence of formal associations and institutions (Aber, Gephart, Brooks-Gunn, and Connell, 1997: 53-54~: ... living in socially organized or "functional" communities should increase parents' contact with their children's friends and with the families of their children's friends.... A weak community organi- zational base impedes participation in local organizations and devel- opment of informal social networks, such as friendship ties among community adults. These, in turn, reduce the collective supervision of youth in the community and diminish the resources available for child care. Such individual-to-group linkages can generate complex feedbacks and path de- pendencies in the life of local communities. Individual behavior can either rein- force local social capital and buttress local institutions, or set diffusion processes into motion that ultimately erode the associational base of the neighborhood. Child-rearing strategies provide one case in point. As noted above, families in risky neighborhoods may choose to restrict their neighborhood ties in favor of cultivating outside ties. Once such strategies become widespread, they effectively disconnect local personal networks and undermine the sense of local community, and this in turn further reduces incentives to invest in local ties. As Furstenberg (1993) writes: The social world of transitional communities contracts as neighbors become strangers. Low participation at school meetings and neigh- borhood improvement associations reduces the networks.... We see in transitional neighborhoods a shift from collective to individualistic strategies of family management. Fewer parents are willing to dele- gate authority to formal institutions which have lost their credibility and command of external resources. Informal networks become at- tenuated as kin and close friends move out and are replaced by new residents who are regarded as outsiders. The perception of normative consensus in the community diminishes accordingly. Sampson (2002) writes in a similar vein on the dynamic feedbacks between crime rates and social capital, whereby crime erodes the bonds of community, and weak- ening community institutions allow for the further penetration of crime. Some de- mographic processes such as residential mobility can undermine attachments to the local community and make it more difficult to preserve local social capital (Chaskin, 1994; Sampson and Morenoff, 1997; Sampson, 2002~. Spatial Segregation In his presidential address to the Population Association of America, Massey (1996) argued that the urban poor and the urban affluent increasingly inhabit
WHY LOCATION MATTERS 43 separate spaces. Massey warned of increasing segregation in high- and low- income countries alike. One spatial expression of segregation can be seen in Figure 2-1, which shows how levels of income (measured here by an index of proxy variables) vary across the metropolitan area of Mexico City. As can be seen, the poorest areas (those with the darkest shading) are concentrated in the city's periphery. Does this spatial patterning of poverty have implications for de- mographic behavior? Spatial segregation can have the effect of enforcing homogeneity in local social network ties and resources, suppressing some of the diversity in social relations that can benefit the poor. To appreciate this point, imagine an alterna- tive organization of Mexico City, one in which poor families are distributed uni- formly across the urban space. If spatial proximity promotes social network ties and heightens the visibility of beneficial role models and reference groups, the redistributed poor would be expected to have more connections to higher-income network partners. The life-course possibilities exemplified by the nonpoor might then be more easily appreciated by the poor and factored into their own family strategies. But when the poor are spatially segregated, such local socioeconomic diversity is diminished, and, as Sampson and Morenoff (1997: 19) write, . . . the social isolation fostered by the ecological concentration of ur- ban poverty deprives residents not only of resources and conventional role models but also of cultural learning from mainstream social net- works that facilitate social and economic advancement in modern in- dustrial society. . In this way, spatial segregation blocks avenues for social learning and severs some of the weak ties that could provide information and pathways to resources.~° Labor market ties are of particular importance. The social networks of adults shape their job-search strategies and, through search, the nature of employment that can be secured. Montgomery (1992) formalizes the role of social network connections in the job-search theory used by economists, developing the idea that weak network ties may supply information about job openings more frequently than do strong ties. By affecting wages and incomes, restrictions on labor i°Granovetter (1973) is usually credited with the distinction between weak and strong ties in social networks. A tie between two network members is said to be strong if they have a long and durable association, if they meet frequently, and if their relationship is multistranded in the sense of joining different types of social relations (e.g., neighbor and kin). Weak ties are generally of shorter duration, can involve infrequent contact, and can be more specialized (say, if two people know each other only as coworkers). Granovetter's contribution was to suggest that weak ties are often those along which novel information is conveyed: they can provide "bridges" leading from dense, strongly tied, and homogenous networks to diverse outside information and resources. Hence, weak ties are thought to be important to innovation, and strong ties to the maintenance of norms and collective identities. tiff weak ties increase the frequency with which job offers are made known, job searchers with more weak ties in their networks would be expected to have higher reservation wage thresholds, and (other things being equal) higher expected wages as well.
44 CITIES TRANSFORMED ;3nl~ln`~,~ t3 Hig!-' IBM t`'l;-(.ieri3tel`,, hI(~] ~ (`,Jo'derate E~ (',]~?t'8~017 10w Mm Lt.~.~ _ 'Jury tow ~ I ["Fuji a\,`311~0 FIGURE 2-1 Mexico City metropolitan area: socioeconomic levels by geostatis- tical areas, 1990. SOURCE: Rubalcava and Schteingart (2000: Map 4.7.2~. Reprinted with permis- sion.
WHY LOCATION MATTERS 45 market information can powerfully (albeit indirectly) constrain demographic . . c .eclslons. In a study of Los Angeles, Stoloff, Glanville, and Bienenstock (1999) explore the links between social network diversity (or range) and women's employment outcomes, focusing on the influence of network ties that reach outside the neigh- borhood and those that cross lines of race, gender, and education. Stoloff and colleagues find that women are more likely to be employed when they have more outside-neighborhood ties and more ties to other women with university school- ing. As Chapple (2002) notes for San Francisco, many poor women face trans- port costs that restrict the geographic reach of their employment contacts and find themselves locked into local low-paying jobs. Poor women may look locally for work because long commutes are incompatible with the child-care options they can afford; with spatially restricted social networks, they are unlikely to hear of better jobs that might be available elsewhere32 In addition to constraining employment information, the spatial concentra- tion of the poor may heighten their exposure to external, macroeconomic shocks. Consider a shock that initially reduces each poor family's income by a given per- centage. If the poor are spatially concentrated, this initial effect will be magnified by local economic multipliers. As the shock is propagated locally, businesses that serve the poor will be placed under stress, and when local businesses give way, some neighborhoods may spiral downward (Massey, 1990~. If the poor were not so spatially concentrated, the effects of such shocks would be dispersed and thereby muted. Not all aspects of spatial segregation need be negative. The isolation of the poor can sometimes enhance their sense of mutual dependence (Lee and Camp- bell, 1999~. The resulting "compression" of social relationships can increase the importance of neighborhoods and other localized ties, creating a possibly belea- guered sense of neighborhood as community.~3 In situations in which groups are segregated mainly along racial or ethnic lines, it is even possible for social com- pression to increase local diversity in the dimensions of income and class, bringing disadvantaged poor households into closer proximity to their better-off ethnic or i2See Hanson and Pratt (1991, 1995) and McDowell (1993a,b). But note that in some cases, poor women seek employment far from home, as discussed in Hondagneu-Sotelo (1994) for Mexican im- migrant women working in the United States. The trade-offs between job search time and the spatial extent of search are well recognized in the human geography literature (e.g., Committee on Identifying Data Needs for Place-Based Decision Making, 2002) as leading examples of what geographers term the space-time prism (Hagerstrand, 1975). i3Comparing white with black households in Nashville, Lee and Campbell (1999) find that black households have more intimate ties to their neighbors and activate these ties more frequently. Among all network partners from whom support might be sought the authors analyze several dimensions of support, including job search, financial assistance, care during illness, and help with important decisions blacks are more likely than whites to list partners who live in their neighborhoods (support in arranging transportation is an exception). To be sure, in the dimensions of social support examined here, neighbors make up only a minority of all network partners. Even for Nashville blacks, most supportive network ties reach outside the neighborhood.
46 CITIES TRANSFORMED racial counterparts (Massey, 1990; White, 2001~. In this way, the compression mechanism could set the stage for the development of social capital in some poor neighborhoods. The spatial element in social interaction and resource distribution should not be overemphasized. Many urban residents participate in multiple social networks and communities, and these can operate at multiple spatial scales. As we have discussed, spatially dispersed social and economic ties could be as important to information flow, access to services, and social comparisons as spatially local- ized ties. Proximity may well facilitate the formation of social network ties and lessen the costs of maintaining such ties, but proximity certainly does not guaran- tee that social linkages will be formed.~4 Nor does spatial proximity necessarily imply socioeconomic homogeneity. But it is reasonable to think that even where localized factors are not dominant, they may still exert appreciable influence. The demographic implications of spatial segregation therefore deserve careful consideration. Neighborhoods and Larger Structures As the foregoing discussion makes clear, neighborhood effects are not wholly at- tributable to neighborhoods. They reflect the situation of neighborhoods within larger social and economic structures and the roles assumed by institutions that operate across the urban space. The crucial interactions among neighborhood and more widely dispersed actors can be illustrated by a demographic example from a century ago, concerning the diffusion of new ideas in the sphere of urban public health. As early as the 1 880s in the United States, proponents of what was then known as "sanitary science" adopted the elements of the germ theory of disease and worked in multiple forums, using multiple media, in an effort to spread new health messages and transform personal hygienic behavior (Tomes, 1998~. There was an outpouring of popular advice on how to keep the home free of germs advice conveyed through manuals on domestic hygiene, pamphlets, newspaper columns, public lectures, exhibits, and posters. Plumbing and toilet manufacturers aided the cause, together with entrepreneurs specializing in disinfectants and ventilat- ing devices. City and town governments assisted by enacting detailed plumbing codes and regulations. Among the affluent and those striving to join their ranks, the fear of germs was fanned by apprehensions about the infective power of newly arrived immigrant populations and was intermixed with social attitudes equating cleanliness with gentility (Tomes, 1998: 62, 111~. i4Mere residence in a neighborhood with superior institutional and social resources does not ensure access to them. See Jarrett (1997) on the experience of black families in affluent U.S. communities, who can be excluded from participation in important community networks and social circles. See also Chapters 5 and 9, where the "proximity but high walls" modes of segregation are discussed.
WHY LOCATION MATTERS 47 Even by the late 1890s, it was still mainly the literate who were in a position to hear the new sanitary advice, and few but the affluent could afford to safeguard their homes. Beginning in the 1890s, however, local and national associations be- gan to be organized with the aim of preventing contagious disease (especially tu- berculosis), adopting commercial advertising techniques to deliver their messages to the poor and illiterate in a vivid and memorable form (Tomes, 1998: 114-21~. Of course, these efforts to improve personal hygiene and home sanitation would have produced little improvement in health were it not for the massive comple- mentary investments made by governments in extending sewerage, cleaning water supplies, and enforcing regulations to keep milk and food free from contamination (Preston and van de Walle, 1978; Ewbank and Preston, 1990; Preston and Haines, 1991; van Poppel and van der Heijden,1997~. This tale illustrates several of the distinctive features of urban social interac- tions. Externalities were at the core of the urban public health problem the com- municable diseases to which urban populations were vulnerable because of their proximity and dependence on common resources. Technological change and the emergence of new theories of disease transmission provided the opportunity for such biological externalities to be fought by way of beneficial social externalities, which took the form of diffusion of information about personal hygiene. In the re- sulting diffusion processes, novel forms of information were propagated through social networks (as in the Taichung experiment), as well as through multiple me- dia. Because urban settings were socially heterogeneous, juxtaposing the literate and the illiterate, the urban poor were exposed to greater heterogeneity in health information than might otherwise have been the case. The process was further aided by two types of social comparison one that linked cleanliness to gentility, and another that linked anxieties about infection to the presence of diverse urban immigrant populations. The economic diversity of urban settings was also evi- dent in the roles assumed by private markets and businessmen. Finally, decisive actions were taken by city governments and by the kinds of associations we would now call NGOs, illustrating the institutional diversity seen even then in the urban arena. As these examples show, macro-level economic, social, and political struc- tures can determine what resources are made available to local communities and rendered accessible through the personal social networks of community residents. In the case of the public health revolution, these larger structures brought enor- mous benefits to individuals and neighborhoods. But as James, Schulz, and van Olphen (2001: 178) note, structural forces can sometimes have the effect of "cir- cumscribing the range of resources that may be mobilized within social networks and decreasing their flexibility in responding to the social environment." The point is that resources cannot simply be expected to materialize, as if by spontaneous generation, in the personal social networks and local associations of poor city residents. Links are needed to the powerful actors and institutions that generate resources and influence how they are distributed.
48 CITIES TRANSFORMED BOX 2.2 A Federation of Low-Income Groups in Mumbai Appadurai (2001) recounts how three local associations in Mumbai, India SPARC (the Society for the Promotion of Area Resource Centres), NSDF (the National Slum Dwellers Federation), and Mahila Milan (an organization of poor women) have formed an alliance to raise the political visibility of issues affecting the poor and to promote creative solu- tions. These three organizations share concerns about the security of land tenure; adequate housing; and access to electricity, transport, sanitation, and related services. They have extensive links to other poor groups across India, and are forming international partner- ships with similarly composed associations in South Africa, Thailand, Cambodia, and the Philippines. (These and other groups are described in the October 2001 issue of the jour- nal Environment & Urbanization.) Their international reach is evident in the founding of Shack/Slum Dwellers International (SDI), an alliance of poor peoples' federations that spans 14 countries across four continents. The Mumbai federation is built upon several principles: (1) a reliance upon local in- formal savings and microcredit groups, which not only give the poor a means of amassing modest amounts of capital, but also instill patience and discipline, thus reinforcing a sense of community; (2) recognition that the conditions of the poor must be documented if they are to be made visible to outsiders, this insight being reflected in a commitment to pe- riodic socioeconomic surveys of the slums in which federation members live; (3) use of "precedent-setting" pilot projects and public exhibitions, which draw attention to feasible, low-cost designs for affordable housing and sanitation (the latter dramatized in "toilet fes- tivals" that showcase functioning public toilets designed by the poor with careful attention to systems of collective payment and maintenance); and (4) adherence to a political stance of resolute nonalignment with any local or national political party, and a commitment to negotiation rather than confrontation in political tactics. Of the three groups involved, SPARC (formed in 1984 by social work professionals) is the best connected to the state and corporate elites of Mumbai and elsewhere. It provides many of the bridges that reach to local and national Indian governments and to the inter- national funders. By presenting a professional face to these outsiders, SPARC gives the funders a means of funneling their resources to its affiliates. SPARC also provides its local allies with a buffer that prevents the agendas of the funders from dominating the concerns of the poor. The question of how resources can be drawn into poor communities can be ad- dressed in terms of the "bridging" role of social networks and local social capital in developing countries. Box 2.2 recounts one of the most successful cases in- volving groups of the poor. Beginning in the mid-1980s, local groups in Mumbai, India, gradually acquired sufficient visibility, reach, and political mass to attract resources from various levels of the Indian government and international agen- cies. As Appadurai (2001) and others have observed, some of the most sophis- ticated alliances of poor groups are making contacts in international networks of like-minded groups and funders, engaging in what some have termed "advocacy without borders." Box 2.3, by contrast, shows how difficult it can be to establish durable institutional mechanisms for delivering resources to poor communities.
WHY LOCATION MATTERS 49 BOX 2.3 Participatory Urban Poverty Programs in Bangalore: Lessons Learned The experience of the Bangalore Urban Poverty Alleviation Programme (BUPP), a Dutch- funded pilot project implemented in 1993-99, illustrates many of the difficulties that face participatory urban initiatives (de Wit, 20024. This ambitious program drew together Ban- galore government departments, local NGOs working in the city slums, and the community- based organizations (CBOs) of these slums. On-the-ground activities were to be carried out by newly formed Slum Development Teams, local groups whose members were to be elected by local inhabitants, with women assuming a full share of the leadership posi- tions. Although the project did record some successes (notably in establishing savings and credit schemes for slum women), the partnerships it fostered among government, NGOs, and CBOs could not be sustained once external funding had ceased. After 1999, these organizations largely reverted to their preprogram, independent modes of operation. According to de Wit (2002), part of the problem had to do with the question of "own- ership" of the project. BUPP was conceived as a separate organizational entity, connected to but standing apart from government departments, NGOs, and preexisting CBOs. This autonomy was meant to encourage experimentation and learning-by-doing, but it had unin- tended side effects. Government departments proved reluctant to channel scarce resources to a program not housed within any unit of government; local NGOs found themselves being asked to play generalist, supervisory roles for which they had neither the experi- ence, the expertise, nor the resources; and the CBOs that had already been in existence in the Bangalore slums often viewed the new program and its slum development teams as competitors for local influence, and even as threats to the interests of their own leaders. (The local community groups tended to be both male-dominated and inegalitarian; they were in some instances the product of local systems of patronage.) As a result, none of the slum development teams could survive after the end of Dutch funding in 1999. As de Wit (2002: 3941) remarks, in retrospect "it might have been better to start from, foster and link to proven well-functioning traditional, endogenous or pragmatic groups (informal credit groups, women's networks, user groups)" and to encourage these groups to gradually adopt more open and participatory approaches. In Bangalore a well-conceived participatory program that enlisted units of gov- ernment, local NGOs, and slum community organizations fell victim to sharply conflicting interests. Summary This section has described features of local social and spatial environments that could influence demographic behavior. In each of the areas mentioned, neigh- borhoods provide the staging grounds upon which social interactions take place; personal networks provide the circuits along which externalities and information flow; and, together with networks, associations and institutions provide the base of local social capital. There is good reason to think that neighborhoods, net- works, and social capital can exert a great deal of influence on demographic be- havior in developing-country cities. It is certainly possible that their effects are powerful and pervasive. To date, however, almost all of the literature on these
50 CITIES TRANSFORMED issues is concerned with the cities of developed countries. Chicago has been well documented, but Cairo has not. In its present form, the research literature can offer little more to those interested in poor countries than intriguing analogies and potentially fruitful lines of inquiry. Even the field of urban health still lacks the multilevel, longitudinal research programs that could identify the effects of neigh- borhoods, social networks, and social capital in developing-country cities. In the literature on the United States, where theoretical propositions about net- works, neighborhoods, and social capital have been in circulation for well over a decade, convincing empirical tests also remain scarce. This chapter is not the place for a thorough critique of research strategies, but several aspects deserve mention.l5 The dominant research strategy used in U.S. studies matches individ- ual data, often drawn from sample surveys, to areal data drawn from censuses. In view of the many difficulties involved in defining neighborhoods, the approach is acknowledged to be imperfect and defensible only as a first pass at the issues. But in most developing countries, no data are available on units akin to census tracts, rendering even this crude approach infeasible. Mixed results have emerged from the pairing of cross-sectional individual and areal data in the first generation of U.S. research. i6 Strong neighborhood compo- sition effects have been found in some studies of adolescent fertility and contra- ceptive use (see the review in Jencks and Mayer, 1990~. However, the estimated effects of neighborhood on measures of child development have generally been small (see Gephart, 1997; Brooks-Gunn, Duncan, Leventhal, and Aber, 1997~. In part this is because in highly segregated cities, it is empirically difficult to sep- arate the effects of individual and family characteristics (e.g., family income) from those of the neighborhood (e.g., the percentage of low-income families in the neighborhood). It is only recently that research studies have been designed explicitly to test neighborhood effect theories (Sampson and Raudenbush, 1999; Sastry, Pebley, and Zonta, 2002) and that randomized experiments have been mined for evidence on neighborhood effects. Large-scale housing voucher experiments notably the "Moving to Opportunity" experiment conducted in Boston, Baltimore, Los Ange- les, and a few other U.S. cities are proving to be a rich source of neighborhood effect studies (Del Conte and Kling, 2001; Sampson, Morenoff, and Gannon- Rowley, 2002~. Nonexperimental longitudinal designs are only now going into the field. Furstenberg and Hughes (1997) suggest that studies of this sort should i5Sampson, Morenoff, and Gannon-Rowley (2002) provide a recent insightful review. Duncan, Connell, and Klebanov (1997) and Durlauf (1999) give a concise account of the statistical difficulties that confront tests of neighborhood effects and diffusion/social interaction theories difficulties that are formidable if only cross-sectional data are available. Manski (1993) develops a special case that nicely exposes these difficulties. i6See Hogan and Kitagawa (1985) for an early effort focused on teen childbearing and the re- cent study of Upchurch, Aneshensel, Sucoff, and Levy-Storms (1999), which addresses the timing of teenage sexual activity.
WHY LOCATION MATTERS 51 begin by mapping local perceptions and definitions of neighborhood, that ques- tionnaires should employ several such concepts to enable comparison, and that neighborhood residents should be asked about social ties extending outside their residential communities. This would appear to be sound advice for new studies in developing-country cities. SUSTAINING DIVERSITY: ECONOMIC INTERACTIONS Much demographic behavior is rooted in individual and family perceptions of eco- nomic costs, benefits, and uncertainties. The lines running from economy to de- mography are readily seen in the case of migration, which is a behavioral response to spatial differences in income and consumption opportunities. Urban economic structure can also be expected to affect women's wage rates and, through wages, the opportunity costs of time spent in child care. These opportunity costs have long been identified as important factors in fertility decisions. Earlier we drew attention to the economic returns to schooling. Models of the "quantity-quality trade-off" suggest that as returns to schooling increase, parents may be persuaded to forego high fertility to have fewer but better-educated chil- dren. By implication, then, the returns to schooling evident in urban areas could induce lower fertility and greater investment in human capital. Incomes obviously affect the full range of urban demographic decisions including decisions about health as well as children's schooling. There are also macro-level effects to be considered. When urban income growth raises government revenues, it provides governments with the means to extend investments in public services (such as water supply and sanitation) and education. Rising incomes support the develop- ment of specialized private markets, such as in the delivery of health services. For all these reasons, economic factors are fundamental to an understanding of urban demography. The preceding discussion has been concerned mainly with what might be termed social and biological externalities, which arise from social interactions in differentiated populations. Proximity matters to the social side of the argu- ment because it reduces the costs of information acquisition and exchange. To economists, these are familiar ideas; indeed, if the theories were to be relabeled as theories of information exchange among firms and workers, involving both pro- duction and search networks, they would be recognizable as the basis for much of urban economics. The fundamental building blocks of urban economic theory are the costs of communication and transport, spatial heterogeneity, returns to scale, and the set of pecuniary and technological externalities known as agglomeration economies. Recently, the special role of economic diversity has attracted attention whether the diversity displayed in a city's portfolio of economic activities (Jacobs, 1969, 1984; Glaeser, Kallal, Scheinkman, and Shleifer, 1992) or that embedded in the preferences of consumers for a range of goods (Fujita, Krugman, and Venables,
52 CITIES TRANSFORMED 1999; Neary, 2001~. Taken as a group, the economic theories seek to explain both the existence and the growth of cities; that is, they have static and dynamic impli- cations. Here we present a sketch of the main arguments, drawing from reviews by Anas, Arnott, and Small (1998), Glaeser (1998), Henderson, Shalizi, and Ven- ables (2000), and Quigley (1998), while reserving a fuller treatment for Chapter 8. Spatial Theories Costs of transport and communication lie at the heart of urban economic theory. When these costs are very high, they have the effect of fixing people, goods, and ideas to specific locations; production and consumption then occur as in autarky. When transport and communication costs approach zero, by contrast, space ceases to matter, and geography might then be said to disappear. It is in the middle range, where costs are appreciable but not prohibitive, that the economics of spatial clus- tering are of interest (Henderson, Shalizi, and Venables, 2000~. Today, with declines in the costs of information exchange due to the tele- phone, the Internet, and other innovations, spatial proximity might be thought to have lost some of its former economic significance. Technological progress pre- sumably removes much of the need for face-to-face exchanges of services and information, and it would appear only logical that exchange mediated by technol- ogy must substitute for personal exchange. Although plausible, this substitution hypothesis needs further scrutiny. Technological and personal exchanges may of- ten be complementary; when freed to conduct their routine interactions by e-mail or telephone, for example, economic actors may increasingly prize and seek to ex- ploit the special advantages of face-to-face exchange (Glaeser, 1998~. Perhaps in personal contacts there are ways of communicating trustworthiness, ensuring con- fidentiality, and displaying interest and commitment that simply cannot be equaled by technological devices, at least in their present form. We return to this point below in the context of "economies of diversity." A very different approach to questions of agglomeration comes from an emphasis on the increased uncertainty of global markets, coupled with the increased speed of electronic markets (see Sassen, 2001a). In a world in which transportation and communication costs remain signifi- cant, the heterogeneities of space present one fundamental rationale for the clus- tering of economic activity. The physical features of land the presence of natural harbors, the confluence of rivers, the sites of mineral deposits confer compara- tive advantages and disadvantages upon locations. Firms move to exploit the ad- vantages of specific sites; trade then links locations much as it does in the standard models of international trade among countries. Spatial clustering and differentia- tion can thus be explained, up to a point, by Ricardian comparative advantage. It is when the productivity benefits of scale and proximity are brought into the picture that urban economic theories take on their distinctive character. Two forms of productivity benefits are of interest: those that are internal to the firm or production unit or that involve links through markets, and those that are external,
WHY LOCATION MATTERS 53 which can take the form of technological externalities and informational spillovers. We consider these in turn. Internal economies of scale and proximity When production is characterized by increasing returns to scale a doubling of inputs yielding more than a doubling of output firms have additional motiva- tion to concentrate their production spatially. And as long as there are significant transport costs involved in moving people, the concentration of production will imply a concentration of the labor force. Many factors determine whether firms choose to locate near supplies of other inputs or near the final markets for their goods. Historically, cities have grown around transshipment points so as to exploit scale economies in the loading and unloading of goods (Anas, Arnott, and Small, 1998~. In an earlier era, the weight of final goods and inputs and the weight lost in the processing of inputs into final goods were regarded as the dominant factors in models of locational choice. But as transport costs for goods have declined, this rationale has lost something of its historical force. Speed and market uncertainty have today replaced weight (Sassen, 2001a). When the firms of a given industry cluster spatially, this permits the suppliers of inputs tailored to that industry to save on transport costs, and may allow for further specialization in the production of inputs. For instance, heavy government investment in urban health facilities can foster specialization among private-sector suppliers of health services. From this base, a broader private health sector can develop, offering services that parallel those in the public sector. As will be seen in Chapters 6 and 7, such private markets in health services are far more prominent in cities than in rural areas, and are a more significant presence in larger than in smaller cities. Similar arguments apply to the cross-industry clustering of firms and the sup- pliers of inputs used across a variety of industries.~7 When firms are spatially clustered, workers can search among them more efficiently, expending less time and fewer resources in locating jobs that are well matched to their human capital. Consumers, too, can benefit from the spatial concentration of differentiated goods (Quigley, 1998~. An important role in the theory pertains to the services that are provided by the public sector water supply, sanitation, electricity which are thought to exhibit significant economies of scale and proximity. The issue is not simply how cost varies with the number of units produced, but also how it varies with distance to consumers of services, whether individuals or private firms (Montgomery, 1987, 1988~. Where these economies of proximity are significant, there are decided cost advantages to spatial clustering. The cost reductions derived from proximity can i7In some accounts of the theory, the within-industry and cross-industry effects are termed, re- spectively, localization and urbanization effects. We reserve the phrase "localization effects" for true externalities associated with spatial proximity, as contrasted with effects mediated through markets.
54 CITIES TRANSFORMED bring further savings; for example, private firms may not be able to secure lower production costs from their own internal scale economies until they are assured of access to reliable public services. The benefits of proximity can also be seen in the desires of workers to protect themselves against firm-specific random shocks. A worker with industry-specific human capital may agree to take a lower-wage position with one firm if there are many similar firms in its immediate vicinity. Likewise, a firm anticipating idiosyncratic shocks in its markets may choose to locate near other firms in the industry if by so doing it can reduce buffer stocks of inventories. Presumably such motives would be evident in industry wage levels and in the nature of in- traindustry, firm-to-firm contracts. If the economic shocks are anticipated to be industry-wide rather than firm-specific, however, equivalent protections may be secured through interindustry clustering. For instance, the "thick labor markets" of cities, involving many firms and workers mutually engaged in search, may con- fer some protection against long layoffs, prolonged unproductive searches, and persistent difficulties in filling selected skilled positions (Glaeser,1998~. In summary, the internal scale and proximity arguments are concerned mainly with the cost savings firms can achieve through spatial concentration of produc- tion. These savings may depend, in turn, on the economies of specialization, scale, and proximity realized by suppliers of inputs, including workers and the providers of public services. These issues are well illustrated in the case of producer ser- vices, described in Box 2.4. External economies of proximity External productivity benefits of- proximity are true externalities in that they in- volve spillovers not directly mediated by markets. Such external effects are generated by economic interactions partly as a response to the complexity of cor- porate services today (Sassen, 2001a), search, and the exchange of ideas (Anas, Arnott, and Small, 1998~. Localization elects arise from the spatial concentra- tion of firms in a given industry. They may be produced when one firm is able to observe the innovations, experiments, production processes, or competitive strate- gies of another. Workers carry such innovative ideas with them as they circulate among firms (Glaeser, 1998~. Striking evidence of such spillovers is seen in the spatial clustering of patent applications in the United States (Jaffe, Trajtenberg, and Henderson, 1993), which suggests that innovations undertaken by one firm can indirectly benefit another (see also Audretsch and Feldman, 1996~. Within this literature, diversity elects refer to the positive externalities gener- ated by information exchange in heterogeneous environments; here the emphasis is on the variety of urban economic activity and the stimuli that are provided by diversity. The concept owes much to Jacobs (1969, 1984) and is expressed by Glaeser, Kallal, Scheinkman, and Shleifer (1992) in terms of cross- industry economic diversity. The argument is that creative, innovative energies i8Quigley (1998) gives a good account of the issues.
WHY LOCATION MATTERS 55 BOX 2.4 Producer Services and High-Skill Labor Markets Although much of any city's daily economic activity must be given over to the routine needs of its residents, there can be analytic returns to a focus on a specialized sector. A narrow focus can shed light on urban dynamics and suggest the course of future developments. When examined from this vantage point, the specialized, high-end business service sector amply repays study (Sassen, 1991, 2001a). In regions that are already well endowed with telecommunications infrastructure, ad- vanced producer services might be expected to locate outside the major cities so as to escape high rents and congestion. Contrary to expectation, however, it appears that pro- ducer services are often concentrated in large cities. The reasons for this may have to do with urbanization economies and the economic rewards to cross-industry, intraurban diver- sification (Jacobs, 1969, 1984; Glaeser, Kallal, Scheinkman, and Shleifer, 1992~. Producer service firms are not necessarily dependent on being close to the businesses they serve. There are cost savings and efficiencies to be found in locating near suppliers and professional collaborators, especially when the service involves a form of joint production. Assembling a financial instrument, for example, can require inputs from firms in fields as diverse as accounting, advertising, law, economic consulting, public relations, design, and printing. An accounting firm may serve its clients at a distance, but the quality of its service depends on proximity to specialists, lawyers, and programmers. Additional motives for concentration arise from the needs of the professionals who are employed in such high- skill jobs; they are often attracted to the amenities and lifestyles that large urban centers can offer, and can be reluctant to live elsewhere (Sassen, 1991, 2001a). The political economy of all this warrants attention. Growing numbers of high-level professionals and specialized service firms can increase urban spatial and socioeconomic inequality. Markets and politicians may prefer to develop housing for high-income pro- fessionals; commercial space may be dominated by firms selling high-priced goods and services; and these tendencies can be reinforced by the growing share of the city's total payroll and tax revenues due to the specialized services core. are stimulated most effectively by the collision of disparate ideas. Just how dis- parate these ideas may be is difficult to formalize, but the theory envisions a cre- ative ferment in which entrepreneurs, scientists, artists, and firms find themselves struck by analogies and by the observation of activity in spheres that overlap but do not wholly coincide with their own. The benefits stemming from urban diversity are often linked to the product development cycle, that is, to the stages in the life of a good or service as it moves from experimentation and design to routine production. The benefits of urban diversity are believed to be greatest in the initial stages of product development. Once a production process becomes standardized, some of its elements can be relocated to suburban or rural sites where wages and rents are lower (Duranton and Puga, 2001~. Services and production in high-technology sectors are thought to derive special benefits from economies of diversity (for Korea, see Henderson, Lee, and Lee, 2001, and Henderson, 2002~. As noted in Box 2.4, there are some services that must be assembled with speed and that require the collaboration of
56 CITIES TRANSFORMED many professionals from diverse fields; services such as these may never become sufficiently standardized to be relocated. Network elects are implicated in these theories in that information exchange and creative partnerships will sometimes form along social network lines and can be facilitated by spatial proximity. The social capital aspect of networks can mat- ter when economic decisions must be made swiftly. To seize opportunities when they arise, economic actors may need assurance that their partners are trustwor- thy and can keep strategic information confidential. Network effects may be es- pecially important in fast-moving sectors such as finance. According to Meyer (1998), in this sector the face-to-face exchanges made possible by spatial con- centration foster trust and allow for the growth of social networks that permit the sharing of private information. Given the importance of education to demographic behavior, it is interesting to consider how it might figure in such theories. It is often said that while ideas depreciate, education has enduring value: in the process of acquiring it, the edu- cated also acquire conceptual flexibility, an openness to innovation, and a capacity to identify and exploit the possibilities that present themselves in disequilibrium. It may be that in urban areas, which are continually roiled by economic shocks and waves of technological change, the economic returns to schooling lie precisely in these abilities. Black and Henderson (1999b) develop this theme in terms of the social returns to schooling in urban areas, arguing that spillovers cause the social returns to exceed private returns. Supportive empirical results are presented by Rauch (1993) and Moretti (2000~. They find that individual returns to schooling increase with the average level of schooling in the local labor market, a result that hints at the benefits to be derived from urban diversity. The diseconomies of proximity As cities grow, the economies derived from spatial proximity are eventually offset by diseconomies. The economic value of space rises with concentration, but this increased value drives up rents. Commuting times grow longer as cities become larger and transport grids seize up with congestion (Glaeser, 1998; Henderson, Shalizi, and Venables, 2000~. Crime rates may also be linked to city size. i9 Levels of pollution can increase with city size as well, although it may be that size is less a root cause than an exacerbating factor. In many dimensions, then, congestion costs and diseconomies would be ex- pected to rise as producers, workers, and consumers become more spatially con- centrated. At some point, the benefits of further growth and concentration are overtaken by the marginal costs. The city size at which marginal benefits just equal marginal costs can be regarded as an optimal city size.20 i9As Glaeser (1998) explains, crime may be associated with city size because the diversity of big- city life lets criminals function with anonymity. He speculates that there may be economies of scale in the distribution of stolen goods and an easier flow of information among diverse networks of criminals. 20At one time it was thought that optimal city sizes might be identified empirically, and a number of heroic efforts to do so made appearances in the literature. This quest has largely been abandoned.
WHY LOCATION MATTERS 57 Summary The theories discussed above provide a framework for thinking about the urban economy its size, diversity, and trajectory of growth in terms closely analo- gous to those used to describe social interaction and neighborhood effects. Indeed, the economic arguments can be applied at spatial resolutions below the level of the city. Operating at micro scales, they can shed light on the organization of neighborhoods and other small clusters of economic activity (Anas, Arnott, and Small, 1998~. As discussed further in Chapter 8, dynamic theories, in which an urban concentration of resources spurs innovation, are now attracting a great deal of research interest. Still, as with neighborhood effects and social interaction, the economic the- ories discussed here have run well ahead of empirical tests. The statistical dif- ficulties that confront dynamic theories of agglomeration are precisely those that confront theories of neighborhood effects. In both cases, unmeasured factors op- erating at an aggregate level (a neighborhood or city) exert influences at lower levels (families or firms) that can easily be mistaken for spillover effects at that lower level. For instance, an unmeasured natural advantage can attract firms in an industry to a particular site and raise the productivity of the firms that cluster about it. The unwary analyst might well take this productivity benefit to be ev- idence of a cross-firm localization effect. The risks of such mistaken inference are much reduced when repeated observations are available on the actors and their contexts, whether these are families in the context of neighborhoods or firms in the context of local industries and markets. Because credible tests of dynamic theo- ries require multilevel, longitudinal research designs, it should be no surprise that progress on the empirical front has been slow. Even in the United States, where the effects of urbanization, localization, diversity, and innovation have been most closely examined, the literature cannot yet offer a consensus as to the size of these effects. Theories inevitably overlook many features of urban economic life. Although we have stressed theories of innovation and dynamics, it can be argued that inertia is of equal importance to the arrangement of urban economic space. Pub- lic infrastructure and private capital are durable, and even networks and patterns of social interaction can be long-lasting.2i These features of the urban scene can continue to attract and concentrate resources even as other elements of the eco- nomic rationale begin to weaken. Durability and sunk costs provide yet another reason to think that the dynamics of cities must be highly path-dependent and difficult to predict. 2iThe cities of South Africa illustrate the enduring imprint of inertia. On this point, Pieterse (2000: 10) quotes Lindsay Bremner, a South African planner: "The marks apartheid left on human lives will fade in the course of time. But its spatial logic will continue to affect people's daily lives for genera- tions to come. Because of apartheid, people live great distances from where they work; standards of urban infrastructure vary enormously; parts of the city are devoid of shops, businesses, entertainment venues and schools, while others are saturated with them; chasms separate one citizen from another, so much so that people feel like tourists in each other's worlds."
58 CITIES TRANSFORMED CITY SYSTEMS AND CITY-REGIONS This chapter began with a focus on families interacting within neighborhoods and through their social networks. In considering the urban economy, it then shifted to a higher level of aggregation at which interactions among firms and workers can be conceived of as taking place city-wide. But we cannot be content to leave off here, with individual cities as the units of analysis. Cities exist within wide- ranging social and economic systems that connect them to other cities and to the rural populations of their regions. Because such a large proportion of the urban population resides in small cities those under 500,000 in population it is espe- cially important to consider the roles of these cities in the wider system. In the chapters that follow, we repeatedly draw attention to the demographic conditions that prevail in small cities and show that, at least on average, their residents gen- erally fare worse on the various demographic indicators than do residents of large cities. To understand such findings, one must consider what theory has to say about systems of cities. It should be acknowledged at once that city and regional systems are enti- ties far too complex to be understood through theory alone.22 As noted above, even the evolution of a single city presents issues of path dependence, historical lock-in, and agglomeration dynamics that render theoretical conclusions indeter- minate. The difficulties involved in explaining a single city's growth trajectory are magnified many times when city systems and regions are considered. A further complication is that the networks in which cities participate are in- creasingly international in character, and therefore involve economic and social forces that operate at supranational scales. Although this chapter is confined to the national scene, international networks in information, trade, and finance are considered in Chapter 3. City Size Distributions and Primacy Henderson (1988) summarizes the implications of localization economies for the size distribution of cities. Consider an industry for which such economies are significant, giving its firms an incentive to cluster. If the benefits of clustering were exhausted at relatively small cluster sizes or rose so slowly with size that they were soon overtaken by diseconomies, this industry would tend to be located in relatively small cities. There would be little payoff to larger-city locations because there the industry's firms (and its workers) would expose themselves to additional penalties from diseconomies without any offsetting productivity gain. Large cities are thus composed of the industries for which clustering confers net benefits even in large concentrations. Acting in this way, localization effects can frame the 22Economists are being guided by models of interaction in complex systems, developed in physics and other disciplines, in their efforts to understand urban change; see Anas, Arnott, and Small (1998) for a recent review.
WHY LOCATION MATTERS 59 architecture of a national system of cities. Firms that benefit from cross-industry, urbanization economies (including economies of diversity) then sort themselves among cities of differing size and composition, seeking sites where the marginal benefits from urbanization economies exceed the marginal costs from congestion and other diseconomies. Are industries with close links to agriculture likely to be found in smaller cities? If so, then by the theory sketched above, these industries should show only mild benefits from spatial clustering. An empirical analysis for Korea by Hen- derson, Lee, and Lee (2001) confirms that localization effects for food, wood and paper products, furniture, textiles, and apparel are significantly smaller than the ef- fects for heavy industries, transport, machinery, and the high-tech sectors. These findings may well apply more generally. Of course, many other factors such as the savings on transport costs achieved by processing raw agricultural outputs and natural site-specific advantages may be just as important as localization effects to the economic life of small cities. Although in poor countries these cities ac- count for a large share of the urban population, researchers have yet to understand just how their economic health is sustained (but see Hardoy and Satterthwaite, 1986a,b). Much more attention has been paid to the other end of the city size distribu- tion. A country's largest city may have attracted much of its core employment by way of localization and urbanization effects, but its character is also likely to have been shaped by the political economy of national development. Discussions of primacy the share of urban population taken by a country's largest city often begin with an argument due to Williamson (1965) that emphasizes sequential de- velopment strategies. Enormous public and private resources are required to or- ganize cities efficiently, and these resources capital and managerial capacities among them are likely to be very scarce in a country's early stages of develop- ment. To conserve resources, policy makers may at first restrict their attention to one region, and turn to others only when the potential of that region appears to have been well exploited. High levels of primacy are thus to be expected in coun- tries in the early stages of development. Eventually, however, Reconcentration and lower levels of primacy should accompany, and will themselves stimulate, sustained economic growth (Henderson, 2002~. Decentralized city systems cannot be expected to evolve until substantial in- vestments are made in interregional and intercity transport. In the case of Korea, a massive program of investment in intercity infrastructure in the 1970s led to significant industrial Reconcentration in the 1980s. Here, political factors- especially access to Seoul bureaucrats proved to be important in delaying decon- centration, as were the time lags involved in extending good-quality education and other public services to areas outside the capital (Henderson, Shalizi, and Venables, 2000; Henderson, 2002~. As discussed in Chapter 8, persistent urban biases in government pricing and social investments can also prolong the life of overconcentrated and inefficient city systems.
60 CITIES TRANSFORMED In highly urbanized regions such as Latin America, an appreciation of urban change requires a reconceptualization of the linkages among cities. The most in- teresting aspect of Mexico's urban transformation between 1980 and 2000 is not that its urban population share rose from 55 to 67 percent, but that its city system evolved from a monocentric structure dominated by Mexico City to a complex polycentric structure linking the country's nine largest cities (Garza, 2002~. In some densely populated regions of the developing world, large cities are overlap- ping as they expand, merging to form extended metropolitan areas. To appreciate the rapidity of such developments, consider that it was just 40 years ago when the first "megalopolis" was identified along the northeastern seaboard of the United States (Gottman, 1961~. Today such city systems are much in evidence in poor countries (Ginsburg, Koppell, and McGee, 1991; Hugo, Champion, and Lattes, 2001~. Spaces and Networks at the Urban/llural Interface increases in rents, prices, and congestion in the central areas of large cities, to- gether with the operation of product development cycles, would be expected to encourage the spatial Reconcentration of manufacturing. Many researchers have noted that city centers tend to cede new manufacturing jobs to the urban periphery. Having done so, the central areas are freed to host a new generation of special- ized services in telecommunications, banking, law, financial management, man- agement consulting, and information technologies (recall Box 2.4~. As reliable transport and communication systems reach into the urban periphery, large-scale projects requiring land and extensive capital whether for new airports, manu- facturing plants, or office space can be relocated in that periphery. Residential housing obeys a similar spatial logic. Something of this dynamic can be seen in the recent history of Sao Paulo, as depicted in Figure 2-2. In the 1970s, rates of (residential) population growth were relatively high across the municipio on the order of 3.7 percent on average. By the 1980s, however, net population declines were already being seen in the central and interior areas, and by the l990s, population growth rates had collapsed in these areas, falling well below2.0 percent. Although population growth rates also declined somewhat outside the core areas of Sao Paulo, growth rates in the outskirts remained well above those in the city core. Yet even while residential growth was declining in the core areas of the city, these areas were accounting for a large share of new employment growth, especially in producer services (not shown in the figure). If the general nature of these developments is predictable, their spatial ex- pressions are surprisingly varied and complex. Many large cities are extending outward in an irregular and seemingly haphazard fashion. Development is not evenly spread, but spatially concentrated at transport nodes (railway stations, river ports, and highway intersections), at scattered industrial sites (such as quarries and
WHY LOCATION MATTERS 1 980-1 991 '..n : I... ,~., / ,.h,, ~2."""""2~:. 61 1 970-1 980 ''''-''c ~ ~'~s '.,,,,,, . .' if ,.... A.,. .......... Population Growth Rate Morethan2.1% ~ From 0.1 %-2% 1~1 From-o.g%_o% 11111 From-1.9%-1% ~ Less than -2% FIGURE 2-2 Population growth rates within metropolitan Sao Paulo, 1970-1980, 1980-1991, and 1991-1996. SOURCE: Taschner and Bogus (2001~. landfill sites), and in residential "bedroom" communities. The villages, towns, and small cities in the vicinity of a large city often develop their own diverse workforces, whose fortunes are tightly bound to the major city. It has been es- timated that almost half of urban development in poor countries occurs outside established city boundaries, where local governments are often fragmented and poorly equipped to handle the demands of city building (Webster, 2000a). Urban and rural activities can be interwoven in ways that thoroughly refashion the social and spatial fabric. In Southeast Asia, for instance, zones have emerged in the intersections between cities and their rural areas that are neither urban nor rural in any conventional sense (Ginsburg, Koppell, and McGee, 1991~. McGee (1991) refers to these as desakota zones, a word derived from the Indonesian words for village (desa) and town or city (kota). The casual visitor to a de- sakota zone sees a landscape that appears to be rural, with almost all land under
62 CITIES TRANSFORMED cultivation. But despite appearances, most income is now derived from nonagri- cultural sources, many of these being urban based. The nature of agricultural pro- duction has also shifted, with subsistence production giving way to higher-value products targeted to urban and world markets (McGee, l991~. Data on such rural areas underscore the importance to rural well-being of nonfarm employment and other nonagricultural activities. Diversification of rural income sources gives the rural poor a means of protecting themselves against agri- cultural risks (Bryceson, 1999; Ellis, 1998; Lerise, Kibadu, Mbutolwe, and Mushi, 2001; Schejtman, 1999~. Risks are also reduced through spells of migration to urban areas (Blitzer, Davila, Hardoy, and Satterthwaite, 1988~. As discussed in Chapter 8, information about the urban jobs open to migrants often filters through to rural areas by way of social and family networks.23 If the rural dependence on the city economy is well recognized, it is less often appreciated that many urban-based households depend on rural-based assets and incomes. Urban migrants frequently make investments in rural housing, land, or cattle, with rural relatives overseeing these assets in the migrant's absence (Afsar, 1999; Kruger, 1998; Smit, 1998~. Like their rural counterparts, the urban poor have little alternative but to rely upon income and asset diversification; better- off urban households also maintain a mix of assets to accumulate capital (Baker, 1995~. The networks of economic flows linking urban and rural households in- clude not only the remittances sent from urban to rural households, but also the foodstuffs that are conveyed in the opposite direction. In sub-Saharan Africa, it is common for urban children to be fostered into the care of their rural relatives when city living conditions make urban child care problematic. On their part, the urban relatives lend critical support to newly arrived rural migrants. As all this sug- gests, it is more than a little artificial to describe households as being wholly rural or urban many of them are deliberately multispatial in their economic strategies, especially in the zones surrounding large cities. As cities extend themselves and bid for rural land, this process inevitably dis- rupts rural economic relationships and redistributes income. Urban expansion can increase rural productivity by giving landed farmers better access to roads, ports, airports, credit, and information about market opportunities. But tenant farmers, sharecroppers and others without land may find their livelihoods curtailed by the city incursions; only some of them will be able to secure more lucrative nonagri- cultural employment. The rural residents most at risk would appear to be those with little or no land, who are dependent on wage or casual agricultural labor and lack the skills, contacts, capital, or freedom of movement to take advantage of new opportunities (Rakodi,1998~. The terms on which rural and urban populations engage are determined in part by national development strategies. A study of Hebei province in China shows Bother types of information also flow across the urban/rural boundary, as was the case with contra- ceptive information in the Taichung experiment discussed earlier.
WHY LOCATION MATTERS 63 how national economic restructuring can impinge upon the rural populations liv- ing near cities (Benziger,1996~. In the proreform era in China, when the prevail- ing ideology of "self-reliance" directed state industries and national investments to relatively small Chinese cities, it was the rural counties near such cities that captured the productivity benefits. But in the 1980s, as economic reforms began in earnest, the role of the Chinese state in resource allocation came to be sup- planted by that of the markets. Larger and more dynamic Chinese cities began to take the lead in attracting foreign investments and engaging with world markets, and, as Benziger (1996) shows, it was their rural counties to which the greatest productivity benefits then accrued. Rural and urban households are knit together in numerous mutually beneficial economic and social networks; as groups, however, rural and urban populations have distinct and at times sharply conflicting interests (Douglass, 1989; Kelly, 1998~. The urban/rural interface is contested terrain, especially where urban and rural populations both make demands on common natural resources, such as water (UNCHS, 1996: 150-151~. Cities and their industries consume great quantities of water; in many countries they also depend on rural rivers and dams to generate electricity. Urban wastewater is deposited in the rural areas near cities and can flow into coastal fishing grounds, which are often overwhelmed by its content and volume. As Showers (2002) documents for sub-Saharan Africa, when such urban demands are imposed upon fragile rural and coastal ecologies, the environmental stress can leave some areas unfit for agriculture. In principle, at least, pricing schedules for water and electricity could be de- vised to take these social costs into account. To ensure that the rural costs are fully appreciated, such pricing policies would probably have to be administered by the governments that oversee the rural and urban populations involved. But the zones around cities would present these governments with vexing problems of equity and efficiency. If water were indeed to be priced according to its social marginal costs, this practice would likely restrain urban growth and increase the costs of living for the urban poor. But slower urban growth would also damage the prospects of some of the rural poor. Regional governments would have to balance the interests not only of urban and rural populations, but also of the poor in each of these populations. The multiple linkages between rural and urban populations imply that rural and urban poverty cannot be viewed as presenting wholly separate and distinct sets of problems. An increase in urban poverty usually leaves fewer city jobs for rural dwellers, reduces flows of remittances from urban to rural households, lessens urban demand for rural products, and may even spur urban-to-rural migra- tion. Likewise, falling crop prices or declining rural productivity may have initial impacts on rural poverty, but urban incomes may suffer as demand declines for the goods and services provided by urban firms to rural areas. Curiously, how- ever, most governments and international agencies operate as if rural and urban poverty were wholly separable.
64 CITIES TRANSFORMED To sum up, with cities having grown and projected their influence across space, the "city-region" now deserves consideration as a unit of analysis for governance and policy. Although difficult to define with precision, a city-region is identifiable by the extent and nature of economic activity in an economic zone surrounding a large city. Many such regions have grown enormously over recent decades. The Extended Bangkok Region, for example, now contains more than 17 million peo- ple; by 2010 it is expected to extend some 200 kilometers from its current center (Kaothien and Webster, 20014. Such new regional forms, with their highly diverse populations, will require innovative approaches to planning and administration. FROM GOVERNMENT TO GOVERNANCE Governments provide the legal and regulatory structures within which social and economic interactions take place; they arrange for the delivery of public services; and they attempt to manage the externalities and conflicts that inevitably accom- pany social interaction. For these reasons, governments are inescapable presences in local urban spaces. As we have just seen, however, cities are assuming com- plex spatial forms, often extending into terrain where the lines of governmental authority are muddled and casting influence across regions that include substan- tial rural populations. These developments are presenting governments with new needs to mediate among diverse demands (Simmonds and Hack, 2000; Scott, 2001). The forms in which governments project themselves into these spaces have also been rapidly changing. In many developing countries, local and regional governments are taking on greater prominence, while national governments are stepping back into indirect and seemingly less intrusive roles. As described in Chapter 9, a process of decentralization is under way, whereby national govern- ments are devolving to lower-level governments many political, fiscal, and admin- istrative powers. Across the developing world, new local governmental forms and units are proliferating at a rate that is little short of astonishing. This phenomenon is in part the result of growing agreement that effective urban management re- quires new formal structures of government (Sivaramakrishanan, 1996~. It also owes a great deal to the introduction of democratic principles in many countries, and to the increased importance being accorded to citizen and community voice (UNCHS, 1996~. These are welcome developments in many quarters; yet they imply that for some time to come, the levers of local policy will be manipulated by new and inexperienced governmental actors. As governments are resealed into smaller spatial units, they are engaging more directly with private-sector actors and NGOs. The term governance describes this engagement. It refers to a set of relationships: between the state and civil society, between rulers and the ruled, and between governments proper and those who are governed. Good governance is, in part, the outcome of government processes that are transparent, as executed by bureaucracies instilled with a professional ethos and accountable for their actions. In a healthy system of governance, these
WHY LOCATION MATTERS 65 structures of government are engaged with a civil society that for its part takes an active role in public affairs; in such a system, all parties adhere to the rule of law (Sivaramakrishanan, 1996~. The reciprocity and mutual engagement entailed in good governance should build trust and lend support to the development of local and national social capital (World Bank, 2000a). As Stren (2002) observes, there is a certain romantic quality to some discus- sions of governance, which imply that moving governments closer to the peo- ple (the "grass roots") must heighten sensitivity to local needs and bring more democracy and transparency to the processes by which these needs are addressed. The current wave of decentralization is far more widespread than its historical predecessors Stren notes two comparable "moments" of decentralization in Africa and Asia, the first in the period surrounding independence and the second in the 1970s but many of the warnings sounded earlier about limits and risks still warrant attention. It is exceedingly difficult to measure the efficiency and responsiveness of lo- cal governments, and empirical evidence on their performance is thus far mixed (World Bank, 2000a). Theories of public finance point to several potential ad- vantages of small, localized governmental units. In decentralized systems, local governments acquire a stake in local economic prosperity. They can arrange the menu of local public goods to suit local preferences, although the quantities sup- plied will still be constrained by local revenue-raising capacities and transfers from other levels of government. In such systems, local consumers can express their preferences for bundles of public goods by voting or by moving to other juris- dictions (Tiebout, 1956~. Under ideal conditions, local politics can then achieve something of the efficiency of markets. The increasingly globalized nature of economic relations is another factor to be considered. Local firms working with nimble, entrepreneurial local governments can collaborate to attract foreign direct investment, sometimes by sidestepping central government authorities or involv- ing them only minimally (UNCHS, 2001~. At the same time, however, small governmental units can suffer from signif- icant disadvantages. Some aspects of governance and regulation may lie well beyond their technical and revenue-raising capacities. Unless transfers from higher-level governments are well designed (see Box 2.5), local governments in have-not regions will rarely be able to marshal the resources available to those in wealthier regions, and if such tendencies are left unchecked, the result can be pronounced regional inequities.24 In decentralized systems, higher-level govern- ments need to devise ways of managing the externalities that spill across local gov- ernmental boundaries.25 In addition, when the national government cedes power 24Discussing how systems of intergovernmental transfers can be designed to promote efficiency and equity, Bird and Smart (2002) note that adverse selection and related behavior on the part of local governments can defeat the good intentions of the system designers. 25Some observers blame weak national states and porous national safety nets for the growth of megacities and the expansion of slums in developing countries, whose cities simply lack the tools to manage national-level demographic and economic flows (Tulchin, 1998).
66 CITIES TRANSFORMED BOX 2.5 Intergovernmental Transfers and Targeted Social Assistance Alderman (2001) describes the case of intergovernmental transfers from the national to the local (commune) level of government in Albania, where a social assistance program is in place to help the poor. The Albanian national government lacks all but the most rudimen- tary data on poverty at the local level. To allocate its transfers among local governments, the national government employs ad hoc criteria that appear to be very weakly related to local poverty rates. Although funds are evidently well distributed once they reach the local level, the system as a whole fails to make the best use of resources. Alderman (2001: 50) concludes that "to take advantage of local governments' assumed access to local informa- tion, there must be a corresponding flow of information to the center as well as an incentive to use this information." For instance, census data can be used to generate poverty rankings at the level of local governments, and such spatially disaggregated data can provide the national government with tools to improve its resource allocation. to local governments, representation is not guaranteed to all local interest groups. In some cases, the Revolution appears to do little more than transfer power from national to local elites. Partnerships of local firms and local governments can in- vite corruption and render local political processes opaque where transparency is the ideal. Finally, decentralization can threaten macroeconomic stability if central governments lose control over total public outlays. The phenomenon of decentralization with all its attendant risks and ben- efits, often heatedly debated in the countries involved does not yet appear to have engaged the attention of the international demographic research community. Perhaps in many countries, health and family planning services are still being delivered through vertically organized ministries of health, much as they have been for decades. But in many other countries, the decentralization of these ser- vices is being actively contemplated, and in some it is already well under way. A recent analysis by Schwartz, Guilkey, and Racelis (2002) in the Philippines employed rare before-and-after data on local governmental units to determine whether decentralization has affected rates of child immunization and the use of family planning. In this case, it appears that the transfer of resources from national to local authorities has increased local resources overall. The additional resources have evidently encouraged the use of family planning, although they do not appear to have had the same impact on immunization. Until researchers can assemble more case studies such as this, the implications of decentralization for reproductive health will remain highly uncertain. When de- centralization confers greater authority over health and family planning services on municipal governments, which have long lacked professional staff and man- agerial expertise, on what basis will these governments make their decisions about resources and policies? Will they possess the requisite technical abilities, and the revenue-rasing capacities, to wield their newly assumed powers effectively? What role should national-level professional associations play, along with the national
WHY LOCATION MATTERS 67 ministries, in seeing that technical expertise is made available to small govern- ments? Perhaps the only certain element in all of this is that the international policy dialogue in reproductive health, which has in the past been a matter of dis- cussion with national ministries and NGOs, will soon have to engage on a broader front with the many new units of government and local NGOs that populate de- centralized settings. WHAT REMAINS OF THE URBAN/RURAL DIVIDE? In concluding, we survey the broad concepts that have entered this discussion and ask whether they point to specific features that distinguish urban from ru- ral landscapes. The urban/rural distinction is one that has been contemplated by generations of thoughtful scholars, few of whom have failed to note its many in- tricacies. The urban concept is an abstraction that involves multiple distinct but interrelated social, economic, political, and ecological factors (McGee and Grif- fiths, 1998; Frey and Zimmer, 20014. Furthermore, when carefully considered, the differences between urban and rural populations are almost always seen to be differences in degree rather than in kind. In almost any aspect that might be con- sidered, urban and rural populations have something in common, and they often overlap substantially. The conceptual challenge, then, is to identify the central tendencies without denying the commonalities. At the outset we referred to five concepts that tap distinctive aspects of ur- ban social and economic relations proximity, diversity, externality, network, and centrality. While giving attention to the first four of these, we have not commented much on the fifth. Centrality is a summarizing concept: in our usage it refers to the multiple strands of economic and social interchange that are knit together in cities (Sassen, 20024. These strands also reach to rural areas, and they have links that extend to the international arena. But they intertwine in cities, and from the many knots and nodes there emanates a quality that might be described as urban- ness. In employing this sort of language abstract and rather tentative we are of course signaling the many difficulties that would be involved in moving from summary concepts to their empirical measures. This report takes the position that urbanness is best conceived in terms of a continuum, or gradient, along which individual populations are arrayed. The dis- cussion earlier in this chapter referred to city-regions and desakota zones, phrases that are suggestive of a blurring between urban and rural populations. Are urban and rural areas now so thoroughly intermixed that the urban/rural distinction has lost its analytic value? We think not. As will be seen repeatedly in the chapters to come, even very crude indicators of position on the urban/rural continuum the definitions of urban and rural that are adopted by national statistical agencies- are empirically powerful in explaining demographic behavior. Whatever these conventional measures may mean, they somehow succeed in capturing impor- tant locational differences. But the empirical performance of crude indicators
68 CITIES TRANSFORMED notwithstanding, the concept of an urban/rural divide appears to be losing what- ever intellectual appeal it may once have had. The difficulty is how to devise satisfactory measures of the alternative concept of an urban/rural continuum- with attention to the many ways in which rural residents are now partaking of urban life. To appreciate the empirical challenge, consider the case of Real Montecasino, a settlement of about a thousand residents located just south of Mexico City, wedged between the Federal District and the Metropolitan Zone of Cuernavaca. The 2000 Mexican census classified Real Montecasino as rural. Yet only 1.8 per- cent of the its labor force is engaged in the primary sector; 83 percent of its houses have electricity, piped water, and waste disposal; 70 percent of its households own telephones; 66 percent own cars; and 92 percent have televisions (Garza, 2002~. Despite its small population size, Real Montecasino is arguably an outpost of Mexico City. To distinguish such fine gradations in the urban/rural continuum, criteria such as the degree to which cities are accessible from rural areas (or remote from them) will need to be explored in some detail, making use of all available census and survey data on commuting times and spells of short-term city residence (Coombes and Raybould, 2001; Hugo, Champion, and Lattes, 2001~. Appendix A gives an account of recent efforts in the United States to rethink urban measurement, and very similar issues face the national statistical services of many developing coun- tries. Advances in geocoding may enable researchers to link many different sorts of data, thus permitting more sophisticated measurement (Hugo, Champion, and Lattes, 2001~. A glimpse of the possibilities is given in Box 2.6 for Cairo, where a combination of remotely sensed and census data permits a gradient of urbanness to be distinguished within the Greater Cairo metropolitan area. But to measure the micro-level aspects of social and economic interaction will surely require en- tirely new forms of data collection; it is doubtful that data gathered routinely by censuses or satellites will suffice. Although the concept of an urban/rural divide should perhaps be readied for the scrap heap, much more research will be required for the concept of a continuum to be put into a useful and operational form. Table 2-1 summarizes the main urban/rural differences as seen from a de- mographic perspective. This chapter has emphasized the social embeddedness of information and behavior (Granovetter, 1985), drawing attention to the ways in which individuals and families are linked to their social networks, neighbor- hoods, and local associations; how they are connected to the larger structures of government; and how they may be engaged as groups in relations of governance. Although a multilevel perspective can be highly informative about rural societies, we would maintain that such that a perspective is essential to an understanding of urban demography. As our review of the U.S. sociological literature shows, this is hardly a novel or controversial perspective, but its insights have yet to be developed in the contexts of developing-country cities.
WHY LOCATION MATTERS 69 BOX 2.6 Using Multiple Data Sources to Define Urbanness: The Case of Cairo ~ ;;;;;;;;;;;;;;;;;;;;;; ..,., .... ............ . ~. I,:......... ................ ............................. i ~ ........ , ., A .. ...... ~ . . ...... . ~ A is, , ............. `, . ~ ~, , ~ ., . t ~ ., \, I ~ ~. Am. ............ - ~ ...... air - . . ........ , ., ~ .............. .b . ?:::::::: ;; At,. ~ . I . ~ s.~ ............ ....... .......... ) ,P,- ,., . .,.-, ,~ ,:;:;:;:;:;:;:;:;:;:;:;:.s:; . ~. ............ ...... ,. i.......... I x This figure, adapted from ongoing research by Weeks (2002), depicts a composite index of urbanness derived from an unusual blend of remotely sensed data on land cover (indicators of vegetation, impervious surfaces, bare soil, and the shade cast by buildings) and census data on population density and the proportion of the labor force in nonagricultural occupa- tions. The spatial units represented are shiakhas, of which there are some 300 in Greater Cairo. The map shows a gradient with the highest values of urbanness (portrayed in dark shading) in the center of the city straddling the banks of the Nile. Urbanness declines as one moves toward the newer urban areas to the west of the Nile (in Giza governorate). Weeks has found that the composite urbanness index is correlated with several demographic measures at the shiakhas level: the areas classified as more urban have lower fertility, later ages at marriage, and greater education.
70 TABLE 2-1 Dimensions in Which Urban Environments Differ from Rural Social: (1) The spatial proximity of social and economic diversity; (2) the range and weak ties of social networks; CITIES TRANSFORMED (6) higher urban incomes on aver- age, possibly with greater income disparities. (3) a social delineation of neighbor- Health <1y Greater inherent risks of com- municable disease, including those that are sexually transmitted; (2) pos- sibly lower unit costs for provision of clean water; (3) a different range of occupational health and safety risks; (4) greater numbers of urban poor at risk from some natural disasters because of population concentration; (5) possible economies of proximity in health media campaigns, and infor- mational spillovers from the educated to the less-educated and through so- cial networks; (6) greater access to health services through private mar- kets and mixed public-private provi- sion; (7) quicker access to emergency services, of great importance to ma- ternal mortality; and (8) composition of disease within the population al- tered by higher incomes and better public provision of services. hoods, with neighborhood ties be- ing subsets of wider network ties; (4) informational spillovers and other externalities; (5) spatial segregation; and (6) distinct forms taken by urban social capital (including gangs and "perverse" forms), and the possibility of "bridges" to government and fund- ing resources. Economic: (1) Scale, spillover, and di- versity effects; (2) far greater special- ization and diversity in private mar- kets, such as in health services; and (3) greater utilization of physical cap- ital and infrastructure. Human Capital: (1) Easier access to middle and secondary schooling; (2) greater visibility of educated ref- erence groups and role models; and (3) greater social risks attending child rearing, implying higher costs in parental time. Prices and Consumption: (1) Costs of living and incomes more monetized; (2) greater exposure to variation in wages and prices, hence greater sub- jective sensitivity to their levels; (3) a greater range of goods and services available; and (4) greater visibil- ity of diverse consumption reference groups. Livelihoods: (1) Nonagricultural occupa- tions far more prevalent; (2) fewer possibilities for own production and consumption of food; (3) possibly greater returns to human capital; (4) for urban households, nothing quite comparable to Green Revolu- tion agricultural technology in rais- ing productivity; (5) greater eco- nomic value of urban housing; and Basic Services: (1) Greater percentage of households with water supply, waste disposal, and electricity; (2) differ- ent dimensions of access, with qual- ity, reliability, and adequacy of ser- vice taking on greater importance, and time costs often of lesser impor- tance; and (3) a greater reliance on il- legal forms of access to basic services and housing. Government: (1) Greater dependence on government implied by urban pop- ulation concentration and diversity; (2) greater exposure to a multiplicity of laws and regulations; (3) possibly greater vulnerability to "bad" govern- ment; (4) especially in large cities, multiple layers and units of gov- ernment; and (5) possibly (in some cities) greater ability of local govern- ments to raise their own revenues.
WHY LOCATION MATTERS 71 Many of the points mentioned in Table 2-1 have already been described at length or are taken up in more detail in later chapters. Our comments here can be brief. A number of the distinctively urban social features stem from one source: spatial proximity brings socioeconomic diversity into focus. Proximity allows in- formation to flow more easily among social network members; it highlights social reference groups and role models; and it puts diverse consumption possibilities on view. As mentioned earlier, spatial segregation is likely to have profound de- mographic implications because it suppresses diversity in the local environs. In the economic realm, proximity enables firms and entrepreneurs to learn from the experiments, successes, and failures of their competitors. The spatial dispersion of rural populations and the greater homogeneity of much agricultural production generally raise the costs of such social and economic exchange. In advancing such broad and general claims, we are mindful of important counterexamples. It was in the context of agriculture, after all, that the early theories of information diffusion and adoption of new technology were formu- lated (Griliches, 1957; Hagerstrand, 1952), and recent data drawn mainly from rural sites provide the most convincing demographic demonstration of diffusion operating through social networks (Behrman, Kohler, and Watkins, 2001; Caster- line, Montgomery, Agyeman, Aglobitse, and Kiros, 2001~. Moreover, some may object to giving urban diversity greater emphasis than city size. Large cities do tend to exhibit greater diversity than small ones (Henderson, 2002), and in the economic arena, scale is something of a precondition for specialization and diver- sity. Nevertheless, scale and diversity are conceptually and empirically separable features of urban environments. As Henderson has shown for high-technology industries in the United States and Korea, it is diversity rather than city size as such that generates productivity advantages for these industries. Where social capital is concerned, there are likely to be many differences between its urban and rural expressions; to our knowledge there has been no sys- tematic study of those differences. Residential mobility and migration are thought to weaken the basis for cooperation in city neighborhoods. Yet it is difficult to know whether urban areas are, in general, sites of high residential mobility. In the panel's own research experience, many city neighborhoods have proven to be residentially stable (see UNCHS, 1996: 206 for confirmation). Certainly urban environments do not prevent the mobilization of social capital. The literature offers numerous examples of strong, effective, and inclusive urban community organizations; recall Box 2.2 on the alliance of SPARC, Mahila Milan, and the National Slum Dwellers Federation in Mumbai. Indeed, urban settings would often appear to provide more opportunities for community organizations to negotiate with government agencies (Appadurai, 2001; Boonyabancha, 2001; Baumann, Bolnick, and Mitlin, 2001; Tostensen, Tvedten, and Vaa, 2001~. The urban engagement between government and civil society is especially apparent in countries with democratic systems, where there are political and legal restraints on the power of government to suppress community mobilization.
72 CITIES TRANSFORMED The differences in human capital are explored further in Chapters 5 and 8. As is shown there, urban residents are more educated on average than rural resi- dents, and would appear to enjoy easier access to middle and secondary levels of education for their children. Less often appreciated, however, is the diversity of educational opportunities for urban adults and children. What matters in cities, we would argue, is not only the higher average levels of educational attainment, but also the greater diversity of educational experiences. As discussed in Chap- ters 5 and 6, in addition to the demands on parental time associated with children's schooling, time costs arise from the distinctive social risks of urban child rearing. Of course, these urban/rural differences should not be exaggerated. Urban returns to schooling vary across economic sector and by city size and diversity; education is known to help rural farmers exploit new agricultural technologies. As is shown in Chapter 5, school enrollment rates among the children of the urban poor often are hardly greater than those among rural children. In both rural and urban settings, the need for child labor can keep children from attending school regularly or at all. It is a commonplace that urban populations rely more heavily than rural pop- ulations on cash income for access to necessities including food, fuel, fresh wa- ter, housing (which is more commercialized in cities), transport, and waste dis- posal. Monetization reduces transactions costs and raises real standards of liv- ing, but with these benefits comes a greater vulnerability to changes in money wages and prices. The fact that most urban goods and services are monetized may also induce in urban populations a keener appreciation of relative costs in general, and may draw special attention to the relative costs of child rearing. To be sure, monetization is probably more characteristic of larger than smaller cities. In many countries, a significant proportion of rural dwellers are also de- pendent on cash income, and they, too, can face variable prices for some goods and services. As discussed in Chapter 5, it is difficult to know just how much the prices of essentials differ between urban and rural populations. The costs no doubt vary enormously among rural areas themselves, among cities, and among dif- ferent neighborhoods within cities. In rich countries, advances in transport and communications, together with sophisticated systems of wholesale and retail trade, have suppressed much spatial variation in prices. These factors operate with far less force in most poor countries. In many of their cities, the urban poor face particularly high costs for such essentials as water and health care because public services are not reliably provided to poor neighborhoods, and the private markets offering substitutes can be highly imperfect or exploitative. Still, rural populations may face high money costs for some goods as well. As recent economic studies have shown at the national level (Limao and Venables, 2001), the costs of transporting goods to rural areas are reflected in two sorts of penalties: higher prices and severe limits on the range of goods available. Much like landlocked countries, rural populations often suffer both of these penalties.
WHY LOCATION MATTERS 73 For many rural dwellers, access is limited by the inconvenience and relatively high cost of transport, which for a given physical distance renders less accessible schools, health centers, emergency services, courts, banks, politicians, and the institutions meant to enforce the rule of law. For the urban poor, it is not so much distance to services and institutions that matters, but rather exclusion from them for economic, social, or political reasons. A squatter household living 200 yards from a hospital, secondary school, or bank can be as effectively excluded as a rural dweller living 20 miles away. Proximity may ease access, but does not guarantee it. Discussions of urbanness often begin by noting the prevalence of nonagri- cultural occupations, and it is certainly true that urban livelihoods are less di- rectly dependent than rural on access to land, water, and other natural resources. Urban residents cannot easily turn to subsistence production to cope with ris- ing prices or declining incomes. However, urban agriculture is more important to low-income residents than is commonly realized. Also, as discussed earlier, many urban dwellers maintain some claims on rural assets. In the same way, rural households can depend on nonfarm income, whether from wages, nonagricultural production, or urban remittances. Housing is a key economic resource for low-income urban residents: it can supply income (from the renting out of rooms or as space for household-based enterprises); it has value as collateral; and it reflects trade-offs made in access to employment, as when the poor accept low-quality or dangerous locations to save on transport costs. Rural housing can also play an economic role (as with food processing and crafts), but generally this role is of lesser importance to household economic strategies. Earlier in this chapter, we described the greater risks of communicable dis- ease faced by city populations in the absence of adequate infrastructure and good governance. Higher levels of health risk are very much to be expected in ur- ban areas lacking provisions for infrastructure, services, and waste management. Dispersed rural populations enjoy a measure of natural protection from much communicable disease. (Some large rural villages can also suffer from urban- like concentrations of population and pollution due, for instance, to livestock and agroprocessing.) Massive public-sector investments are required to convert an in- herent urban health disadvantage to the urban advantage that is often taken as a given in modern populations. Cities exhibit a different range of occupational health and safety risks than is seen in rural areas as a result of differences in the kinds of work undertaken (involving industrial chemicals and wastes, dust, heat, or dangerous machinery). Particular groups (such as waste pickers) face especially high occupational risks. But rural occupational health and safety risks should not be understated. There are high levels of health risk in many rural areas due to poorly managed irriga- tion (schistosomiasis, malaria), agricultural chemicals, dangerous machinery, and excessive physical demands. Finally, the spatially concentrated urban poor are
74 CITIES TRANSFORMED vulnerable to natural disasters because they live on land at high risk from floods, landslides, or earthquakes. Rural populations are also vulnerable to natural disas- ters in many countries. A central theme in this discussion is the pervasive influence of governments. Cities are marked by a multiplicity of laws, official norms, rules, and regula- tions that can be applied to land use, construction, economic enterprises, and production. To many observers, it appears that a regrettably common use of these regulations is to render illegal many of the means by which the urban poor gain access to their housing and livelihoods (Hardoy and Satterthwaite, 1989~. Because access to services is less a matter of distance than of ability to pay and political clout, one finds in cities a greater reliance on illegal solutions for access to services such as illegal taps of piped water and electricity and this carries over to illegally occupied or subdivided land. Illegal or informal settlements are often concentrated on land sites subject to flooding or at risk from landslides or other natural hazards, especially where these sites offer low-income settlers the best chance of establishing a home or avoiding eviction. Often these sites also prove to be difficult to equip with basic infrastructure (Hardoy, Mitlin, and Sat- terthwaite, 2001~. The spatial concentration and visibility of urban populations may well leave them at the mercy of bureaucracies and powerful vested interests. On occasion, however, spatial concentration can also confer on the poor a certain political mass and even a measure of power. Stren (2002) notes that in Latin America, a common strategy among poor groups was to stage mass "land invasions" in an effort to secure access to urban land. Although not always immediately successful, this strategy enabled some poor groups to voice effectively their claim to a share of public resources. In summary, Table 2-1 shows the main elements the panel believes lend ur- ban landscapes their distinctive character. In considering each of these elements and in highlighting exceptions and counterexamples, we have endeavored to show how along each dimension, the urban/rural distinction is mainly a question of degree. The quality of urbanness which eludes definition, but is somehow easy to sense emanates less from any single dimension listed than from their combination.