Residential segregation and unemployment: The case of Brussels

Transcription

1 Residential segregation and unemployment: The case of Brussels Claire DUJARDIN *, Harris SELOD **, and Isabelle THOMAS *** November 16, 2005 * FNRS Research Fellow, Department of Geography and CORE, Université catholique de Louvain, Place L. Pasteur, 3, B-1348 Louvain-la-Neuve, Belgium. Fax: dujardin@geog.ucl.ac.be ** INRA Researcher (PARIS-Jourdan), CREST and CEPR, France. Fax: selod@ens.fr *** FNRS Senior Research Associate, Department of Geography and CORE, Université catholique de Louvain, Belgium. Fax: isabelle@geog.ucl.ac.be 1

2 Acknowledgments The authors would like to thank Patrick Deboosere and the Point d Appui Démographie (Vrije Universiteit van Brussel) for providing them with the data, as well as four anonymous referees for useful comments. Harris Selod acknowledges the support of the European Commission while he was visiting CORE as a Marie Curie Fellow from the EC (Grant HMPF-CT ). 2

3 Residential segregation and unemployment: The case of Brussels Summary This paper investigates the causal effects of the spatial organization of Brussels on unemployment propensities. Using 1991 Census data, we estimate the unemployment probability of young adults while taking into account personal, household and neighborhood characteristics. We solve the endogeneity of residential locations by restricting our sample to young adults residing with their parents, and evaluate the potential remaining bias by conducting a sensitivity analysis. Our results suggest that the neighborhood of residence significantly increases a youngster probability of being unemployed, a result which is quite robust to the presence of both observed and unobserved parental covariates. Keywords: Neighborhood effects, Residential segregation, Unemployment, Endogeneity bias, Sensitivity analysis. 3

4 Introduction For decades, sociologists, economists and geographers have written extensively on how the spatial structure of cities reflects socioeconomic differences in the population (see the seminal contributions of Burgess, 1925, Hoyt, 1939, or Harris and Ullman, 1945, and their respective descriptions or models of urban stratification). Strikingly, most cities today are characterized by stark disparities opposing city centers and peripheries. In the U.S. for instance, inner cities are usually poor and the catalyst of many social problems, whereas suburbs are more well-off. Brussels exhibits a similar spatial structure since its inner city concentrates unemployed workers and disadvantaged communities, including many unskilled workers and ethnic minorities (Vandermotten et al, 1999; Thomas and Zenou, 1999; Goffette-Nagot et al, 2000; Kesteloot et al, 2001). To explain the general pattern of residential segregation in urban areas, the economics literature has often stressed the role of location choices through which individuals spontaneously sort themselves into a city according to their different socioeconomic characteristics. In this respect, residential segregation is a standard result in both public and urban economics (see Tiebout, 1956, for whom individuals vote with their feet for the provision of local public goods, or Fujita, 1989, for a formalization of the standard monocentric model with heterogeneous agents). However, urban stratification should not just be considered as a process whereby households with different socioeconomic backgrounds sort themselves in the city. As a matter of fact, an abundant literature has investigated the reverse 4

5 causality, arguing that the spatial organization of cities could explain differences in social and economic outcomes. In the U.S. context, it has been argued that the high level of unemployment among inner-city minorities could be explained by residential segregation and/or disconnection from job opportunities (see Kain, 1968; Kasarda, 1989; Jencks and Mayer, 1990; Wilson, 1987 and 1996, Cutler and Glaeser, 1997). Even though the theories which link labor-market outcomes to the spatial organization of cities have mainly inspired empirical papers on U.S. metropolitan areas, the mechanisms put forward clearly have a general validity. But only a few works have focused on European cities (see for instance Fieldhouse, 1999, on London, or Gobillon and Selod, 2006, on Paris). The objective of the present paper is to test whether city structure can be a source of unemployment in the Brussels metropolitan area. Despite the huge interest on the topic, the bulk of empirical studies still has not reached a consensus regarding the role of spatial factors in explaining individual labor-market outcomes. As argued by Ginther et al (2000) or Dietz (2002), this lack of consensus can probably be explained by the great diversity regarding the methods used to test for the existence of spatial factors. In particular, as individuals sort themselves into different parts of the urban space on the basis of their personal characteristics (including their labor-market outcomes), studies encounter an endogeneity bias, for which no perfect solution exists at present (Glaeser, 1996). Among the few empirical studies that try to deal with the endogeneity issue, most choose to restrict their sample to young adults residing with their parents, arguing that the location choice of parents can be thought of as fairly exogenous to the employment status of young adults (see e.g. O Regan and 5

6 Quigley, 1996, 1998). However, it has been argued that this solution does not completely eliminate endogeneity as there may exist unobserved parental characteristics that influence both the residential choice and the employment status of young adults (Glaeser, 1996). In the present paper, we nevertheless choose to focus on young adults living with their parents. But in order to get better confidence in our results, we resort to a strategy inspired by Ginther et al (2000) and Harding (2003) which enables us to analyze the sensitivity of our estimated spatial effects to the presence of both observed and unobserved parental covariates. The paper is structured as follows. Section 1 presents a brief synthesis of the economic literature that links the formation of unemployment to city structure. Section 2 describes the database and presents the studied area. Section 3 describes our methodological approach. Section 4 presents some stylized facts about the Brussels metropolitan area. Section 5 presents and discusses the main results. Section 6 concludes. 1. Urban unemployment and city structure: a brief review of the literature An abundant literature in sociology and urban economics suggests that the spatial organization of cities can exacerbate unemployment among disadvantaged communities. In this perspective, labor-market outcomes should depend on individual characteristics (age, education, ethnicity, ) but also on the exact location within the city. Two types of factors have been put forward to explain the existence of such spatial effects, focusing either on the role played by the physical 6

7 disconnection from jobs or the harmful effects of residential segregation and the social composition of neighborhoods. The first works purporting to demonstrate the influence of space on individual labor-market outcomes were based on the spatial mismatch hypothesis put forward by Kain in 1968 (see Gobillon et al, 2005, for a mainly theoretical survey of the literature, and Ihlanfeldt and Sjoquist, 1998, for an empirical one). In theory, there are several mechanisms according to which distance to job opportunities can be problematic. One important mechanism is that job-seekers residing in areas disconnected from job opportunities are likely to refuse job offers if commuting costs are too high in view of the offered wages (Brueckner and Martin, 1997). Another mechanism is that workers job-search efficiency may decrease with distance to jobs since it is obviously more difficult to search far away from one s place of residence (as modeled in Wasmer and Zenou, 2002). The intensity of the search effort may also decrease with distance from job opportunities, for instance if workers residing far away from job centers face lower rents and thus feel less pressured to quickly find a job (as in Smith and Zenou, 2003). Similarly, high search costs may also deter workers from searching far away (as in Ortega, 2000). Finally, firms could discriminate against distant workers, for instance if distance makes them less productive because of long and tiring commutes (as in Zenou, 2002). General empirical tests confirm that the disconnection between places of residence and job locations exacerbates unemployment (see Weinberg, 2004, or Martin, 2004), but few of the abovementioned mechanisms have been specifically tested (see Gobillon et al., 2005, for more details on this issue). 7

8 Other works have focused on the role of residential segregation and more generally the quality of the social environment on individual socioeconomic outcomes. In this respect, several mechanisms can account for an adverse effect of residential segregation, either directly on unemployment, or indirectly through low employability. One mechanism is that residential segregation can be a hindrance to human capital acquisition: in neighborhoods which concentrate low-ability students, human capital externalities can deteriorate school achievements and employability (Benabou, 1993). Social problems which deteriorate the employability of workers can also spread through neighborhood interactions. For instance, Crane (1991) develops an epidemic theory of ghettos in which the propensity of youngsters to adopt a socially deviant behavior depends on the proportion of same-behavior individuals in the neighborhood. This contagion is all the more prevalent as adults are themselves unemployed and do not provide a figure of social success to which youngsters could identify (Wilson, 1987). Another mechanism whereby residential segregation can exacerbate unemployment is that it can deteriorate social networks in disadvantaged communities. This is a crucial point since a significant proportion of jobs are usually found through personal contacts (Mortensen and Vishwanath, 1994) and since low-skilled workers, young adults, and ethnic minorities often resort to such informal search methods (Holzer, 1987 and 1988). In particular, in neighborhoods were local unemployment rates are higher than average, local residents know fewer employed workers that could refer them to their own employer or provide them with professional contacts. In this respect, Reingold (1999) concludes that the poor quality of social networks explains a significant portion of unemployment 8

9 problems in disadvantaged urban areas in the U.S. (see Selod and Zenou, 2001 and 2006, for formalizations). Another mechanism which links labor-market outcomes to segregation involves the reluctance of employers to hire workers residing in disadvantaged neighborhoods. The stigmatization of these neighborhoods is at the root of redlining, a practice in which employers draw an imaginary red line around a stigmatized neighborhood and beyond which they discriminate against residents (see Zenou and Boccard, 2000, for a model). As for the spatial mismatch hypothesis, there are general tests which show that segregation has an adverse effect on the labor market (see Cutler and Glaeser, 1997), but distinguishing which particular mechanisms explain this adverse effect remains on the research agenda. 2. Data and studied area 2.1. Studied area This paper focuses on Brussels which, like Flanders and Wallonia, forms one of the three institutional regions of Belgium: the Brussels-Capital Region consisting of 19 municipalities (called communes in French) and hosting around 1 million residents on a 163 km² area. However, as in most cities, Brussels functional metropolitan area extends far beyond its institutional limits. Several studies have tried to measure the spatial extension of its urban area applying various methods (see Thomas et al, 2000 for a synthesis). In the present paper, we chose to use the so-called Extended Urban Area (E.U.A., see Figure 1), which 9

10 perfectly reflects the social dualism between the city center of Brussels and its close suburbs. 1 It hosts 1.4 million inhabitants and extends over a 723 km² area. [Insert Figure 1] The smallest spatial unit for which census data are usually available is the statistical ward, a subdivision of the municipality defined in 1971 according to social, economic and architectural similarities (Brulard and Van der Hagen, 1972). Statistical wards that present a common functional or structural character (for example a common attraction pole like a school or a church) can further be grouped into larger entities which we will refer to as neighborhoods, and which constitute an intermediate level between the statistical ward and the municipality. It is the spatial level of analysis chosen in this paper. There are 328 such neighborhoods in the E.U.A., grouping on average 4,250 inhabitants. For statistical reasons, neighborhoods of less than 200 inhabitants were not considered in the subsequent analysis. 2.2 The data The empirical analysis conducted in this paper is based on two databases extracted from the 1991 Census of Population which differ according to their level of aggregation (Institut National de Statistiques, 1991). In the first database, the basic statistical unit is the individual. For all individuals aged 19 to 64 and residing in the Extended Urban Area, the database provides the main personal 10

11 characteristics, including age, gender, education, citizenship, employment status, statistical ward of residence, and kinship with household s head. The database also contains several household characteristics (for instance whether parents own a car) as well as an identification number, which makes it possible to identify individuals who belong to a same household. In the second database, the basic statistical unit is the statistical ward. It includes various indicators of the socioeconomic composition and average housing characteristics of statistical wards. This database was complemented with the average income of households in the statistical wards computed from 1993 fiscal sources (Institut National de Statistiques, 1993). These ward-level data were further aggregated to form neighborhood-level variables. 3. Methodological approach The objective of the present paper is to investigate whether the spatial structure of Brussels may have an effect on unemployment. In that purpose, we estimate unemployment probabilities at the individual level taking into account personal, household, and neighborhood characteristics, using the following logistic model: Pi Log = P 1 i α + β γ + δ Ii + Hi Ni [1] where P i is the unemployment probability of individual i, I i is a vector of personal characteristics, H i is a vector of household characteristics, N i is a vector of neighborhood characteristics (social composition and physical access to jobs), and α, β, γ and δ are parameters to be estimated. In particular, δ, when significantly 11

12 different from zero identifies an impact of spatial factors on unemployment, which we will refer to as neighborhood effects in the rest of the paper. Using [1], the individual probability of unemployment P i is given by: P i ( α + βii + γhi + δni ) e = [2] ( α + βi ) 1 i + γh i + δn + e i 3.1. Definition of neighborhood characteristics Two types of indicators are used to account for the neighborhood characteristics which may potentially influence individual unemployment probabilities. The first type relates to the spatial mismatch theory and characterizes the disconnection of neighborhoods from jobs, while the second type relates to the social composition of neighborhoods. Measuring spatial access to job opportunities is a difficult task as data on job offers are generally not available. In the absence of data on job openings, we use the census individual declarations of workplaces to compute the number of occupied jobs in each neighborhood. 2 Two indicators are then computed for each neighborhood i: distance to jobs and job density. First, distance to jobs D i is defined as the average distance from neighborhood i to each neighborhood j in the E.U.A. (d ij ) weighted by the number of jobs located in each one of these neighborhoods (E j ): j D i = [3] E j d ij E j j 12

13 with the intra-neighborhood distance d ii being equal to 2/3 the radius of a disc of an area equivalent to the area of neighborhood i, which comes down to assuming that the population in each neighborhood is uniformly distributed around a central point which concentrates all jobs. Second, we consider that the relevant job density for residents of neighborhood i is the ratio of the number of jobs located in that neighborhood and in the adjacent neighborhoods to the overall labor force residing in the same areas. This definition has the advantage of smoothing job density over space and attenuating extreme values. These two indicators were computed both for all jobs and for low-skilled jobs (i.e. jobs occupied by workers having at most a diploma of the lower secondary education segment Ordinary Level equivalent in the U.K. system). Regarding the variables which characterize the social composition of neighborhoods, researchers generally resort to one or several quantitative measures of the aggregate characteristics of residents. However, even though it is likely that individual outcomes are determined by a wide variety of neighborhood characteristics (employment, education, racial composition, ), considering all these characteristics together into a single regression may cause collinearity problems (making parameter values and significance levels unstable) as many indicators of neighborhood composition are highly inter-correlated (O Regan and Quigley, 1996; Johnston et al, 2004). To circumvent this problem, we use standard Factorial Ecology methods (see e.g. Johnston, 1978) to summarize these multiple characteristics into different types of social environment within Brussels (see Section 4.2 below for the construction of the typology). 13

14 3.2. The endogeneity of residential location Linking individual labor-market outcomes to residential location raises the issue of the endogeneity of location choices (Glaeser, 1996; Dietz, 2002). It is indeed well-known that individuals with similar socioeconomic characteristics, notably similar labor-market outcomes, tend to sort themselves in certain areas of the urban space. For instance, individuals with well-paid jobs will choose to reside in neighborhoods with a good social environment. There is thus a two-way causality: on the one hand, residential location influences individual labor-market outcomes, and on the other hand, individual outcomes influences the choice of a residential location. Stated differently, we are allowed to think that individual characteristics that influence labor-market outcomes may also influence residential choices. Of course, standard models like equation [1] make it possible to control for some individual and household characteristics which might influence both neighborhood choice and individual outcomes. However, it is likely that some individual and household characteristics which are unobserved to the researcher (and therefore not included in I i nor H i ) influence both the outcome of interest and neighborhood choice. For example, individuals with a low labor-market attachment (which directly influences the probability of unemployment) may choose to reside in poor neighborhoods for some economic or social reason. As a consequence, what the researcher perceives as a neighborhood effect through the estimated parameter δ may simply stem from a correlated effect reflecting common residential choice. 14

15 Various strategies have been developed to correct for the endogeneity of neighborhood choice. For example, the existence of quasi-experimental situations (such as government subsidized relocation programs in the U.S. like the Gautreaux or the Moving To Opportunity programs) makes it possible to obtain more reliable estimates of neighborhood effects (as households are moved from one neighborhood to another through an exogenous intervention; see Oreopoulos, 2003, for a review). However, due to the scarcity of such experiments, researchers are often constrained to resort to more questionable strategies. Most studies restrict their sample to young adults residing with their parents, arguing that location choices were previously made by the parents and can thus be thought of as fairly exogenous to the employment status of young adults (see e.g. O Regan and Quigley, 1996, 1998). This is also the strategy adopted in this paper, as our studied sample is restricted to young labor-force participants aged 19 to 25 and residing with their parents. However, this approach does not completely eliminate the endogeneity bias. Indeed, there may still exist parental unobserved characteristics which determine their residential choice and also influence the employment outcomes of their adult children (Glaeser, 1996). For example, lack of commitment to work may induce parents to locate in high poverty neighborhoods but may also influence the motivation of their children to search for a job as well as the intensity of their job search effort. In this context, one cannot distinguish neighborhood effects from the effect of those unobserved parental characteristics when estimating the unemployment probability of young adults living with their parents. 15

16 3.3. Sensitivity analysis Following a suggestion made by Glaeser (1996, p. 62), we evaluate the potential remaining endogeneity bias by conducting a sensitivity analysis in order to assess the robustness of our estimated neighborhood effects. In that purpose, we resort to a two-step strategy. In a first step, we follow the approach used by Ginther et al (2000) and estimate several models of individual unemployment probability which incorporate various sets of household characteristics H i, moving away from a model with no household variable towards a model including an extensive set of parental controls (parental employment and professional status, parental education, male or female household head, possession of an automobile). The comparison of the estimated neighborhood effects in these models enables us to test the robustness of our results to the presence of observed parental covariates. In a second step, we test the sensitivity of our results to the endogeneity bias which results from the omission of an unobserved covariate which is correlated to both the probability of unemployment among young adults and parental residential choice, following the approach developed by Rosenbaum and Rubin (1983) and recently applied by Harding (2003) in the context of neighborhood effects. In order to carry out the sensitivity analysis, we build a dummy variable which equals to one if the individual resides in a deprived neighborhood and zero otherwise (see Section 5.2 for the precise definition of deprived neighborhoods). The goal of our analysis is to assess how an unobserved binary covariate which affects both the probability of unemployment among young adults and the choice of parents to 16

17 reside in a deprived or a non-deprived neighborhood would alter our conclusions about the magnitude and significance of neighborhood effects. This is done by generating a series of unobserved binary variables U that vary according to their degree of association with the neighborhood dummy variable X and with the binary outcome measure Y. The degrees of association between U and X and between U and Y are classically measured by odds ratios. In practice, we use the Sensuc function in the Design S library written by Harrell (2003) to generate a binary variable U sampled according to the following logistic model: U i Log = + log( b) Yi + log( c) 1 U i α X [4] where Y i is a binary variable indicating whether individual i is unemployed or not, X i is a binary variable indicating whether i resides in a deprived neighborhood or not, and b and c are chosen odds ratios measuring the strength of the association i that we impose between U and Y and U and X respectively. 3 α is determined so that the overall prevalence of U=1 is 0.5 (or any other given value). Once a value for U has been generated for each individual so that the distribution of U verifies the three constraints (on the two sensitivity parameters b and c and on the overall prevalence of U=1), new estimates of neighborhood effects are obtained by including the U variable in the unemployment probability model given by [1]. This is repeated for increasing values of the sensitivity parameters b and c in order to investigate what level of endogeneity bias (i.e. the strength of the association between U and Y and between U and X) would be needed to invalidate the results and render our estimated neighborhood effects not significant. 17

18 4. Statistical description of Brussels 4.1. Stylized facts The Brussels Extended Urban Area presents a well-marked spatial structure characterized by important disparities opposing its city center to the periphery. Figure 2 maps the percentage of unemployed workers among labor-force participants, highlighting a zone of very high unemployment rates (above 20%) in the central part of the urban area, along the former industrial corridor (the Charleroi-Willebroek canal). On the contrary, unemployment is much lower (below 10% or even 7%) in the suburbs (notably in the municipalities of Tervuren, Overijse, Grimbergen, Dilbeek and Sint-Peeters-Leeuw). [Insert Figure 2] Figure 3 maps the percentage of North Africans (Moroccans, Algerians and Tunisians) and Turks among the whole population. These nationalities correspond to the latest wave of labor immigration in the second half of the sixties (Kesteloot and Cortie, 1998). We consider them as a single group since they usually face the same type of problems on the labor market. 4 Figure 3 shows a high level of ethnic residential segregation: North Africans and Turks are mainly located around the center of Brussels, with a concentration above 10% or even 25% in some neighborhoods. Moreover, as can be seen from the computation of dissimilarity 18

19 indices (Duncan and Duncan, 1955) 5, North Africans and Turks form the most segregated group of foreigners in Brussels: 64% of them would have to be relocated to obtain a uniform mix with the Belgian population. For non-belgian citizens of the European Economic Community (consisting of 12 countries in 1991), the dissimilarity index goes down to only 32%. [Insert Figure 3] It is widely acknowledged that the concentration of North Africans and Turks in Brussels central neighborhoods is mainly due to the functioning of the local housing market. Indeed, the promotion of home-ownership -an important housing policy goal in Belgium- led to the quasi-absence of a social housing sector and to the massive suburbanization of high- and middle-income households towards the periphery. In this context, suburbanization accelerated the degradation of central zones as housing units were rented to low-income households and maintenance was neglected by owners mainly concerned with the additional profit they could derive from their investments. This gave rise to a residual rental sector which concentrates the oldest, most poorly equipped and cheapest housing units of the city, which are the only dwellings poor foreigners can afford (Kesteloot and Van der Haegen, 1997; Kesteloot and Cortie, 1998). By comparing Figures 2 and 3, it can easily be seen that neighborhoods with a high proportion of North Africans and Turks also exhibit high unemployment rates, the correlation between the two variables being equal to This high value is partly explained by neighborhood composition (foreigners are usually more 19

20 likely to be unemployed because they have a lower education or because they are discriminated against in the labor market). It could also be explained by the exacerbating effect of residential segregation on labor-market outcomes. Figure 4 maps the density of jobs for all jobs (on the left-hand side) and for low-qualified jobs (on the right-hand side). It shows a zone of very high job densities in the center of the E.U.A., where some neighborhoods have more than five jobs per labor-force participant. These neighborhoods constitute what is usually called The Pentagon, where one can find the Administrative City as well as many financial institutions and headquarters of national and international firms. This zone of very high job densities is surrounded by a zone of high, though lower, job densities (more than two jobs for one resident) comprising the European institutions and extending towards the northeastern periphery where the national airport as well as some industries are located. Note that the spatial distribution of job densities is quite similar for all jobs and for low-qualified jobs. However, the very high density zone of the center (more than five jobs per labor-force participant) and the high density zone of the northeastern periphery are spatially less extended when it comes to low-qualified jobs. By comparing Figures 2 and 4, it can be seen that central zones with high unemployment rates also have relatively high job densities, which seems in contradiction with the spatial mismatch theory. Indeed, the correlation between unemployment rate and total job density is significant and positive (0.35 for all jobs and 0.31 for low-qualified jobs) and the correlation between unemployment rate and average distance to jobs is significant and negative (-0.52 for both all and low-qualified jobs). In other words, in 20

21 Brussels, individuals living in high job density neighborhoods close to the city center tend to be more unemployed. [Insert Figure 4] 4.2. Typology of neighborhoods Standard Factorial Ecology methods are used to identify sociallyhomogeneous areas within Brussels, which will be subsequently used in the regression analyses in Section 5. We first run a Principal Component Analysis which defines a limited number of non-correlated factors summarizing the information carried by a set of neighborhood variables (see Table A.1 in Appendix 1 for the list of variables and their contribution to the retained factors). Then, neighborhoods are grouped according to their coordinates on the factorial axes, using a hierarchical ascending classification (with the Ward method which minimizes intra-group variance). We obtained five neighborhood types 6 which are presented on Figure 5 below (see Table A.2 in Appendix 1 for their mean characteristics). The first neighborhood type gathers very deprived areas in the center of Brussels and corresponds to neighborhoods with high proportions of North- Africans, Turks, and female-headed households as well as with very high unemployment rates. These neighborhoods are characterized by low educational levels and have the lowest income levels of the whole agglomeration. They are surrounded by a group of deprived neighborhoods presenting similar characteristics 21

22 but with a smaller proportion of North Africans and Turks and a less severe situation in terms of education and unemployment. The third group of neighborhoods extends to the Southwest and Northeast of the inner city, along the former industrial axis. It groups neighborhoods that also have a lower socioeconomic status, in particular an overrepresentation of blue collars and individuals with a lower education. But the unemployment rates and income levels are closer to the city s average. The two remaining groups are characterized on average by higher levels of education and professional statuses. The first, which we labeled well-off, corresponds to the periphery of the Brussels E.U.A., while the second, which we labeled very well-off, occupies the Southeast part of the Brussels Capital Region as well as its continuation in the periphery (notably the municipalities of Waterloo and Lasne). The latter differ from the former by its high proportion of executives and its very high average income. [Insert Figure 5] 5. The effect of spatial structure on unemployment: main results The previous section has shown that Brussels exhibits a high level of residential segregation associated with high local unemployment rates but that disadvantaged neighborhoods are usually close to job locations. However, a statistical analysis is required before one can conclude on the role of spatial factors on unemployment in Brussels. The aim of the present section is (i) to investigate the role played by Brussels spatial structure on individual unemployment 22

23 probabilities and (ii) to test for the robustness of our results. As explained in Section 3.2, in order to limit the endogeneity problems associated with residential sorting, the sample is restricted to young labor-force participants living with their parents, i.e. a sample of 34,250 individuals. 7 In Section 5.1, we consider a model including only individual characteristics and then progressively add other neighborhood characteristics to the specification. Section 5.2 tests the robustness of the results to the presence of both observed and unobserved parental characteristics The role of spatial factors in explaining unemployment probabilities Table 1 presents the parameter estimates as well as the odds ratios for five different models explaining the individual unemployment probability. Model Ia considers only the role of individual characteristics (gender, age, education and citizenship), while the four other models add various combinations of neighborhood characteristics to the specification. In Model Ia and in all other regressions, men or educated workers are less likely to be unemployed than women or workers with a lower education. The probability of unemployment also decreases with the age of the individual and when the individual is a foreigner. North Africans and Turks are more disadvantaged than EEC citizens. Interestingly, young Belgian adults born of foreign parents are also more likely to be unemployed than young Belgian adults born of Belgian parents. This result suggests that besides citizenship, the name or visible characteristics associated with foreign origin are a handicap on the labor market. This is consistent with both the 23

24 existence of labor-market discrimination as well as social networks of lower quality for individuals of foreign origin. As can be seen from the likelihood ratios in Table 1, introducing neighborhood characteristics significantly increases the fit of the regression. Model Ib uses only the unemployment rate to represent neighborhood influences. It shows that the local unemployment rate significantly increases the unemployment likelihood of young adults. As suggested by the theory, this may hinge upon deteriorated local social networks: the higher the proportion of unemployed neighbors, the more difficult the insertion of a young adult in the labor market. Model Ic substitutes the type of neighborhood for the unemployment rate, which increases the fit of the model, thus confirming that the unemployment probability is affected by a wide variety of neighborhood characteristics, instead of only by the unemployment rate. All things else being equal, the unemployment probability is the lowest for young adults residing in the industrial axis and in well-off areas. In accordance with the theory, living in a central area with a socioeconomic environment of lower quality (deprived or very deprived) significantly increases the unemployment probability of young adults (with odds ratios of and respectively) in comparison with living in well-off areas. Surprisingly however, living in a very well-off area also significantly increases the probability of unemployment. However, we do not wish to conclude that living in a wealthy area is detrimental to finding a job. A possible interpretation for this unexpected result could be that unemployed youth in very well-off areas do not feel pressured to intensively search for a job if they get enough financial support from their parents. Another possible explanation could be the existence of a selection bias if living in a 24

25 wealthy area enhances the chances to find a well-paid job so that young adults that originate from these areas are likely to quickly move out of their parents dwelling. This would leave an over-representation of unemployed workers living with their parents in very well-off areas. Investigating the veracity of this story would require to model unemployment probabilities jointly with the process of leaving parental home through a selection model. However, as the choice to leave home is probably influenced by parental characteristics (such as household financial resources) as well as neighborhood characteristics, this would require more specific data. In particular, we would need to be able to identify the household and neighborhood of origin of young adults who do not live with their parents. This is unfortunately not feasible using the Belgian census data. Model Id complements the specification by adding indicators of the physical disconnection from jobs. As could easily be suspected from the statistical description of Brussels (Section 4.1), the effect of distance to jobs is not significant. As for job density, its impact is significant but plays in the wrong direction: controlling for all other variables, young adults who reside in areas with the highest job densities are less likely to hold a job. The same obtains when one considers distance to low-skilled jobs and low-skilled job density. This last result is not consistent with the spatial mismatch hypothesis and suggests that spatial mismatch is not a problem in Brussels, a city in which the unemployed reside close to the jobs they could occupy 25

26 5.2. Sensitivity analysis In order to test the robustness of our estimated neighborhood effects, in a first step, we conduct a sensitivity analysis to the presence of parental observed characteristics. In that purpose, we estimate different models which incorporate various sets of parental and household characteristics. Table 2 presents the results of our sensitivity analysis, moving away from a model with no household characteristics (Model I) towards models including an increasingly comprehensive set of parental controls (Models II to VI). Note that distance to jobs and job density were not included as they played little role in the previous models. Models II to VI show that the unemployment probability of a young adult is higher when the household head (or spouse) is not participating to the labor-force or is unemployed than when he is employed. This effect is highly significant and is consistent with social network theories (at the household level, unemployed parents being little able to help their job-seeking children) and socialization considerations (unemployed parents failing to provide their children with an image of social success to which they could identify). Living in a female-headed household (a proxy for single mother households) also significantly increases the likelihood that a young adult is unemployed, suggesting that these households are more frequently prone to social problems detrimental to finding a job. Living in a household which does not own a car (an indirect measure of lack of financial resources and low mobility for those who do not) also significantly increases the unemployment probability. The effects of parental professional status and educational level are not always significant and, when significant, seems counterintuitive. Indeed, all other 26

27 things equal, having a parent with an executive profession or a higher level of education increases the unemployment probability of young adults. This effect mirrors our counterintuitive finding on residence in very well-off areas. As mentioned previously this could be explained if rich kids are less pressured to search intensively for a job (because of their parents financial support) or in presence of a selection bias. With the present dataset, we are unfortunately not able to distinguish between these two explanations. By comparing the parameters and significance levels of the neighborhood types across the different models, one can assess the sensitivity of our estimated neighborhood effects to the inclusion of a more comprehensive set of parental controls. Table 2 shows that although the inclusion of parental and household characteristics significantly increases the fit of the model (see the likelihood ratios), the estimated neighborhood effects change very little (for example, the odds ratio associated with residence in very deprived areas varies from to between Model I and Model VI), and all parameters remains significant at a 1% level. In a second step, we complement the sensitivity analysis by testing the robustness of our estimated neighborhood effects to the presence of unobserved characteristics, using the methodology developed by Rosenbaum and Rubin (1983). Since this methodology requires the use of a binary neighborhood variable we redefined our five-modality variable by grouping together the two least favored neighborhood types (very deprived and deprived) while considering all other neighborhood types as another category. Doing so yields a dummy variable equal to one if the individual resides in a deprived neighborhood and equal to zero 27

28 otherwise. Table 3 presents the estimated effects of living in such a deprived neighborhood, for three combinations of parental and household characteristics: no parental characteristics (as in Table 2 s Model I), only the parental employment status (i.e. unemployed or not-participating to the labor-force; as in Table 2 s Model II), and in the last case the full set of parental controls (as in Table 2 s Model VI). For each one of these models, an artificially created binary variable U is included and a sensitivity matrix is obtained by varying the sensitivity parameters b and c, which measure the associations of U with the neighborhood type and the employment status. As expected, Table 3 shows that in the three model specifications, the estimated neighborhood effect decreases as the values of b and c increase. This means that accounting for a previously omitted variable correlated both to the neighborhood type and employment status does indeed reduce the intensity of the neighborhood effect. However, the effect seems fairly robust to the presence of unobserved covariates since the values of b and c would have to be very high to make the neighborhood effect not significant. Indeed, for the model including no parental or household characteristic (Model I), an unobserved covariate which multiplies both the odds of living in a deprived neighborhood and the odds of unemployment by 3.0 would be required to totally erase the neighborhood effect. This is also true when observed parental characteristics are included in the analysis. For example, when the full set of parental characteristics are used (Model VI), the neighborhood effect becomes insignificant when at least one of the two odds ratios reaches 3.0 and the other one equals 2.5. These values are very high in comparison with the odds ratios estimated on observed covariates. As can be seen from Table 2, none of the observed parental 28

29 and household characteristics produces odds ratios as high as 3.0 or even 2.5, the highest odds ratio for a parental characteristics being in Model VI (for having an unemployed parent). In Table 2, even individual characteristics do not produce odds ratios above 2.5: the highest odds ratios are for North-African or Turkish citizenship and for Belgians born of foreign parents. In other words, for the neighborhood effect to become insignificant when considering an omitted variable, the effect of this omitted variable on unemployment would have to be stronger than that of parental employment status and at least as strong as that of citizenship. This provides some relatively strong evidence on the robustness of neighborhood effects. 6. Conclusion The explanation of unemployment usually revolves around well-known determinants (notably the lack of formal education or the skill mismatch between supply and labor demand). Alternative theories suggest that, in predominantly urban economies, unemployment may also be determined by the spatial organization of cities which concentrates disadvantaged families in poverty zones and/or distance them from job opportunities. This paper studies the particular case of the Brussels metropolitan area and investigates the extent to which residential location can affect the unemployment probability of young adults. As in many other studies, we address the endogeneity of residential choices by restricting our studied sample to the case of young adults residing with their parents. However, his solution is imperfect as some unobserved parental characteristics may still 29

30 influence both the residential choice of parents and the employment status of young adults. In this context, the originality of our work is to evaluate the potential remaining endogeneity bias by conducting a sensitivity analysis to the presence of both observed and unobserved parental characteristics. Results mainly show that urban unemployment is exacerbated by residential segregation but not by disconnection from jobs. Indeed, distance to jobs does not exacerbate unemployment. This result is coherent with the spatial structure of Brussels, which concentrates disadvantaged households in areas close to jobs. We show however that urban unemployment is strongly exacerbated by social, economic, and ethnic segregation at the place of residence. In particular, all things else being equal living in a neighborhood with a socioeconomic environment of a lower quality significantly increases the unemployment probability of young adults. This result is quite robust to the presence of both observed and unobserved covariates. Indeed, neighborhood effects remain statistically significant whatever the combination of parental and household characteristics introduced in the regression. Moreover, our sensitivity analysis to the presence of unobserved covariate shows that the amount of selection on unobservables would have to be unreasonably high to render the estimated neighborhood effect insignificant. Our study is a first step that sheds light on the formation of unemployment in Brussels. However, further research is needed to identify particular mechanisms mediating the effect of segregation on unemployment. 30

31 Notes 1. The Extended Urban Area of Brussels consists of the 36 municipalities of the operational metropolitan area defined by Van der Haegen et al (1996) to which 5 municipalities have been added (Lasne, La Hulpe, Rixensart, Hoeilaart and Overijse) in conformity with Mérenne-Schoumaker et al (1998) and Thomas and Zenou (1999). 2. Raphael (1998) argues that employment growth measures (defined as the difference in employment levels between two given years) should be used instead of employment levels as the latter only reflect vacancies created by job turnover and not vacancies created by job growth. However, the Belgian Census of Population takes place only every ten years, a time period we believe too important to adequately reflect the number of jobs openings that may potentially matter for unemployed workers in a cross-section study. 3. The odds ratio measures how much more likely it is for the outcome to occur (i.e. to be equal to 1) for individuals whose independent variable X is equal to 1 compared to individuals with X equal to 0 (or for a marginal increase in the case of a continuous independent variable). An odds ratio greater than 1 indicates that a value of 1 for X increases the odds of observing the outcome in comparison with a value of 0. For example, the sensitivity parameter c which measures the association between U and X is given P( U = 1, Y = 1) by: P( U = 1, Y = 0) P( U P( U = 0, Y = 0, Y = 1) = 0) 31

32 4. Note that the nationality criterion underestimates the size of extended ethnic group because many individuals have acquired the Belgian nationality or were simply born Belgian from a mixed marriage (Kesteloot and Van der Haegen, 1997). However, since Belgian census data do not contain any information on the ethnic origin of people, nationality is the only way to approach the issue of ethnic segregation. 5. The dissimilarity index is given by: 1 Fori i Bel where For 2 For Bel i and Bel i are the respective numbers of foreigners and Belgians in neighborhood i and For and Bel are the respective numbers of foreigners and Belgians in the whole urban area. 6. This was the optimal number of clusters according to several criteria, including the Cubic Clustering Criterion, the pseudo-f and pseudo-t values. 7. The number of observations indicated in the different tables might differ as individuals for whom some individual or household characteristics were missing had to be set aside. Individuals residing in neighborhoods of less than 200 inhabitants were also erased from the sample. i 32

33 References BENABOU, R. (1993) Workings of a city: location, education, and production, Quarterly Journal of Economics, 108, pp BRUECKNER, J. and MARTIN, R. (1997) Spatial mismatch: an equilibrium analysis, Regional Science and Urban Economics, 27, pp BRULARD, T. and VAN DER HAEGEN, H. (1972) La division des communes belges en secteurs statistiques. Le point de vue des géographes, Acta Geographica Lovaniencia, 10, BURGESS, E.W. (1925) The growth of the city, in: R.E. PARK, E.W. BURGESS and R.D. McKENZIE (Eds) The City, Chicago: University of Chicago Press, pp CRANE, J. (1991) The epidemic theory of ghettos and neighborhood effects on dropping out and teenage childbearing, American Journal of Sociology, 96, pp CUTLER, D.M. and GLAESER, E.L. (1997) Are ghettos good or bad?, Quarterly Journal of Economics, 112, pp DIETZ, R.D. (2002) The estimation of neighborhood effects in the social sciences: an interdisciplinary approach. Social Science Research, 31, 4, pp DUNCAN, O.D. and DUNCAN, B. (1955) A methodological analysis of segregation indexes, American Sociological Review, 41, pp FIELDHOUSE, E. (1999) Ethnic minority unemployment and spatial mismatch: the case of London, Urban Studies, 36, pp FUJITA, M. (1989) Urban Economic Theory. Land Use and City Size, Cambridge: Cambridge University Press. 33

34 GINTHER, D., HAVEMAN, R. and WOLFE, B. (2000) Neighborhood attributes as determinants of children s outcomes. The Journal of Human Resources, 35, 4, pp GLAESER, E.L. (1996) Spatial effects upon employment outcomes: the case of New Jersey teenagers. Discussion. New England Economic Review, May/June, pp GOBILLON, L. and SELOD, H. (2006) Accès à l emploi, ségrégation résidentielle et chômage : le cas de l Ile-de-France, under revision for Economie et Prévision. GOBILLON, L., SELOD, H. and ZENOU, Y. (2005) The mechanisms of spatial mismatch, CEPR Discussion Papers Series GOFFETTE-NAGOT, F., THOMAS, I. and ZENOU, Y. (2000) Structure urbaine et revenus des ménages, in: C. BAUMONT, P.-Ph. COMBES, P.-H. DERYCKE and H. JAYET (Eds) Economie Géographique: Les Théories à l Epreuve des Faits, Paris: Economica, pp HARDING D.J. (2003) Counterfactual models of neighborhood effects : the effect of neighborhood poverty on dropping out and teenage pregnancy, American Journal of Sociology, 109, 3, pp HARRIS, C. and ULLMAN, L. (1945) The nature of cities, Annals of the American Academy of Political and Social Science, 242, pp HOLZER, H. (1987) Informal job search and black youth unemployment, American Economic Review, 77, pp HOLZER, H. (1988) Search method used by unemployed youth, Journal of Labor Economics, 6, pp

35 HOYT, H. (1939) The Structure and Growth of Residential Neighborhoods in American Cities, Washington D.C.: Government Printing Office. IHLANFELDT, K. and SJOQUIST, D. (1998) The spatial mismatch hypothesis: a review of recent studies and their implications for welfare reform, Housing Policy Debate, 9, pp JENCKS, C. and MAYER, S. (1990) Residential segregation, job proximity, and black job opportunities, in L.E. LYNN and M.G.H. McGEARY (Eds) Inner-City Poverty in the United States, Washington, DC: National Academy Press, pp JOHNSTON, R.J. (1978) Residential area characteristics: Research methods for identifying urban sub-areas analysis and factorial ecology, in: D.T. HERBERT and R.J. JOHNSTON (Eds) Social Areas in Cities. Processes, Patterns and Problems, Chichester: John Wiley and Sons, pp JOHNSTON, R.J., JONES, K., BURGESS, S., PROPPER, R., SARKER, R. and BOLSTER, A. (2004) Scale, factor analyses, and neighborhood effects. Geographical Analysis, 36, 4, pp KAIN, J. (1968) Housing segregation, negro employment and metropolitan decentralization, The Quaterly Journal of Economics, 82, pp KASARDA, J.D. (1989) Urban industrial transition and the underclass, Annals of the American Academy of Political and Social Science, 501, pp KESTELOOT, C. and CORTIE, C. (1998) Housing Turks and Moroccans in Brussels and Amsterdam: the difference between private and public markets, Urban Studies, 35, pp

36 KESTELOOT, C., DE TURCK, A., VANDERMOTTEN, C., MARISSAL, P. and VAN HAMME, G. (2001) Structures sociales et quartiers en difficulté dans les régions urbaines belges, Report to Charles Piqué, Ministre chargé de la Politique des Grandes Villes. KESTELOOT, C. and VAN DER HAEGEN, H. (1997) Foreigners in Brussels : spatial continuity and social change, Tijdschrift voor Economische en Sociale Geografie, 88, pp MARTIN, R. (2004) Can black workers escape spatial mismatch? Employment shifts, population shifts, and black unemployment in American cities, Journal of Urban Economics, 55, pp MÉRENNE-SCHOUMAKER, B., VAN DER HAEGEN, H. and VAN HECKE, E. (1998) Urbanisation, Recensement Général de la population et des Logements, Monographie n 11A, Bruxelles: Institut National de Statistiques. MORTENSEN, D. and VISHWANATH, T. (1994) Personal contacts and earnings: It is who you know!, Labour Economics, 1, pp O REGAN, K.M. and QUIGLEY, J.M. (1996) Spatial effects upon employment outcomes: the case of New Jersey teenagers, New England Economic Review, May/June, pp O REGAN, K.M. and QUIGLEY, J.M. (1998) Where youth live: economic effects of urban space on employment prospects, Urban Studies, 35, 7, pp OREOPOULOS, P. (2003) The long-run consequences of living in a poor neighborhood, Quarterly Journal of Economics, 118, 4, pp ORTEGA, J. (2000) Pareto-improving immigration in an economy with equilibrium unemployment, Economic Journal, 110, pp

37 RAPHAEL, S. (1998) The spatial mismatch hypothesis and black youth joblessness: evidence from the San Francisco Bay Area. Journal of Urban Economics, 43, 1, pp REINGOLD, D. (1999) Social networks and the employment problem of the urban poor, Urban Studies, 36, pp ROSENBAUM, P.R. and RUBIN, D.B. (1983) Assessing sensitivity to an unobserved binary covariate in an observational study with binary outcome. Journal of the Royal Statistical Society B, 45, 2, pp SELOD, H. and ZENOU, Y. (2001) Social interactions, ethnic minorities and urban unemployment, Annales d Economie et de Statistique, 63-64, pp SELOD, H. and ZENOU, Y. (2006), City structure, job search, and labor discrimination. Theory and policy implications, Economic Journal, forthcoming. SMITH, T. and ZENOU, Y. (2003), Spatial mismatch, search effort and urban spatial structure, Journal of Urban Economics, 54, pp THOMAS, I., TULKENS, H. and BERQUIN, P. (2000) Quelles frontières pour Bruxelles? La réponse d un exercice statistique, géographique et économique, in: CIFOP (Ed), Quatorzième Congrès des Economistes belges de Langue française, Commission 3, Liège, Sart Tilman, pp THOMAS, I. and ZENOU, Y. (1999) Ségrégation urbaine et discrimination sur le marché du travail: le cas de Bruxelles, in: M. CATIN, J.Y. LESUEUR and Y. ZENOU (Eds) Emploi, Concurrence et Concentration Spatiale, Paris: Economica, p TIEBOUT, C. (1956) A pure theory of local expenditure, Journal of Political Economy, 64, pp

38 VAN DER HAEGEN, H., VAN HECKE, E. and JUCHTMANS, G. (1996) Les régions urbaines belges en 1991, Etudes Statistiques, 104. VANDERMOTTEN, C., VERMOESEN, F., DE LANNOY, W. and DE CORTE, S. (1999) Villes d Europe. Cartographie comparative, Bulletin du Crédit Communal, WASMER E., ZENOU Y. (2002) Does city structure affect search and welfare?, Journal of Urban Economics, 51, pp WEINBERG, B. (2004) Testing the spatial mismatch hypothesis using inter-city variations in industrial composition, Regional Science and Urban Economics, 34, pp WILSON, W.J. (1987) The Truly Disadvantaged: The Inner City, the Underclass, and Public Policy, Chicago: The University of Chicago Press. WILSON, W.J. (1996) When Work Disappears. The World of the New Urban Poor, New York: Alfred A. Knop. ZENOU, Y. (2002) How do firms redline workers?, Journal of Urban Economics, 52, pp ZENOU, Y. and BOCCARD, N. (2000) Labor discrimination and redlining in cities, Journal of Urban Economics, 48, pp

39 Appendix 1: Building a typology of neighborhoods Table A.1: Variables used in the principal component analysis and their contributions to factors Factor 1 Factor 2 Eigenvalue Percent of variance explained 56.33% Loadings % North-Africans and Turks % female-headed households Average household income % students in technical classes % individuals with lower education % individuals with at least intermediate education % individuals with a higher level of education % blue-collars among employed workers % executives among employed workers Unemployment rate Unemployment rate of the under Only factors with eigenvalues superior or equal to one were retained. Figures presented are those obtained after a Varimax rotation. Table A.2: Mean characteristics of the five neighborhood types Very Deprived Indus- Welloff Very Total deprived trial axis well-off % North-Africans and Turks % female-headed households Average household income % students in technical classes % with lower education % with at least intermediate educ % with a higher level of education % blue-collars % executives Unemployment rate Unemployment rate of the under The five neighborhood types were defined by a hierarchical ascending classification with the Ward criterion. 39

40 Figure 1: The Brussels Extended Urban Area 40

41 Figure 2: Percentage of unemployed workers among labor-force participants 41

42 Figure 3: Percentage of North Africans and Turks 42

43 Figure 4: Density of total jobs and low-qualified jobs 43

44 Figure 5: Typology of neighborhoods 44