Levitt - Growing Up in the Projects: The Economic Lives of...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: Growing Up in the Projects: The Economic Lives of a Cohort of Men Who Came of Age in Chicago Public Housing By STEVEN D. LEVITT AND SUDHIR ALLADI VENKATESH* by the sociologist allows for a richer understanding of the problem. The costs of such an approach are that it is time-intensive, cannot be replicated, and is of questionable generalizability. In this paper, we report preliminary findings of a joint study undertaken by an economist and an ethnographer that aims to integrate the usual economic and ethnographic approaches to empirical research.1 Focusing on the residents of one high-rise housing-project building in Chicago, we have attempted to reconstruct retrospectively the economic and social histories of all young males who spent their adolescence in the housing project in the early 1990’s. Through the use of an extensive net of community contacts, we have to date succeeded in locating 105 of these 118 men, who are now in their late twenties and early thirties. For almost all of these men, we have obtained not only extensive self-reported information, but also, evaluations of these men by other community members on a wide range of characteristics. I. Background and Methods Empirical research by labor economists relies overwhelmingly on data collected in largescale, government funded surveys. The availability of such data dramatically lowers the cost of conducting research for any individual scholar. Large sample sizes and extensive breadth of variables allow precise testing of a wide variety of hypotheses. Combined with enormous increases in computing capabilities and the development of new techniques to analyze such data, these data sets have transformed the field of labor economics (James Heckman, 2000). The use of large-scale survey data, however, has not come without a cost. The set of questions that can be addressed is constrained by the variables that have been collected. These studies are typically nationally representative, with sample sizes too small to undertake detailed analysis in particular localized geographic areas or very specific subsets of the overall population. Because the researcher is removed from the data-collection process and the analysis is national in focus, understanding of institutional detail is de-emphasized. Ethnography, used by sociologists and anthropologists, represents the polar opposite of the standard economic approach. An ethnographer integrates himself or herself into the community being analyzed, paying careful attention to the context in which the actors operate. Such studies tend to rely almost exclusively on qualitative analysis, rarely gathering systematic quantitative data. The benefits to such an approach are that the researcher structures the analysis to answer precisely the desired question, and the depth of specific knowledge acquired * Levitt: Department of Economics, University of Chicago, 1126 E. 59th Street, Chicago, IL 60637; Venkatesh: Department of Sociology, Columbia University, New York, NY 10027. Joseph Krupnick provided outstanding research assistance on this project. Financial support was provided by the American Bar Foundation, the National Science Foundation, and the Sloan Foundation. 79 One of us (Venkatesh) began researching the lives of housing-project residents in Chicago in 1989, using what is known as the participantobservation approach (see e.g., Venkatesh, 2000). As part of that earlier research, Venkatesh assembled a list of all 118 males aged 17–26 residing in one particular high-rise housing-project building in 1991, along with a limited set of background information on these individuals. This building 1 Although we are unaware of previous research partnerships involving empirical economists and ethnographers, there are a number of earlier studies whose analysis is similar in spirit to ours, namely, extending standard economic approaches to incorporate a better appreciation of institutional detail or to integrate nonstandard variables into the analysis (e.g., Anne Case and Lawrence Katz, 1991; James Rosenbaum, 1995; Kathryn Edin and Laura Lein, 1997; Jeff Biddle and Daniel Hamermesh, 1998; Greg Duncan and Rachel Dunifon, 1998). 80 AEA PAPERS AND PROCEEDINGS TABLE 1—BASELINE CHARACTERISTICS Overall sample (N 9 0) 20.9 (0.3) 0.93 (0.03) 2.9 (0.2) 0.44 (0.11) Gang members (N 2 9) 20.0 (0.5) 0.93 (0.05) 3.3 (0.4) 0.41 (0.09) OF THE MAY 2001 SAMPLE is part of a large public housing complex that included some of the poorest census tracts in the entire nation in 1990. Virtually all of the residents of the building are African-American. A well-entrenched street gang controls the local drug trade. Average homicide rates in the census tracts encompassing the housing project over the period 1990 –1995 were over 100 per 100,000 residents annually, or more than ten times greater than the national average. Roughly ten years later, we have tried to locate each of these individuals in an attempt to document the labor-market experiences of these men who came of age at the peak of the crack epidemic in a neighborhood profoundly affected by drugs and gangs. We began surveying these individuals in May 2000. The first round of field work, which we report on in this paper, was completed in September 2000. A nominal sum of money was provided to those who participated, as well as to community members who aided us in locating members of the target group. All of the subjects were given a structured survey containing a common set of questions, administered orally. In addition, as part of ongoing fieldwork, we have been asking a wide range of community members who knew individuals in our sample when they were growing up (e.g., teachers, clergy, social workers, etc.) to rate them on a seven-point scale along a range of different dimensions: physical strength, work ethic, popularity, propensity to get into trouble, and attitude toward school. We were able to locate all but 13, or 89 percent, of the people on the original 1991 list of residents. Eleven of the men were deceased. Four refused to participate. Thus, we have a total of 90 interviews. II. Analysis of the Survey Results Variable Age in 1991 Lived with mother growing up Number of siblings Employed in 1991 Non-gang (N 6 1) 21.3 (0.3) 0.93 (0.03) 2.8 (0.2) 0.46 (0.06) Differencea 1.3* (0.6) 0.00 (0.06) 0.6 (0.4) 0.05 (0.11) lowest, 0.4 (0.4) 0.0 (0.5) 0.5 (0.4) 0.4 (0.4) 0.8† (0.4) Evaluation by community members on seven-point scale (1 7 highest): Physical strength Stays out of trouble Hard-working Serious about school Well-liked 4.1 (0.2) 3.6 (0.2) 3.7 (0.2) 3.6 (0.2) 4.5 (0.2) 4.4 (0.4) 3.7 (0.4) 3.4 (0.4) 3.3 (0.4) 4.0 (0.4) 4.0 (0.2) 3.6 (0.2) 3.8 (0.2) 3.7 (0.2) 4.8 (0.2) Notes: Our sample is the 90 living members of the initial 118 housingproject residents in 1991 who have been reinterviewed in 2000. Values in columns are within-group means. Standard errors of the mean are in parentheses. Data collection of the community-member evaluations is ongoing. Consequently, the number of observations for these measures is lower: 74 total, 20 for gang members and 54 for non-gang individuals. a Gang minus non-gang. † Significantly different from zero at the 10-percent level. * Significantly different from zero at the 5-percent level. Table 1 presents the overall means and standard deviations, as well as the breakdown by gang status, for the set of background characteristics we have collected. For the purposes of this paper, an individual must have been actively engaged in gang life in the late 1980’s– early 1990’s to be classified as a gang member. Those with only a nominal social affiliation with the gang are treated as non-gang. All of the results in Table 1 pertain to life circumstances on or before 1991, although it is important to note that most of the information was collected retrospectively. Roughly one-third of the individuals in the sample were active in the gang in 1991. The average age in 1991 was 20.9 years, with gang members on average about one year younger than those outside the gang. Over 90 percent of the men grew up living with their mother in both the gang and non-gang sample. Neither the number of siblings (roughly three) nor the likelihood of working in the legitimate sector in 1991 (between 40 percent and 50 percent did) differed systematically with gang status. In addition to self-reported data, Table 1 also shows the results of community-member ratings of the individuals on a seven-point scale (1 being the lowest, 7 the highest). Gang members are on average physically stronger, less hardworking, less serious about school, and less well liked than those not in the gang, although only for the last of these measures is the difference statistically different at the 0.05 level. Surprisingly, community members perceived no differ- VOL. 91 NO. 2 TABLE 2—ECONOMIC Variable Highest grade completed Graduated from high school Currently employed No source of income at current time Total annualized income, if income greater than zero Annualized legal income, if total income greater than zero Annualized illegal income, if total income greater than zero Owns a car Some money saved AND ECONOMICS AND SOCIAL BEHAVIOR EDUCATIONAL OUTCOMES Non-gang 11.8 (0.2) 0.66 (0.06) 0.80 (0.05) 0.11 (0.04) 23,966 (1,145) 21,397 (1,438) Differencea 0.6* (0.2) 0.24* (0.11) 0.18† (0.11) 0.17† (0.09) 2,935 (2,651) 8,446* (2,214) 81 Overall sample 11.6 (0.1) 0.58 (0.05) 0.74 (0.05) 0.17 (0.04) 23,144 (1,063) 19,032 (1,214) Gang members 11.2 (0.1) 0.41 (0.09) 0.62 (0.09) 0.28 (0.08) 21,031 (2,391) 12,950 (1,684) 4,113 (986) 0.64 (0.05) 0.43 (0.05) 8,081 (2,299) 0.63 (0.09) 0.50 (0.10) 2,569 (975) 0.64 (0.06) 0.39 (0.06) 5,512* (2,498) 0.01 (0.11) 0.11 (0.12) Notes: Our sample is the 90 living members of the initial 118 housing project residents in 1991 who have been re-interviewed in 2000. Values in columns are within-group means. Standard errors of the mean are in parentheses. Annualized income is calculated by multiplying reported income for the month preceding the interview by 12. a Gang minus non-gang. † Significantly different from zero at the 10-percent level. * Significantly different from zero at the 5-percent level. ence in the likelihood of staying out of trouble by gang status, although the criminal-justice experiences of these two groups are radically different. Table 2 presents means and standard deviations for measures of educational attainment and a wide range of economic indicators.2 Educational attainment in the sample is low: the average number of grades completed is 11.6, and less than 60 percent of the respondents 2 Because of space constraints, the analysis presented here is limited to economic and educational outcomes. We have also assembled a great deal of information on criminal justice and social outcomes as well. We note in passing that the exposure of these men to the criminal justice system and violence is remarkably high: Seventy-seven percent of the men in the sample report having been arrested, and 60 percent have been incarcerated (for an average of a little more than two years each). Almost 10 percent of the individuals in the original sample are dead, mostly as a consequence of violence. The average number of gunshot wounds among those surviving in our sample is 0.41 per person. Gang involvement dramatically increases the amount of contact with violence and the criminal justice system. completed high school. None of the individuals completed more than two years of post-highschool education. Educational attainment among gang members is significantly lower than for non-gang individuals. High-school graduation rates, for instance, are 25 percentage points lower among gang members. The members of our sample have not fared particularly well in the labor market. Approximately three-quarters of the sample is currently employed in the legitimate sector. This rate is extremely low for prime-age males, especially when the overall unemployment rates in the Chicago metropolitan area was around 4.0 percent in 2000. More than one-sixth of the respondents report no source of income (either legitimate or illegal); slightly more than half of those with no income are currently incarcerated. Excluding those with no income, average annualized income from all sources is slightly over $23,000, with 80 percent of that income from the legitimate sector. Of those with some income, 91 percent report a legal source of income, and 72 percent report only legal sources of income. Roughly 20 percent report both legal and illegal income. There are no systematic differences in total income for those with only legal, only illegal, or both sources of income. There are important differences between the gang and non-gang sample. Those who were active in the gang in 1991 are much less likely to be currently employed in the legitimate sector (62 percent versus 80 percent) and are almost three times as likely to have no source of income in 2000. Although total income from all sources is similar for gang members and others (conditional on having some source of income), those who grow up affiliated with the gang garner almost 40 percent of their income from illegal activities, compared to less than 15 percent for others. There is no difference in car ownership (the major asset for most of these men) across the groups. Surprisingly, gang members were more likely to report having some money saved, although the difference was not statistically significant. Research of this nature cannot definitively isolate a causal role for gang involvement. The fact that gang and non-gang individuals look similar on all of the background characteristics in Table 2, yet have very different educational and labor-market experiences, suggests that being active in the gang has 82 AEA PAPERS AND PROCEEDINGS MAY 2001 long-term consequences. This result parallels the earlier findings of Terence Thornberry et al. (1994). Table 3 explores the relationship between background characteristics of individuals in the sample and the decisions regarding gang involvement and educational attainment. Each column of Table 3 represents a separate regression. The set of explanatory variables comprises those characteristics that are predetermined as of 1991 (employment status in 1991, lived with mother growing up, number of siblings, and age), gang status, educational attainment, years imprisoned, and community-member assessments of physical strength and the average over the other four traits (hard working, stays out of trouble, serious about school, and well-liked). We take an average over the other four traits because the pairwise correlations of those traits are all very high, between 0.4 and 0.7. In the educational-attainment regressions, gang status is also included as a covariate. In all instances, the estimation approach is ordinary least squares. Background characteristics do a poor job of predicting gang status. The R 2 value of the regressions is less than 0.10. Although many of the variables carry the expected sign (more siblings, greater strength, and lower levels of other desirable traits are all associated with higher levels of gang involvement), none is statistically significant. Those who are younger are less likely to be in the gang. The regressions predicting educational attainment perform only slightly better. Not being in the gang, growing up with one’s mother, having fewer siblings, being less physically strong, and having higher levels of other desirable traits all carry point estimates implying increased years of education. The statistical significance of these measures, however, is weak. Panel B of Table 3 reports economic outcomes as the dependent variable. The last column in panel A is a regression using an indicator variable for whether an individual has any source of current income. The only statistically significant predictor is years of education: each year of schooling increases the probability of having positive current income by 10 percentage points. Living with one’s mother while growing up, having been employed in 1991, and staying out of the gang all carry sizable point estimates but are not statistically significant. The first column of panel B TABLE 3—PREDICTORS OF GANG INVOLVEMENT, EDUCATIONAL ATTAINMENT, AND ECONOMIC OUTCOMES A. Variable Active in gang Years of education Years incarcerated Employed in 1991 Lived with mother growing up Number of siblings Age in 1991 Assessed physical strength Assessed composite of other attributes Constant R2: Number of observations: B. Variable Active in gang Years of education Years incarcerated Employed in 1991 Lived with mother growing up Number of siblings Age in 1991 Assessed physical strength Assessed composite of other attributes Constant R2: Number of observations: Annualized Annualized Annualized total incomeb legal incomeb illegal incomeb 1,090 (2,639) 2,973* (1,048) 892 (855) 357 (2,358) 8,383† (4,720) 254 (594) 661 (442) 1,169† (621) 471 (794) 40,357* (16,350) 0.364 63 2,329 (2,129) 3,712* (925) 2,783* (754) 3,745† (2,081) 4,733 (4,165) 24 (524) 249 (390) 262 (549) 1,532* (700) 26,514 (14,428) 0.598 63 1,239 (2,233) 740 (887) 1,891* (724) 3,388† (1,995) 3,650 (3,994) 279 (503) 412 (374) 1,432* (526) 1,060 (672) 13,843 (13,837) 0.383 63 Active in gang, 1991 — — — 0.10 (0.12) 0.09 (0.21) 0.03 (0.03) 0.03 (0.02) 0.03 (0.03) 0.05 (0.04) 1.09 (0.69) 0.085 74 Years of education 0.41 (0.30) — — 0.32 (0.31) 0.47 (0.53) 0.12 (0.08) 0.03 (0.05) 0.05 (0.08) 0.11 (0.10) 10.41* (1.77) 0.127 74 Incomea 0.08 (0.10) 0.10* (0.04) 0.03 (0.03) 0.14 (0.09) 0.24 (0.16) 0.03 (0.16) 0.01 (0.02) 0.01 (0.03) 0.03 (0.03) 1.38 (0.67) 0.209 74 Notes: The dependent variables are listed at the head of each column. All regressions estimated using ordinary least squares. Standard errors are in parentheses. See text for further description of variables. Columns 4–6 are restricted to those reporting positive earnings. a Total income greater than zero (indicator variable). b If total income greater than zero. † Statistically different from zero at the 10-percent level. * Statistically different from zero at the 5-percent level. presents results for total annualized income for the subset of the sample with positive income. Each year of education is associated with an VOL. 91 NO. 2 ECONOMICS AND SOCIAL BEHAVIOR 83 increase in total income of almost $3,000 among those with income, implying an unusually high rate of return on an additional year of schooling of over 10 percent. Combined with the fact that an additional year of education increases the probability of having some source of income, the total return to a year of schooling is over 20 percent. This result is in line with some previous estimates of the returns to schooling for this part of the ability distribution (Heckman, 2000). Growing up with one’s mother in the household is associated with an increase in income of over $8,000 annually, consistent with earlier research suggesting that the quality of the home environment is an important determinant of later outcomes (e.g., Robert Sampson and John Laub, 1993). Less time incarcerated, being older, and having greater physical strength are all positively correlated with income, although none of these variables is statistically significant at the 0.05 level. Examining legal and illegal income separately (the last two columns in panel B) suggests that the factors that predict success across the two categories are very different. With the exception of two variables (living with one’s mother growing up and age), each of the regressors carries a different sign when predicting legal versus illegal income. Education is an important determinant of legal income, as is holding a legal job in 1991. Physical strength, not being employed in 1991, and years incarcerated are the variables most strongly correlated with illegal income. Interestingly, the composite indicator (reflecting an individual’s hard work, seriousness about school, staying out of trouble and being well liked) is positively related to illegal income but strongly negatively related to legal income. This puzzling result may reflect the fact that assessments of such traits by community members may not correspond to the views of broader society. In other words, what makes one appear to be hardworking or well-liked in the housing project may not effectively translate into a job setting but is valuable in the illegal sector. III. Conclusion that an ethnographer has established with a community to obtain information about a particular subgroup of the population: young men who came of age at the peak of the crack epidemic in a neighborhood that is economically marginalized and heavily influenced by gangs and drugs. The data we have assembled are tailored to the specific research questions of interest and are of a type and detail unavailable in standard surveys. These data allow us to address questions that would otherwise be foreclosed. A number of findings emerge from the research. At a time when unemployment rates in Illinois are near 4.0 percent, more than onequarter of this sample (now in their twenties and thirties) are not currently employed in the legitimate sector, and one-sixth report no source of income at all. Among those with some source of income, annualized earnings are approximately $23,000, with 20 percent of that income from illegal sources. The returns to education are quite large in the legal sector. Physical strength is an important determinant of illegal income. Although those who were gang members as adolescents appear similar to non-gang individuals on observable characteristics, gang members obtain less education, fare much worse in the legitimate economy, but are much more successful in generating illegal income. Thus, it appears that youthful gang involvement has a long-run impact on an individual’s economic trajectory. REFERENCES Biddle, Jeff and Hamermesh, Daniel. “Beauty, This paper represents an attempt to enrich the traditional empirical economic approach by combining it with elements of sociology. In particular, we exploit the unique relationship Productivity, and Discrimination: Lawyers’ Looks and Lucre.” Journal of Labor Economics, January 1998, 16(1), pp. 172–201. Case, Anne and Katz, Lawrence. “The Company You Keep: The Effects of Family and Neighborhood on Disadvantaged Youths.” National Bureau of Economic Research (Cambridge, MA) Working Paper No. 3705, 1991. Duncan, Greg and Dunifon, Rachel. “Soft Skills and Long-Run Labor Market Success,” in Solomon W. Polachek, ed., Research in labor economics, Vol. 17. Stamford, CT: JAI Press, 1998, pp. 123–56. Edin, Kathyrn and Lein, Laura. Making ends meet: How welfare mothers survive welfare and low-wage work. New York: Russell Sage Foundation, 1997. 84 AEA PAPERS AND PROCEEDINGS MAY 2001 Heckman, James. “Microdata, Heterogeneity and the Evaluation of Public Policy.” Nobel Prize Address presented in Stockholm, Sweden, December 2000. Rosenbaum, James. “Changing the Geography of Opportunity by Expanding Residential Choice: Lessons from the Gautreaux Program.” Housing Policy Debate, 1995, 61(1), pp. 231– 69. Sampson, Robert and Laub, John. Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press, 1993. Thornberry, Terence; Lizotte, Alan; Krohn, Marvin; Farnworth, Margaret and Jang, Sung Joon. “Delinquent Peers, Beliefs, and Delin- quent Behavior: A Longitudinal Test of Interactional Theory.” Criminology, February 1994, 32(1), pp. 47– 83. Venkatesh, Sudhir. American project: The rise and fall of a modern ghetto. Cambridge, MA: Harvard University Press, 2000. ...
View Full Document

This note was uploaded on 05/13/2010 for the course ECON 323 taught by Professor Jakes during the Spring '10 term at Alcorn State.

Ask a homework question - tutors are online