14 Points

For a chapter of my dissertation, I calculated the extent to which blacks and whites vary according to their exposure to homeownership, unemployment, and college education in 49 of the largest metropolitan areas in the United States. This represents over half of the total U.S. population and almost 60% of the black population. The results are clear: in most MSAs with above average segregation, blacks were less likely to live in the same neighborhood as homeowners, people with jobs, and people with college degrees. Segregation is a strong and reliable predictor of these inequalities, even after controlling for a long list of metropolitan characteristics.

The question is: how does this affect the lives of black people? It’s reasonable to suspect that these things matter, and matter greatly, in the upbringing of children or in the probability of adult success. Homeownership, for instance, is strongly associated with civic participation, as the work of economists Denise DiPasquale and Edward Glaeser (1999) demonstrates. They find that homeowners are more likely to work on solving local problems, attend church, volunteer, and know more about local politics. This homeowner effect is true even for people who switch from renters to homeowners.  In that sense, homeowners are like the administrators of the private welfare state; they are, in many respects, a fountain of charity and advocacy on behalf of communities.

Meanwhile, young black children are disproportionately deprived of contact with employed, college educated adults. These are people who could otherwise teach them the habits and pathways of success and self-determination. A famous paper by Anne Case and Lawrence Katz (1991) found that the behavior of neighbors affects the behavior of young adults and teenagers, with respect to church attendance, education, and out-of-wedlock birth.

Indeed, segregation is proven to have severe consequences on the accumulation of human capital. In an extraordinary paper, Cutler and Glaeser (1995/97) present evidence that blacks would graduate from high school at the same rate as whites if residential segregation were eliminated. Their main strategy is to estimate individual level outcomes data conditional on segregation and the effect of segregation on blacks, while controlling for some basic metropolitan characteristics -namely the log of city population, percentage black, the log of median household income, and the share of city employment in manufacturing.  They find that eliminating segregation would eliminate all differences between blacks and whites on high school graduation, income, and idleness, and two-thirds of the difference in out-of-wedlock births.

Some critic may ask, doesn’t doing poorly on these factors cause blacks to live in poorer and more segregated cities? It turns out that Cutler and Glaeser (1997) get the same results when they allow for this possibility. They do this by using the number of metropolitan governments in 1962 and the degree that school districts are locally financed in 1962 as external sources of variation in segregation. Said otherwise, metropolitan governance in 1962 is not caused by black poverty and education attainment in the 1990s, but it does predict segregation. So, if they use the portion of segregation that is correlated only with metro governance in 1962, they have an unbiased measure of segregation that can be causally interpreted.

These results are consistent with Card and Rothstein (2007), who find that residential segregation significantly reduces black SAT scores. Moreover, segregation at the neighborhood level has more consistent and negative effects than school segregation.  In their most basic specification, 70% of the SAT score gap between blacks-and whites would be eliminated if residential segregation were wiped out. Card and Rothstein’s work relies on an even more thorough methodology, controlling for a wide variety of factors, including the overall level of income inequality.

Card and Rothstein (2007) also examine how segregation affects education attainment more directly. Controlling only for MSA-characteristics, they find that segregation decreases the gap between the percentage of blacks and whites who graduate high school by 12%. This falls to 6% after adding parental education attainment but remains significant.

In a recent working paper of my own, I ask the question: how does segregation distort the market for human capital? That is, given a clear economic incentive to become educated, do blacks respond differently in a segregated metropolitan area as opposed to one that is integrated? The answer is definitely yes. The share of blacks with college degrees increases 2 to 5 percentage points for every standard deviation decrease in segregation. For big chances in desegregation (a .5 reduction on a 1 point scale), this effect amounts to a 7-16 percentage point increase in the share of educated blacks.

To calculate the economic incentive to get a college degree, I obtained individual data from IPUMS for a broad sample of MSAs. For young black and white adults, I calculated the average return on a college education, as measured by annual wages per week, after adjusting for sex, age, household head, and whether or not the respondent is currently in school. Then, I used lagged values of the college premium to predict college attainment ten years in the future for young adults. I controlled for a variety of metropolitan characteristics correlated with education attainment, including previous education attainment. This yielded the results reported above.

Finally, I tested if changes in lagged segregation and the college premium predict changes in future attainment, controlling for unobserved state variables that didn’t change from 1980 to 2000 (e.g. climate, governance structure, history). The results remained significant. For big changes in segregation, the share of blacks with college degrees increased by 2-3 percentage points every 10 years.

In future research, I intend to explore the extent that social capital is mitigated by living in segregated communities. That is, I suspect that trust and cooperation between neighbors is inhibited greatly in these communities. Scholars of segregation have suspected this for decades. St. Clair Drake and Horace Clayton’s (1945) classic study found that, among other things, participation in social clubs and church membership was lower for the poorest wage earners. In a recent book, the Berkeley sociologist Wacquant (2008) finds that church membership and participation in voluntary activities is lower for blacks in segregated black ghettos compared to those living in more integrated communities. Friendship and intimate relations also reportedly suffer according to the Urban Family Life Survey, which is analyzed by Wacquant.

These results fit in neatly to more theoretical literature on cooperation, which is thought to be enhanced by repeated and predictable interactions. These are lacking in communities with high percentages of renters and low marriage and employment rates.  Furthermore, wealth and income are facilitators of bargaining agreements because they offer insurance that informal repayments will be made, and favors are worthwhile.

If this logic is valid, we already have a profound explanation for the persistence of racial inequality in the U.S.A. Cooperation is vital to establishing and improving public goods through collective action. No one household can single-handedly be responsible for the quality of local institutions -both formal and informal -including schools, journalism, social services, professional organizations, educational opportunities, child-rearing duties, and policing.

Public Policies to Promote Desegregation

Unfortunately, the role of the federal government has often been on the side of promoting or permitting segregation, rather than seeking to mitigate or prevent it. Historian Arnold Hirsch (1993) deems the period from 1933 to 1968 second of three stages in the formation of the modern ghetto. He characterizes this period as heavily marked by federal government policy, which maintained and accelerated racial segregation. Previously the government’s role was only enforcing racial covenants and zoning ordinances (the latter until 1917), during this period it actively employed discriminatory policies in its support for private housing, public housing, slum clearance, and urban renewal (Hirsch, 1993). Cutler, Glaeser, and Vigdor, 1999 present econometric evidence that supports Hirsch’s main contention that collective action on the part of whites, whether official or quasi-official, was the primary cause of segregation, at least until 1970, when private discrimination of whites appears to drive the results, according to their data. Since the Civil Rights Movement, the federal government has generally supported policies that promote desegregation, whether through housing vouchers or anti-discrimination laws, but even these laws did not have appropriate teeth until the late 1980s (as Massey and Denton 1993 argue).

Yet, I’d argue that there have been very few efforts on the part of the federal government to deal with the primary source of segregation: local land use regulation. Using survey data from Cornell University Professor Rolf Pendall, Doug Massey and I go a long way towards proving that metropolitan level segregation is perpetuated by anti-density and anti-development land regulation. These laws, which simultaneously lead to sprawling low-density metropolitan areas, inhibit migration out of the central city and suburban ghettos of the Northeast and Midwest. We feel confident that we have identified a causal effect because we can replicate our results using year of statehood as an exogenous source of variation. As is happens, year of statehood does not predict segregation in the 1920s, before zoning laws were widely adopted, but it does very well in predicting density zoning from 1980 through 2003.

Federal policy makers should start taking local governance seriously. The effects of anti-density zoning on housing production cross state boundaries and the disparate harm to blacks and low income groups necessitates suspicion. There is one recent precedent. The 2000 Religious Land Use and Institutionalized Persons Act mandates a very high standard (“strict scrutiny”) for laws that poses a substantial burden on religious exercise (e.g. forbidding the construction of a religious building). This act recognizes that local land use laws can not infringe on important civil rights, and the federal government has an obligation to intervene.

While the logic of religious discrimination cannot be applied, we can look to Supreme Court’s rulings on racial discrimination for guidance. In a case with direct applicability to land use zoning, Village of Arlington Heights (outside of Chicago, Illinois) vs. Metropolitan Housing Corporation, 1977, the developer argued that it was denied rezoning for higher density racially integrated construction because of racial bias in the municipalities anti-density zoning laws. In a 5-3 decision that reversed the Court of Appeals rationale, the Supreme Court ruled that “protecting property values and the integrity of the Village’s zoning plan” was a legitimate rationale and that there was no evidence of intent to discriminate as to race. The majority proceeded to establish a standard for determining racial discrimination:

“A racially discriminatory intent, as evidenced by such factors as disproportionate impact, the historical background of the challenged decision, the specific antecedent events, departures from normal procedures, and contemporary statements of the decision makers, must be shown.”

In the majority opinion, Justice Powell argued that the court may have ruled differently if Arlington had switched its zoning law after finding about the developer’s intention to build multi-family racially integrated housing or if minutes from the administrators of the law revealed an anti-black animus.

In a related case involving hiring, the court has another standard even if no intent to racially discriminate is established. The Supreme Court ruled in Griggs v. Duke Power Company, 1971, under Title VII of the Civil Rights Act of 1964, that no rules made by an employer that have a disparate impact on blacks can be used if they do not have a reasonable basis, which may include higher profits. The court upheld that rules that harm blacks can in fact meet this standard in Washington v. Davis, 1976.

Since, higher property values would likely be in accord with such a rational standard, the Supreme Court is unlikely to rule against anti-density zoning in all but the most obviously racist cases. The burden on the prosecution would be immense in such circumstances, which, in addition to the fact that state constitutions usually have authority over local laws, probably explains why no case has been won against anti-density zoning in federal courts (see Mt. Laurel v. NAACP 1975 for a triumph in the New Jersey State Supreme Court).

Yet, there would be nothing preventing Congress from revising the Fair Housing Act to outlaw any housing regulation that causes disparate harm to minorities and the poor unless, in the language of the Religious Land Use and Institutionalized Person Act of 2000, it is the least restrictive means of furthering a compelling governmental interest. The constitutional grounds for writing such legislation are easily met since zoning laws affect interstate housing markets, by distorting housing supply and prices (See Rothwell 2008 for evidence). Anti-density zoning laws effectively discriminate against renters or anyone who cannot afford large single family homes by preventing their demand for housing from being realized. At the same time, they significantly contribute to racial inequality.