State-level "Fair Housing" laws sought to eliminate discrimination based on race, religion, and national origin in the sale, rental, and financing of housing, and were the direct antecedents of the federal Fair Housing Act of 1968. By the time When the federal government passed the Fair Housing Act of 1968, there were 22 states, covering 57 percent of the entire US. While fair housing laws are typically considered among the most significant legislative achievements of the Civil Rights Movement, as well as their impact on blacks.
All of this leaves the door open for a significant effect of fair housing policies, but it is far from a direct evaluation of the hypothesis that fair housing laws have helped improve housing market outcomes for Black people. In section 2, I review the history of fair housing laws and outline the economics of their potential impact in section 3. As suggested above, the question remains whether fair housing laws have actually contributed to the observed relative improvement in the housing market outcomes for black people an unanswered question. empirical question.
The empirics below proceed as if the 1968 Fair Housing Act had a negligible (or an even-across-states) effect on African Americans' relative housing market outcomes by the time of the 1970 census.4. In theory, the potential impact of a fair housing law depends on how discriminatory attitudes and institutions affected housing market equilibrium prior to the law's implementation.
Empirical Strategy: Household Data
I also examine fair housing effects on the probability of home ownership and the size of households. Essentially, DDD estimation compares blacks' housing in states with relatively extensive fair housing laws ("experimental states") to blacks' housing in states with little or no fair housing laws ("nonexperimental states"), relative to whites' outcomes within each state group (a second difference), before and after the laws' adoption (a third difference). Ohio, Illinois, and Wisconsin, all of which had fair housing initiatives narrower in coverage than those of the experimental states.
The non-experimental group B consists of Delaware, Kansas, Maryland, Missouri, and West Virginia, all of which had weak or no fair housing laws applicable to private homes. For β8 to accurately measure the impact on fair housing, the laws must of course be exogenous to changes in the racial gap in housing market outcomes. But if there is a race-specific change in the nature of selection into homeownership that differs across state groups, perhaps due to fair housing laws themselves, then interpreting the results obtained by using only owners or only renters would use may be distorted.
Using non-experimental group A, there is no evidence that the fair housing laws facilitated movement into the ownership category. Again, there is no evidence that the fair housing laws have been associated with significant relative changes. By this measure, there is no evidence that whites in fair housing states have had a greater decrease in the probability of living in relatively old housing than whites elsewhere.
But by this standard, relative to whites, blacks in fair housing states appear to have lost ground over the decade relative to blacks in nonexperimental states. The regressions of Table 4 already directly control for family size; and in Table 3, where household size is treated as a dependent variable, there is no evidence that fair housing laws have been associated with significant relative changes. A third hypothesis is that whites in fair housing states responded to the laws by moving to relatively large suburban homes.
In this case, blacks in fair housing states may have apparently fared poorly relative to whites in the DDD framework, even though they benefit from the laws by gaining access to better housing than before. In a similar black-only DD framework, there is no support for the idea that blacks in fair-housing states increased their number of rooms relative to other blacks: black owners and renters both lost ground in the fair-housing states relative to blacks in other states. Among owners, there is no evidence that the Fair Housing Act facilitated improvements in black housing; the coefficients are consistently negative and insignificantly different from zero.
Suburbanization, Segregation, and Neighborhood Racial Composition
This finding may seem somewhat confusing given the results in Table 4, which suggested that there were no significant positive effects on the quantities of the different housing components, and that in fact there was a slight relative decrease in the number of rooms per household in reasonable proportions. housing states. 21 Due to the coverage of the metro variable, Colorado was dropped from the sample entirely in 1960, but in the 1970 sample. In contrast, black renters in states with strong housing status appear to have slightly increased their relative likelihood of living in a county. the suburbs.
By using the consistency maps and tables included in the published census books for each metropolitan area in each year, it is possible (usually) to identify the "new" areas added in 1970 and thus make more consistent comparisons over time. Louis (MO and IL) and Weirton-Steubenville (WV and OH) are treated as one unit in the dissimilarity indices and are included in both nonexperimental groups in Table 7 . Given that this paper focuses on state law, if part of a metropolitan area in a state not included in the study, these tracts are omitted.26.
A simple comparison of the change in the level of segregation between experimental and nonexperimental state groups during the 1960s suggests that segregation fell slightly more in experimental states than in nonexperimental states, although the differences are small. With this in mind, table 7 regresses the change in the level of segregation between 1960 and 1970 on nonwhite population growth, white population growth, and the fair housing dummy variable for a sample that includes metro areas with at least 1,000 nonwhite residents in 1960. The "predominantly nonwhite" row focuses on the other end of the racial composition distribution, examining changes in the proportion of tracts that were predominantly nonwhite (more than 50 percent).28 Appendix Table A3 reports summary statistics for census tract data.
When considering the non-white share of the population of each metropolitan area (as described in equation 2), the decline was slightly greater in experimental states than in non-. While the share of nonwhite residents in the relatively high human capital sector increased between 1960 and 1970 (from 3.0 to about 4.6 percent), it did not increase any more in states with strong fair housing than in comparison groups of states. Finally, in Panels C and D, I examine changes in house prices or rent levels in “majority nonwhite” neighborhoods relative to others under DDD.
29 The 1970 data show that the “new,” mostly non-white areas had higher property values and rents than the original, mostly non-white areas.
Conclusions
But to make sure that "new" predominantly non-white tracts are not confounding the results in panels C and D, it helps to compare the geographic coverage of predominantly non-white tracts in the 1960s and 1970s using maps in the census volume of each metropolitan area and then identify tracts that were predominantly non-white in both census years.30 After doing this, I re-ran the regressions, assigning tracts the status of “predominantly non-white” only if they were clearly predominantly non-white in both the 1960s and 1970s - the idea is to compare (roughly) continuous coverage areas for non-white and white tracts over time. Although fair housing policymakers and advocates were clearly concerned with the outcomes of the black housing market, studies of the political history of state fair housing legislation do not suggest that differences in the timing of adoption across states were driven by differential changes in relevant housing. of blacks. market results. Change in the size of the black population would be a prime candidate to link policy changes and changes in average black housing market outcomes, but the results are generally insensitive to adding controls for the black percentage of each state's population. .
On the basis of the existing evidence, it is difficult to argue that their impact was significantly uneven across the state groups compared here. A more direct examination of federal influence would certainly be worth undertaking, but based on. In The Politics of Fair Housing Legislation: State and Local Case Studies, edited by Lynn W.
Notes: Nonwhite share of population, change in share of condominiums, and share of units built during the 1950s are calculated using the published population and housing census amounts. As noted, state-level data were taken from different volumes of the published population and housing censuses of 1950 and 1960. Each reported coefficient is from a separate regression and relates to the interaction between the fair housing, black, and 1970 indicator variables.
The text describes the construction of the hedonic indices; Please note: the figures are for owners. As discussed in the text, the availability of the suburb–central city distinction is compromised in some places for confidentiality reasons. Notes: In panel A: the “very white” row reports are the result of an OLS regression where the dependent variable equals one if the white share of the tract population is greater than 0.99; the “mostly non-white.”
The regressions include fixed effects for metropolitan areas, the share of the area's population that was nonwhite each year, a 1970 dummy variable, and the interaction between the 1970 dummy and a dummy for states with good housing. In Panel C: the log of average property value or monthly gross rent is regressed on metropolitan area fixed effects, log of metropolitan area population, a dummy variable for predominantly nonwhite areas, a 1970 dummy variable , an interaction of the mostly nonwhite variables and 1970s variables, an interaction of the mostly nonwhite variable and a dummy for experimental conditions, an interaction of the experimental and 1970s dummies, and (reported in table) an interaction of the 1970 dummy, the experimental condition dummy, and the predominantly nonwhite tract dummy . See the notes to Tables 4 and 6 for full descriptions of the variables and regression specifications.
