In the decades following World War II, the center of gravity in American urban areas shifted from the inner city to the suburban ring.¹

Suburbanization occurred alongside an influx of black migrants from the rural South, the majority of whom settled in downtown areas.² My dissertation begins by assessing the causal relationship between this growing black urban population and white relocation to the suburbs. I then explore the motivation of households that left racially diverse cities. The existing historical literature on “white flight” emphasizes white distaste for living near black neighbors.³ However, because cities were already highly segregated by neighborhood, avoiding black neighbors did not require a suburban address.⁴ I focus instead on the civic costs of living in a diverse jurisdiction. These include compromising on public goods and sharing a single school system. Lastly, I consider the effect of black migration on the wages of similarly skilled workers in the northern labor market.

The first chapter documents a correlation between black in-migration and the relocation of white households to the suburbs within metropolitan areas over time. This pattern is consistent with—but not necessarily proof of—white flight. First, any in-migration to a downtown area could bid up urban housing prices, inducing some residents to move out. However, black in-migration was associated with falling prices in cities relative to their suburbs, which implies that the migrants themselves reduced the perceived value of urban living.

A correlation between black migration and white suburbanization could also arise if black migrants were attracted by the same economic factors that underlie the demand for suburban living—for example, rising incomes and centrally located manufacturing jobs—or by cheaper urban housing left in the wake of white departures.⁵

To address this possible source of reverse causality, I design an instrumental-variables procedure that predicts black out-migration from southern counties using local agricultural conditions and assigns these flows to northern cities based on settlement patterns observed during the first black migration (1915–1934).⁶ These predictions are correlated with black population growth in northern cities both because black mobility was sensitive to southern agricultural change, and because black migration followed persistent channels from particular southern states to specific northern cities.⁷

Even after accounting for migrant location choices, I detect a strong, positive relationship between changes in a city's racial composition and the white suburban share of the surrounding metropolitan area. My estimates suggest that, if not for southern black migration, the growth of suburban areas would have been 20 percent lower. This magnitude leaves room for factors that pulled residents out to the suburbs, including rising incomes, the building of interstate highways, and increasing family size, but also suggests that racial changes within cities played an important role.⁸

By remaining in the city, whites had to vote in municipal elections and send their children to public school with black newcomers, whatever the demographics of their immediate neighborhood. The suburbs may have been attractive not only for their homogeneity, but also for their political autonomy. The second chapter of my dissertation posits that a desire for political distance should be reflected in higher housing prices in predominately white municipalities. Evaluating this hypothesis is an empirical challenge. Suburban housing differs from its city counterpart along a number of dimensions, including larger lots and newer construction, all of which may command a positive price. To isolate the contribution of local government, I exploit the division of urban space into separate jurisdictions and compare the prices of neighboring units located on either side of municipal borders. The underlying identification assumption is that, while local policy and school assignments change abruptly at borders, housing and neighborhood quality shift more continuously.⁹

To account for any fixed differences in housing quality, for example due to local zoning regulations, I also compare changes in the housing price gap with changes in jurisdiction attributes.

Block-level data on housing prices are available from the Census of Housing for over 50 city-suburban pairs in 1960. I collect data along these borders over three Censuses (1960–1980), matching housing units to demographic, socio-economic, and political features of their respective jurisdictions.¹⁰

In this period, homeowners were consistently willing to pay to avoid living in jurisdictions with poor residents. For example, the average border in 1970 divided a town with a 5 percent poverty rate from a city with a 9.3 percent poverty rate. Crossing this border reduced home values by 1.5–2.0 percent. This relationship can be partially attributed to the fact that higher poverty is associated with higher property taxes.

In contrast, the relationship between housing prices and racial composition was not stable over time, but appears to have been tied to particular historical events. In the 1960s, city-level racial diversity only affected housing prices in areas that experienced race-related riots.¹¹

Although these border areas were, on the whole, far from riot damage, riots may have shifted the racial balance of political power. In the 1970s price declines were concentrated in cities that enacted court-ordered desegregation plans.¹² On average, housing prices fell by 1 percent for each mandated action in a desegregation remedy, whereas some steps—particularly forced bussing and school reassignment—reduced prices by up to 6 percent.

Taken together, these findings nuance our view of white flight. Black in-migration triggered the relocation not only of immediate neighbors but also of residents seeking to escape the changing city electorate. While this out-flow may be better termed a flight of the middle class, race itself mattered in the schools and when underscored by local violence.

The third chapter turns to the effect of black migration on the northern labor market. Black migration represented a substantial increase in the labor supply above the Mason-Dixon line, particularly for the low-skilled workforce. Although migration benefited those who left the South, the arrival of this “great reservoir of labor”—so termed, presciently, by W. E. B. Du Bois in the 1920s—may have had negative consequences for blacks already in the North.¹³

To evaluate the impact of this labor flow, I begin with the premise that migrants should have the strongest effect on workers who were close substitutes in production. Following George Borjas, I partition the northern economy into skill groups according to education and work experience. Workers are assumed to be perfect substitutes with other members of their skill group and imperfect substitutes (to an equal degree) with members of other cells. Each skill group starts in 1940 with a stock of black southern migrants and receives a new flow in each subsequent decade. I use this over-time variation to identify the effect of migrant supply shocks on the wages of competing northern-born workers.

My estimates imply that a flow of black southern migration large enough to increase labor supply by 5 percent (the mean across skill groups) would have reduced the annual earnings of existing black workers by 3–5 percent, but would have had no effect on similarly skilled whites. This pattern bolsters case study evidence, which documents that black workers were channeled into “negro jobs” and faced limited opportunities for promotion.¹⁴

The ongoing migration also helps to explain the fact that racial convergence was faster in the South than in the North during this period.¹⁵