VoxEU Column Economic history Migration

The rise of residential segregation

Racial disparities in socioeconomic conditions remain a major policy issue throughout the world. This column applies a new neighbour-based measure of residential segregation to US census data from 1880 and 1940. The authors find that existing measures understate the extent of segregation, and that segregation increased in rural as well as urban areas. The dramatic decline in opposite-race neighbours during the 20th century may help to explain the persistence of racial inequality in the US.

Academics have approached residential segregation primarily by measuring the distribution of minority populations across the political units in a city. Traditional indices of segregation are similar to measures of income inequality – they attempt to assess how evenly that minority population is spread across those units or how isolated the minority population is by looking the racial composition of typical ward in which minorities live. These traditional measures are used in the seminal work of Cutler et al. (1999) documenting the rise and decline of the American ghetto. As black residents moved to cities during the Great Migration, those cities became increasingly segregated. Between the late 19th century and 1940, cities developed cores that were almost entirely black as whites moved out of the city to the suburbs. The size of those cores then grew over the subsequent decades.

This white flight from urban areas has become the focus of much of the academic and policy debates surrounding racial inequalities. With the flight of wealthier white residents to the suburbs, the resources available to the community also fled to the suburbs, leaving minorities in the urban cores with fewer resources and effectively creating conditions for a poverty trap. However, several aspects of this story remain unresolved. We know little about the motivations underlying this rise in residential sorting. Further, traditional segregation measures tell us how that sorting was occurring across political units but cannot speak to residential patterns within those units. Consequently, questions about the impacts residential segregation at a very local level have remained beyond our grasp.

At the same time, economists and other social scientists increasingly stress the influence of peers and social networks on a variety of outcomes. Similarly, urban economists stress the presence of agglomeration economies in urban areas. Each of these ideas has interaction (and its externalities) at its core. While segregation would naturally be related to these ideas, developing a measure that could successfully proxy for residential interactions between races has proved difficult.

Measuring segregation using historical census data

With the recent digitisation of historical census manuscripts, we have an opportunity to push beyond traditional segregation indices and measure residential segregation at the finest level possible, that of next-door neighbours. Federal census enumerators would go door-to-door when enumerating the residents of a community. Consequently, next-door neighbours appear adjacent to one another on the census manuscript pages. With the availability of complete census manuscripts for 1880 and 1940, this allows us to identify the races of all household heads and of their next-door neighbours. With this information, we have constructed an intuitive new measure of residential segregation that compares the number of black households with white neighbours in a community to the number expected under complete integration (where household location is completely independent of race) and under complete segregation (where all black households are concentrated in one continuous neighbourhood) (Logan and Parman 2015). The resulting segregation index tells us the extent to which households tend to sort themselves within a community given the overall racial composition of the community. This new measure does not depend on the boundaries of geographical subunits like wards, can be calculated at any geographic level, and serves as a much stronger proxy for levels of social interaction between races.

Residential segregation at the local level

When we apply this new measure to the entire United States in 1880 and 1940, several striking features of American residential segregation emerge. Consistent with the story told by traditional segregation indices, residential segregation rose dramatically throughout the US over the first half of the 20th century. Figure 1 demonstrates this dramatic rise, showing county-level segregation in 1880 and 1940 for the Eastern US. All areas of the US experienced rising residential segregation levels, both North and South and well as urban and rural.

Figure 1. Neighbour-based segregation by county in the Eastern US, 1880 and 1940

Much of this residential segregation was obscured under traditional segregation measures. For example, contrary to results using traditional measures, we find that Southern cities were actually more segregated than their Northern counterparts. This can be seen in Figure 2, in which Southern cities represent the most segregated cities in both 1880 and 1940. While political units in Southern cities appeared integrated in terms of the percentage of black residents, looking at next-door neighbours reveals that there was substantial sorting within those units. That segregation occurs within even small geographic units such as a ward or enumeration district suggests that traditional segregation measures may actually overstate the level of social interactions occurring between members of different races.

Figure 2. Segregation by city in 1880 and 1940

Notes: The dashed lined is a 45-degree line. Cities lying above the line experienced a rise in segregation from 1880 to 1940.

Segregation in rural areas

The increase of residential segregation in areas outside of cities is a new finding with important implications for our understanding of the origins of segregation and policies designed to mitigate its impacts. Previous segregation measures, with their reliance on geographic subunits like wards, cannot be easily applied to rural areas that lack such divisions. Our neighbour-based measure can be calculated for rural areas and, equally important, has the same interpretation as for urban areas.

When estimating segregation in rural counties, we find that it increased in a similarly dramatic fashion. This finding forces a reevaluation of the mechanisms underlying the rise in residential segregation. The Great Migration and white flight to the suburbs cannot explain the rise in rural segregation we identify. Counties that experienced inflows of black residents experienced rising segregation levels, but so did counties that lost black residents, and counties whose racial compositions remained relatively unchanged. White flight to the suburbs may account for the growth of black urban cores, but it cannot account for similar levels of sorting in more sparsely populated counties.

This broad rise in segregation suggests that while white migration out of cities to suburbs may be the proximate cause of rising segregation and racial inequalities in urban centres, it is far from a complete story. There was a more fundamental and sweeping change in the US that led to the rise in residential segregation and impacted residents of every community size. Focusing on urban-specific accounts of segregation neglects the far broader scope of segregation and racial inequalities.

Implications for the persistence of racial inequality

The sweeping rise in both urban and rural segregation across communities in every region implies a significant drop in social interactions between black and white residents. Consider a city like Chicago that moved from a moderate level of segregation in 1880 to a high level of segregation in 1940 (Figure 2). During this period, the black residents rose from 1.2% to 7.7% of the total city population. Unsurprisingly, only 1% of white households had a black neighbour in 1880. However, despite the nearly seven-fold increase in the black population share, the percentage of white households with a black neighbour declined to only 0.4% in 1940, hence the enormous rise in the segregation index. The percentage of black residents with a white neighbour declined from 66% to only 5% over this period.

This dramatic decline in opposite-race neighbours could have profound impacts on the evolution of racial biases and ultimately become self-reinforcing. Dating back to the work of Williams (1947) and Allport (1954), social scientists have theorised that interpersonal contact provides a critical avenue for reducing prejudice between groups. An evolving literature, most recently Carrell et al. (2014), is providing empirical support for this. Changing racial attitudes requires interactions between the races. In this respect, the sweeping rise of residential segregation patterns over the 20th century may very well have created an environment that has allowed racial prejudices to harden and lead to the stubbornly persistent racial inequalities we see today.


Allport, G W (1954), The Nature of Prejudice, Addison-Wesley.

Carrell, S E, M Hoekstra, and J E West (2014), “The Impact of Intergroup Contact on Racial Attitudes and Revealed Preferences”, NBER Working Paper 20940.

Cutler, D M, E L Glaesar, and J L Vigdor (1999), “The Rise and Decline of the American Ghetto”, Journal of Political Economy 107(3).

Logan, T D and J M Parman (2015), “The National Rise in Residential Segregation”, NBER Working Paper 20934.

Williams, R M Jr (1947), “The Reduction of Intergroup Tensions: A Survey of Research on Problems of Ethnic, Racial, and Religious Group Relations”, Social Science Council Bulletin 57.

5,039 Reads