VoxEU Column Frontiers of economic research

A novel explanation for the black-white wage gap: Productivity spillovers require social interactions between workers

The black-white wage gap persists. Year after year, data tell us that black workers in the US earn less than their white counterparts. This column presents new evidence focused on the notion of race-specific social networks and ‘knowledge spillovers’. Data suggest that a black worker may be less able than an otherwise similar white worker to enjoy knowledge spillovers that arise in predominantly white work environments, suppressing their earnings.

A longstanding literature documents the existence and persistence of a black-white gap in labour market earnings, which cannot be entirely explained by traditional measures of endowments or human capital (Lang and Manove 2011; Black et al 2006). These wage differences have been attributed to labour market discrimination (Bertrand and Mullainathan 2004, Holzer and Ihlanfeldt 1998), segregation into lower quality neighbourhoods (Cutler and Glaeser 1997, Ananat 2011), concentration in central cities with poor access to jobs (Ihlanfeldt and Sjoquist 1990, Ross 1998, Weinberg 2000) and low quality schools (Card and Rothstein 2007, Hanushek and Rivken 2009), and early childhood experiences (Fryer and Levitt 2006). 

Many studies document strong patterns of homophily – the tendency of individuals to associate and bond with those similar to themselves – in friendships (Weinberg 2007, Fletcher et al 2013). Recent work has suggested that race- or ethnic-specific social networks might play a substantial role in explaining differences in labour market outcomes (Hellerstein et al 2011, Dustman et al 2009).1

Knowledge spillovers and race

A previously unexplored source of racial differences in the labour market is the productivity returns that arise from agglomeration economies – the benefits that come when firms and people locate near one another together in cities and industrial clusters -- and human capital externalities in cities. One of the three major sources of agglomeration economies is the increased productivity arising from knowledge spillovers. For example, Ellison et al (2010) find that productivity gains from co-agglomeration are substantially larger when manufacturing firms are more connected as measured by technology flows and patent citations. Glaeser and Maré (2001) argue that the urban wage premium arises in part from worker learning, since workers tend to retain their wage premium even after moving away from a large metropolitan area.2 If productivity gains arising from the physical concentration of economic activity are due to exchanges of knowledge and ideas between work location peers, it is quite conceivable that these gains may accrue along racial lines, because social relationships and interactions operate primarily along these lines.

Explaining why you earn less if you’re black

In Ananat et al (2013) we examine this possibility, and find that African-Americans appear to receive substantially smaller wage gains than whites from being exposed to high concentrations of economic activity, both across and within metropolitan areas. In a simple regression across metropolitan areas, we find that a one standard deviation increase in employment density is associated with a 1.4 percentage point increase in the black-white wage gap, even after controlling for residential racial segregation. Within metropolitan areas, we find very robust racial differences in the returns (in wages) to working in locations where employment density is greater and where the concentration of college graduates is higher.  These results are robust to allowing the return to agglomeration and human capital externalities to vary based on other demographics, industry, occupation and metropolitan area, as well as to including controls for unobserved ability based on the information revealed by workers’ residential location decisions (Fu and Ross 2013). After these controls, one standard deviation increases in exposure to employment density and to share workers with a college degree among the local workforce are associated with 1.7 and 1.3 percentage point increases, respectively, in the black-white wage gap. Together these changes account for 43% of the unexplained black white wage gap in the data.3

More significantly, we provide several pieces of evidence that point to social interactions with nearby workers of the same race as a key explanation for the differential return to the work location environment. 

It’s harder if knowledge doesn’t spill over

First, we find that a worker obtains a higher return in wages to employment density and share of college-educated workers when the share of workers of the same race is higher in that worker’s employment location.  Next, we estimate models of firm productivity in manufacturing and show that a firm’s productivity gain from exposure to higher employment density or a larger share of college educated workers is substantially larger when the workers at that firm tend to be the same race as the workers at surrounding firms.4 Finally, using survey responses for the General Social Survey – a large sociological survey in the US – we examine whether black workers at predominantly white firms achieve closer relationships with whites than blacks who work at firms with few white workers.

Black workers and white workers in general report significant social distance from each other,5 and we find no evidence that blacks at white firms are closer to whites than are other black workers. This set of findings reinforces the notion that a black worker may be less able than an otherwise similar white worker to enjoy knowledge spillovers that arise in predominantly white work environments.


Ananat, E (2011), “The Wrong Side(s) of the Tracks: The Causal Effects of Racial Segregation on Urban Poverty and Inequality”, American Economic Journal: Applied Economics 3(2): 34-66.

Ananat, E, S L Ross and S Fu (2013), “Race-Specific Agglomeration Economies: Social Distance and the Black-White Wage Gap”, NBER Working Paper 18933.

Behrens, K and F Robert-Nicoud (2015), “Agglomeration theory with heterogeneous agents”, in G Duranton, V Henderson and W Strange (eds.) The Handbook of Urban and Regional Economics Vol 5, Elsevier BV.

Bertrand, M and S Mullainathan (2004), “Are Emily And Greg More Employable Than Lakisha And Jamal? A Field Experiment On Labor Market Discrimination”, The American Economic Review 94(4): 991-1013.

Black, D, A Haviland, S Sanders and L Taylor. (2006), “Why Do Minority Men Earn Less? A Study of Wage Differentials among the Highly Educated”, Review of Economics and Statistics 88(2): 300-13.

Card, D and J Rothstein (2007), “Racial Segregation and the Black-White Test Score Gap”, Journal of Public Economics 91(11-12): 2158-2184.

Carlino, J and B Kerr (2015), “Agglomeration and Innovation”, in G Duranton, V Henderson, W Strange (eds.) The Handbook of Urban and Regional Economics Vol 5, Elsevier BV.

Combes, P P and L Gobillon (2015), “The Empirics of Agglomeration Economics”, in ibid.

Cutler, D and E Glaeser (1997), “Are Ghettos Good or Bad?”, Quarterly Journal of Economics 112: 827-72.

Dustman, C, A Glitz, and U Schonberg (2009), “Job Search Networks and Ethnic Segregation in the Workplace”, Working Paper.

Ellison, G, E Glaeser, and W Kerr (2010), “What Causes Industry Agglomeration? Evidence from Coagglomeration Patterns”, The American Economic Review 100: 1195-213.

Fletcher, J, S L Ross and Y Zhang (2013), “The Determinants and Consequences of Friendship Composition", NBER Working Paper 19215.

Fryer, R G and S D Levitt (2006), “The Black-White Test Score Gap Through Third Grade”, American Law and Economics Review 8(2): 249-281.

Fu, S and S L Ross (2013), “Wage premia in employment clusters: How important is worker heterogeneity", Journal of Labour Economics 21: 271-304.

Glaeser, E L and D C Mare (2001), “Cities and Skills”, Journal of Labour Economics 19: 316-42.

Hanushek, E A and Steven G Rivkin (2009), “Harming the best: How schools affect the black-white achievement gap”, Journal of Policy Analysis and Management 28(3): 366-393.

Hellerstein, J K, M McInerney, and D Neumark (2011), “Neighbors And Co-Workers: The Importance Of Residential Labour Market Networks”, Journal of Labour Economics 29(4): 659-95.

Holzer, H J and K R Ihlanfeldt (1998), “Customer Discrimination and Employment Outcomes for Minority Workers”, Quarterly Journal of Economics 113(3): 835-867. 

Ihlanfeldt, K R and D L Sjoquist (1990), “Job Accessibility and Racial Differences in Youth Employment Rates”, The American Economic Review 80(1): 267-276.

Lang, K and M Manove (2011), “Education and Labour Market Discrimination”, The American Economic Review 101(4): 1467-96.

Ross, S L (2011), “Social interactions within cities: Neighborhood environments and peer relationships”, in N Brooks, K Donaghy, G Knapp (eds.) Handbook of Urban Economics and Planning, Oxford University Press.

Ross, S L (1998), “Racial Differences in Residential and Job Mobility”, Journal of Urban Economics 43: 112-36.

Weinberg, B (2007), “Social Interactions with Endogenous Associations”, NBER Working Paper 13038.

Weinberg, B (2000), “Black Residential Centralization and the Spatial Mismatch Hypothesis”, Journal of Urban Economics 48: 110-34.


1 See Ross (2011) for a broader discussion of the effects of social networks on labour market outcomes.

2 See Behrens and Robert-Nicoud (2015), Carlino and Kerr (2015), and Combes and Gobillon (2015) for recent surveys of the literature on agglomeration economies.

3 Following Fu and Ross (2013), Ananat, Fu and Ross (2013) include residential location fixed effects in order to control for aspects unobserved worker ability that affect location choice.  They find that these controls reduce the unexplained black-white wage gap by 53% to about 7 percentage points, which is comparable to the 48% reduction in the black-white wage gap found by Lang and Manove (2011) from the inclusion of the ‘AFQT’ score.  They also find no correlation between race and either employment density or share college exposure after conditioning on residential location.

4 Specifically, they estimate a Constant Elasticity of Substitution model of firm Total Factor Productivity controlling for equipment capital, structural capital, low skill labour and high skill labour in the firm. 

5 Relative to their reported closeness to members of their own race.


2,415 Reads