Earlier (2006): Andrew Young Was Right, Not That Anyone Dare Admit It

From the National Bureau of Economic Research:

WHOSE JOB IS IT ANYWAY? CO-ETHNIC HIRING IN NEW U.S. VENTURES

Sari Pekkala Kerr William R. Kerr

ABSTRACT
We explore co-ethnic hiring among new ventures using U.S. administrative data. Co-ethnic hiring is ubiquitous among immigrant groups, averaging about 22.5% and ranging from <2% to >40%. Co-ethnic hiring grows with the size of the local ethnic workforce, greater linguistic distance to English, lower cultural/genetic similarity to U.S. natives, and in harsher policy environments for immigrants. Co-ethnic hiring is remarkably persistent for ventures and for individuals. Co-ethnic hiring is associated with greater venture survival and growth when thick local ethnic employment surrounds the business. Our results are consistent with a blend of hiring due to information advantages within ethnic groups with some taste-based hiring.

… We use these empirical findings to consider the extent to which three conceptual models of co-ethnic hiring are likely to be responsible for the patterns observed. A first model focuses on enhanced communications in the workplace if co-ethnic individuals are able to better communicate with each other and those with similar native languages. A second model emphasizes how co-ethnic hiring can result from better information flow within ethnic groups. To the extent that ethnic networks and concomitant resource access allow superior hiring within ethnic groups, co-ethnic hiring could enhance venture performance. A final conceptual model features taste-based rationales, where members of an ethnic group favor working together, that do not result in a performance advantage and may even lead to weaker outcomes.

… We find striking levels of co-ethnic hiring. The average new venture with five or more workers has a co-ethnic share of about 22.5%, with an enormous range from as low as 1.8% (top earners from Germany) to as high as 45% (top earners from Vietnam). The variation within language groups is also striking, such as the co-ethnic hiring rates for top earners from Mexico being about four times larger than those evident for top earners from Guatemala, Colombia, and Cuba. Our analyses show how these co-ethnic hiring rates systematically relate to the local workforce composition surrounding new firms, the native languages spoken by ethnic groups, and the cultural similarity of groups to the United States (Spolaore and Wacziarg 2016). We observe remarkable persistence in co-ethnic hiring. Even five and ten years after the establishment birth, the co-ethnic share tends to increase slightly, rather than abate. We also follow top earners when they move to new ventures, finding substantial persistence at the level of individuals in terms of hiring patterns.

Turning to the dynamics of ventures, we find that higher initial co-ethnic shares are associated with an increased survival of the firm to five years. We also observe, conditional on survival, greater employment growth to five years with a higher initial co-ethnic share, although some of this effect appears connected to a reduced likelihood of the firm to shrink in size as it ages. Importantly, most of this survival and growth effect happens when the firm conducting co-ethnic hiring is situated in a city-industry setting where the ethnic group makes up a substantial share of local employment. Extensions consider differences by wage levels of firms, estimations using genetic and linguistic differences across immigrant nationalities, and local policy environments with E-Verify implementation. Returning to the three conceptual models, our empirical results are mostly inconsistent with the model focused on overcoming communication barriers. A common language is a feature in hiring, but distinctly second-order to factors like thick local ethnic labor pools. Instead, the weight of evidence suggests that the co-ethnic hiring more typically follows from factors operating within ethnic groups. We find evidence that supports both the information advantage and group taste frameworks, concluding both play important roles. There is overall more support for the information advantage rationale as, on average, venture performance improves with co-ethnic hiring. We observe, however, negative employment growth with co-ethnic hiring when the surrounding local ethnic labor pool is very small, which is more consistent with the group taste rationale. The striking empirical connections evident in this study suggest that the economic consequences of greater workplace and entrepreneurial diversity deserve closer attention.

I’d like to see a study of how many African-Americans are hired by immigrants.