An Experimental Study of Diversity Management Techniques

A novel field experiment is used to evaluate two competing strategies for managing ethnic diversity: (1) assimilationist strategies, which encourage the construction of shared, superordinate social identities (e.g., based on a team or nation), and (2) multiculturalist strategies, which aim to foster beliefs that ethnic differences are valuable. Results from Kenya indicate that assimilationist approaches are more effective at minimizing diversity’s social costs, such as discrimination and reduced cohesion. But multiculturalist approaches are better at maximizing diversity’s economic benefits, such as increased creativity and problem-solving capabilities. I also provide evidence that the productivity gains from ethnic diversity are larger than the gains from age, birthplace, gender, and religious diversity, and driven by skill complementarities between ethnic groups. These findings provide an empirical foundation for adjudicating between competing theories about diversity’s costs and benefits, and strategies for managing them. ∗Co-Founder at Groundtruth.io, 2582 Great Highway, San Francisco, California 94116. Email: [email protected]. This research was made possible by support from an SBE Doctoral Dissertation Improvement Grant from the National Science Foundation (SES-1160303), a DDRO grant from the Stanford Vice-Provost of Graduate Education, a GRO grant from the Stanford School of Humanities and Sciences, and a fellowship from the Stanford Institute for Innovation in Developing Economies. I am particularly grateful to David Laitin for very helpful feedback and guidance at all stages of this project. I also thank Jonathan Wand, Saumitra Jha, Joshua Cohen, Simon Jackman, Eran Bendavid, and participants at the Stanford Comparative Politics Workshop for useful suggestions. Romeo Castro, Peter Kariuki, Kevin Kimani, Chacha Maroa, Joseph Njoro, Stacey Ondimu, Franc Tiago, and the staff at the Busara Center for Behavioral Economics provided excellent research assistance. Supplementary materials for this paper are available at http://sangickjeon.io.