Revision requested, American Economic Review
Developing country entrepreneurs face family pressure to share income. This pressure, a kinship tax, can distort capital allocations. I combine evidence from a lab experiment-which allows me to estimate an individual-level sufficient statistic for the distortion-with data I collected on a sample of Kenyan entrepreneurs, to quantify the importance of the tax. My data reveal high kinship tax rates for a third of entrepreneurs in my sample. My quantitative analysis makes use of a simple structural model of input allocation fitted to my data, and implies that removing distortions from kinship taxation would increase total factor productivity by 26%, and increase the share of workers in firms with five or more employees from 9% to 56%. These effects are substantially larger than those coming from credit market distortions, which I estimate using a cash transfer RCT. My analysis also implies strong complementarities between kinship taxation and credit constraints.
Linking Mobile Money Networks to “e-ROSCAs”: An Experimental Study (with Patrick Francois)
In press, Science Advances
We present results from the first experimental attempt to implement a Mobile Money version of Rotating Savings and Credit Associations (ROSCAs), which are ubiquitous across the developing world. We test an extreme version of such an “e-ROSCA” by having anonymous group members unable to sanction cheaters. Strikingly, we find high rates of contribution, suggesting the potentially broad scope for this technology to reach the unbanked globally.
Work in progress
Economic consequences of intensive kinship: Evidence from US bans on cousin marriage (with Arkadev Ghosh and Sam Hwang)
Close-kin marriage, by sustaining tightly knit clan-like structures, may impede development. We use 19th and 20th century US state-level bans on cousin marriage to study the causal effect of tight kinship on economic outcomes. We show that these bans reduced rates of in-marriage, and that affected descendants are therefore more urban, have higher female labor force participation, more education and higher incomes.
Family ties and migration: Evidence from historical U.S. census data (with Arkadev Ghosh and Sam Hwang)
While social networks are a key determinant of migration decisions, useful measures of these networks are hard to come by. This is especially so in large, population-scale datasets. We show that surnames are a useful proxy for kin-based social networks using multiple full-count rounds of the US census from the late 19th century. We validate this novel use of surnames to identify kinship ties by showing that (1) Americans are more likely to migrate to states where more people share their surname, and (2) people with more common surnames, having larger such networks, are more likely to migrate across states.
Census Linking: A Bounds Approach (with Hu Fu, Arkadev Ghosh and Sam Hwang)
Linking historical data at scale typically requires substantial human effort and subjective individual judgement on the quality of links. We propose a method to identify bounds on statistics of interest that requires minimal assumptions. This method is complementary to state-of-the-art approaches to linking census records. We implement our method to compute an upper and lower bound on inter-state migration rates between the 1850 and 1860 US Census. We ﬁnd a lower bound that is higher than existing estimates of inter-state migration in the literature. We discuss why our estimate is larger than past estimates, that are based on smaller and more selected samples.
Selection and Impact of Modern Industrial Employment: Field Experimental Evidence from a Chinese Factory in Tanzania (with David Yang and Noam Yuchtman)