Research
Working Papers
-
Seeds of Market-Based Capitalism: Missionaries, Banks, and China's Modernization Job Market Paper
What determines the adoption of modern financial institutions?
Abstract
What determines the adoption of modern financial institutions? We study this question in the context of China, which transitioned from traditional to modern banking following its opening to foreign powers in the nineteenth century. We explore the role of Christian missionaries in shaping the adoption of modern banks. Missionaries were among the first foreigners to reach China’s interior and played an important role in educating future bankers and translating the Western knowledge needed to operate these new institutions. Before the Communist takeover in 1949, we use a staggered difference-in-differences design to show that missionary arrival first increased the number of modern banks by 31% and later the number of enterprises by 27%. We then estimate the long-run effects of missionary exposure after 1949 using variation induced by the timing of railway construction relative to national missionary inflows. The effect of missionary activity on banking disappeared under Maoist central planning but re-emerged as markets reopened after 1978. Locations with greater historical missionary exposure have higher GDP, more enterprises, and greater foreign investment, indicating broader effects of missionaries on market-based capitalism. The dissipation of these effects under Mao suggests that the spread of new economic institutions relies on a complementarity between the local human capital and knowledge needed for their adoption and a policy environment conducive to market development.
Selected Work in Progress
Same Faith, Different Politics: Religious Broadcasting and Voting in Italy and Spain
Can religious identity shape elections?
Abstract
How does exposure to the same religious message affect voting across different political contexts? We study Radio Maria, one of the world’s largest Catholic broadcasting networks, in Italy and Spain. We combine municipality-level election results with confidential transmitter data and exploit variation in radio reception induced by local topography, together with time variation from the expansion of the radio network. We find that radio exposure increases support for religiously aligned parties in both countries, but in different political directions. In Italy, exposure shifts votes toward parties in the center, while in Spain it shifts votes toward the right. These differences reflect the historical political representation of Catholic identity, which has been associated with the center in Italy and the right in Spain. The results show that a common religious message can produce different electoral responses depending on the political context in which voters receive it. More broadly, our polarization measures suggest that religious media can have very different implications for political polarization, potentially moderating it in one context while intensifying it in another.
Restricting Online Pornography and Violence Against Women: Evidence from India and the United States
Does the adult film industry affect violence against women?
Abstract
Governments increasingly restrict online pornography in an effort to reduce violence against women. The effect of such restrictions is theoretically ambiguous. Pornography may increase aggression through exposure to degrading or violent content, but it may also displace behaviors that create opportunities for violence. We study two large restrictions in two countries and find opposite effects. In the United States, state age-verification laws raised the cost of access, and induced the largest platforms to withdraw from the treated states. Exploiting the staggered adoption of the age-verification laws, as well as a comparison of adjacent counties across states, we find a decline of 1.3-1.4 (3%) monthly non-sexual assaults against female victims per 100,000 residents. In India, a 2018 nationwide order blocked 827 pornographic websites. We compare places with higher internet access before and after the policy, and instrument with a measure of how costly it was for a geographic location to access internet based on the number of lightnings in the area. In this context, we find that the policy ban increased annual rape incidents by 1.05 per 100,000 residents, while crimes not specifically directed at women do not rise. Both policies materially reduced access. Evidence on circumvention and substitution, offender age, and local age composition and household structure suggests that the balance between exposure and behavioral substitution differs across settings. Pornography restrictions therefore can have a differential effect on violence against women, and evaluating them requires attention to how users adjust to the regulations and to the local context.
Increasing Female Employment in the Automotive Sector in India
How can women’s participation in male-dominated industries be increased?
Abstract
We study whether workplace amenities can help firms attract women into traditionally male-dominated industries. In partnership with a major automotive industry association in India, we conduct a randomized recruitment experiment comparing job offers that vary in the amenities associated with a factory job. These amenities include greater female coworker representation, the option to bring a friend or relative, female supervision, and female human-resources personnel and grievance support. Across 1,190 offers, the peer treatment increases take-up by an estimated 7.2 percentage points, while the other treatment arms have smaller point estimates. Estimated treatment effects are larger in periods with shorter average job-search duration, consistent with amenities having a greater influence on recruitment when outside employment options are better. The findings suggest that peer recruitment can help firms attract women and provide preliminary evidence that workers’ outside options shape their responses to workplace amenities.
Religious Broadcasting and Belief Formation in Tanzania
Are religious media effective at influencing beliefs and civic engagement?
Abstract
Can mass religious broadcasting change religious identity and social behavior? We study the expansion of Radio Maria across Tanzania between 2001 and 2024. We combine the staggered rollout of the network’s transmitters with predicted signal strength, individual-level data from Afrobarometer and the Demographic and Health Surveys, and geocoded conflict data from ACLED. Our empirical strategy compares locations with similar proximity to transmitters but different signal reception because of local topography, while accounting for free-space signal, other radio coverage, and location and year fixed effects. Exposure to Radio Maria increases Catholic affiliation and reduces Protestant affiliation, with larger effects in places farther from historical Catholic missions. It also reduces in-person contact with religious leaders, consistent with radio partly substituting for traditional religious institutions. We find little change in most social and intergroup attitudes, but some reduction in conventional political engagement and an increase in protests and religiously connected events. These additional events are not accompanied by more fatalities or violence against civilians. The results show how religious media can reshape affiliation while also affecting civic mobilization.