“Do Venture Capitalists Venture?” [Draft coming soon]
Innovators learn from others, but they may also follow others simply because others do. This paper asks whether the resulting convergence in venture capital reflects useful learning or costly herding. Using text similarity between startup descriptions, I measure whether VCs converge on recently funded business models, and estimate how exposure to peers' convergence in other markets affects their own. When those peers converge elsewhere, focal VCs converge on the similar business models, even though the resulting investments earn lower financial returns. Startups funded through this imitation also show weaker revenue growth, innovation, and follow-on funding. The results hold when I use only peer convergence triggered by earlier shifts in unrelated product markets, which are unlikely to affect the focal startup directly. Heterogeneity in peer influence reveals a mismatch between who attracts imitation and who is profitable to follow: investors respond more strongly to network-central peers even though following them predicts worse returns. Imitation creates no offsetting benefits for VCs' other portfolio startups or subsequent investments, yet VCs who herd most deploy more capital and become more central in future syndication networks.
“Missionaries and the Birth of International Development” [Draft (July 2026)] Submitted
The United States used to be the world's leading provider of foreign aid. Was it merely a geopolitical strategy pursued by foreign policy elites, or did it rest on a grassroots constituency for international development? This paper studies this question through the Student Volunteer Movement, the largest American Protestant missionary movement of its era. To isolate missionary exposure, I use denominations' exposure to the movement's unexpected 1886--87 campaign, recentered on YMCA networks, as an instrument. I find that missionary exposure increased congressional support for major foreign aid bills between 1945 and 1989, the period in which the modern aid regime and its domestic coalition were built. The effect follows the movement's quasi-random timing variation closely: it is driven by YMCA exposure already in place just before 1886 and is absent for YMCA exposure accumulated only afterward. I then trace the social infrastructure through which this exposure persisted and diffused into postwar politics. Missionary-linked denominations organized foreign aid advocacy in their own name, and legislators from exposed areas described aid in cooperative and developmental terms. Exposed places continued to produce more Peace Corps volunteers and overseas-aid nonprofits, extending this orientation beyond Congress. Individual-level evidence shows that missionaries generated knowledge of non-Western societies, visible in publications and WWII expert roles. Missionary exposure predicted greater foreign aid support precisely where it produced more of this expertise, whose value rose during the postwar era of decolonization and international development.
“Beyond Demo Day: Sorting and Value Added in Startup Accelerators” (w/ Deepak Hegde) [NBER WP (April 2026)] [Draft (June 2026)] Submitted
We study who joins startup accelerators, how founders sort across programs, and which accelerators improve startup outcomes. Using a comprehensive sample of about 750,000 U.S. startups linked to 329 accelerators, we adapt the teacher value-added framework from education economics to estimate accelerator value added (AVA) while accounting for sorting. Selection is systematic: observably better ventures are more likely to enter accelerators and to sort into higher-AVA programs. Yet accelerator performance is highly dispersed. Most accelerators have negative value added relative to a no-accelerator benchmark, while a small right tail generates large gains. High-AVA accelerators predict better long-term outcomes, including acquisition, employment, revenue, and valuation, and are also more likely to accelerate the shutdown of weaker ventures. We validate AVA using internal applicant data from a large U.S. non-equity accelerator.
“Estimating the Supply Elasticity of Innovation” [Draft (April 2025)]
Government R&D may pass through to higher scientists' wages rather than an increase in innovation when the labor supply of scientific workforce is inelastic. This paper leverages granular data on scientists’ productivity, wages, and research funding to estimate the impact of government R&D on both inventive output and wages. To identify these effects, I exploit variation in city size driven by universities’ historical ties to defense funding, which influenced the geographic distribution of scientists through the placement of graduates into research-intensive cities. I find that government R&D increases inventive output, particularly among scientists who do not receive funding themselves. The monetary value of increased inventions outweighs the increase in labor costs. While R&D subsidies do lead to higher wages, the results indicate that government investment in research can enhance overall innovation output through the entry of new scientists and subsequent knowledge spillovers, and that the elasticity of innovation with respect to government funding is large at least in the long run.
“Industrial Spillovers from Agricultural Processing: Evidence from the Beet Sugar Industry” [Published version (Aug 2026)] [Draft (April 2026)] [Replication package] Journal of Urban Economics
This paper investigates the agglomeration spillovers from the beet sugar industry, which was supported by the U.S. government as an industrial policy to encourage rural development during the early twentieth century. To estimate the effects of plant openings, I identify suitable locations for beet sugar plants from a historical trade journal and find that these plant openings had large and long-lasting effects on population and manufacturing activities over one hundred years. The local jobs multiplier was significantly larger in less populated areas, suggesting that low congestion in sparsely settled regions enabled a sizeable impact. The agglomeration spillovers benefited industries not only directly linked through input-output linkages but also extended to broader, less related industries outside the production chain of agricultural processing.