Evidence on Firm-to-Firm R&D Collaboration Intensity [Latest Version] [SSRN]

Presented at: IIOC 2026 - Rising Stars, Econometric Society NASM, EARIE 2026 - Rising Stars, Federal Reserve Bank of Philadelphia

Abstract: Firms collaborate with other firms on R&D but allocate different amounts of resources to each collaboration. Because their R&D resources are limited, firms face a trade-off between many weak collaborations and fewer, stronger ones. Yet existing research on collaboration subsidies considers only whether firms collaborate, not how intensively they collaborate. I develop a model in which firms choose R&D collaborators, the intensity of each collaboration, and their R&D investment. I estimate the model using U.S. patent and accounting data. Counterfactual analysis shows that over a broad range of subsidy budgets, targeting subsidies to strong collaborations yields higher producer surplus than subsidizing all collaborations uniformly. The uniform subsidy becomes preferable only when the subsidy budget exceeds $26 billion---more than three times observed private spending on collaborations.

The Hidden Inflation in Homeowners' Insurance (with Mallick Hossain, Natee Amornsiripanitch, Siddhartha Biswas) 

Presented at: ASSA 2026 (by coauthor), Urban Economics Association European Meeting

Abstract: Rising homeowners’ insurance prices increase housing and debt service costs, while household coverage adjustments may increase exposure to expense shocks. Using novel data linking mortgages to insurance policies, we develop a quality-adjusted price index showing that the menu of insurance costs increased 18 percent more than premium expenditure growth from 2013-2022. During this period, households responded by increasing deductibles and reducing coverage relative to structure values, leaving them more exposed to financial shocks. Our estimates reveal low household demand elasticities for insurance features, though these elasticities increase significantly during home purchases, refinancing, or when households receive updated property valuations. These findings suggest inattention and information frictions substantially influence insurance decisions. Policy implications favor behavioral nudges and regular property assessments over price-based interventions to reduce household financial vulnerability in insurance markets. Credit utilization results suggest that households self-insure when faced with higher insurance costs. 

Disaster Risk and Insurer Competition (with Mallick Hossain, Conor Ryan)

Identifying Causal Effects of Information Interventions under Network Interference (with Ju Hyun Oh)

Abstract: Providing information to targeted individuals may affect not only their own behavior but also that of their peers as the information spreads through networks. This paper develops a partial-identification framework for the direct and peer effects of such information interventions. We focus on key causal parameters, including the average treatment effect and the persuasion rate---the probability that exposure to new information changes behavior. Using plausible monotonicity assumptions, we derive bounds on these parameters without relying on restrictive functional forms, parallel trends, or fully exogenous instruments. We apply the framework to study how pharmaceutical detailing visits affect the prescribing behavior of targeted physicians and their peers. We find evidence of positive direct and peer effects for some levels of exposure. At higher levels of peer exposure, the lower bounds flatten, suggesting that peer effects may saturate as exposure increases. These findings highlight the importance of accounting for network spillovers and allowing the effects of peer exposure to be nonlinear.