Abstract
Why do VC funds delay entry into high-uncertainty sectors? I structurally estimate a dynamic model with two frictions (an LP-GP scale wedge and a GP-entrepreneur information wedge) embedded in diffusion that separates learning from adoption. Fund-1 exploration and monitoring map into Fund-2 portfolio composition and scale. Using simulated method of moments on matched fund pairs, I find: (i) the information wedge depresses composition by killing early signals; (ii) the LP-GP wedge rations investment in hot markets, with dominance reversing in cold markets. Counterfactuals imply combined frictions reduce surplus by about $37 billion (approximately 2.8% of VC and 12% of new-sector capital). Policies targeting composition (exploration-linked carry) and state-contingent scale tilts expand investment; implemented jointly, they deliver super-additive gains.