Venture Capital (Mis)Allocation in the Age of AI

We use machine learning to study how venture capitalists (VCs) make investment decisions. Using a large administrative dataset on French entrepreneurs that covers VC-backed as well as non-VC-backed firms, we use algorithmic predictions of new ventures' success to uncover potential sources of inefficiencies in venture capital allocation. We estimate the shadow cost of various frictions in VCs' decisions and show that inefficiencies arise as VCs rely too heavily on a few signals about entrepreneurs and not sufficiently on some firm characteristics. Algorithmic decision aids show promise to increase the scope of VC investments and founder diversity