Causal Inference for Asset Pricing

Abstract: This paper provides a guide for using causal inference with asset prices and quantities.Our framework revolves around an elementary assumption about portfolio demand: homogeneoussubstitution conditional on observables. Under this assumption, standardcross-sectional instrumental variables or difference-in-difference regressions identify therelative demand elasticity between assets with the same observables, the difference betweenown-price and cross-price elasticity. In contrast, identifying aggregate elasticitiesand substitution along specific characteristics requires joint estimation using multiplesources of exogenous time-series variation. The same principles apply to the estimationof multipliers measuring the price impact of supply or demand shocks. Our assumptionmaps to familiar restrictions on covariance matrices in classical asset pricing models,encompass demand models such as logit, and accommodate rich substitution patternseven outside of these models. We discuss how to design experiments satisfying thiscondition and offer diagnostics to validate it.