Latent Heterogeneity in the Marginal Propensity to Consume

Abstract

We estimate the distribution of marginal propensities to consume (MPCs) using a novel clustering approach that generalizes the fuzzy C-means algorithm to regression settings. We apply the estimator to the 2008 stimulus payments, exploiting the ran- domized timing of disbursements, and find considerable heterogeneity in MPCs that varies by consumption good. We document observable determinants of this heterogeneity, without imposing ex ante assumptions on such relationships; MPCs correlate positively with income and the average propensity to consume, but much heterogene- ity remains unexplained. The partial equilibrium response to the stimulus is twice that based on homogeneous estimates, highlighting the quantitative importance of heterogeneity. joint with Daniel Lewis and Davide Melcangi