Spurious Correlations in Quantile-Based Consumption Spillover Tests

Abstract

Consumption spillovers are difficult to estimate. Many tests in the literature argue that spillovers cause positive correlations between individual consumption levels and aggregate income quantiles. This paper develops simulation-based procedures for evaluating reduced-form tests for consumption spillovers. I find that the correlation found in prior tests may be spurious, arising from the mechanical relationship between a household’s income in a given period and a quantile of the income distribution in that period. This paper also explores the mechanical correlation’s determinants and proposes strategies for estimating unbiased consumption spillover effects.