Title:Testability of Reverse Causality without Exogeneous Variation

Abstract:This paper shows that testability of reverse causality is possible even in the absence of exogenous variation, such as in the form of instrumental variables. Instead of relying on exogenous variation, we achieve testability by imposing relatively weak model restrictions. Our main assumption is that the true functional relationship is nonlinear and error terms are additively separable. In contrast to existing literature, we allow the error to be heteroskedastic, which is the case in most economic applications. Our procedure builds on reproducing kernel Hilbert space (RKHS) embeddings of probability distributions to test conditional independence. We show that the procedure provides a powerful tool to detect the causal direction in both Monte Carlo simulations and an application to German survey data. We can infer the causal direction between income and work experience (proxied by age) without relying on exogeneous variation.