An expanded version of the front door criterion under the potential outcome framework: A guide to empiricist

Abstract

In recent years, the front-door criterion (FDC) has been increasingly noticed in economics. However, using regression to apply the FDC under the potential outcome framework (RCM) is not purely the same as the original non-parametric method in the structure causal model (SCM). This article aims to exposit the vital additional factors of using the FDC under the RCM. It first defines a new type of causal effect: path average causal effect (PATE), which is used for further exposition. Then, the key assumptions of the FDC are redefined with the language of the RCM. Four further assumptions are made specifically for the FDC under the RCM framework. After that, the causal connotations of the FDC estimates are elaborated in detail with PATE, and the estimation bias caused by violating some new assumptions is theoretically derived. Rigorous simulation data are used to confirm the theoretical derivation. It is proved that the FDC can still provide useful insights into causal relationships even when some key assumptions are violated. Finally, the FDC is also comprehensively compared with the instrumental variables (IV). The analyses of this paper prove that the FDC can provide new insights into causal relationships compared with the conventional methods in economics.