Endogeneity and Causal Inference in Marketing

Abstract

The posted pdf is a pre-publication draft. Please see https://doi.org/10.1142/13305 for the published version.

In this chapter, we trace the history of how endogeneity came to be recognized as an important issue in marketing analysis and the widespread adoption of methods to address endogeneity. To quantify the rise of econometric methods for addressing endogeneity in marketing, we begin with a systematic literature review which counts the number of papers using one of seven causal inference methods: instrumental variables, latent instrumental variable, Gaussian copula, propensity score matching, difference in differences, regression discontinuity and synthetic controls. Based on this summary of the literature, we then trace the evolution of methods over three decades, discussing which papers were (in our view) important landmarks that had an influence on how the community viewed and handled endogeneity in the analysis of observational marketing data. After that, we conclude with a brief summary of the causal inference methods in common use today, which we hope marketing researchers can use as a "menu" to tackle their own endogeneity problems. We then briefly discuss the future of endogeneity in the field and present our conclusions.

Keywords: Endogeneity, Causal Inference, Instrumental Variables, Gaussian copula, Latent Instrumental Variable, propensity score matching, difference in differences, regression discontinuity, synthetic control

JEL Classification: B23, C36