Assessing endogeneity issues in international marketing research

Abstract

Purpose – Endogeneity is a potential threat to the validity of international marketing (IM) research. The purpose of this paper is to draw the attention of IM researchers to issues of endogeneity, to provide a comprehensive overview of the sources of endogeneity, and to discuss the statistical solutions. Design/methodology/approach – The authors conduct the research in two steps. In the first step, the authors review the nature and sources of endogeneity specifically in IM research. In the second step, the authors review 60 IM papers on endogeneity published in the period 1995-2014 and assess the current practice of addressing endogeneity in the IM literature. Findings – Sample selection bias and simultaneity are prevalent sources of endogeneity in IM research. Internationalization-performance relationship and innovation-export nexus are the two most frequently adopted models subject to potential endogeneity. Simply lagging the main independent variable is statistically flawed in dealing with endogeneity despite its popularity in IM research. Research limitations/implications – First, a careful choice and application of methods are critical when addressing endogeneity. Second, the authors suggest the employment of multiple study methods to address endogeneity robustly. Third, to prevent or solve endogeneity in structural equation modeling, researchers may either collect data on independent and dependent variables from different respondents or employ a two-stage least squares approach. Finally, it is helpful to design dedicated models to prevent proactively potential endogeneity a priori. Originality/value – The contribution of this study is twofold. First, it is the first in the literature to discuss the endogeneity issue specifically in IM research. In particular, the study elaborates the origins and consequences of the three most frequently confronted types of endogeneity in IM research. Second, the authors assess the four major methods of addressing endogeneity in IM research with a systematic discussion of the literature from the last two decades. The authors offer suggestions on how to minimize endogeneity in model design and empirical implementation for future IM research.