Endogeneity Bias in the OLS Estimates of Basu’s Model

Abstract: Basu (1997) is the main reference in the market-based accounting literature related to the measurement of conditional conservatism. However, the econometric specification of Basu’s model involves an endogeneity problem due to simultaneity in the variables used in its empirical application: earnings and returns. In this paper we analyse the effect of endogeneity on the Basu’s conditional conservatism measure by using two alternative estimation techniques: the usual Ordinary Least Squares, affected by endogeneity, and System Generalized Method of Moments, specially designed to deal with the endogeneity problem. We apply these estimation procedures to a comparative model of the conditional conservatism in order to analyse the effect of IFRS first adoption on conditional conservatism. Our results evidence that endogeneity problem exists, that it induces bias in the OLS estimations, and that biases can be large enough to modify empirical conclusions.