Multiple-Predictor Regressions: Hypothesis Testing

Abstract

We propose a new hypothesis-testing method for multipredictor regressions in small samples, where the dependent variable is regressed on lagged variables that are autoregressive. The new test is based on the augmented regression method (Amihud and Hurvich, 2004 ), which produces reduced-bias coefficients and is easy to implement. The method's usefulness is demonstrated by simulations and by testing a model where stock returns are predicted by two variables, income-to-consumption and dividend yield.