Estimation of Hybrid Phillips Curve: A Source of Conflicting Empirical Results

Recent empirical studies of the hybrid Phillips curve draw conflicting conclusions about the relative role of forward‐looking behavior and the role of marginal cost as the driving force. It is difficult to identify the source of the conflicting results in past studies because of differences in the measure of inflation, the estimation equation, and the instrumental variables. This article presents evidence that a source of the conflict is the use of suboptimal sets of instruments. When the optimum set of instruments is selected by using the L₂‐boosting method from the principal components of a large number of valid instruments, all conflicting conclusions in previous studies vanish. Regardless of the choice of inflation measure and estimation equation, we identify the dominant role of forward‐looking behavior and a significant effect of the real marginal cost on inflation.