TRACING THE ROLE OF FORESIGHT IN THE EFFECTS OF U.S. TAX POLICY: EVIDENCE FROM A TIME-VARYING STRUCTURAL VECTOR AUTOREGRESSION

Abstract

Recent empirical research stresses the importance of foresight in tax policy analyses, because failing to model foresight adequately leads to two types of biases: a bias that relates to a mismatch of information sets in the presence of foresight, and one that arises when time variation in foresight is ignored. This paper incorporates tax foresight into a framework of time-varying structural vector autoregression models with stochastic volatility to account for both types of biases. This reveals the effects of changing degrees of tax foresight on U.S. tax policy. Two findings stand out: First, anticipated and unanticipated tax shocks show considerable time variations, suggesting that some tax reforms were more anticipated than others. Second, the time-varying bias reveals that tax foresight was higher during the 1980s and 1990s than during the 1960s and 1970s. The results compare well with the literature and find support in documented U.S. tax episodes.