Abstract:
Recent empirical research emphasizes the importance of foresight for tax policy analyses. According to Leeper et al. (2013a), failing to model foresight adequately can lead to biased inference in empirical models. The authors reveal this bias by augmenting the SVAR model of Blanchard and Perotti (2002) with a measure for tax foresight. I extend this finding by transforming their model into a time-varying SVAR with stochastic volatility. This approach allows to study the time-variations of tax foresight, and resulting effects on tax policy shocks over time. Two findings stand out: First, both anticipated and unanticipated tax shocks show considerable movements over time. The magnitude of these shocks suggests that some tax reforms are more anticipated than others. Second, I find that the bias in the tax shocks is more pronounced during the 1980s and 1990s, suggesting that tax reforms in these decades were to a higher degree anticipated than in the 1960s and 1970s. The results compare well to other studies and find support in anecdotal evidence on documented U.S. tax reforms.
