The Effects of Government Spending Shocks with Fiscal Foresight

Abstract

This thesis investigates the effects of government spending with consider- ation of fiscal foresight. In chapter 2, we suggest that a part of the contro- versy in the literature may be explained by the difference in the nature of the shocks identified by Ramey (2011b). We show evidence of inconsistency in the results in Ramey (2011b)

we also explain the mixed results in follow-ups in Ramey (2011b). Moreover, we study the effects of the news shock by construct- ing changes in the expected present value of changes in government spending based on the SPF data, as in the first method. With this modified method, we investigate the effects of anticipated government spending shocks and discuss whether the empirical results are consistent with the neoclassical view. In chapter 3, we study the effects of unanticipated government spending shocks. Our focus is on verifying whether a specific VAR model has fundamental representation by a theoretical model and empirical test. First, we dis- cuss three widely used specifications of VAR models—conventional structural VAR (SVAR), expectation-augmented VAR with forecast error (EVAR FE), and expectation-augmented VAR with future news (EVAR FN)—with theoret- ical examples. We conduct the formal presentation of fundamentalness based on two types of news processes: observable news process and noisy news process. We find that i) assuming that agents can observe the news clearly, SVAR may not be fundamental. Also, EVAR FE may still be non-fundamental, but EVAR FN be fundamental. ii) With noisy news assumption, a simple remedy of adding relative information may not be working in terms of theoretical models. EVAR FN become also nonfundamental where the news is noisy. In the second part of chapter 3, we empirically examine the fundamentalness of three VAR specifications. We conduct the orthogonality test of Forni and Gambetti (2014c) to check whether the shocks from each VAR are fundamental. We find that unanticipated government spending shocks estimated from SVAR and EVAR FE are likely to be non-fundamental, but EVAR FN is not.