Journal of Monetary Economics 49(4), 709-746

May, 2002

Abstract. In this paper, I consider whether postwar fiscal policy has helped stabilize the U.S. economy. I do this by adding to the stochastic growth model fiscal policy feedback rules estimated from postwar data. These rules allow fiscal policies to respond to current and lagged output and labor hours. I use the estimated policy rules to see if postwar fiscal policy reduces output volatility and/or lengthens expansions and shortens recessions. I find that fiscal policy in general provides little stability on either count. I also find that the endogenous feedback rules, by themselves, are at best moderately stabilizing and are in some cases destabilizing.

Data: The text file loaded into the GAUSS program is here. The spreadsheets underlying these data, including the construction of the quarterly tax rates, are here.

Supplemental Material: Background calculations and solution method (also see below).

Computer Codes: Underlying Codes (in GAUSS). Among other things, this includes an older version of the simplex maximization algorithm written by Honore and Kyriazidou and downloaded from the electronic GAUSS library at American University, and a version of the HP filter written by Matheny, Norden and Vigfusson and downloaded from Christian Zimmerman's QMRBC web site. The programs rely on GAUSS translations of LAPACK routines for the real generalized Schur decomposition, which were graciously provided by Paul Söderlind.

Solution Method: Everybody and their brother has their own special method for solving linear expectational difference equations. Here's mine. First, the formal description is here. My approach is based on the work of (especially) Paul Klein and Chris Sims. The procedures for the routine are contained in the GAUSS files rgsar1.src, rgsar1.lcg, and the underlying LAPACK routines, all of which are described in the readme.txt file that accompanies the computer code. None of this would work without GAUSS translations of LAPACK routines for the real generalized Schur decomposition, which were graciously provided by Paul Söderlind.