ESTIMATED MACROECONOMIC EFFECTS OF THE U.S. STIMULUS BILL

Abstract

I. INTRODUCTION This paper uses a structural multicountry macroeconometric model, denoted the MC model, to analyze the macroeconomic effects of the U.S. stimulus bill passed in February 2009. The policy changes are taken from a report issued on March 2, 2009, by the Congressional Budget Office (CBO 2009). A baseline simulation is first run under the assumption that the stimulus bill passed (which it did), and then a simulation is run with the stimulus taken out. The difference between the predicted values from the two simulations for each variable and each quarter is an estimate of the stimulus effects on that variable. The simulation period is 2009:1-2020:4. Because the model is a multicountry model, the effects on other countries are estimated in addition to the effects on the United States. There is considerable controversy about the stimulus effects, and a number of methodologies have been followed to estimate them. The CBO (2010) uses results from two commercial forecasting models and the FRB-US model of the Federal Reserve Board to choose ranges for a number of government spending and tax multipliers on output. These multipliers are then used to compute stimulus effects on output. Additional equations are used to link changes in other variables, like employment and the unemployment rate, to output changes. The estimates are partial in that they are not the result of solving a complete model. Many potential endogenous effects are ignored. Also, as will be seen, the ranges chosen for the multipliers are large, which leads to large ranges for the estimated stimulus effects. Romer and Bernstein (2009) follow a similar methodology. They use a commercial forecasting model and the FRB-US model to choose government spending and tax multipliers on output. They use these multipliers to compute stimulus effects on output, and they have an additional equation linking employment changes to output changes. They present results for 2010:4. Again, these estimates are not the result of solving a complete model. Another procedure for estimating multipliers is what might be called a procedure. The change in real GDP is regressed on the change in a policy variable of interest and a number of other variables. The equation estimated is not, however, a true reduced form equation because many variables are omitted, and so the coefficient estimate of the policy variable will be biased if the policy variable is correlated with omitted variables. The aim using this approach is to choose a policy variable that seems unlikely to be correlated with the omitted variables. Hall (2009) and Barro and Redlick (2010) are concerned with government spending multipliers and focus on defense spending during wars. (1) Romer and Romer (2009) are concerned with tax multipliers and use narrative records to choose what they consider exogenous tax policy actions, that is, actions that are uncorrelated with the omitted variables. This paper uses a model of the economy that captures many important features of the world economy. It has been extensively tested, and it appears to be a good approximation of the economy. It is briefly outlined in the next section. The stimulus experiment that is performed is based on the solution of the entire model. All the endogenous effects in the model are accounted for, including the effects of the stimulus bill on the rest of the world and the effects of the rest of the world responses back on the United States. The methodology of structural macroeconometric modeling, which goes back at least to Tinbergen (1939), does not have the problem of possible omitted variable bias in reduced form equations, because reduced form equations are not directly estimated. What is required is that the structural equations be consistently estimated. Take, for example, a consumption or investment equation. If there are right hand side endogenous variables, like current income or a current interest rate, and thus correlation between these variables and the error term in the equation, this has to be accounted for. …