Fiscal Stimulus

Abstract

Fiscal Stimulus”, the size of “fiscal multipliers” and the impact of discretionary fiscal spending on GDP and unemployment, has once again become central to policy debates in wake of the financial crisis of 2008 and fiscal policy responses in a number of countries. In this paper, we therefore seek to quantify the size, uncertainty and sensitivity of fiscal multipliers in response to a “fiscal stimulus” as in the American Recovery and Reinvestment Act (ARRA) of 2009 in the United States, using an extension of a benchmark New Keynesian model. From a purists' perspective, this may be the wrong question to ask. Policy should care about welfare, rather than derivative measures such as GDP or unemployment.

American Recovery and Reinvestment Act of 2009 (Keynesian Fiscal Stimulus) as a success or failure based on its stated goals

Introduction

Moreover, it should seek to solve a Mirrlees-Ramsey problem, and use the best combinations of available tools and taxes to maximize welfare, subject to constraints imposed by markets and the asymmetry of information. We do not disagree. Indeed, there is a considerable literature on these topics. Indeed, many public debates focus on the effects of fiscal spending on GDP and unemployment. Economists have the tools to answer these questions, and therefore, perhaps they should. Several recent papers have addressed these issues. In essence, we seek to understand how much of the rather negative perspective on long-run multipliers in Uhlig (2010b), due to distortionary taxation in a neoclassical growth model, survives in a model that takes a very Keynesian perspective.

In a nutshell, the answer is: while the benchmark long-run multiplier is now modestly negative rather than substantially negative and while the precise answer is sensitive to some key assumptions and uncertain parameters, much survives indeed. We view the following elements as important. First, “fiscal stimulus” takes time in practice, despite calls for immediate actions as in e.g. Spilimbergo et al. (2008).

Analysis

The American Recovery and Reinvestment Act or ARRA (2009) therefore serves as a useful benchmark and example for the speed at which fiscal policy tools can be deployed, as emphasized by Cogan et al. (2010). Second, government expenditures are financed eventually with distortionary taxes, creating costly disincentive effects, a point emphasized by Uhlig (2010b).

Third, monetary policy and its restrictions due to the zero lower bound (ZLB) on interest rates can matter substantially for the effectiveness of “fiscal stimulus”, as emphasized by Eggertsson (2010) as well as Christiano et al. (2009), in particular if there are sticky prices and wages. Fourth, transfers are a substantial part of the ARRA and similar programs: the degree to which they are given to credit-constrained households may matter considerably, see Coenen et al. (2010).

Finally, model coefficients are uncertain and results are sensitive to specific assumptions. For that reason, we use a reasonably tractable “small-scale” model rather than a larger “black box”, employing Bayesian estimation techniques as well as sensitivity analysis to quantify the uncertainty in our answers. As Leeper et al. (2011) have pointed out, the New ...