VoxEU Column Global crisis Macroeconomic policy

Measuring the output responses to fiscal policy

The return from a fiscal stimulus – the fiscal multiplier – remains one of the most controversial topics in economics today. This column considers the influence of expectations, of variation in recessions and expansions, and of different components of government spending. It finds that the size of the multiplier varies considerably over the business cycle: between 0 and 0.5 in expansions and between 1 and 1.5 in recessions.

The effect of fiscal policy on output – and its components – has long been a central part of fiscal policy analysis. And yet, as has been made clear by the recent debate over the likely effects of fiscal stimulus in the US and elsewhere, economists are a long way from reaching a consensus. Indeed economists remain divided over areas such as:

  • the strength of fiscal policy’s macroeconomic effects,
  • the channels through which these effects are transmitted, and
  • the variations in these effects and channels with respect to economic conditions.

The issue at the centre of this debate is the size of fiscal multipliers when the economy is in recession. On this site for example, Barro and Redlick (2009) find a multiplier of 0.6 to 0.8 using data on US defence spending while Almunia et al. (2009) find a multiplier of greater than one when looking at the Great Depression.

Recent theoretical work by Christiano et al. (2009), Woodford (2010), and others emphasise that government spending may have a large multiplier when the nominal interest rate is at the zero bound, which occurs rarely and only in recessions.

These novel theoretical findings for market-clearing models echo earlier Keynesian arguments that government spending is likely to have larger expansionary effects in recessions than in expansions. Intuitively, when the economy has slack, expansionary government spending shocks are less likely to crowd out private consumption or investment.

To the extent that discretionary fiscal policy is heavily used in recessions to stimulate aggregate demand, the key empirical question is how the effects of fiscal shocks vary over the business cycle. The answer to this question is not only interesting to policymakers in designing stabilisation strategies but it can also help the economics profession to reconcile conflicting predictions about the effects of fiscal shocks across different types of macroeconomic models.

The fiscal multiplier over the business cycle

Despite these important theoretical insights and strong demand by the policy process for estimates of fiscal multipliers, there has been little, if any, empirical research trying to assess how the size of fiscal multiplies varies over the business cycle.

In part, this dearth of evidence reflects the fact that much of empirical research in this area is based on linear structural vector autoregressions (SVARs) or linearised dynamic stochastic general equilibrium (DSGE) models, which by construction rule out state-dependent multipliers.

The limitations of these two approaches became evident during the recent policy debate in the US, when government economists relied on neither of these approaches, but rather on more traditional large-scale macroeconometric models, to estimate the size and timing of US fiscal policy interventions then being undertaken (e.g., Romer and Bernstein 2009, Congressional Budget Office 2009). This reliance on a more traditional approach, in turn, led to criticisms based on contrary predictions from SVAR and DSGE approaches.

Our recent work (Auerbach and Gorodnichenko 2010) addresses this issue, providing estimates of state-dependent fiscal multipliers. Building on Blanchard and Perotti (2002) and subsequent studies, our paper extends the existing literature in three ways.

First, using regime-switching SVAR models, we estimate effects of tax and spending policies that can vary over the business cycle.

We find large differences in the size of fiscal multipliers in recessions and expansions, with fiscal policy being considerably more effective in recessions than in expansions.

Figure 1 presents estimates for the historical effects of shocks to government purchases on output. For each period, we consider a policy shock equal to 1% increase in government spending and report a dollar increase in output per dollar increase in government spending over 20 quarters (see the paper for details).

Figure 1. Historical multiplier for total government spending

Notes: shaded regions are recessions defined by the NBER. The solid black line is the cumulative multiplier computed as 

, where time index h is in quarters. Blue dashed lines are 90% confidence interval. The multiplier incorporates the feedback from G shock to a measure of the business cycle conditions. In each instance, the shock is one% increase in government spending.

Plainly, the size of the multiplier varies considerably over the business cycle. For example, in 1985 an increase in government spending would have barely increased output. In contrast, a dollar increase in government spending in 2009 could raise output by about $1.75. Typically, the multiplier is between 0 and 0.5 in expansions and between 1 and 1.5 in recessions.

Note the size of the multiplier tends to change relatively quickly as the economy starts to grow after reaching a trough. Thus, the timing of changes in discretionary government spending is critical for effectiveness of countercyclical fiscal policies.

Second, to measure the effects of a broader range of policies, we estimate multipliers for more disaggregate spending variables, which often behave quite differently in relation to aggregate fiscal policy shocks.

Specifically, we find that defence spending has the largest multiplier, with the maximum response of output being $3.56 for every dollar in defence spending in a recession.

Third, we provide a more precise measure of unanticipated shocks to fiscal policy.

We use information from the Survey of Professional Forecasters, forecasts from the University of Michigan’s RSQE model, forecasts prepared by the staff of the Federal Reserve Board for the meetings of the Federal Open Market Committee, and military spending news shocks constructed in Ramey (2009). We include these forecasts and news shocks in the SVAR to purge fiscal variables of predictable “innovations.” We find that controlling for predictable “innovations” increases the size of estimated multipliers in recession by about 30 percent.


Our findings suggest that the extensions developed in our paper – controlling for expectations, allowing responses to vary in recession and expansion, and allowing for different multipliers for different components of government purchases – all have important effects on the resulting estimates. In particular, policies that increase government purchases have a much larger impact in recession than is implied by the standard linear model, even more so when one controls for expectations, which is clearly called for given the extent to which independent forecasts help predict VAR policy “shocks.”

While we have extended the SVAR approach, our analysis still shares some of the limitations of the previous literature. We have allowed for different economic environments, but there may be still other important differences among historical episodes that we lump together. In particular, some recessions such as the recent one are associated with financial market disruptions and very low nominal government interest rates, while other recessions were induced by monetary contractions (such as the one in the early 1980s).

Our predictions are also tied to historical experience concerning the persistence of policy shocks, and therefore may not apply to policies that are more permanent – or indeed less permanent. Moreover, the effects of taxes, even if purged of expected changes, are still probably too simple as they fail to take account of the complex ways in which structural tax policy changes can influence the economy. And finally, as we enter a period of unprecedented long-run budget stress, the US postwar experience, or even the experience of other countries that have already dealt with budget crises, may not provide very accurate forecasts of future responses.

These limitations of our analysis should motivate future theoretical work to develop realistic DSGE models with potentially nonlinear features to understand more deeply the forces driving differences in the size of fiscal multipliers over the business cycle, the role of (un)anticipated shocks for fiscal multipliers in these environments, and implications of levels of government debt for the potency of discretionary fiscal policy to stabilise the economy.


Almunia, Miguel, Agustín S Bénétrix, Barry Eichengreen, Kevin H O’Rourke, Gisela Rua (2009), “The effectiveness of fiscal and monetary stimulus in depressions”, VoxEU.org, 18 November.

Auerbach, Alan, and Yuriy Gorodnichenko (2010), “Measuring the Output Responses to Fiscal Policy,” NBER Working Paper 16311.

Barro, Robert and Charles Redlick (2009), “Design and effectiveness of fiscal-stimulus programmes”, VoxEU.org, 30 October.

Blanchard, Olivier, and Roberto Perotti (2002), “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output”, Quarterly Journal of Economics, 117(4):1329-1368.

Christiano, Lawrence, Martin Eichenbaum, and Sergio Rebelo (2009), “When is the government spending multiplier large?”, NBER Working Paper 15394.

Congressional Budget Office (2009), “Estimated Macroeconomic Impacts of the American Recovery and Reinvestment Act of 2009”, 2 March.

Ramey, Valerie A (2009), “Identifying Government Spending Shocks: It’s All in the Timing”, NBER Working Paper 15464.

Romer, Christina, and Jared Bernstein (2009), “The Job Impact of the American Recovery and Reinvestment Plan”, Council of Economic Advisors, 9 January.

Woodford, Michael (2010), “Simple Analytics of the Government Expenditure Multiplier”, NBER Working Paper 15714.

3,465 Reads