VoxEU Column Macroeconomic policy

Government spending multipliers and the business cycle

There is no consensus on the effectiveness of government spending as a measure for boosting output. This column suggests that increasing government spending is highly effective exactly when it is most needed – when the economy is experiencing a deep recession. But the finding does not imply a one-size-fits-all recommendation. There are potential dangers in increasing spending in countries whose level of debt might be perceived as unsustainable.

How effective is government spending in boosting output when the economy is in bad shape? The question has been at the centre of intense policymaking discussions in the aftermath of the Great Recession and has spurred a remarkable amount of work from researchers.

As recently stressed by Ramey and Zubairy (2015), despite the importance of this question, there is no consensus on whether increasing government spending is an effective policy measure to lift the economy up during recessions. Estimates of the so-called fiscal multipliers – i.e., how much output is generated by an extra dollar of government spending – range from values that are close to zero to values that are larger than four!

Fiscal multipliers quantification: Lack of consensus

There are two main factors behind the lack of consensus on the quantification of fiscal multipliers. First, as stressed by Parker (2011), most studies look at the effects of increased government spending using linear models. By their nature, linear models mix up the effects of an extra dollar of spending during recessions – something that might help the economy to re-start and get out of troublesome times – with the possibly milder bang that an economy gets out of an extra buck during expansions.

However, theoretical models going from the old IS-LM-AD-AS framework to recent micro-founded ones – accounting for the interaction between low policy rates due to recessionary shocks and rational expectations – suggest that the effects of fiscal spending might be larger if the economy is slack because of a lower crowding out effect on private spending (for an example of the latter, see Christiano et al. 2011).  

Recent studies focusing on US data and dealing with this issue have, again, pointed to different results. Auerbach and Gorodnichenko (2012) work with a nonlinear version of the linear fiscal vector autoregressive (VAR) framework popularised by Blanchard and Perotti (2002). They find evidence of countercyclical fiscal spending multipliers.

However, in computing the multipliers, a modelling assumption they entertain is that of ‘absorbing states’, i.e., an expansionary fiscal spending shock is assumed to be unable to pull the economy out of a recession. This assumption may somewhat lead to an overestimation of fiscal spending multipliers whose returns, according to some theories, should become more moderate as the economy exits a recession. Ramey and Zubairy (2014) also work with nonlinear techniques but relax this assumption. They find that, on average, fiscal spending multipliers in recessions are not larger than in good times.

Importantly, Ramey and Zubairy (2014) study fiscal multipliers by focusing on anticipated fiscal shocks. Virtually all changes in government spending are partly anticipated by economic agents. This is due to the intrinsic decision and implementation lags of fiscal policy. Any time a democratic government decides to intervene by increasing spending, it must get the approval of the Parliament and must then actually implement the stimulus package. As a consequence, the change in the official data will show up months after the public has come to know that an increase in fiscal spending will take place.

However, the expectation of future increases in public spending might have an immediate effect on households’ consumption plans and firms’ hiring and investment decisions. Not accounting for such anticipation effects, i.e., fiscal foresight, might then lead to important econometric problems and severe quantification errors, as also shown by Leeper et al. (2013).

The role of nonlinearities: New evidence

In a recent contribution (Caggiano et al. 2015), we further elaborate on the role of nonlinearities for the quantification of fiscal multipliers when anticipated fiscal shocks are at work. We deal with fiscal foresight by employing a measure of changes in expectations by professional forecasters about future government spending recently developed by Forni and Gambetti (2014). To account for the role played by the stance of the business cycle in the transmission of fiscal shocks to the real economy, we use a smooth transition vector auto-regression (ST-VAR) model.

Key for our analysis is the computation of a ‘generalised’ version of impulse responses to anticipated fiscal shocks. This allows us to deal with ‘non-absorbing states’, so that an expansionary anticipated fiscal policy shock may indeed lead the economy out of a recession. Moreover, it allows us to isolate fiscal multipliers conditional on ‘extreme events’, i.e., particularly deep recessions.

Recessions are indeed heterogeneous. For instance, the recession of the early 2000s was relatively short and mild compared with the Great Recession. It may very well be that fiscal multipliers vary not only between states – i.e., good and bad times – but also within states – i.e., between the mild recession that occurred early this millennium and the 2007-09 recession, which was by any recorded standard an ‘extreme event’.

Fiscal multipliers in extreme events

Interestingly, our findings confirm that this presumption holds in the data.

  • Fiscal multipliers in deep recessions turn out to be much higher and statistically different from fiscal multipliers in strong expansionary periods. Treating recessions and expansions as homogenous would hide such a difference.

Not only do we find that the deeper the recession, the more output is generated by increasing government spending;

  • We also find that government spending is highly effective exactly when it is most needed.

According to our estimates, one extra dollar spent by the US government during the Great Recession would have generated resources up to $2.50 in three years’ time.

Our findings also show that increasing government spending when economic conditions are less harsh generates quite different effects – with expected increases in government spending when the economy is in an expansionary period having mild positive effects at most for one year, and zero or negative afterwards.

As documented in our paper, our results are robust to a number of controls, including the inclusions of revisions of expectations over future real GDP realisations, different sample lengths, and the debt/GDP ratio. 

Concluding remarks and policy implications

Our research suggests that an increase in government spending is more effective exactly when it is most needed – i.e., when the economy is experiencing a deep recession.

We stress here that our main conclusion hinges on the employment of a model that is able to discriminate among multipliers associated with different recessions. That is why our results are different to those of Ramey and Zubairy (2014), who focus on the distinction between average multipliers in good times and bad.

Our main conclusion points to large fiscal multipliers associated with deep recessions. While having a clear policy implication on the use of the fiscal pedal in harsh economic conditions, such a conclusion comes with a 'but'. One of the underlying assumptions of our analysis is that fiscal debt is sustainable even in presence of increases in public spending which are not immediately backed up by increases in fiscal revenues. Indeed, different multipliers may emerge in countries that are more fiscally fragile than the US.

Hence, our findings do not imply a one-size-fits-all recommendation. Increasing government spending in countries whose level of debt might be perceived as unsustainable is a fiscal move whose pros and cons require, in our opinion, further research.

References

Blanchard, O and R Perotti (2002), “An empirical characterization of the dynamic effects of changes in government spending and taxes on output”, Quarterly Journal of Economics, Vol 117(4): 1329–68.

Caggiano, G, E Castelnuovo, V Colombo, and G Nodari (2015), “Estimating Fiscal Multipliers: News from a Nonlinear World”, Economic Journal, 125(584): 746-776.

Christiano, L J, M Eichenbaum, and S Rebelo (2011), “When is the government spending multiplier large?”, Journal of Political Economy, Vol 119(1) : 78-121.

Forni, M and L Gambetti (2014), “Government spending shocks in open economy VARs”, CEPR Discussion Paper 10115.

Leeper, E M, T B Walker, and S C S Yang (2013), “Fiscal foresight and information flows”, Econometrica, 81(3): 1115-1145.

Parker, J A (2011), “On measuring the effects of fiscal policy in recessions”, Journal of Economic Literature, 49(3): 703-718.

Ramey, V A and S Zubairy (2014), “Government spending multipliers in good times and in bad: evidence from US historical data”, mimeo, University of California at San Diego and Texas A&M University.

Ramey, V A and S Zubairy (2015), “Government spending multipliers in good times and in bad: Evidence from US historical data”, VoxEU.org, 23 January. 

5,459 Reads