A major question facing many governments in the rich world today is whether we should try to stimulate the economy by increasing government spending. The professional opinion of economists regarding this question is sharply divided. While many economists believe that increases in government spending can have large ‘multiplier’ effects – ie increase output by more than the increase in spending – many others are sceptical of this and some even believe that increases in government spending may harm the recovery.
A major reason for this disagreement is that it is notoriously hard to construct convincing empirical evidence on the effects of fiscal stimulus. The empirical challenge is a familiar one; correlation does not imply causation. In the case of fiscal stimulus, the simple-minded approach of seeing whether output is high in periods of high government spending doesn’t work since governments tend to systematically increase spending when output is low for some other reason, eg because of a financial crisis. The simple correlation will then ‘confound’ the effects of government spending with the effects of other factors. What is needed is some sort of ‘natural experiment’, ie pseudo-random variation in government spending.
One approach to overcoming this problem is to study how US output has responded to wartime military spending in the 20th century (Ramey and Shapiro 1989, Ramey 2011, Barro and Redlick 2011, Hall 2009). The idea here is that these wars were caused by geopolitical factors that were largely unrelated to the state of the US economy at the time. However, large wars are few and far between. Also, wars often involve a surge of patriotism and controls on economic activity that can directly impact the economy. And to muddle the water even further, there is a large degree of variation in the extent to which taxes are raised contemporaneously to finance the war. For these reasons there is only so much we can learn from these wars.
Using regional variation in military spending to identify the multiplier
These issues motivate the quest for other sources of pseudo-random variation in government spending. In recent work, our approach has been to exploit regional variation in military spending in the US (Nakamura and Steinsson 2011). We use the fact that when the US embarks upon a military buildup, there is a systematic tendency for spending to increase more in some states than others. For example, when aggregate military spending in the US rises by 1% of GDP, military spending in California on average rises by about 3% of California GDP, while military spending in Illinois rises by only about 0.5% of Illinois GDP. Under the assumption that the US doesn’t embark upon military buildups like the Vietnam War because states like California are doing badly relative to states like Illinois, we can use regional variation associated with these buildups to estimate the effect of a relative increase in spending on relative output. Our conclusion is that when relative spending in a state increases by 1% of GDP, relative state GDP rises by 1.5%.
A number of other authors have recently exploited other sources of sub-national variation in spending to estimate similar relative multipliers. For example, Shoag (2011) studies increases in spending associated with windfall returns to state pension plans; and Acconcia et al (2011) study reductions in provincial-level spending in Italy associated with legally mandated crackdowns on the mafia that were triggered by evidence of political corruption. See also studies by Chodorow-Reich et al (2011), Clemens and Miran (2011), Cohen et al (2011), Fishback and Kachanovskaya (2011), and Wilson (2011). Most of these papers have estimated effects of relative spending on relative output that are of a similar magnitude to those we estimate or somewhat larger – multipliers between 1.5 and 2.5.
Are multipliers of 1.5 too large to be true?
At first glance, these multiplier numbers may seem quite large (eg relative to the estimates of Barro and Redlick 2011 and Ramey 2011) and thus favourable to advocates of additional fiscal stimulus. However, some care is required in interpreting these empirical results.
One difference between our estimates and older evidence based on aggregate data is that in our setting, the region getting the spending is not paying for it. Could this be the reason why we are getting such a high multiplier estimate? Neoclassical models would actually suggest the opposite. The reason is that the negative wealth shock that accompanies an aggregate government spending shock causes an increase in labour supply. The fact that no such wealth shock occurs in our setting should thus lower the multiplier, not raise it.
Another important difference is that when spending increases in California relative to Illinois, national government policy is held fixed across these states. For example, the Fed is not able to respond by raising interest rates in California relative to Illinois, and Congress does not respond by raising tax rates in California relative to Illinois.
In sharp contrast, monetary and tax policy is not constant in response to aggregate government spending shocks. “Normal” monetary policy – eg the policy practiced by the Fed under the leadership of Paul Volcker and Alan Greenspan – is to ‘lean against the wind’ quite aggressively by raising real interest rates – or decreasing them by less– in response to aggregate government-spending shocks. The tax policy response to aggregate government-spending shocks varies more over time. During the Korean War, taxes were raised by a large amount. This is less true for more recent military buildups.
This difference between the response of national policy to a regional spending shock and an aggregate spending shock implies that the government spending multiplier we estimate, in effect, is conditioning on a relatively accommodative monetary and tax policy response. This likely explains why our multiplier estimate is higher than those of, eg Barro and Redlick (2011) and Ramey (2011).
Do we need additional stimulus today?
So, what does our analysis imply about the effects of additional stimulus today? One lesson that our analysis illustrates is that there is no ‘single multiplier,’ but rather the multiplier is highly sensitive to the stance of monetary and tax policy (see also Woodford 2011 on this point). An important special feature of the current situation in many economies is that nominal interest rates are very close to their lower bound of zero. This constraint implies that nominal interest rates are likely higher at the moment than the monetary authorities in these countries would like them to be. This means that these central banks are unlikely to respond to fiscal stimulus by raising rates the way they would in normal times. In other words, monetary policy is likely to be more accommodative in response to fiscal stimulus today than in normal times.
It turns out that our analysis is particularly well suited to help us draw inference about this situation. As we discuss above, we know from the fact that the US is a monetary and fiscal union that the Fed can’t differentially increase interest rates in one region versus another and that Congress doesn’t raise tax rates in one region relative to another. This pins down an important ‘moving part’ when it comes to interpreting our estimate of the fiscal multiplier.
For estimates based on aggregate variation in spending, it is much less clear what the monetary and tax policy response was at the time and it is therefore much harder to interpret these estimates and much harder to distinguish between the Neoclassical and the New Keynesian view of how government spending affects the economy.
The fact that we can pin down relative policies allows us to show that our estimates are much more consistent with New Keynesian models in which ‘aggregate demand’ shocks – such as government spending shocks – have large effects on output when monetary policy is sufficiently accommodative than they are with the plain-vanilla Neoclassical model. In particular, our results support the view that aggregate fiscal stimulus should have large output multipliers when the economy is at the zero lower bound.
Acconcia, A., G. Corsetti, and S. Simonelli (2011) “Mafia and Public Spending Evidence on the Fiscal Multiplier from a Quasi-Experiment,” CEPR Discussion Paper 8305.
Barro, R. J., and C. J. Redlick (2011) “Macroeconomic Effects from Government Purchases and Taxes,” Quarterly Journal of Economics, 126(1), 51-102.
Chodorow-Reich, G., L. Feiveson, Z. Liscow, and W. G. Woolston (2011) “Does State Fiscal Relief During Recessions Increase Employment? Evidence from the American Recovery and Reinvestment Act,” Working Paper, University of California at Berkeley.
Clemens, J., and S. Miran (2011) “The Effects of State Budget Cuts on Employment and Income,” Working Paper, Harvard University.
Cohen, L., J. Coval, and C. Mallow (2011) “Do Powerful Politicians Cause Corporate Downsizing?,” Journal of Political Economy, forthcoming.
Fishback, P., and V. Kachanovskaya (2010) “In Search of the Multiplier for Federal Spending in the States During the New Deal,” NBER Working Paper No. 16561.
Hall, R.E. (2009) “By How Much Does GDP Rise if the Government Buys More Output?,” Brookings Papers on Economic Activity, 2009(2), 183-249.
Nakamura, E. and J. Steinsson (2011) “Fiscal Stimulus in a Monetary Union Evidence from US Regions,” NBER Working Paper 17391.
Ramey, V. A. (2011) “Identifying Government Spending Shocks It's All in the Timing,” Quarterly Journal of Economics, 126(1), 1-50.
Ramey, V. A., and M. D. Shapiro (1998) “Costly Capital Reallocation and the Effects of Government Spending,” Carnegie-Rochester Conference Series on Public Policy, 48(1), 145-194.
Serrato, J. C. S., and P. Wingender (2011) “Estimating Local Fiscal Multipliers,” Working Paper, University of California at Berkeley.
Shoag, D. (2011) “The Impact of Government Spending Shocks Evidence on the Multiplier from State Pension Plan Returns,” Working Paper, Harvard University.
Wilson, D. J. (2011) “Fiscal Spending Jobs Multipliers Evidence from the 2009 American Recovery and Reinvestment Act,” Working Paper, Federal Reserve Bank of San Francisco.
Woodford, M. (2011) “Simple Analytics of the Government Expenditure Multiplier,” American Economic Journal Macroeconomics, 3(1), 1-35.