The effectiveness of fiscal and monetary stimulus in depressions

Barry Eichengreen, Kevin Hjortshøj O’Rourke, Miguel Almunia, Agustín S. Bénétrix, Gisela Rua, 18 November 2009

a

A

The debate over the effectiveness of stimulus rages on (Barro and Redlick 2009). Fewer than two years of data – that being the amount of time since monetary and fiscal measures to counter the crisis were put in place – are not enough to pin down the effects. And different theoretical models, for better or worse, predict different results. Strongly held priors rule the roost.

There is, however, one important source of information on the effectiveness of monetary and fiscal stimulus in an environment of near-zero interest rates, dysfunctional banking systems and heightened risk aversion that has not been fully exploited: the 1930s. This column – based on a paper presented at the 50th Economic Policy Panel Meeting held in Tilburg on 23-24 October 2009 – draws out some of the lessons for today’s crisis (Almunia et al. 2009).

Parallels: the Great Depression and the Great Recession

In previous columns, two of us documented the strong parallels between the early stages of the Great Depression and the early stages of our Great Recession. The causes of the two episodes were quite similar. In the earlier episode they included an unsustainable real estate boom (centred in Florida), lax supervision and regulation, and global imbalances (known then as “the transfer problem”). Similar circumstances suggest similar effects of policy, whether positive, negative or none.

The problem is that the policy response then was limited. The Keynesian argument for expansionary fiscal policy – whether right or wrong – was not known in this pre-Keynesian era. Hence there was relatively little variation in fiscal stance, with conservative policies being the default option. The aggressive use of discretionary monetary policy was also relatively unusual, as central banks were wedded to gold-standard ideology.

But there were exceptions. Japan’s aggressive use of monetary policy under Takahashi was one, Italy’s large budget deficits under Mussolini another. And there were good – exogenous – reasons for this variation. Fiscal impulses were generally governed by forces other than immediate economic conditions; Italy’s war in present-day Ethiopia, Hitler’s rearmament, the approach of World War II. Who responded to the crisis with monetary stimulus depended heavily on prior monetary experience; counties that had suffered high inflation in the 1920s tended to be reluctant to abandon the gold standard in the 1930s.

New research

Cross-country comparisons can thus help us untie the Gordian Knot and move the debate from the realm of ideology to that of evidence. Our project therefore focuses on assembling annual data on growth, budgets and central bank policy rates, mainly from League of Nations sources, for 27 countries covering the period 1925-39.

This leaves the question of what model or empirical technique to apply. Rather than prejudging the answer, we employ a battery of empirical methods. We use panel vector autoregressions (VARs) with conventional assumptions about the “ordering” of the variables (whether a variable affects the others contemporaneously or only with a lag). We consider alternative orderings, and also run panel VARs with defence expenditure entering the equations as an exogenous variable. We run panel instrumental variables regressions using defence spending as an instrument for fiscal policy and gold standard membership as an instrument for monetary policy. And we run alternative panel regressions looking at the impact of fiscal and monetary shocks, the latter calculated by running simple autoregressions and extracting the residuals.

Where tried, fiscal policy was effective in the 1930s

The details of the results differ, but the overall conclusions do not. They show that where fiscal policy was tried, it was effective.

Our estimates of its short-run effects are at the upper end of those estimated recently with modern data; the multiplier is as large as 2 in the first year, before declining significantly in subsequent years. (Figure 1 shows this in the case of the panel VAR estimates with the conventional ordering assumptions.) This is, in fact, what one should expect if one believes that the effectiveness of fiscal policy is greatest when interest rates are at the zero bound, leading to little crowding out of private spending. It is what one should expect when households are credit constrained by a dysfunctional banking system. Given similar circumstances in 2008, this underscores the advantages of using 1930s data as a source of evidence on the effects of current policy.

Figure 1. Impulse response functions, shock to defence spending (1% of GDP)

 

Note: Solid lines are the point estimates of the impulse-response mean. Dashed lines are the 16th and 84th percentiles from Monte Carlo simulations based on 1000 replications. Vertical axis indicates defence spending (G), GDP (Y), revenues (T) and central bank discount rate (R). Each equation in the system includes country fixed effects, country-specific linear trends and year dummies.

Other methodologies may yield somewhat smaller fiscal multipliers. But the message is the same.

Monetary policy in the 1930s was not powerless

The results for monetary policy are less robust but point in the same direction. A positive shock to the central bank discount rate leads to a fall in GDP (Figure 2). The fall in output just misses statistical significance at conventional levels (that is, the confidence bands just barely span the horizontal line denoting no change.) Under alternative assumptions about the ordering of the variables, however (Figure 3), the direction of the effect is the same, and this time it is significant.

Figure 2. Impulse response functions, shock to discount rate

 

Note: Solid lines are the point estimates of the impulse-response mean. Dashed lines are the 16th and 84th percentiles from Monte Carlo simulations based on 1000 replications. Vertical axis indicates defence spending (G), GDP (Y), revenues (T) and central bank discount rate (R). Each equation in the system includes country fixed effects, country-specific linear trends and year dummies.

Figure 3. Impulse response functions, shock to discount rate (alternative ordering)

 

Note: Solid lines are the point estimates of the impulse-response mean. Dashed lines are the 16th and 84th percentiles from Monte Carlo simulations based on 1000 replications. Vertical axis indicates defence spending (G), GDP (Y), revenues (T) and central bank discount rate (R). Each equation in the system includes country fixed effects, country-specific linear trends and year dummies.

This result is notable, given the presumption, widespread in the literature, that monetary policy is ineffective in near-zero-interest-rate (liquidity trap) conditions. On the contrary, in the 1930s it appears that accommodating monetary policy helped, by transforming deflationary expectations (Temin and Wigmore 1990) and by helping to mend broken banking systems (Bernanke and James 1991). Given the prevalence of both problems circa 2008, we suspect that the results carry over.

For others with different priors, these results may sit less easily. But the time for priors is over. Policy should rest on an evidentiary basis. The evidence we have marshalled so far speaks clearly.

References

Almunia, Miguel, Agustín S. Bénétrix, Barry Eichengreen, Kevin H. O’Rourke, and Gisela Rua (2009), "From Great Depression to Great Credit Crisis: Similarities, Differences and Lessons", presented at the 50th Economic Policy Panel Meeting, 23-24 October.

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

Bernanke, Ben and Harold James (1991) The gold standard, deflation, and financial crisis in the Great Depression: An International Comparison. In R. Glenn Hubbard (ed.), Financial Markets and Financial Crises. Chicago: University of Chicago Press.

Temin, Peter and Barrie A. Wigmore (1990) The end of one big deflation. Explorations in Economic History 27: 483-502.

Topics: Macroeconomic policy
Tags: global crisis, Great Depression, stimulus

2008-2009 Shapiro Fellow, Department of Economics, University of California, Berkeley
Post-Doctoral Fellow at the Institute for International Integration Studies (IIIS), Trinity College Dublin
Professor of Economics and Political Science at the University of California, Berkeley; and formerly Senior Policy Advisor at the International Monetary Fund. CEPR Research Fellow
Chichele Professor of Economic History, All Souls College, University of Oxford; and Programme Director, CEPR
Ph.D. Candidate in the Department of Economics at University of California, Berkeley