Does the recent paper by Barry Eichengreen, Miguel Almunia, Agustin Benetrix, Keven O’Rourke and Gisela Rua, deserve the praise it has been getting from Paul Krugman and other Keynesian fans of fiscal stimulus? I have my doubts. But before expressing them, let me make a few points.
I share many of Eichengreen’s views on the Great Depression. We both argued that monetary stimulus (especially currency devaluation) was helpful. We both have spoken out in favor of “competitive devaluations” during the current crisis. He is more optimistic about the efficacy of fiscal stimulus than I am, but I would concede that the military buildups of the late 1930s probably boosted RGDP, at least by a little. However, I was far from convinced by their most recent paper on fiscal stimulus during the Great Depression. For instance, they start by admitting that in many of the most important economies the fiscal shocks were too small to produce statistically significant results:
Ritschl (2005) similarly finds that fiscal deficits were too small to have made an economically consequential difference in Nazi Germany. Not even in Sweden, a country where Keynesian ideas were circulating avant la lettre, were fiscal deficits big enough to make a significant difference (Schön 2007). Appendix 2 shows that what was true for the United States and Germany was true for other countries: in most cases budget deficits were
moderate, and even remained below the per cent threshold that has become familiar to European readers since the 1990s. The decade that saw the publication of The General Theory did not see the widespread adoption of Keynesian pump-priming measures.
So they use VAR estimation with panel data, which is something I intensely dislike for all sorts of reasons. Let’s start with this assumption:
we have chosen to use the central bank discount rate as our measure of monetary policy.
That’s right; low interest rates mean easy money. Now I will admit that on any given day a central bank rate cut is more expansionary that a rate increase, but their data set uses annual data. There is a massive identification problem here. Indeed in my work on the Great Depression I found the devaluation of 1933, which might have been the most powerful monetary shock in American history, had almost no contemporaneous impact on nominal interest rates or the money supply. Because I am highly suspicious of VAR studies, I wanted to see the stylized facts that they believe best illustrate their hypothesis. Here is the “money quote” that I have seen cited by pro-stimulus bloggers:
Individual country experience with large fiscal stimulus was rare in this period, but where it occurred the evidence points in the same direction. One of the biggest fiscal stimuli in this sample occurred in Mussolini’s Italy during 1936-7, as a result of the war in Ethiopia. Italy ran a deficit in excess of 10 per cent of GDP in 1936 and 1937. Italian GDP grew by 6.8 per cent in 1937, by a marginal amount in 1938, and by 7.3% in 1939. According to Toniolo (1976), the Italian economy moved to full employment during this period. In France, the budget deficit increased substantially beginning in 1935, and GDP grew by 5.8 per cent in 1936. The deficit exploded in 1939, during which year the economy grew by no less than 7.2 per cent.
This set off all sorts of alarm bells in my mind.
1. 1936 was the year that France left the gold standard and devalued the franc (I think by about 30%.) Italy also left the gold standard in 1936, and I recall the devaluation was closer to 40% percent. These are enormous monetary shocks that don’t necessarily show up in the lower interest rates. Even if rates did fall, the change in interest rates would dramatically understate the size of the monetary shock.
2. Those real GDP growth rates are actually very disappointing for the year after a major devaluation in a severely depressed economy. The US grew much faster in the period immediately after the dollar was devalued in 1933. And the US had to contend with an adverse wage shock. On the other hand France was also hit by a wage shock in 1936, and contemporaneous observers attributed the weak French performance in 1936-37 to the fact that their wage shock preceded devaluation, whereas in the US it followed devaluation.
If you want to know why this is so important, read some old Krugman posts. Over the last few months he has had to continually address fiscal stimulus critics who pointed to various small countries that were able to recover nicely despite some sort of fiscal austerity program. And each time Krugman dismissed them as being irrelevant, as they were able to offset the contractionary effects of fiscal austerity with currency depreciation. (Although in at least the Canadian case, the timing of the currency depreciation was all wrong for Krugman’s argument.) Reading Krugman’s posts, one began to wonder how anyone could be so stupid as to argue that these were examples of successful fiscal austerity. OK, but Krugman is doing exactly the same thing by hyping this Eichengreen, et al, study. Their showcase examples occurred right in the midst of large currency devaluations, something we already know from the American context were highly expansionary in a deeply depressed economy.
Their regression did include a dummy for whether or not countries were on the gold standard, but that doesn’t really address the problem. It would have been much better to include a variable for the change in the exchange rate. But even that wouldn’t address the much more fundamental identification problem here. This entire project rests on the implicit assumption that “the” multiplier is some sort of deep parameter that is waiting to be discovered. It rests on the assumption that the following is a scientific question that should have an objective answer:
What is the fiscal multiplier when there is an exogenous fiscal shock?
But there is no answer to that question, because it all depends on how monetary policymakers respond. Did the French fiscal expansion of 1935 cause the French government to devalue the franc in 1936? VAR can’t answer that question, and thus can’t tell us the independent impact of a fiscal shock in a world where central banks can, in principle, determine the path of NGDP. And it certainly can’t provide any useful information about fiscal multipliers in a non-gold standard economy where the central bank seems determined to peg inflation at about 1%.
Krugman might say none of this matters if rates are stuck at zero. But we have just seen that Italy and France found ways to enact powerful monetary shocks in a near-zero rate environment. So zero rates don’t magically take money out of the equation, at least when estimating multipliers using interwar data. One can argue that modern central banks would not behave the same way, but that’s completely beside the point. The issue here is whether the interwar studies are reliable.
Now I will ask a really stupid question, to give Keynesians readers a chance to ridicule me. Why is the multiplier estimated by looking at the impact of spending on real GDP? (Heh, I never claimed to understand Keynesianism.) I thought in the Keynesian model higher government spending impacted AD, i.e. NGDP, and the impact on RGDP and prices depended on the slope of the SRAS. Wouldn’t it make more sense to see how NGDP changed in response to a fiscal shock? Higher NGDP is the proximate goal, isn’t it? Obviously you’d hope to also increase RGDP; indeed if the AS curve were vertical there would be no point in even having any fiscal stimulus. But the proximate goal of demand stimulus is higher NGDP, so it seems to me that we should be estimating the impact of stimulus policies on NGDP.
You might recall that I view NGDP growth expectations as the only sensible indicator of the stance of monetary policy. All other monetary policy indicators (such as r and M) are profoundly unreliable. If you follow my logic, you might begin to understand why I view the entire fiscal multiplier debate as being about as meaningful as the question of how many angels can dance on the head of a pin. The question is not whether fiscal stimulus can change RGDP, it is whether fiscal stimulus can trigger monetary stimulus.
Update: I just noticed that Bob Murphy did a similar post.