Winner of the New Statesman SPERI Prize in Political Economy 2016

Tuesday 9 October 2012

Multipliers: using theory and evidence in macroeconomics.

Paul Krugman and Jonathan Portes have picked up on the IMF’s recent analysis (see Box 1.1) of multipliers. The IMF (or more specifically Olivier Blanchard and Daniel Leigh) say

“... earlier analysis by the IMF staff suggests that, on average, fiscal multipliers were near 0.5 in advanced economies during the three decades leading up to 2009. If the multipliers underlying the growth forecasts were about 0.5, as this informal evidence suggests, our results indicate that multipliers have actually been in the 0.9 to 1.7 range since the Great Recession.”
Jonathan paraphrases this as the IMF saying: "Delong et. al. were right; we were wrong". They, and many others besides, underestimated the impact of austerity because multipliers were underestimated.

The point I want to pick up on is why 'Delong et. al' got it right. In most cases it was not because we had undertaken a superior analysis of the empirical evidence. Instead we were thinking about basic macroeconomic theory. In particular, we recognised that a world where nominal interest rates were fixed, either because they cannot go below the Zero Lower Bound (ZLB), or because the rate is determined for the Eurozone as a whole, is very different to a world where monetary policy is unconstrained.

One of the things I did when I recently spent a week at the European department of the IMF was talk about multipliers. What I said was a version of this post. With fixed real interest rates the starting point for the government spending multiplier in New Keynesian theory is one, and most elaborations make it larger than one. These elaborations include that austerity will reduce inflation, which with fixed nominal interest rates could mean higher real interest rates.[1]  You can then add something for hysteresis effects, or any supply side effects if the government spending is investment rather than consumption. You can also add an effect from credit constrained households which, as Paul Krugman points out, are a key feature of deleveraging. So theory suggests something significantly larger than one, which is exactly what the IMF now finds.

It would be overstating things to say that the IMF's analysis proves this theory is correct. It is just not detailed enough to be a very good test. New Keynesian theory suggests austerity achieved through temporary income tax increases will have a smaller multiplier, as I discussed here.  Tax changes that impact directly on inflation will have different impacts, as I suggest here. The IMF's analysis looks at the budget deficit as a whole, so it cannot discriminate in this way.  However New Keynesian theory does suggest that multipliers can be large, and the IMF analysis suggests they have been large.

Those of us who got it right may therefore have simply had the insight to use standard theory. Those that used multipliers of around a half for fiscal consolidation packages that leaned heavily on spending cuts seemed to discount this theory, and instead may have been following evidence that was not applicable at fixed nominal interest rates.[2] . While there is plenty wrong with New Keynesian theory, it is also important to note when it gets things right. It is also important to note an occasion where thinking about macroeconomic theory can be rather more useful than naively following the evidence of the past.

[1] As my post made clear, government spending on domestically produced goods will have the same multiplier in an open economy. Although the multiplier will be lower because government spending will have some import content, any real interest rate effect will be larger in a flexible exchange rate open economy because it they influence the real exchange rate.
[2] Antonio Fatás argues that there were plenty of earlier empirical studies of multipliers that suggested numbers well above 0.5. In my view it makes little sense estimating multipliers that do not control for the reaction of monetary policy. 

1 comment:

  1. Depressingly, getting the theory right doesn't help all that much on the way to policy, as this from the EC staff shows:


Unfortunately because of spam with embedded links (which then flag up warnings about the whole site on some browsers), I have to personally moderate all comments. As a result, your comment may not appear for some time. In addition, I cannot publish comments with links to websites because it takes too much time to check whether these sites are legitimate.