The IMF and Austerity

Yesterday I discussed the IMF’s admission that they have blatantly miscalculated the effects that austerity would have on European economies. In an intellectually very honest research paper, chief IMF economist Olivier Blanchard and his co-author Daniel Leigh explain that the Fund seriously under-estimated the negative effects on GDP that would result from austerity measures.

I took the opportunity to gloat a bit yesterday. I felt I had the right to do so since I have been pointing to the disasters of austerity since the summer of 2010 when I published my book Remaking America: Welcome to the Dark Side of the Welfare State. There, I presented the disastrous effects of austerity policies on the Swedish economy, including evidence of a serious decline in private economic activity.

Today I would like to take one more step in the analytical direction. The IMF did a good job in explaining its errors, and given how serious the consequences are of their miscalculations, it is important that we understand what they did wrong. This insight can help legislators and policy analysts avoid making the same mistake all over again.

In plain English: those who still endorse austerity better pay very close attention.

The IMF mea culpa paper is called Growth Forecast Errors and Fiscal Multipliers. It explains two things:

  • The Fund made policy recommendations on erroneous data; and
  • The Fund’s data errors resulted in a miscalculation of the multiplier effects of austerity by a well defined number.

As for the policy recommendations, there was a lot of damage done. I am not going to go into details on that damage – though as far as Greece is concerned I suggest this paper which I published this past summer – but I will suggest that the effects of austerity are clearly visible in European GDP data.

That data, however, is the ex-post austerity data. It presents the numbers that show what the effects are; obvoiusly it is not the data that the IMF would use for its forecasting of what the effects would be of their recommended austerity policies. That data has to come from earlier periods in the economy.

And it is right here that the Fund made its mistake.

In order to forecast the effects of fiscal policy measures, economists use models that show how various sectors of the economy affect each other over a defined period of time. The most critical of those “transmittors” of economic activity is the multiplier, a centerpiece of modern macroeconomics. British economist Richard F Kahn was the first to analytically define the multiplier; his 1931 article The Relation of Home Investment to Unemployment played a crucial role in the development of new fiscal policy measures in Britain.

John Maynard Keynes gave the multiplier a very important macroeconomic context in this book General Theory of Employment, Interest and Money. Based on his work, in turn, John Hicks developed the formalized foundations for modern macroeconomic modeling, now known as the IS-LM model. Any use of this model – in any of its modern forms – incorporates the use of the multiplier.

A multiplier shows how the spending of an extra $100 by consumers will spread through the economy and magnify the initial spending. By the same token, a multiplier also shows how the removal of $100 worth of spending from the economy will spread through the economy. The spreading effects, so to speak, vary depending on, primarily:

  • Who increases or decreases spending – consumers, government, businesses or foreign markets buying our export products;
  • What kind of spending is involved – an increase or decrease in government spending on bureaucrat salaries has less immediate effects on the economy than a change in consumer spending at retail stores; and
  • When the spending change takes place.

It was this last item that the IMF messed up. Here is how they summarize their new findings:

Robustness checks indicate an unexpected output loss, relative to forecast, that is for the most part near 1 percent and typically above 0.7 percent, for each 1 percent of GDP fiscal consolidation. We obtain similar results when we extend the analysis to forecasts for all advanced economies. … Looking within the crisis, we find evidence of more underestimation of fiscal multipliers early in the crisis (for the time intervals 2009-10 and 2010-11) than later in the crisis (2011-12 and 2012-13).

The meaning of this is, in a nutshell, that if the IMF recommended spending cuts and tax increases equivalent to one percent of a country’s GDP, they knew that this would have a negative effect on GDP growth. What they did not know was that the effect was going to be a whole percent of GDP larger than they estimated.

This may sound like a mere technicality, but it is not. The practical, everyday-life effects of such a forecasting error almost reach catastrophic levels.

Consider the following example. A country has a GDP of $100 billion. Let’s say taxes and government spending comprise 30 percent of GDP at $30 billion, and that there is a three-percent-of-GDP deficit. That would be $3 billion. The IMF and its European partners, the EU and the ECB, fly in and recommend an austerity package with tax hikes and spending cuts totaling $1 billion. They do this knowing that there will be a contraction of economic activity. According to their models that contraction will amount to $1.5 billion (with a multiplier of 1.5).

Enter the forecasting error of one-to-one percent: for every percent of proposed spending cuts there is another percent of spending losses, on top of the $1.5 billion forecasted. The total drop in GDP, as a result of the austerity package, is now $2.5 billion. That extra billion of lost GDP means that unemployment goes up more than expected, corporate investments fall more than expected – as does private consumption – and that tax revenues shrink more than expected.

Obviously, the larger the initial cut in spending, the larger the forecasting error. A program to cut the entire deficit in one year would lead to an extra, unanticipated loss of GDP amounting to $3 billion.

The most immediate effect of this would be a loss of a full $1 billion of tax revenues. The IMF would in other words over-estimate the ability of its austerity policies to close the budget gap. Instead of resulting in a balanced budget their package will result in a $1-billion tax revenue shortfall.

When the IMF and national policy makers see that their budget is still exhibiting a significant deficit, they double down on their austerity policies and launch another round of tax hikes and spending cuts. The same forecasting error now compounds over two years, resulting in a chain reaction of persistent deficits. Not realizing that the medicine is killing the patient, the IMF would then recommend yet another round of austerity.

One major reason why the IMF apparently got the multipliers wrong is that the multipliers change over time. They are larger when the economy is growing strongly and just enters a downturn than they are when the economy is at the depth of a recession. There are two explanations for this: there is more spending out there that consumers can easily repeal when they sense a change in economic conditions; and there is widespread uncertainty and resistance to spending in a recession.

I developed the latter point in my doctoral dissertation back in 2000 (published in 2002 by British academic publisher Ashgate). The mainstream of macroeconomics has yet to take into account the role of uncertainty, which means that its practitioners – especially in the forecasting business – are vulnerable to making forecasting errors of the kind the IMF has made. I do not blame them for it; modeling uncertainty is almost an oxymoron. Uncertainty is very difficult to quantify, though it can be done. But the methods for doing so are incompatible with current practices in econometrics.

Back to reality now. The compounded forecasting errors by the IMF have in all likelihood played an instrumental role in putting the Greek economy through five straight years of shrinking GDP. Regardless of why the IMF made its mistakes, the consequences are enormous. In fact, I suspect the repercussions will be felt through the economics community for years to come. Leading academic economists need to ask themselves if their almost theological obsession with training graduate students in econometrics – and nothing else – really is the right thing to do.

For policy makers, the IMF mea culpa is a huge red flag. The worst thing to do for U.S. Congress at this point is to embark on a European-style austerity crusade. A far better strategy is structural reform that gradually but permanently will phase out costly, runaway entitlement programs.

My fellow free-market think-tankers also better pay attention. This entire issue is a big package of evidence that it is always worth doing the hard, analytical work rather than pumping out pointless talking points. But it is also yet more evidence that Keynesian economics, as opposed to Austrian theory, is still a superior analytical instrument in economics.