Economists are pretty hopeless at predicting sharp downturns, leaving them open to the sort of derision handed out by Tim Harford and reported by Sam Roggeveen. Actually, Harford went on to give us economists an excuse. Picking up Keynes' hope that economists would become simple technicians, such as dentists, Harford observes that we don't expect dentists to predict when our teeth will fall out – just to help stop it happening.
In fact, things aren't quite as bad as Harford makes out. If he had quoted his source more fully, he would have had to note that forecasters do see normal business recessions coming, although not far in advance. It's the big downturns associated with financial collapses that they miss altogether.
It's easy enough to see why. These are 'tail events', meaning they are quite rare. Economies routinely operate with many tail risks, but if you predict that the sky is falling every time you see a tail-risk problem, you will soon have neither credibility nor audience.
There is, however, more than meets the eye in the article Tim Harford quotes. The authors, both IMF economists, use 2009 (a deep recession associated with a financial crisis) as a definitive illustration of economists' forecast failings, giving Harford his storyline: 49 out of 77 economies were in recession in 2009, but in a large sample of economic forecasts made in the previous year, not even one recession was predicted. The conclusion is that all economists, whether private sector, official or from international institutions such as the IMF, are equally blameworthy.
Was it a coincidence that this IMF-authored article appeared at around the same time that a report examining the Fund's own forecasting record was published by the IMF Independent Evaluation Office (IEO), a body set up to assess the IMF's performance? [fold]
The IEO made many of the same points about the challenge of forecasting. By and large, it assessed that the Fund forecasting performance was about the same as private-sector forecasters (in fact all forecasters tend to cluster together). But the IEO does make a substantive observation on the Fund's performance: 'they tended to be optimistic in high-profile cases characterized by exceptional access to IMF resources; these cases represented over eighty percent of the dollar amount of IMF resources disbursed.'
It's easy to see how this bias might happen (I have written on this in the past). In cases where the IMF has a substantial assistance program, it's awkward for policy-makers to predict a poor GDP outcome because it would raise the question: 'why don't you do something to make it better?' This is especially fraught when there are political constraints surrounding the assistance package: either it is too small to do the job properly or there has been disagreement about its terms. It would be unhelpful, to say the least, to deliver a dismal forecast at a moment like that. Unlike a weather forecast, an economic forecast can actually change the future.
Examples abound. At the end of 1997, when the Indonesian economy was already being swept into an economic maelstrom, the Fund continued to forecast positive growth. More recently, each of the European rescue operations has been accompanied by forecasts which were absurdly optimistic. How could those rescues, which stretched the limits of the Fund's lending powers, have gone ahead without such rosy outlooks?
Equally, it would have been unhelpful to the Fund's 2010 efforts to get countries to tighten their fiscal policies if the accompanying forecast had showed the recovery to be sluggish (which proved to be the case).
So Tim Harford has written a slick piece asking why economists bother to make forecasts. But the focus on the uniformity of forecast failure avoids the more serious point, made very gently in the IEO Report: it's one thing to be wrong randomly, but it is another to be wrong mostly on one side and in only certain circumstances. The Fund's forecasts are biased at exactly the time when it matters most for policy-makers to have a clear-eyed view. Whatever the answer to this quandary, Harford's article has nothing to offer.
Photo by Flickr user Allan Foster.
