Buiter on the Failure of Economics
For those who enjoyed Paul Krugman's attack on modern economics last Sunday, Willem Buiter goes him one better in today's FT:
http://blogs.ft.com/maverecon/2009/09/i-know-i-know-nothing-but-at-least-i-know-that/
The essence of his argument is that once you take away all the mathematical mumbo jumbo, there's really very little solid empirical data upon which to base a forecast.
Of course, all economists know there are problems with the data they use on a daily basis--just look at how often they are revised. As a friend once put it after one of BEA's periodic revisions of the GDP, "Not only can't economists forecast the future, they can't even forecast the past."
The classic book on this topic for those interested in the gory details is Oskar Morgenstern's On the Accuracy of Economic Observations. There's a gem on practically every page. For example, on page 21, I read this:
"When the Marshall Plan was being introduced, one of the chief European figures in its administration (who shall remain nameless) told me: 'We shall produce any statistic that we think will help us to get as much money out of the United States as we possibly can. Statistics which we do not have, but which we need to justify our demands, we shall simply fabricate.'"
I have no doubt that a vast amount of data from Africa falls into the same category. Even countries that do not need aid, like China, produce data that is deeply suspect. But economists still use it because they assume that any data is better than no data.
Recently I came across a blog entry by someone I know that made use of a 50-year old GDP data series to make his point. As far as I could see, his only reason for doing so is that the old data proved his point while the newer data didn't.
The point is simply that mathematical elegance and precision does not necessarily translate into accuracy or understanding of economic phenomena. Remember the advice given to a young economist working in India many years ago:
"The government are very keen on amassing statistics--they collect them, add them, raise them to the nth power, take the cube root and prepare wonderful diagrams. But what you must never forget is that every one of those figures comes in the first instance from the chowty dar [village watchman] who just puts down what he damn pleases."
But is this really a "failure of economics"?
But is this really a failure of economics? Of any kind of meaningful test?
In my experience, attempts at econometric modeling to seriously forecast the future went loudly kerplunk back in the 1980s, so this should hardly any kind of news today. And the kerplunk happened for multiple good reasons.
1) As mentioned, the data problem: economists can't even forecast the present. In the middle of just last summer oil was at $140, inflation was booming, commodity speculators were the villains of the day, I have a copy here of the Economist from then warning the Fed was at risk of losing its credibility as an inflation fighter. That was seven months after the recession started.
2) The grossly understated, under-appreciated, random factors in the economy.
Today is the opening of the pro football season. Fully 50% of NFL game outcomes are determined by random chance. Not just close games, also routs, as butterfly effects have their way. No supercomputer programmed by the top physical science modelers in the world can predict more than 75% winners (simple w-l, not against the spread) -- only 50% correct picks, splitting the rest even. Is this a failure of supercomputers and physicists?
And that's with possession in advance of almost all possible relevant knowledge about a simple finite system. Aren't there rather more random processes, unknowns and unknowables at play in an evolving economy than in a football game?
3) The human factor. E.g.: many mavens say today that the recession turned deep only due to the failure to bail out Lehman. But a sale of Lehman was agreed to, until a British regulator killed it at the last minute. (Something for those who put their faith in regulators to think about.) What econometric model was supposed to forecast that?
Moreover, the economy is actively steered. If policy makers forecast a recession they take steps to avoid it.
Thus, as Scott Sumner has pointed out, to the extent policy makers do a perfect job all shocks to the economy become unpredictable -- recessions will still occur (albeit with decreasing frequency) and always as a surprise, unforeseen, caused by random and previously unknown factors.
That's the optimum scenario -- and in it the punditry will castigate the policy makers for ineptitude ("Those morons never see a recession coming! How blind and incompetent can they be?").
So with all this, correctly forecasting recessions is pretty difficult! When they actually arrive after being forecast, policymakers are doing a bad job. It is only when they arrive out of the blue, unforeseen, that forecasters are doing a good job. The Zen of forecasting. What is foreseen should be avoidable. But one can't foresee and avoid that caused by the random and unknown.
(Of course, correctly forecasting good times doesn't count. Even for years on end. Who cares? They're taken for granted.)
Yet economics makes countless correct predictions every day, many subtle and intriguing, and many clearly affecting the quality of life of billions of people.
Altogether, forecasting doesn't seem to be the real test of it, IMHO.
And being that all this, 1, 2 & 3, has been known since the failure of the big forecasting models of the 1980s, for decades, I tend to take press stories expressing shock, shock, that economics doesn't forecast well as resulting from the naiveté of journalists who don't know better, and glibness, disingenuousness or agenda-pushing by economists who do.
Hey, it's the middle of hurricane season now, but forecasters have no idea when or where the next one will hit the US coast -- so I don't know what to do with my homeowners insurance. Is this the failure of meteorology? (And meteorological models don't have to predict whether regulators will decide to sink Lehman.)
It is a failure of
It is a failure of MACROECONOMICS. I see no evidence of failures of microeconomics.