Why don't economists make better forecasts? Daniel Gross fills us in:
The Forecast for the Forecasters Is Dismal, by Daniel Gross, Economic View, NY Times: ...If a recession were imminent, would .... economist[s] be able to forecast it?
The answer, based on recent experience, is a resounding no. “I don’t think we, as a profession, ever had an ability to forecast recessions,” said Jeffrey A. Frankel ... of ... Harvard and a member of the National Bureau of Economic Research’s Business Cycle Dating Committee... “It’s hard enough to know when a recession has started, looking at it with hindsight.”
Indeed..., the ... professional fraternity [does not] have a particularly good record of forecasting recessions. ...
Economists offer several explanations as to why their fellow dismal scientists collectively make such lousy forecasters. Nouriel Roubini ... of ... New York University ... believes that there are institutional reasons. Many forecasters surveyed ... work for Wall Street investment banks or asset management companies, which tend to argue that it is always a good time to invest. There are powerful incentives and pressures not to be unduly bearish about the economy. “When your firm is bullish on everything else, and is peddling all kinds of stocks and bonds, nobody will be foolish enough to go the other way,” he said. Of course, Mr. Roubini is perfectly willing to go the other way. Last summer, he boldly predicted a recession for the first half of 2007. ...
The complexity, dynamism and diversity of the United States economy also make forecasting recessions difficult. In small countries, which may depend on a single export, like oil, or where a natural disaster can wreak catastrophic results for the entire economy, it is comparatively easy to determine when and how one of these factors can cause a contraction, Mr. Frankel ... said. But in the United States, whose overall economy has responded so well in recent years to a series of external shocks — from 9/11 to Hurricane Katrina — it’s rarely sufficient to focus on a single factor.
Christina Romer ... of ... the University of California, Berkeley, says economists can’t predict recessions for the same reason stock market analysts can’t accurately predict market crashes. “Both kinds of events, by their nature, are not predictable events,” she said. Almost all the postwar recessions were preceded by a shock, like a spike in short-term interest rates, or a sharp rise in oil prices. “It’s impossible to see the shocks coming,” Ms. Romer said.
The very infrequency of recessions in the United States may make it more challenging to detect their imminent arrival. An entire generation of economists has grown up believing that the business cycle is largely something of the past, like black-and-white TV. Since March 1991, there has been only one recession, which lasted eight months. It’s like asking people who spend their time in Alaska to start forecasting tropical storms.
As a group, forecasters certainly don’t see a recession coming. On Feb. 13, those of the Federal Reserve Bank of Philadelphia collectively raised their estimates for real gross domestic product growth for 2007 to 2.8 percent, from 2.6 percent.
But just because they’ve been wrong in the past doesn’t mean forecasters are wrong now. “There is no reason at the moment why the steady momentum of the economy, with gains in employment feeding back into consumption growth, should falter,” said Robert J. Gordon ... of ... Northwestern University and a member of Business Cycle Dating Committee at the National Bureau of Economic Research. ...
Update: A few quick thoughts as a follow-up:
Using Economic Models Even if our predictions are inaccurate, that doesn't mean we have nothing to
offer. There are two uses of economic models, and the discussion above is about just one of the
Using models to forecast the future with precision is an area that has proven difficult either because our models are inaccurate, or because of unpredictable noise in the economic system that makes any forecast inherently noisy even with a perfect model.
The second use of models is to understand how the world works and this is, I hope, an area where we have done a bit better (though I expect that statement will be challenged). Even if we can't predict the future of the economy very well, that doesn't mean we can't use our understanding of the economy to design policies and institutions to minimize economic fluctuations.
Let me give an example. Where I live, rain is hard to predict. If you don't like the weather, wait ten minutes. Because of the volatility and unpredictability of storms coming off the Pacific, the weather people are routinely wrong, particularly in their hour by hour forecasts for the next day. I coached baseball here for a lot of years, and I can't tell you how many times the weather took an unexpected turn. The accuracy was pretty frustrating.
But that doesn't mean science can't help. We can still ask what to do when it
does rain, and take the appropriate precautions, from obvious things like
carrying an umbrella just to be safe to less obvious precautions such as
ensuring that structures like outdoor decks are free from dry rot and the danger
of falling and injuring people. Buildings can be constructed and materials used can be designed to avoid dry rot and other problems
and all sorts of other measures grounded in science can be taken to ensure that rain doesn't do any damage
(this may sound strange to, say, Californians who sometimes don't even have rain
gutters - here that would be inviting costly damage to your house and
foundation). Going back to baseball, you'd be amazed how well bags of cat litter can dry a pitcher's mound, batter's box, or a wet area of the infield. Some fields here are built with underground drainage to take surface water away quickly and you can can play on these fields even when it's fairly wet. There are a lot of precautions we can take to minimize the consequences of the rain even though we can't accurately predict when it will come. [Note: Comments tell me this isn't the best example I could have used. Agreed, it isn't since the policies aren't directly connected to the model - but the larger point that there are two uses of models remains - see this reply and the surrounding discussion.]
My point is that the inability to predict accurately does not mean we are powerless to react to changes in economic conditions any more than we are powerless to react to the weather when it changes. With models to help us understand how the world works, we can design policies to automatically stabilize the economy when shocks hit, and we can design institutions that minimize the chance of unexpected shocks occurring (it's not possible to stop the rain through institutional design - at least in the short-run when things like global warming don't matter - but other types of shocks do depend upon how the economy is organized). If shocks do hit, we can minimize the chance they'll spread if we understand how the shocks affect the economy (e.g. financial regulation to minimize shocks, think about the mortgage market for example, and injections of reserves by the Fed to stabilize the financial system when shocks do hit the financial system).
I won't claim we can predict well or that we understand perfectly how the economy operates, but I don't think it's as bad as many of you seem to believe either. There are things we understand, correlations in the data that are reliable, policies that appear to work, and other successes that make me believe we are making progress in understanding how the economic world functions and how our institutions and actions affect economic conditions.