[Viewpoint] The misleading nature of forecasts

Just how bad is the outlook for the United States’ economy? Unfortunately, you cannot tell from the forecasts.

These days, it is common to read forecasts predicting that the U.S. economy will grow at a 3 percent annual rate in the coming year. But just what does that mean?

The forecaster is not saying he is confident that growth will be exactly 3 percent. Every forecaster know that the actual growth rate may be higher or lower than the number he states. There is a distribution of possible growth rates, and the forecaster is telling us just one of the outcomes he can contemplate.

But if a forecaster tells us that he “expects” a growth rate of 3 percent, does that mean that he thinks that it is as likely to be above 3 percent as it is to be below 3 percent - the “median,” as he sees it, of the distribution of possible growth rates? Or could it mean that the he thinks the most likely growth rate will be close to 3 percent (the “modal” value), even though he may believe that it is much more likely to be less than that value than to be greater?

Many forecasters currently believe that there is a significant probability that the economy will slump during the next 12 months - a “double dip” in the expansion process. It is possible for them to hold that view and still forecast 2 percent growth for the next 12 months as the most likely outcome, or the “median,” of their probability distribution.

Any decision maker who depends on forecasts - a businessman, an investor, or a government official - needs to know the probability of very low or very high growth rates, as well as the median forecast. But that information remains hidden.

Some surveys of forecasts report the distribution of the forecasts of the different individuals being surveyed. We might read that the average forecast of 20 different forecasters is 2.8 percent, with the five highest forecasts above 3.2 percent and the five lowest forecasts below 2.5 percent. While that is useful information, it does not say anything about the extent to which each forecaster believes that growth might turn out to be less than 1 percent, or even less than zero.

Recent U.S. data have clearly raised the probability that the economy will run out of steam and decline during the next 12 months. The key reason for increased pessimism is that the government stimulus programs that raised spending since the summer of 2009 are now coming to an end. As they have wound down, spending has declined.

The government programs failed to provide the “pump-priming” role that was intended. They provided an early spark, but it looks like the spark did not catch. For example, a tax subsidy for car purchases caused GDP to rise in the third quarter of 2009, with more than two-thirds of the increase attributable to motor-vehicle production. But now that the subsidies have ended, the level of both sales and production has declined. A recent survey of consumers reported the lowest level of intended car buying in more than 40 years.

Although annual GDP growth was 3 percent in this year’s first quarter, almost all of it reflected inventory accumulation - some of which was unwanted build-up caused by disappointing sales. When inventory accumulation is excluded, first-quarter growth of “final sales” was just 0.8 percent in annual terms - and 0.2 percent compared to the fourth quarter of 2009.

The second quarter benefited from a surge in home purchases, as individuals took advantage of the tax subsidy for home buyers that expired in April. But what will happen in the third quarter and beyond now that the program has ended?

While it would be rash to forecast a double dip as the most likely outcome for the economy during the rest of this year, many of us are raising the odds that we attribute to such a downturn. Unfortunately, those forecasts of what might happen remain hidden.

*Copyright: Project Syndicate, 2010.
The writer is a professor of economics at Harvard and former President of the National Bureau for Economic Research.

By Martin Feldstein