Sunday, July 24, 2011

Economists are terrible forecasters - why trust them anyway?

The herd that is 'consensus' clings to this hope that GDP will bounce back smartly in Q3 and Q4, where all the while some pretty miserable data and financial conditions are staring us in the face. On May 26 I worried about the 'soft patch'. On July 24, the data doesn't look too much better.

This week we get the Q2 GDP report from the Bureau of Economic Analysis, of which the mean growth forecast from 69 polled economists by Bloomberg is 1.8% Q/Q SAAR (Bloomberg terminal, no link). That may change as we near Friday (unlikely by much), but those same economists (on Bloomberg) are forecasting a 3.2% rebound in Q3.

Alas, after the huge forecast miss in the first half of the year, I no longer put much stock in what economists expect by way of GDP growth just one quarter ahead (some yes, but broadly no).

See, economists are generally terrible forecasters.

The chart illustrates the one-step-ahead quarterly GDP forecast of the Philadelphia Fed's Survey of Professional Forecasters (SFP) alongside the ex-post GDP growth rate (GNP prior to 1992 for continuity with the SFP data). You can download the GDP data here or here, and the historical SFP data here. As an example, 2011 q1 represents the SFP mean real GDP growth forecast (% Q/Q, SAAR) from the fourth quarter of 2010, 2.5%, and the actual q1 GDP growth rate, 1.9%.

They do okay looking broadly at the business cycle; but the SFP point-to-point forecast error is huge. In sum, the average forecast error is 0.3% on a Q/Q basis spanning the years 1968 to 2011. That means economists miss the next quarter forecast by an average of 0.3%. And they're really terrible recession forecasters: the average error is -2.1%. Spanning the historical expansion periods - of which we are in right now - the SFP miss was 0.8% on average.

On May 13 the SFP predicted a Q2 growth rate of 3.2% SAAR, where 1.8% is the Q2 forecast for Friday.

Economists completely underestimated the following factors in the first half of 2011:

(1) the adverse effects of rising oil prices in the first third of 2011,
(2) the supply disruption from the Japanese earthquakes, and
(3) the underlying weakness in the economy.

Remember David Altig's July 8 post on the implication of sub 2% GDP growth? Mark Thoma expanded on the commentary as well. Mark Thoma also points us to a Macroeconomic Advisers piece that forecasts a 'brief growth recession' if the US economy gets downgraded. S&P puts the probability of downgrade at 50%.

I don't trust economic forecasts. I'll take the under on Q3 and Q4, especially given the way the budget talks are going and recent high-frequency data is looking.

Rebecca Wilder

1 comment:

  1. I've wondered this: Do econ grad students even study forecasting as part of their PhD program?

    ReplyDelete