Arnold Kling  

Fed Forecasting Errors

Friday Night Video: Henderson ... Getting Rich in America...

Evan Soltas writes,

A disparaging but not unfair comparison would be to little orphan Annie. "The sun'll come out tomorrow," Annie sang. "Bet your bottom dollar that tomorrow there'll be sun." The 3-to-4 percent recovery growth we've been long promising will come out tomorrow, the FOMC basically says every quarterly meeting. Bet your bottom dollar that tomorrow there'll be lower unemployment.

Pointer from Phil Izzo. Let me make two comments.

1. The Fed forecasting team, which includes at least three dozen Ph.D economists, just got pwned by a high school student!

2. I have remarked before that one hypothesis to explain the 1970s inflation is that the Fed consistently underpredicted inflation and consequently over-estimated the amount of monetary ease that was appropriate. A similar hypothesis for a Sumnerian today would be that the high unemployment since 2008 is a result of the Fed consistently over-predicting aggregate demand* and consequently under-estimating the amount of monetary ease that is appropriate.

*No, I don't personally like the aggregate demand concept. But there is some non-zero probability that I am wrong and that Scott Sumner and the rest of the economics profession that relies on AD are correct.

Comments and Sharing

COMMENTS (12 to date)
Costard writes:

The Fed "didn't try hard enough"? Ah wonderful -- that clears it up. I'm so glad we can stop worrying about the right and wrong of it, and start grading on effort. Evan, you get an A.

If only we had infallible people to fill our infallible institutions....

Ted Craig writes:

They've been "pwnd" on this for a long time:

Which gets back to this idea:
"I have, for many years, been in favor of replacing the Fed with a computer," Milton Friedman.

Jim Rose writes:

At the practical level, forecasting requires that there are regularities on which to base models, such regularities are informative about the future and these regularities are encapsulated in the selected forecasting model

Forecast errors are so large relative to the mean rate of change that, on average, forecasters cannot distinguish slow growth from a deep recession. Policy makers cannot distinguish recession from booms, so, reliance on such information may mislead policy-makers are crucial times.

The limitations of forecasting are well-known. Forecasts are conditional on a number of variables; there are important unresolved analytical differences about the operation of the economy; and large uncertainties about the size and timing of responses to macroeconomic changes. Shocks to the output, prices, employment and other variables are partly permanent and partly transitory.

We have very little reliable information about the distribution of shocks or about how the distributions change over time. Forecast errors arise from changes in the parameters in the model; mis-specification of the model, estimation uncertainty, mis-measurement of the initial conditions and error accumulation.

Most early discussions argued against the forecasting in principle. Forecasting was not properly grounded in statistical theory, it presupposed that causation implies predictability and the forecast themselves were invalidated by the reactions of economic agents to them.

Richie writes:
Which gets back to this idea: "I have, for many years, been in favor of replacing the Fed with a computer," Milton Friedman.
Who programs the computer? Who sets the parameters? Is this not still central planning of the money supply? If the growth of the money supply is set to 2 percent per year, who is to say it could not be increased to 5 percent?

Why would a free market in money not work?

steve writes:

The Fed has consistently overestimated GDP and overestimated inflation. It is prioritizing price stability.


Wexler writes:

Soltas is an undergrad college student, no?

Chris Koresko writes:

Three possible alternative explanations for the Fed's excessive optimism:

* Those PhD economists are making predictions based on a bad model. For example, a Keynesian might predict that the large fiscal deficits must be creating jobs and growth, when the real economy doesn't work that way;

* The Fed deliberately over-predicts economic performance to create an optimistic outlook in order to improve "animal spirits" and create a self-fulfilling prophesy of growth and prosperity;

* The Fed is doing to make its policies and those of its political masters look more successful than they are.

travis writes:

FED forecasting is useless unless your a short term trader. In that case even a good report can cause asset prices to decrease because there would be less of a chance for more QE. Remember Bernanke in 20005 talking about the stability of housing? Yes, me too. Predicting FED policy is useless other than predicting that they are indeed useless.

Jim Glass writes:

As to Fed "forecasting errors"...

Let us not forget the example of 2008 when deflation started running in July and accelerated to a 13% annual rate in the 4th Q -- the worst deflationary plunge since the collapse days of the Great Depression -- during which, at its meeting *after* Lehman failed, the Fed refused to drop rates citing risk of inflation.

In light of which, we might start by considering the Fed's record at "predicting the present".

Philip George writes:

A measure of money that accurately captures the inflation of the 1970s can be seen at

It also shows that there has been very large monetary expansion since the end of 2009. This monetary expansion has revealed itself not in higher real income but in higher prices of financial assets.

Les Cargill writes:

Does NGDP-centric/Sumnerian thinking actually depend on AD? If monetary velocity declines, you'll get (relative) deflation - "everywhere and always a monetary phenomenon".

Velocity is presumed to be measurable ( even if badly ) directly, so it seems somehow less an "aggregate"/inferred quantity. It's at least no worse than GDP.

Surely I missed something....

Les Cargill writes:

Phillip George:

According to Russ Roberts and others, the Fed has increased money supply, but is paying interest on reserves. That means that said issuance of that money has been effectively cancelled out. The net effect is to cause shrill declamations of inflationary peril while we experience something closer to deflation.

one explanation is that the bernank does not wish to be seen unilaterally causing a presumably politically unpalatable inflation.

ObDisclosure: I am an adherent to Scott Sumner and have been since before I'd heard what he was saying - I thought that actual money supply was weak as early as 1996. So I am biased that way.

Comments for this entry have been closed
Return to top