The Atlanta Fed's macroblog provides commentary and analysis on economic topics including monetary policy, macroeconomic developments, inflation, labor economics, and financial issues.

Authors for macroblog are Dave Altig, John Robertson, and other Atlanta Fed economists and researchers.

« The end of buy and hold? | Main | A home price index gets a new name (but it’s telling the same story) »

October 22, 2008

The evolving economic outlook

I understand that economic forecasts are notoriously inaccurate. The last time I looked, the root mean square error of economists’ year-ahead growth forecasts was 1.4 percentage points. In other words, you’d expect the consensus forecast to be within about 1½ percentage points of the actual outcome only about two-thirds of the time (and individual forecasts tend to be even worse than the consensus).

Accurate or not, predictions of the future are an essential input into decisions made throughout the economy, and the negative tone in the incoming data has led most forecasters to sharply cut their growth expectations through 2009. The first chart below shows how the Blue Chip consensus growth forecast for 2008–2009 has evolved.

Evolution of the 2008 Blue Chip Consensus Growth Forecast

As of October 10 (before forecasters had seen most of the September data), the Blue Chip consensus forecast showed the economy entering a period of negative growth last quarter and remaining that way (or near it) until the second quarter of next year with subpar growth continuing through the end of next year.

Indeed, the current Blue Chip growth projection for the U.S. economy is a sharp deterioration from what economists had been anticipating when the year began. In January, the consensus outlook called for the year to start off a little sluggishly before gradually resuming a more typical growth path. As the year unfolded, however, and the deep problems affecting the housing market became more evident, growth prospects waned.

Want some good news? Economic forecasts are notoriously inaccurate. And they are especially inaccurate at turning points, which, according to the Blue Chip consensus, we’ve recently passed. Why do forecasts perform worst just when needed most? The shocks that hit the U.S. economy from time to time are complex and not very well behaved. The likelihood of extreme outcomes—so-called “tail risks”—vary from shock to shock. Moreover, there may be “nonlinearities” in response to these shocks, meaning economic relationships that operate one way when the economy is in an expansionary state may operate differently when the economy is in a recessionary state. So a model that may have worked well during the good times may not work so well during the bad.

Perhaps the evolution of the economic outlook during the previous business cycle is a useful example. Consider the next chart, which shows the evolution of the consensus forecast for 2001–2002, beginning with January 2002. As the data revealed the economy was in some distress, the economic outlook deteriorated—severely so when it became clear the U.S. economy was in recession (and in November 2002, the NBER officially declared that a recession had actually begun the previous March).

Evolution of the 2001-2002 Blue Chip Consensus  Forecast Recessionary Phase

But remember, there are (at least) two turning points in the business cycle. There’s the downturn and a recovery. So the same problems that make spotting recessions hard also make it hard to anticipate the rebound. Again, let’s look at what happened during the last business cycle. The next chart shows how the consensus forecast evolved as the economy transitioned from recession to expansion, which, coincidently, began in November 2001.

Evolution of the 2001-2002 Blue Chip Consensus  Forecast Expansionary Phase

Between November and May of that time, the incoming data revealed that the consensus forecasts made during the recessionary period were overly pessimistic. The onset of expansion occurred sooner, and with more strength, than economists initially expected.

I am not aware of any individual forecast that can consistently outperform the consensus. And this is not an argument denying the economy is facing difficult times. But perhaps it’s useful to keep in mind the limitations of economic forecasts in times like these.

By Mike Bryan, vice president and economist of the Federal Reserve Bank of Atlanta

October 22, 2008 in Data Releases , Forecasts | Permalink


TrackBack URL for this entry:

Listed below are links to blogs that reference The evolving economic outlook :


I like many things about the University of Michigan consumer sentiment data that came out a few days ago.

They use partial rotating paired groups in their methodology which makes their data a little more useful.

The latest results from them, the people who actually PURCHASE goods and services on a daily basis were not at all pretty.

It's a useful comparator with the data set above.


Posted by: Matt Dubuque | October 23, 2008 at 08:42 PM

Greenspan made it very clear last week that computer models could not be used to perfectly predict the future. And we all know this intuitively- we actually play athletic games rather than award the win to the higher ranked team because you never know what can happen in real life. (Evidence: Phillies and Rays- no computer would have predicted this!)

The question now becomes what do we do to soften the slump and get back to recovery. The choice requires assumptions because we don't have perfect information. We can look at the patterns of past recessions. Steve Forbes does this in his excellent new article (http://www.forbes.com/hcome/forbes/2008/1110/018.html). It is important to remember how powerful and resilient the market is when it is given the opportunity to preform. Like an undervalued player, the market can come through in the clutch when the manager gets out of the way and lets the play do its thing.

Posted by: Randy | October 26, 2008 at 09:29 PM

Post a comment

Comments are moderated and will not appear until the moderator has approved them.

If you have a TypeKey or TypePad account, please Sign in

Google Search

Recent Posts



Powered by TypePad