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.

« An agnostic gets a little religion | Main | Prospects for a small business-fueled employment recovery »

October 02, 2009

Economic troughs, changes in the unemployment rate, and fed policy

Recent data on the U.S. economy have been mixed. But today's weak labor market report (discussed more here, here, and here) provides a reminder that thinking of the economy as being in anything other than a technical recovery at this time is likely an exaggeration. The report showed the unemployment rate inching higher to 9.8 percent—and it would have been even higher absent a measured decline in labor market participation. Those active in the labor market declined by an estimated 571,000 in August. Discouragement about job prospects is a likely explanation for at least part of this decline.

In the face of such a weak labor market, it is interesting to consider the relationship between the timing of a recession's end and the peak in the unemployment rate. The following table shows the National Bureau of Economic Research's recession dates, lining those dates up with the month when the unemployment rate peaks. (Note that I exclude the 1980 recession since the unemployment rate did not peak before the 1982 recession.)

Historical lag between end of recession, unemployment rate peak, and beginning of funds rate tightening cycle
End of Recession

rate peak

Beginning of funds rate tightening cycle

Months from end of recession to unempl. peak

Months from unempl. peak to beginning of funds rate tightening cycle

Nov 2001

Jun 2003

Jul 2004



Mar 1991

Jun 1992

Feb 1994



Nov 1982

Dec 1982

Jun 1983



(Jul 1980)





Mar 1975

May 1975

May 1976



Nov 1970

Aug 1971* Mar 1972



*Following the 1970 recession, the unemployment rate was 6.1 in December 1970 and again in August 1971. If the December 1970 peak is used, months from end of recession to unemployment peak is 1 and months from unemployment peak to beginning of funds rate tightening cycle is 15.
Source: Bureau of Labor Statistics, National Bureau of Economic Research, and Federal Reserve Board

From the table, it is clear that there is considerable variation in how long it takes for the unemployment rate to move lower after the economy enters recovery mode. The typical explanation for the lag is that, as the economy shows signs of improvement, more people enter the labor force. These additional people raise the denominator in the unemployment rate calculation that often more than offset actual declines in unemployment (more on the labor force from Calculated Risk).

The past two recessions stand out as cases where the months from the end of the recession before the unemployment rate peak were very long—more than a year.

Of related interest is the historical relationship between the peak in unemployment and the beginning of a policy tightening. (Tim Duy discusses prospects for Fed policy and contrasts some recent economic data with recent Fed commentary.) Of course, the Fed does not base policy solely on movements in the unemployment rate, and the always useful advice to not casually extrapolate future decisions based on the past is even more important given the unusual circumstances of this recession. Nonetheless, historically, the beginning of a tightening cycle has lagged behind the peak in the unemployment rate by many months. This pattern is true for expansions when the FOMC was felt to have done a poor job in managing inflation, such as the post-1975 period, and it is equally true for periods when the Fed is believed to have done a very good job of managing inflation, such as the post-1991 episode.

Let me be clear that I am not offering a forecast but instead a reminder that the dynamics of unemployment do not always follow the dynamics of recessions. And for what it's worth, Federal Reserve policy has not historically responded immediately to declines in the unemployment rate—for both better and worse.

By John Robertson, a vice president in the Atlanta Fed's research department

October 2, 2009 in Data Releases , Labor Markets , Monetary Policy | Permalink


TrackBack URL for this entry:

Listed below are links to blogs that reference Economic troughs, changes in the unemployment rate, and fed policy :


John - a useful perspective and helpful for looking at the future. There's a chart running around that looks at comparisons of Unemployment in various recessions (fyi - it was first concieved by CR but rapidly percolated until the NYT blog was getting credit). By eyeball it looks like every one fits into a similar genus of curve but two highly differentiated species. Short and deep vs shallow and long. One with sufficient math skills would be tempted to create and estimate a set of parametric curves.
This one is "shallow" and very long but shallow is already deeper then the prior deeps. All the curves are symmetrical as well, self-similar. So if unemployment peaks around 10% or better in the Spring when it peaks that'll mean it lasted for ~27 months. That puts positive growth around Q312. Now that's optimistic in the sense that we get growth sufficiently far north of 2.5% in real GDP at some point, which on the odds is unlikely. Which might push off positive employment growth even farther.
One could then speculate about Fed policy, credit conditions, troubled loans and sustained weakness in the banking sector; particularly if one were concerned with monetary policy and the soundness of the banking system and noted that very few seem to be contemplating these scenarios.
FWIW - some of these using some simple graphics just got considered in a recent post:

Posted by: dblwyo | October 03, 2009 at 07:39 AM

NB: forgot so let me add that the recent OMB mid-session update seems to mirror these arguments almost exactly.

Posted by: dblwyo | October 03, 2009 at 07:41 AM

the employment data were no surprise, especially the benchmark revision 00 for anyone who has been paying attention to the birth/death adjustment. in addition, the key variable is cap utilization not employment. you can read all my remarks at econmkts.blogspot.com


Posted by: steven blitz | October 03, 2009 at 02:07 PM

Dear John, NFP data were disappointing but in any case they show an improvement of the job market that should produce a first positive reading in the first quarter of 2010. Market expectations of a below 200k NFP reading were biased by an exceptionally good result in August. If September had confirmed August, we would have probably had the first positive reading in October, in sharp contrast with the lead-lag data that you mentioned.
I would like also to add an observation about monetary policy. Today, monetary policy anticipated (with the minimum in fed funds rate) the end of the recession by many months. That was not true for some of the previous crises that you mentioned. Between the end of recession and the peak in unemployment in the last two recessions (2001 and 1991) interest rates were lowered by c.a. another 2% according to your timing. This change in monetary policy during the loosening phase means in my opinion that we don't have to take for granted another 36 months of zero rates. Much more likely a faster exit like the ones of the comparable big crises of the '70s and '80s.

Posted by: exp | October 04, 2009 at 07:14 AM

Oh, what is the cause & the effect? In the area of economics, those who believe that history is certain to repeat, are certain to error.

Posted by: flow5 | October 10, 2009 at 01:30 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