We are unable to display your institutional affiliation without JavaScript turned on.
Browse Book and Journal Content on Project MUSE

Find using OpenURL

Comments and Discussion

From: Brookings Papers on Economic Activity
Fall 2012
pp. 118-131 | 10.1353/eca.2012.0024

In lieu of an abstract, here is a brief excerpt of the content:

Comment by jan Hatzius and Sven Jari Stehn

We like the focus of this paper by Regis Barnichon and Christopher Nekarda, because the unemployment rate is our "desert island" U.S. economic indicator, the one we would choose if we had to choose only one indicator to provide information about the economy. One reason is its value for estimating the amount of slack in the economy. As Federal Reserve Chairman Ben Bernanke noted in his August 2012 speech at Jackson Hole, "following every previous U.S. recession since World War II, the unemployment rate has returned close to its pre-recession level." Although there are certainly more sophisticated ways to measure the output gap, one could do worse than start from the observation that the unemployment rate in the fall of 2012 is about 2 percentage points above its postwar average of 5.8 percent.

But an equally important reason why we view the unemployment rate as so important is its value in diagnosing recessions. This is one of the most important, yet most daunting, jobs for business forecasters such as ourselves. On the one hand, we are all painfully aware that we as a profession have a terrible track record in recognizing recessions when they occur—not to mention forecasting them successfully. On the other hand, we also know that nothing destroys one's reputation as reliably as crying wolf about a recession that then fails to materialize. And of course, that is the reason why forecasters tend to be late in calling recessions. Being human, we want to be reasonably sure before we stick our necks out.

Thus, any tool that allows forecasters to get a leg up in forecasting or diagnosing a recession is highly valuable, and one of our favorite tools is the plain old unemployment rate. Over the 64 years since monthly unemployment data first became available, the (unrounded) unemployment rate has never risen more than 0.35 percentage point on a 3-month moving-average basis without being followed or accompanied by a recession (see our figure 1). We call this the "0.35 rule."

Click for larger view
Figure 1. 

Unemployment Rate and the 0.35 Rule, 1948-2012a

Often this 0.35 rule has done a better job than actual human forecasters. Our table 1 compares the timing of the 0.35 rule with the month in which the median participant in the Philadelphia Federal Reserve Bank's Survey of Professional Forecasters (SPF) first predicted a recession, and the month when the Federal Reserve Board staff did so. At least in the last three business cycles, the 0.35 rule signaled a recession before either of these predicted it.

Note that in order to provide a fair comparison, we use the reporting month for the 0.35 rule, not the reference month. For example, the 0.35 threshold was reached in the December 2007 employment report, which was released at 8.30 am on Friday, January 4, 2008. And as it happens, December 2007 was ultimately chosen by the Business Cycle Dating Committee of the National Bureau of Economic Research as the peak of the cycle.

Given the usefulness of the unemployment rate, anything that allows one to do a better job in forecasting the unemployment rate could be very useful. This is where the Barnichon and Nekarda study comes in. They show that labor market flows hold a significant amount of information about the future of the unemployment rate. They also show that their approach results in smaller mean squared forecast errors than either the real-time forecasts of the private business economists surveyed by the SPF or those of the Federal Reserve staff.

Click for larger view
Table 1. 

Business Cycle Peaks, Actual and Predicted, 1948-2012

Given this success, a natural question is whether this general approach might help in improving and refining the 0.35 rule. At first glance, the chances look quite promising. Our figure 2 plots the actual unemployment rate against Barnichon and Nekarda's estimated flow-consistent rate through all 11 postwar recessions. Each series is indexed to zero 4 months before the start of the recession (the month when, on average, the trough of...

You must be logged in through an institution that subscribes to this journal or book to access the full text.


Shibboleth authentication is only available to registered institutions.

Project MUSE

For subscribing associations only.