Selected in the top 100 Economics Sites

Follow me on Twitter

Friday, May 22, 2009

Dr. Rothman hits the bull's eye.

My colleague and friend Phil Rothman has started forecasting the weekly unemployment insurance claims numbers at the web site below. As he explains, the last six recessions have ended within 8 weeks of this series reaching its peak. This is a site worth keeping a close eye on...if this relation holds up, you'll know the recession has bottomed out and recovery has started here first.


John Seater said...

Several people recently have become excited about the unemployment insurance claims series as a business cycle indicator. This is not my area of expertise, but I have a question. As Phil Rothman notes, the initial unemployment insurance claims series is part of the Index of Leading Indicators. There are nine other series included in the Index. The Index is constructed as a weighted average of the 10 constituent series, with weights chosen by statistical methods that determine the relative information content of each series. So my question is, Why are we suddenly singling one out for special attention?

Phil Rothman said...

Good question. I was strongly influenced by what Jim Hamilton has written, and while I can’t speak for him, it seems to me that he found interesting/intriguing Robert Gordon’s claim that: "Recently I have discovered a surprisingly tight historical relationship in past US recessions between the cyclical peak in new claims for unemployment insurance (measured as a four-week moving average) and the subsequent NBER trough" (quote from As Gordon also notes, some may find this to be "a will o' the wisp." Further, we know very well that a “tight historical relationship” is no guarantee that the future will be like the past, but lots of attention is frequently directed at such "relationships."

As to “why this series and not any other of the ILI series,” I don’t know what led Gordon to focus on initial ui claims. My hunch is that it was the result of an extensive search through lots of data.

FYI, I've also started providing forecasts for some other business cycle indicators: