Is There Any Good News in Recent Jobless Claims Numbers?

Last Thursday, the Jobless Claims numbers released by the Department of Labor were pretty grim. The weekly Initial Jobless Claims were 467,000 representing 0.35% of those workers covered by unemployment compensation (i.e., insured work force). These numbers are down a bit from those reached two weeks earlier (586,000 and 0.44%), which were the worst since 1992, late in the administration of President George H. W. Bush. The continuing unemployment claims came in at 4,611,000, the worst since 1982 during the second year of the administration of President Ronald Reagan. This number represents 3.4% of the insured work force. This is the worst since the fall of 1983, when this value had been dropping since a peak of 5.4% late in 1982.

So, the initial jobless claims have dropped for two weeks, while the continuing claims are still rising. Is there any room for optimism in these numbers? Based on recent historical patterns, is there any reason to believe that the current recession, now 14 months old, is at or nearing an end?

Let’s look at the initial claims first. Is a drop for two straight weeks indicative of the end of the recession? No. In fact, this has already happened six times in this recession. In fact, there is so much “noise” (short term up and down movement) in this number that any consistent change of even three or four weeks isn’t really a signal of anything. At the end of the recessions that have occurred since 1970, there is still a great deal of noise in the individual weekly numbers. This is why this number is often presented as a four-week running average. This has the effect of smoothing out some of this week-to-week volatility.

This four-week running average has now declined two weeks in a row. Is this good news? Not by itself – this has already happened seven times in this recession alone. In fact, we have seen four straight weeks of decline in this number early during the current recession.

What has happened historically in this value? The pattern has not been consistent. At the end of the last five recessions, the longest “runs” of consecutive weeks where the four-week running average of weekly initial claims declined were four weeks (in 2001) and six weeks (in 1970). During recent recessions, there have been runs of five, six and seven weeks that did not occur at the end of those downturns.

So, even a consistent decline in this four-week running average of initial claims doesn’t seem to provide much information about the end of recessions. Perhaps we need yet more “smoothing” of the data? Let’s look changes in this four-week running average over more than a one week period. I examined the historical patterns of the past six recessions (those since 1970) and found that only by looking at the eight-week change in four-week running average of initial claims was there a pattern at the end of recessions that was not found within them.

That was quite a mouthful! Let me explain what I did in a bit more detail. I created a data set of the four-week running average of the weekly initial jobless claims. I then looked at the change in this value each week compared to the value from a point in the past. Finally, I looked to see if there were consistent periods in which this difference was declining, and that these periods were longer at the end of recessions than at any time during but NOT at the end of recessions.

The critical time period over which to examine the change in four-week running average of initial claims turned out to be eight weeks. That is, lets’ look at the four-week running average now compared to what it was eight weeks ago. Let’s call this the 8-4 number. At the end of the last six recessions, there are consecutive periods where the 8-4 number declined at least 17 weeks in a row – the average was 21 weeks. During, but no at the end of, any of these recessions, the longest consecutive period where the 8-4 number declined was 11 weeks – usually, it occurred no more than five weeks in a row.

This means that we need to see the 8-4 number decline every week for about three or four months. So far, it has not declined even one week. To get the first decline in the 8-4 number, we’ll need to see initial jobless claims this Thursday of no greater than 483,000. Then we’ll need to see it continue to decline for three or four months.

Next, I’ll look to see what kind of “signals” might be deduced from the continuing claims numbers.