Jobless claims suggest US GDP growth is peaking

Today’s commentary

(outside the paywall today)

Summary: Right now, US jobless claims are very low by recent historical standards. I believe this represents a peak for growth in this business cycle. But there are almost certainly seasonal adjustment problems with this data series. Despite problems with the jobless claims series in the US, it is a useful tool for tracking the progress of the US economy. Here’s why.

The jobless claims series is a US economic series that I have tracked closely for more than a decade through various jobs. I find it a useful real-time indicator of the strength of the US economy and of directional changes from recession to recovery or from expansion to recession. For example, in 2008, I saw both the initial and the continuing jobless claims series as sending an unambiguous recession indicator. Another example comes from April 2009, when I wrote how the initial claims series was signalling the end of recession.

The problem with the series today is two-fold. First, because of the deep downturn in employment, the continuing claims series became unusable for me for a large period of time. The rise in unemployment and job market weakness was masked by the continuing claims data as so many people were unemployed for long periods. The continuing claims series was unable to capture this as people exhausted their unemployment insurance.

Second, the depth of the downturn has made seasonal adjustment especially tricky. It is hard for statisticians to determine what the right seasonal adjustments for the data are when the  seasonal patterns are so distorted by the extreme nature of this particular business cycle. That means that seasonally-adjusted initial jobless claims numbers are not as useful on a week-to-week basis.

Nevertheless, I think the data are still useful. Increasingly I have turned to the unadjusted data for signals in order to remove the seasonal adjustment distortion and I have stuck to initial claims as an indicator as well. Below are the latest charts for the year-on-year change in unadjusted initial jobless claims and for the adjusted level in jobless claims.

Combining those charts leads to a few conclusions. First, the actual level of jobless claims on a seasonally adjusted basis is largely consistent with a cyclical peak in the economy. While jobless claims can fall further still, I believe they are unlikely to fall much further from here. As this time series does track the change in GDP growth inversely quite well, it suggests that US GDP growth will decline from here.

I am cautious about this conclusion however given the unique nature of this business cycle. Two weeks ago I went to a business conference where Vincent Reinhart, Morgan Stanley’s chief economist, made a compelling case for the US economy accelerating from here. The member post on that commentary is here. Bloomberg has a good post up today on Reinhart’s view, which is dependent on improving capital spending by business. Reinhart’s views notwithstanding, I expect GDP growth to continue to slow.

Second, the change in jobless claims at a period well into the business cycle also supports my contention that the peak in growth is here or coming soon. The unadjusted year-on-year change in the rolling average initial claims is now at -56,000. This is a mean-reverting time series that is negatively correlated to the change in GDP growth. In plain English, this means that as the rate at which people get laid off each week becomes slower, GDP growth rises. But when you compare these layoffs to the year ago period and the comparison becomes more and more favourable, you hit a ceiling. Basically, there are always frictional layoffs and so there is a lower bound to where we should expect jobless claims to go. Given this, we should expect the change in claims to mean-revert and GDP growth to slow. This is the mirror image of the data signals we saw in January 2009, when jobless claims were going to stratospheric levels.

None of this says recession, I should note. What I am talking about is a peak in growth, a peak in the economic cycle, not recession. And clearly, all of this doesn’t take into consideration what the government shutdown or a debt default might do to the US economy.