Following a 2-week plunge totaling 93k, jobless claims increased30k to 554k in the week ended July 18th, roughly as expected. Another increase is likely next week for the same reasons behind today’s snapback and the recent decline. Specifically, the widely-held view is that the plunge has been because of atypical patterns in the timing of factory shutdowns in the automobile industry. While most agree that there has been impact, the degree of impact is debated and nary is an economist prepared to put a precise figure on it.
Nevertheless, one can derive a reasonable approximation. Through the noise a peak in jobless claims is apparent—a classic sign of a recession’s end.
Data reported last week, which were for the week ended July 10th, indicated that jobless claims fell 45k to 524k following a decline of 48k the previous week (claims data for previous weeks showed no unusual pattern). There was no actual decrease in claims; the tally actually increased. Only after applying the seasonal adjustment factor did claims post a decline.
Claims typically increase on an unadjusted basis in the first two weeks of July, as did happen. The percentage increase in past years was about 40%.
Accordingly, seasonal adjustment factors assume a 40% rise (take note: seasonal factors for claims are multiplicative and not stated in absolute numbers of people), which means that the unadjusted level of about 570k seen throughout June translates into an adjustment of about 228k (570k x 40%). This is a much larger adjustment in absolute terms than was seen during the four of the past five years for which the seasonal factors for this year were derived (because in four of the past five years jobless claims were very low).
Claims for the two weeks they declined averaged 547k, which represents a decrease of 65k from June’s average. The 2-week cumulative dip of 130k (65k x 2) thus far is in keeping with the 228k adjustment that should be expected relative to trend. A further alignment occurred today with the print of 554k, which is 58k below the weekly average for June; hence most of the alignment is complete. One more week will complete it.
For the latest week and the week ahead, seasonal factors "expect" decreases in claims by 18% and 17%, respectively, which are expected to result in sharp increases in claims, as was seen today. If no meaningful increase occurs next week (say about 30k or so), a lower underlying trend will become more apparent.
If we justifiably assume that the amount of furloughed workers returning to make cars and light trucks this time around is smaller for past years when production increased we can assume that the claims tally would be even lower than has been reported. In other words, looking away from the automobile sector conditions in the labor market are probably better than the headline data suggest.
As for continuing claims, the record 591k drop reported a week ago (to 6.313 million) and the 88k reported for the latest week are the result of both the multiplicative adjustment I mentioned earlier (data are adjusted by multiplying the raw data by the percentage change that seasonal factors anticipate; the actual level increased by 64k in the latest week) and the expiration of benefits (benefits last 26 weeks).
Keep in mind that claims are reported with a 1-week lag relative to the initial claims tally. The tally in extended benefits is released with a 2-week lag.
Through the Noise
Through the noise, claims have been falling slowly ever since the 4-week moving average peaked at 659k in early April. This is obviously important given the widespread belief that a peak in claims signals recession’s end. Professor Robert Gordon, who has been a member of the cycle-setting National Bureau of Economic Research (NBER) since 1978, contends as much, concluding in his research that the 4-week moving average for claims leads by a few weeks the time stamp the NBER places on troughs for economic recessions.
Faulty seasonal adjustment is likely for the automobile-related component of the monthly jobs report. Specifically, a boost of about 100,000 or more is possible (relative to what would otherwise be reported). This figure results from adding the 50,000 workers that will be added back from seasonal adjustment factors "looking for" 50,000 workers to be furloughed from factory shutdowns that do not occur, to the 50,000 that will have actually have been called back in atypical fashion for July.
It is reasonable to assume that there are many on the non-automotive side of the equation whose jobs are affected by the automotive sector, which could add jobs to the tally and raise the odds of an upside surprise. Forecasts are leaning toward a below-trend decline.
The market response to these and other data tainted by the ups and downs of the automobile sector depends heavily upon the extent to which market participants believe any improvements in the economy stick and become self-reinforcing, as well as the extent to which there are improvements that stretch beyond both autos and the broader effort to align production with sales following steep drops in inventories. Whatever the case, a cavalcade of news showing the economy making strides will set better with risk assets than Treasuries, at least for a time.
Eventually, focus will shift back to the many constraints to growth and the under-utilization of resources, tempering the market response to data on the production end of the economy. Only if the production rebound kindles final demand via income effects will it have lasting effect. This is unlikely, as there is not much of an inventory rebound ahead and because of all of the many other known constraints to growth.
One could say that the automobile sector’s production rebound and the realignment of production with sales (production had fallen well below sales across many industries) is enough to spark a dim fire not strong enough to stand the prevailing winds.
More: Click for Latest Economic coverage ...
Tony Crescenzi is Senior VP, Strategist, Portfolio Manager Pimco.
Crescenzi makes regular appearances on financial television stations such as CNBC and Bloomberg, and is frequently quoted across the news media.