Weekly Jobless Claims

The weekly jobless claims will be coming out shortly. Last week, a sell side analysts put out a note saying that jobless claims might be less than expected because of seasonal adjustments. Many seasonal workers are laid off after Christmas every year, so the government seasonally adjusts the number downwards. The analyst surmises that since there was less seasonal hiring this year there will be less seasonal firings and this week's number should come in better than expected.

This logic seems sound to me although I am not familiar with the data. Therefore I would take any improvement in the reported numbers with an even larger grain of salt than other government figures.


Market Owl said...

Here is what Briefing has to say aboutit:

ECONX Initial Claims Preview

For two consecutive weeks, the initial claims broke through the 450,000 new claims barrier and are now holding steady at roughly 435,000 claims. During those same two weeks, most economists believed claims would hold at a slightly higher pace. However, they have now adjusted their bias and expect claims to hold at 436,000 for the third consecutive week. We feel slightly differently. During those two weeks that the initial claims figures fell below 450,000, there were two major holidays -- Christmas and New Year?s -- that wreak havoc on the normal statistical adjustments. Given that nonfarm payrolls dropped by 85,000 in December, even though new initial claims fell by 42,000 during the month, we believe that the statistical discrepancies brought initial claims down to 435,000. Now that the statistical bias has gone away this week, we expect initial claims to rise back to the previous level of 450,000. Continuing claims may have also suffered from the same statistical problems, which should continue in this week?s data. The consensus expects continuing claims to fall from 4.802 million to 4.750 million for the week ending January 2.

PJ said...

Actually, the reason is that the government uses a multiplicative factor for seasonal adjustments. This week the factor is 1.806.

In fact, a combination of a smaller multiplicative factor plus an additive factor would work better, but they keep it simple.

So, the simple multiplicative factor exaggerates adjustments when job losses are high and underestimates them when job losses are low.

In 2008, NSA number was 547,000, the SA factor was 1.746, and so the SA number was 547,000/1.746 = 314,000, for a removal of 233,000 jobs.

In 2009, NSA number was 956,791, the SA factor was 1.79, and so the SA number was 956,791/1.79 = 535,000, for removal of 421,791 jobs.

In 2010, like 2009, we'll remove a large number of jobs in SA.

PJ said...

Very disappointing report. Not only were first time claims much higher than expected, given weak seasonal hiring, but also hiring off the unemployment rolls was the weakest of the whole recession. Maybe people just didn't want to look for work over the holidays ... or maybe we're heading for a double-dip.

Combine that with the weak retail sales, and the economy looks very fragile.