Seasonal Posting: NYTFail, Part 2

First, David Leonhardt argued that this recession was good for workers.

Now, Floyd Norris apparently has decided to mix and match data. (I wonder if the fact many NYT employees who are looking at their 45-day severance offers is having an effect on its economic coverage.)

One of the standard “economist jokes” is about the one who died because he forgot to “seasonally adjust” his pool. In that tradition, Norris declares:

The adjustments are for seasonality. For some reason, October is the month with the largest seasonal adjustment down in jobs. So the increase in the unemployment rate does not reflect people actually losing jobs. It reflects the belief that seasonal factors should have added more jobs than they did.

All this may be very reasonable, and there is no way I can think of to test whether the seasonal adjustments are reliable. [emphases mine]

Gosh, I wonder why October would have a larger seasonal adjustment, and whether there is any BLS data to support that adjustment?

Apparently, employers traditionally hire a lot of people in October for “the Holiday Season.” And while it’s possible that they will be doing all that hiring in November this year, it hasn’t been the way to bet during this millennium.

Norris continues:

But I suspect seasonal factors are less important this year, when the economy may be changing directions, than they normally are.

It was with such optimism that Napoleon went to Russia, people bought VA Linux at $100 a share, and the Bush/Cheney/Rumsfeld axis decided to run a two-front war in Afghanistan and Iraq. With statements similar to Norris’s:

In reality, the government report says unemployment rates remained steady at 9.5 percent. And the number of jobs actually rose, by 80,000. And the number of jobs for college-educated Americans rose more than in any month in the last six years.

Well, the number of jobs rose (as one would expect, given the Holiday Sales push) but Table B-1 is closer to 40,000 than 80,000:

Where we do see an 80,000 job increase is in the private sector, which is more than 500,000 workers lower than it was in August. If you want to play a non-seasonally adjusted, private-sector only game with the data, you should at least be honest about it.

More vitriol and data below the break.

The details of that 80,000 look even worse: declines in all Goods-producing areas (except about 200 new jobs in primary metals, 300 in “miscellaneous manufacturing,” and 1,100 in motor vehicles and parts; cash for clunkers, anyone?) which are balanced by the Service sector, most notably the 63,500 new Retail jobs. Can you say “seasonal employment”? Floyd Norris apparently cannot.

The rest of the Non-Seasonally Adjusted figures are even less encouraging. Table A-8 of the report shows more than 100,000 people added to “not on temporary layoff”:

while Table A-9 is depressing: a larger number of unemployed at all durations, with the median duration of unemployment increased by more than one month (in a month):

And while the BLS has not updated their Job Openings data for October, the graphic through September isn’t exactly pointing to a decline in that median (or a robust recovery):

Is there a recovery in process? Maybe, though I’m not convinced, since most of the positive data seems, as Paul Krugman noted, “unrepresentative.”

But things are not so good as Floyd Norris wants to pretend, even (or especially) using the data he chooses to highlight.