Monday, July 9, 2012

Are seasonal adjustments obsolete? Is the U.S economy not really slowing down?

Something else grabbed me from the June jobs numbers that were released last Friday, and it related to the seasonal adjustments that go into those monthly figures. Now it's assumed that "Summer jobs" will add a number of people to the job force and employment rolls, so the BLS tries to deflate this seasonal adjustment to find out if we're truly seeing growth, or if the new jobs are short-term seasonal ones that'll go away with the Fall and the start of the school year.

On Page 28 of the jobs report, you see the non-adjusted and seasonally-adjusted jobs numbers, and it's quite the contrast in gains.

Jobs added May and June 2012
May non-seasonally adjust.- 780,000, seasonally-adjust.- 77,000
June non-seasonally adjust.- 391,000, sesonally-adjust.- 80,000

So even though 1.17 million more people are working, the seasonal adjustment assumes that over 1 million of those jobs are because of Summer Work, and therefore, it only counts 157,000 as growth. The converse is true in the winter months of December and January- jobs are traditionally shut down in some manner for these months, so when fewer are lost than normal, it translates into a "gain" of jobs.

Job change December 2011 and January 2012
Dec. 2011 non-seasonally adjust.- -207,000, seasonally-adjust. +223,000
Jan. 2012 non-seasonally adjust.- -2,668,000, seasonally-adjust. +275,000

But maybe these huge seasonal adjustments are behind the times in 2012. That's the hypothesis of the New York Times' Floyd Norris, and he says that maybe the seasonal adjustment shouldn't be as big as it is.
The reason that few doubt that the jobs picture is getting worse is that they look at the Labor Department’s seasonally adjusted figures. But those adjustments most likely overstate reality these days. Employers are acting more cautiously than they did in previous cycles. They add fewer seasonal jobs than they used to, and they therefore get rid of fewer seasonal workers when the season is over.

The result is that the seasonal adjustments make things look better than they are in the winter, when fewer workers are being let go than the government expects, and worse in the spring and summer, when the workers who were not let go cannot be rehired. There is, of course, more than seasonal adjustment going on, but I suspect that the underlying swings are far more modest than the monthly figures seem to indicate....

June is the month when the seasonal adjustments call for the largest upward adjustment in private sector jobs, as students find summer work. If there are fewer summer jobs than there used to be, we would expect the June seasonally adjusted numbers to be disappointing. They have been so in each of the last three years.[Ed. Note- It actually seems to be May, as I note above, but the point still stands]

If that analysis is correct, the job numbers are likely to seem poor for the next two months, but to pick up with the September report on Oct. 5, and then to look impressive in the October report, which will appear on Nov. 2, four days before the election.
Interesting theory, and past jobs reports indicate the August and September "seasonal deflators" are much smaller than the one you get in April and May. So even if the same amount of jobs get added in real-life in August and September, the seasonally-adjusted stat will look much better than the disappointing 68,000 and 77,000 figures we got for April and May. And all Republicans will be crying heavily in their cups of tea.

The same happens in Wisconsin, except on a more extreme level because of our more extreme weather. Take a look at the Winter vs. Summer months the last couple of years, and you'll notice how the jobs numbers change based on the seasonal adjustment.

Wisconsin job changes, selected months, 2010-2012
May 2010 non-season. prvt. +32,000, seasonal adjust. prvt. -7,900
June 2010 non-season. prvt. +38,300, seasonal adjust. prvt. -1,000
Jan. 2011 non-season. total -71,000, seasonal adjust. total +4,400
Feb. 2011 non-season. prvt. -6,100, seasonal adjust. prvt. +2,500
May 2011 non-season. total +36,200, seasonal adjust. total -1,900
June 2011 non-season. total +18,900, seasonal adjust. total -9,000
Jan. 2012 non-season. total -62,000, seasonal adjust. total +12,500
Feb. 2012 non-season. prvt. -2,400, seasonal adjust. prvt. +4,000
May 2012 non-season. total 39,000, seasonal adjust. total +2,600

You may remember that June 2011 number, as it was originally released as a major gain for state jobs despite warnings to the Governor's Office that the gains were all in seasonal jobs like tourism and were "very questionable." Walker held an "It's Working" press conference anyway, in no small part because it was 3 weeks before last Summer's Senate recall elections. In fact, it was a dress rehearsal for how Walker's DWD released preliminary jobs numbers before his own election this year which also turned out not to be all they were cracked up to be.

I've also mentioned before that it's interesting to note that 4 of Scott Walker's best job-gaining months by the seasonally-adjusted standards have been in January and February. These were months that actual jobs were lost, but because there were fewer cutbacks than usual, it meant that it tricked the seasonal adjustment into showing a "gain." If Floyd Norris' theory is correct, maybe there's less seasonality and that the state really wasn't gaining ground in Winter (or losing ground in Summer) as much as we thought. It also means that we may well see down jobs numbers in the next few months once that seasonal adjustment is played in (even more than the increasing unemployment claims in the state would indicate).

Sounds like we better keep our eyes peeled for more shenanigans and wordplay with next week's seasonally-adjusted Wisconsin June jobs report, shouldn't we?

And it also means that we may not be heading toward any type of nationwide jobs slowdown or recession, but maybe instead we're just moderating our seasonal hiring, and that we're remaining on a slow-but-steady recovery. Maybe we need to follow Norris' lead in the article and look at year-over-year jobs numbers (which have steadily stayed in the 1.7-2.0 million range for the U.S over the last 10 months), as that seems to give a better overview instead of seasonally-adjusted month-to-month figures.

Don't count on our regular media to have a clue, and watch for them to casually give seasonally-adjusted numbers for the next 4 months (especially if it keeps Romney close to Obama). So we'll have to work even harder to get the truth out over what we know will be numerous deceptions by the right-wing propaganda machine in Wisconsin and in the rest of the country.

No comments:

Post a Comment