Tuesday, January 03, 2006

Fooled by seasonality 

I read Mike Meyers' piece on Minnesota jobs while waiting for a haircut appointment yesterday, and I thought to myself it didn't sound right, or at all like our most recent observations on St. Cloud. Yet there are two state economists I respect highly quoted as saying the jobs outlook has looked dismal.

Since July, the state has created a not-so-grand total of 1,100 jobs -- a number that is all but imperceptible in an economy where more than 2.7 million people are employed.

At the rate the job market has been going, the state would be lucky to add 3,000 jobs for this whole year -- a number that state forecasters have been hoping Minnesota could add every month in 2006.

"There's no growth anywhere across the board," said Minnesota state economist Tom Stinson. "This is a concern because if we don't start seeing signs of a little stronger growth in employment, that could lead us to bring down the [economic] forecast in February."

..."The last few months certainly have raised concerns about the vigor and longevity of the current expansion," said Steve Hine, labor market research director at the Minnesota Department of Employment and Economic Development.
I wondered if I remembered the data correctly, so I looked it up. Without waiting for the December number, year-over-year growth in nonfarm employment in Minnesota was a +1.3%, exactly the number Stinson's group forecasts for 2006. So why so gloomy? They say it is because all the growth happened in the first half of the year.

And that's when my light flipped on. The graph draws each year as a separate line, since 2001. That uplift in the first of each year is a normal seasonal effect, and as you can see there's a drop every July. If you pick the end of June as your reference point you will always get that movement, because it's a normal seasonal effect. I'm more used to seeing it in June figures up here in St. Cloud, because that's the month the state lets a number of its seasonal employees go. They come back in October and November. The July drop is almost all local employees, which I think are mostly teachers. About 30,000 go off the payrolls each July.

Suppose we take government out and look only at private sector employment? A very similar pattern will emerge.We will still see the increase in springtime as weather improves -- you cannot get away from weather effects on an economy, even if it has less than 10% employment in manufacturing. In summer, leisure and hospitality employment rise substantially, accounting for perhaps 25,000 jobs across the state.

As you can see by the 2005 line (the top one) we are up for 2005, by 1.4% over 2004 at this time, in private nonfarm employment. We will find a normal seasonal effect as Christmastime hires in retail are laid off this month (it's about 7,000-8,000 jobs -- we get a good bit of that up here in Cloudytown too), which may cause those alarms to go off in St. Paul. They shouldn't. There's sufficient data now to do some seasonal adjustment of the figures -- though I don't necessarily blame Hine and DEED for not doing that yet, since the patterns may change and most government statisticians are leery about using Census seasonality methods on data with a short lifespan (most of the industry-level state employment data start only in 1990.)

My basic message here is that the data that Meyers is suggesting says there is a real slow down in the economy is tainted by seasonal patterns and that in a normal year would see numbers similar to that expressed in the article. I would quibble as well with the notion that we need 3000 jobs as some sort of steady-state calculation. As I noted with the national data last week, an aging baby boomer cohort may be exiting the labor force at such rates, and young enough, that labor force participation is naturally falling. Those increased exits would decrease the net gain in employment. The one thing that argues against that in my view is that productivity gains are still quite large. You would think experienced workers leaving at faster rates would lower productivity, wouldn't you?

Categories: ,