Beware analysts torturing jobs data to fit model
NEW YORK — Traders, take your mark. Go! The monthly employment stakes has begun.
At precisely 8:30 a.m. Eastern Standard Time Friday, the Bureau of Labor Statistics will release the Employment Situation report for February. (Note: It is not the “unemployment report,” a misnomer favored by almost every TV and radio commentator.) Then the real fun begins: dissecting the numbers and spinning them to reflect one’s political biases or entrenched view on the U.S. economy. In other words, attempting to make reality conform to the model.
How much of the job creation is real? How much of it is the result of the unseasonably warm winter weather, which boosts seasonally adjusted employment in industries such as construction, mining and real estate? (Remember, what the seasonal adjustment factors add in one month, they take away in others.) What about the birth-death model, which uses sample- based estimates to capture new business formation and the creative destruction of others?
Last month, the BLS incorporated new data from the 2010 census into its January statistics, which created a break in the series and a headache for anyone trying to fend off the conspiracy theorists.
Those arguing that recent job gains — an average monthly increase of 185,000 in private payrolls in the past year — are statistical anomalies have an increasingly wobbly leg to stand on. Start with the plunge in weekly jobless claims, a tally of real people lining up at real state unemployment offices to file real claims. The Labor Department publishes both unadjusted and seasonally adjusted numbers. Take your pick. Both measures are near a four-year low, reflecting the improvement in the labor market.
Then there’s the small matter of income earned from employment. When the Bureau of Economic Analysis took a second look at fourth-quarter gross domestic product last week, it found an additional $82.4 billion of wage and salary disbursements in the third quarter for a total increase of $107.2 billion. Wages and salaries are compensation for work performed, not phantom jobs.
For its revisions, the BEA uses the BLS’s quarterly census of employment and wages. The QCEW, which will be released to the public on March 28, is a comprehensive count of jobs and wages derived from tax reports submitted by every employer subject to unemployment-insurance laws. It covers 99.7 percent of civilian employers, according to the BLS.
Additional income in the third quarter is no guarantee of continued increases, but the message is one of a gradually improving labor market.
Finally, there’s the geeky-but-hot question of the labor- force participation rate, which peaked at 67.3 percent in 2000 and declined steadily to 63.7 percent in January. (What appears to be a sharp, 0.3 percentage-point December-January drop is the result of applying new population controls to January’s numbers.)
Just as the mass influx of women into the labor force beginning in the mid-1960s boosted the participation rate, so is the aging and retirement of the baby boomers causing a structural decline. The last recession clearly hastened the process — or perhaps the housing bubble masked the decline, according to a March 1 analysis by economists at Barclays Capital.
To the extent that the labor-force participation rate is mainly a function of age distribution, the aging population means more people moving from a high participation age bracket (45-55 years old) to a lower one (55 and over). Barclays economists dispute the conventional wisdom that the influx of jobless workers to the labor force as the economy improves will prevent further declines in the unemployment rate. Using 45 years of available data, they failed to turn up a single period when the jobless rate was driven mainly by re-entrants, rather than job losers.
Among the skeptics — not of the data per se but of their interrelationship — is Federal Reserve Chairman Ben Bernanke. Last week, he told House and Senate oversight committees that the decline in the unemployment rate this past year from 9.1 percent to 8.3 percent is “somewhat more rapid than might have been expected, given that the economy appears to have been growing during that time frame at or below its longer-term trend.”
No one would take issue with Bernanke’s observation that 1.6 percent real GDP growth in 2011 is below trend. But given the sluggish-to-nonexistent labor-force growth since 2008, it doesn’t take much to reduce the unemployment rate, as demonstrated by a new jobs calculator created by the Atlanta Fed.
Assuming a labor-force participation rate of 63.7 percent and monthly population growth of 0.07 percent, the past year’s average rate, it would take monthly payroll growth of 242,558 to reduce the unemployment rate to 7 percent in 12 months. Nonfarm payrolls rose 243,000 in January. And yes, some of that increase was due to unseasonably warm weather, for which there will be a giveback this spring. Still, 243,000 net new jobs a month isn’t out of the realm of possibilities.
All it would take is 152,775 new jobs a month for the unemployment rate to breach 8 percent by the time voters go to the polls on Nov. 6, according to the jobs calculator.
The decline might be “somewhat more rapid” than Bernanke would have expected, but President Barack Obama would be grateful for the surprise.
Caroline Baum, author of “Just What I Said,” is a Bloomberg View columnist.