It's hardly surprising, then, that no economic report is more keenly watched than the Labor Dept.'s monthly employment data. And that explains why December's news, showing a measly 1,000 increase in payrolls, downward revisions to jobs in October and November, and a suspect drop in the December unemployment rate, has created such a stir. The data seemed at odds with other recent trends and economists' expectations that payrolls would rise by a sturdy 150,000 last month.
But are things really as bad as the latest employment data imply? Probably not. The fact is, the recent payroll data are giving the wrong impression of the strength of the labor markets. It has happened before, especially during the early stages of the recovery from the 1990-91 recession. Back then, the payroll numbers significantly understated how fast the job markets were recovering, a fact underscored by subsequent revisions. "To be blunt, we do not believe these data," says Brian S. Wesbury, an economist at Griffin, Kubik, Stephens & Thompson Inc. Indeed, almost every other labor-market indicator points to bigger and broader improvement.
Start with the declining volume of both new and continuing unemployment claims, which have dropped to levels consistent with solid monthly job gains. That's also true for surveys of the nation's purchasing managers, which show substantially improved hiring across a wider range of industries in both manufacturing and nonmanufacturing. Soundings from small businesses are especially upbeat, echoing very strong hiring plans. New temp jobs, always a precursor of hiring of full-time workers, are also up sharply. Plus, looked at over a longer period and ignoring the one-month drop in the labor force in December, the decline in the unemployment rate in the second half of last year is also encouraging.
The Labor Dept.'s payroll numbers, derived from a widely followed survey of some 400,000 companies, also run counter to the economy's sizzling growth in the second half of 2003, apparently about a 6% annual rate. Yearend readings on retail sales, car buying, new orders, and production show momentum heading into 2004.The Limits of Productivity
To be sure, the U.S. labor markets are changing in response to emerging technologies and globalization. But the recession magnified the downtrend in jobs. Now, powerful business-cycle forces are driving the economy upward, and job growth will follow. In addition, only so much blame for the lack of new jobs in recent months can be heaped on improved productivity, which lessens the need for workers. Given that hours worked in the fourth quarter rose for the first time in three years, the unsustainable 7.5% annual pace of productivity in the second and third quarters of 2003 is already moderating. Businesses can't extract those kinds of productivity gains indefinitely. "If output growth remains strong and productivity growth returns to more normal levels, employment must begin to rise," noted Federal Reserve Governor Ben S. Bernanke in a recent speech.
The Labor Dept.'s other measure of job creation, which is based on a poll of some 60,000 households, may be closer to reality right now than the payroll survey. The household survey shows job growth in the second half of 2003 averaging a healthy 134,000 per month, vs. a paltry 37,000 for the payroll survey. And during the most recent four months -- even after removing self-employment and other areas from the household-employment data to make the two head counts line up conceptually -- monthly job growth in the household data has averaged a sturdy 229,000 per month, vs. just 61,000 for the payroll data. Like most economists, Neal Soss at Credit Suisse First Boston (CSR
) ordinarily gives more weight to the payroll numbers because it is a larger survey and, after all revisions are in, it tends to track job trends better. "But given all the other evidence, it's possible -- and we are inclined to believe -- that the payroll numbers are understating job growth," says Soss.
Labor's household survey, which also collects the unemployment data, shows improvement on the jobless front as well. Since the jobless rate peaked at 6.3% in June of last year, the data show an 800,000 rise in employment and an 800,000 drop in unemployment. During that same period, the labor force, comprised of both employed and unemployed workers, held about steady, implying that the decline in the unemployment rate, to 5.7% by December, was mostly genuine and not the result of people dropping out of the labor force.Odd Discrepancies
Some of the payroll data just don't make sense. Consumer spending is expected to have grown 3.8% last year, the strongest year since 2000, yet the data say retail payrolls are down from a year ago. Historically, when consumers have bought at that clip, retailers have added jobs. In the past, capital spending and hiring have been highly correlated. But while business outlays for equipment last year posted an estimated gain of 9%, also the strongest increase in three years, manufacturing payrolls in both basic machinery and high-tech gear are down from a year ago. Indeed, production of high-tech equipment, such as computers, peripherals, telecommunications gear, and semiconductors, is up 27.5% from a year ago. Productivity gains can't account for all of these discrepancies.
One big problem is that the payroll number has a history of miscounting jobs when the economy is at a turning point. In business upswings, the Labor Dept. misses a lot of hiring at small businesses and startups. That happened in the early '90s. Since then, Labor has improved its sampling methods to better capture these jobs, but officials concede that accounting for new businesses remains "problematic." So some hiring may still be falling through the statistical cracks. The National Federation of Independent Business, which represents some 600,000 small businesses, says hiring plans are the strongest since the boom times of 1999, with positive readings in every major industry category. And in December, the NFIB says average new jobs per company were double the November level.
Eventually, the Labor Dept.'s revision process, which can take years, catches up to reality. But the miss can be enormous. In the second year of recovery from the 1990-91 recession, for example, payrolls were said to have increased by just 300,000. After more complete information came in, especially from state records and tax reports, the data had been revised to show that 1.6 million jobs were added. Even by the third year of that recovery -- the point the U.S. economy is at in this one -- job growth was originally estimated at 2.2 million. Today, the revised number stands at 3.2 million.Convergence
Over the long haul, the two employment measures tend to show the same trends. The payroll numbers will start to be revised in early February, when the data get updated. Another revision will come in the fall, and a reworking of the household data is due in March. The household job numbers may be overstated because the Labor Dept. has to estimate immigration in its total. However, new border restrictions after September 11, 2001, may mean that Labor's estimates are too high. So the truth probably lies somewhere between the two job tallies.
Still, most analysts believe that the payroll data will eventually look stronger, as they did with the early '90s recovery. But more important, even ignoring potential revisions, other labor-market trends already strongly imply that the payroll numbers will start to look a lot healthier this spring. By James C. Cooper