The statistician John Tukey once said: "An approximate answer to the right problem is worth a good deal more than an exact answer to an approximate problem."
For the national economy, the right problem of concern these days is jobs. To deal with the problem, the answers we need require reasonably accurate employment data.
How exact or approximate are the official data on employment? The job numbers reported monthly by the U.S. Bureau of Labor Statistics (BLS) and publicized by the mass media give the superficial appearance of precision. However, a statistician would say that the data are approximate because, among other reasons, the definition of employment involves judgment and survey data are subject to sampling and non-sampling errors. Still, even without exactness, the data are sufficiently accurate to enable assessment of the job market and guide policy decisions.
At least most of the time. Sometimes it's questionable. Take the latest employment data for August.
There are two major employment series reported monthly by the BLS - payroll jobs collected by employer survey and civilian employment collected by household survey. Mostly it's the payroll jobs count that is featured in the government's press release and in the media. Nevertheless, as the BLS has stated, both series should be taken into account in assessing the employment picture.
In August, nonfarm private sector payroll jobs continued to trend gently upward, rising by a modest 67,000, seasonally adjusted. Total nonfarm jobs logged in at minus 54,000 - "little changed," as the BLS reported.
By comparison, and not mentioned in the BLS summary press release, employment of nonfarm private industries wage and salary workers as measured by the household survey soared by 891,000 last month after seasonal adjustment. A one-month fluke you might say. Not so. That series has been trending upward nicely all this year, increasing by an average 291,000 a month. Even excluding the August jump, the average rise has been more than 200,000 a month.
There's further evidence of a healthy upward trend. When household-measured employment is adjusted to be consistent with the definition of payroll jobs, the increase this year has averaged 220,000 a month.
If it's real, we've had a stronger than expected employment recovery that supports the interpretation that our high and sticky unemployment rate has been due in good part to reflation of the labor force. People who have been waiting on the sidelines for a pickup in job growth have become encouraged by the signs of expanding employment and have taken up the job hunt.
In contrast, so far this year the media-featured private sector payroll job count, as measured by employer survey, has risen on average by an anemic 95,000 a month, not enough to keep up with the normal increase in the labor force from population growth or make a dent in unemployment.
Which employment series are we to believe? Most economists tend to favor the payroll over the household count, though both series have merit. Looking down the middle of the estimates suggests that private-sector employment has been growing by about 200,000 a month this year - a respectable trend.
The reasons for periodic discrepancies between the payroll and household series are not fully understood. Explanations that favor the accuracy of the household survey are that it's more timely in picking up the new hires of startup companies, it counts some off-the-books workers excluded in the payroll count, and it's subject to smaller annual data revisions. (The latest payroll benchmark revision was a negative 930,000 compared to a population control revision to employment of minus 243,000 in the household survey.) On its side, the payroll survey has a larger sample size.
In sum, when the results of both surveys are taken into account, the employment picture becomes brighter and prospects for the rest of this year and next year look more encouraging. If even partial credence is given to the household data and employment continues to rise faster than population additions to the labor force, unemployment will decline in the months ahead. But the decline will likely be slow until more of the backlog of discouraged jobless outside the labor force is absorbed into employment. A reduction in payroll taxes and the corporate income tax would help speed things up.
Alfred Tella is a former Georgetown University research professor of economics.
Read Full Article »