Phantom jobs and job losses

Public Interest, Wntr, 2005 by Tim Kane

What do all these percentages mean in terms of real numbers of jobs? In August 2004, in the same paper in which the BLS admitted that payroll double counting was a problem, it released a new payroll employment time series that tried to correct for this methodological flaw. To give just one example, in March 2001 the BLS had estimated the total number of jobs to be 132.51 million. But after subtracting its estimate of double-counted job-changers, the BLS now says the true figure was 130.85 million.

The change does not represent nearly two million jobs that we now know were lost. Rather, it signifies the purging from the record of nearly two million jobs that never existed to begin with, that appeared only as the result of a statistical fluke. And there may be more such "phantoms." It is worth noting that these phantoms are still included in official payroll employment estimates every month. Even though its acknowledgment of sample problems is a vital first step, the BLS still uses overly cautious assumptions when trying to correct the error, presuming that only a fraction of job-changers are double counted.

Companies track their payrolls on several schedules, the most common options being monthly, biweekly, or weekly. Therefore, on the payroll survey questionnaire the BLS phrases the question in terms of how many people have been on the payroll during the company's reference period. For monthly, or even biweekly, tallies it is easy to see how double-counting would occur--so easy to see that even the BLS is now conceding the point. But the bureau still holds that a worker who leaves a weekly payroll job for another weekly one cannot be counted twice. This neglects the use of "terminal leave," when an employee overlaps two paychecks. Such real-world complications are likely to amplify the turnover effect.

This statistical evidence is made all the more compelling by the existence of an obvious theoretical explanation for the sudden, and recent, drop in turnover: terrorism. On reflection, it is hard to imagine that the September 11 attacks are not the main rationale driving lower job turnover. In the aftermath of the attacks on the World Trade Center and Pentagon, conventional wisdom held that Americans would re-orient their priorities toward family and away from work, and seek security wherever they could find it. They would see the big picture about what mattered in their personal lives, and act accordingly. "Cocooning" is the phrase market researchers coined to describes this tendency for families to stay closer to home for dining, entertaining, and working, and all these trends were developing before September 11, but accelerated after the attacks. The hospitality industry can certainly attest to this, having witnessed a precipitous decline in leisure (and even business) travel in the wake of September 11. It does not take a great leap of imagination to suppose that the impact of terrorism concerns would extend from questions of where to vacation into questions of whether or not to change jobs. If the overall trend has been for people to display more risk aversion, we should not be surprised that they are not changing jobs as frequently as they did in the go-go 1990s, thus preventing the BLS from over-counting them in the payroll survey.