NCPA - National Center for Policy Analysis

Understanding Job Statistics

November 29, 2012

It is important to understand how job statistics are derived and what that means for public policy, says Robert McTeer, a distinguished fellow with the National Center for Policy Analysis.

To-date, the United States has only regained about half of the almost 9 million lost jobs in 2008 and 2009. Job growth has improved recently, but is still below the rate of past recoveries. Consider:

  • Payroll employment increased by 171,000 jobs in October and the unemployment rate increased to 7.9 percent, representing 12.3 million people.
  • The payroll numbers were revised up by 50,000 in August and by 34,000 in September.

Unemployment averaged below 5 percent before the recession. It peaked at 10 percent in late 2009, and declined to 7.9 percent as of October. It is still about 3 percentage points above what would reasonably be considered normal or full employment.

The situation, however, is worse than these numbers portray. Not only is unemployment high as a percent of the labor force, but the labor force itself has been shrinking as job seekers give up and drop out.

  • Indeed, labor force participation (job seekers and the employed) has declined from 66 percent before the recession to 63.8 percent.
  • Moreover, employment, as a percent of the population, has declined from over 63 percent before the recession to 58.8 percent in October.

These numbers are estimates based on samples, which make them subject to sampling error. The payroll number comes from reports submitted by 141,000 business establishments and government entities representing about 486,000 workers. Statistically, at the 90 percent confidence level, the "true" payroll change is the estimated number plus or minus 100,000. In October, that is between 71,000 and 271,000. A larger sample size would narrow that range by reducing the sampling error.

The unemployment rate is derived from a separate household survey, which has an even smaller sample size and a wider confidence range. Because the household survey's sample size is smaller than the establishment survey, its confidence range is larger. At the 90 percent confidence level, the "true" number of unemployed persons is the estimated number plus or minus 280,000, and the "true" unemployment rate is the estimated rate plus or minus 1.9 percentage points. Thus, the estimated unemployment rate of 7.9 percent in October means that the "true" rate was between 6 percent and 9.8 percent.

Source: Robert McTeer, "Understanding Job Statistics," National Center for Policy Analysis, November 29, 2012.


Browse more articles on Economic Issues