The CES surveys around 140,000 businesses each month. But it does not capture those who are self-employed, working under the table, or on unpaid leave during the sample period. An individual with more than one job will show up on multiple payrolls. An individual who works two part time jobs, but gives up one to work full time would show up as a net loss of one payroll job.
An issue arises from the fact that businesses are created and fail all the time. Without proper accounting of business creation and failure, the CES's sample will become less representative over time. The BLS attempts to correct for this in its benchmarking process, where they use more accurate UI tax data (available bi-annually) to adjust the sample. They also use econometric models to predict business creation and failure to adjust the sample between benchmarks.
The CPS surveys 60,000 households each month to estimate the size of the labor force, employment, weekly hours, earnings, and demographic information. The labor force is defined as individuals who are employed or unemployed – those who are not employed but have actively looked for a job within the past month.. Those who are self-employed, farm workers, or unpaid workers in a family business are included in the CPS. The CPS has a smaller sample size than the CES, and is more volatile from month to month.
Due to its larger sample size and lower volatility, the CES is often considered to be the more accurate source. But the relative usefulness of the two sources depends on what data is needed, as well as when and for what time period. Much of the relative monthly volatility of the CPS is removed by averaging over a quarter or year. And less volatility is not necessarily a sign of greater accuracy. Both series often go through large revisions.
No comments:
Post a Comment