Corrections to Current Population Survey Estimates for January through July 2020

On September 23, 2020, BLS corrected data errors in the Current Population Survey (CPS or household survey). These errors were associated with the introduction of a new occupation classification system in January 2020. The errors did not affect most major series, including the official unemployment rate. The impact was negligible for the vast majority of the corrected series. The error primarily affected a small number of specific occupations and the aggregate occupation groups associated with those detailed occupations. At the major and intermediate occupation group level, the changes were minor relative to the large size of the categories. (See the summary tables of the impact for each occupation group in terms of levels and percent change.)

New industry and occupational classification systems were introduced into the CPS beginning with data for January 2020. A new edit check was added that looks at the combination of occupation and educational attainment for specific occupations. If the reported educational attainment does not support the occupation claimed, the occupation is edited to an occupation that is in line with the education. For example, someone who says that they are a physician but also reports they have completed some college but not a bachelor’s degree will be reassigned to a medical care occupation that does not require a professional degree. Thus, a person without a medical degree would not be included in the radiologists occupation group but would instead be included in the radiologic technologists and technicians group.

This education-occupation edit included programming errors that resulted in incorrect data for January 2020 through July 2020. As mentioned previously, changes at the major and intermediate occupation group level were very minor. Many of the changes occurred within aggregate occupation groups and did not involve relocating workers to other aggregate categories.

At the same time as the new classification systems and education-occupation edits checks were introduced, another edit check was inadvertently discontinued. This edit—which ensures that people who report working for the postal service are coded as federal employees, rather than as private wage and salary workers or as self-employed—has been restored and is incorporated in the revised data.

Because of the interaction between the corrected occupation edit, other industry and class of worker edits, and the allocation methods used in constructing CPS estimates, all estimates by occupation, industry, and class of worker were subject to change. However, most series either did not change or had very minor changes.

The Employment Situation news releases for January 2020 through August 2020 have been reissued with tables containing corrected data for January through July. (See archive.) The Usual Weekly Earnings news releases for the first and second quarter of 2020 have also been reissued. (See archive.) Data series in two BLS databases (LN and LE) were corrected.

Last Modified Date: September 22, 2020