Accounting for Business Births and Deaths in CES: Bias vs. Net Birth/Death Modeling

Jurgen Kropf, Sharon Strifas, and Monica Traetow

Abstract

The Current Employment Statistics (CES) Survey, conducted monthly by the Bureau of Labor Statistics, estimates payroll employment, hours, and earnings. The survey originated as a quota-based sample, but currently is transitioning to a probability-based sample. Because business births and deaths cannot be captured in a timely manner by the CES survey, they are accounted for by modeling. The bias adjustment factor used for the quota sample is discussed and compared to the net birth/death model used for the probability sample. While the latter sample uses imputation of business deaths to account for business births, an ARIMA time series model, or the net birth/death model, estimates the residual not accounted for by the imputations. In contrast, the bias adjustment factors are derived solely from a regression-adjusted mean-error model. Although both models account for business births, the bias adjustment factor model also accounts for other elements of non-sampling error. The numbers derived from the net birth/death model are compared to the results from the bias adjustment factor model.