Reducing Sample Sizes in the National Compensation Survey in Response to Budget Cuts

Christopher J. Guciardo, Lawrence R. Ernst, Gwyn R. Ferguson, Yoel Izsak, and Erin E. McNulty


Recently, the budget for the National Compensation Survey was reduced, leading to a reduction in the number of sampled establishments. Reductions were made to updates, initiations, and units from future samples. There are 3 phases of sampling: localities, establishments, and jobs. The establishment-sample has 2 parts: a "wage sample" for locality-wage publications, and an "Index subsample" for the Employment Cost Index (ECI) and other publications that use benefits data. Reductions were made to the private industry wage sample sizes, with some constraints. First, since the ECI is a principal federal economic indicator, all reductions were designed to preserve Index sample sizes, and most reductions involved subsamples of "wage-only" samples (wage samples less Index units). Second, reductions mostly were made only to those localities for which separate estimates are needed for federal pay setting purposes, since the allocations for these localities had been previously supplemented. Several scenarios are discussed: most attempt to eliminate all or part of this supplement. The allocation, subsampling, and weighting adjustment methodologies for the chosen scenario are detailed.