The Effect of Some Design and Estimation Issues on the Variance Estimates of the Employment Cost Index

Steven P. Paben


The Employment Cost Index (ECI) is a measure of the change in the cost of labor, free from the influence of employment shifts among occupations and industries. The calculation of the quarterly change in the ECI involves the multiplication of the previous quarter's cost weight in each industry-major occupational group cell by the estimate of the quarterly change to obtain a current cost weight for the cell. The cost weights themselves are therefore variable, and the variability generally is an increasing function of the number of quarters from the base period. The impact on the variance of the variability of the cost weights is examined. The variance of an alternative calculation of the ECI that is less affected by the variability of cost weights is also examined. Additionally, the ECI is currently in transition from a national-based sample to a geographic area-based sample. The impact on the variance of this design change, which adds an additional level of sampling while increasing the number of sample establishments is studied. Finally, the effect of using Fay's method of BRR rather than the standard BRR for operational advantages is investigated.