Abstract

Julie Gershunskaya and Alan H. Dorfman " Calibration and Evaluation of Generalized Variance Functions"

The work of this paper is prompted by the particular case of the Current Employment Statistics (CES) Survey conducted monthly by the U.S. Bureau of Labor Statistics. Besides estimates at the national level, the survey yields estimates of employment for numerous domains defined by intersection of industry and geography, providing important information about the current status of the local economy. Variances of the employment estimates are estimated from the sample. However, the sample based estimated variances can be unstable, especially in smaller domains. More stable variance estimates can be obtained using a model-based generalized variance function (GVF). The modeling is based on past years of the survey and, assuming a satisfactory model fit, the result can be applied to predict variances for the current period. However, some features of the design or population characteristics may change from one year to another, making it necessary to adjust the model parameters. We here give a method for evaluating the suitability to current data of a GVF model based on past years' data and suggest ways to calibrate the GVF to the current data.