Small Area Estimation for Business Surveys

Small area estimation techniques are becoming increasingly used in survey applications to provide estimates for local areas of interest. These techniques combine direct survey estimates with auxiliary data to produce estimates for areas with very few sampled units. Most of the applications have been in social surveys where the areas of interest are geographical regions, with fewer applications to business surveys, although tax information is usually available as auxiliary data. Statistics Canada has been investigating small area estimation for its Survey of Employment, Payrolls and Hours (SEPH). Preliminary results have been encouraging but some issues with negative variance component estimates have been noted. In this presentation, alternative methods to estimate the variance components will be investigated and evaluated using a population generated from SEPH data. Statistics Canada’s Survey of Employment, Payrolls and Hours (SEPH) is a monthly survey designed to produce estimates of levels and month-to-month trends of payrolls, employment, paid hours and earnings. The target population is composed of all employees in Canada except for those in a few select industries (ex. agriculture, fishing and trapping, etc.). The program makes extensive use of administrative data with the aid of a monthly survey. The administrative source is the Canada Revenue Agency’s (CRA) Payroll Deduction Accounts (PD7) file, which includes the number of employees and the gross monthly payroll for the approximately one million employers in Canada. The administrative data is combined with data from the monthly Business Payroll Survey (BPS) through the use of the Generalized Regression (GREG) estimator. Taking advantage of the administrative data has allowed SEPH to produce quality estimates at a moderately detailed industry by province level. However, data users are asking for levels of detail for which the SEPH sample is unable to support estimates of reliable quality. To address these demands, SEPH is investigating the use of small area estimation techniques which would produce quality estimates at domains where there are very few sampled units.