Using small area estimation to increase the resolution and accuracy of employee compensation statistics
Analyzing how compensation varies across job characteristics is critical to understanding the labor market. Previously, these statistics had only been provided at an aggregated level, limiting their usefulness for detailed analyses.
The Bureau of Labor Statistics (BLS) hired Westat to develop a new methodology to estimate the employee compensation for large numbers of detailed job classifications.
Westat developed a comprehensive small domain estimation methodology, which creates detailed wage and non-wage compensation estimates, based on data from the National Compensation Survey.
With 500,000+ possible combinations of job types and detailed characteristics, efficient implementation required a combination of novel model-based approaches and state-of-the-art computing resources, such as Bayesian hierarchical modeling, multicore processors, R, and JAGS.
The model Westat developed will enable BLS to produce improved compensation estimates for the wide range of job types that exist in the U.S. labor market.
These estimates may serve the BLS as key data in producing economic indicators, such as the quarterly Employment Cost Index and the Employer Costs for Employee Compensation, and serve as input in economic analyses for other government agencies and institutions.