On the state of the geography in the U.S. Bureau of Labor Statistics Covered Wages and Employment (ES-202) series

This article discusses the strengths and weaknesses of the confidential establishment-level Covered Wages and Employment (ES-202) data series for small area economic analyses and other kinds of quantitative geographical research. The article examines the improvements in the geographic identifiers in the file over the past decade based on analysis of confidential micro data for forty-six states. It also examines the extent of spatial censoring in the ES-202 file stemming from two sources: (1) missing or corrupt physical address information in the raw data and (2) failure to attach spatial identifiers (geocode latitude/longitude coordinates) to physical addresses. Results of geocoding tests show that samples of address-matched units from state ES-202 files are likely to be biased along several dimensions. The article argues that a key area of geographical research in the future is methods to address bias in administrative samples. Research along these lines would substantially improve the usefulness of confidential micro data series for regional science research.