OEWS Wages – About the Data
- Designed as Point-in-Time Estimates
- The Methodology Itself Affects the Comparability of Estimates
- Procedural Changes
- Other Changes
Designed as Point-in-Time Estimates
The Occupational Employment and Wage Statistics (OEWS) Survey data are not a continual time series, instead it was designed to provide accurate point-in-time wage estimates for the nation, states, and each Metropolitan Statistical Area (MSA).
The use of OEWS data for time series analysis, or comparing onetime period to another is not encouraged.
Some reasons that make the use of OEWS less useful for comparisons of two or more points in time include changes in survey methodology, improvements in data collection, and changes in the structure of the coding system used to classify occupations in OEWS.
If you choose to make such comparisons, you should understand the changes in both the survey methodology and procedures over the last several years that will affect the comparability between time periods.
The Methodology Itself Affects the Comparability of Estimates
When preparing wage estimates, employer surveys are pooled from the last three years. The weights assigned to each observation are adjusted and then the estimates are benchmarked to a recent count of employment by size of establishment, industry, and area. This recent count is
established from the Quarterly Census of Employment and Wages file, also called the ES-202. The full three-year sample allows the production of estimates at fine levels of geographic, industry, and occupational detail. Because the observations cover a three-year period, it is
less likely to capture year-to-year changes.
A second methodological issue is that the survey updates wage data for all but the recent data collection period using the change in the BLS Employment Cost Index (ECI) at the national level. In California, the OEWS wage estimates are updated using the latest quarter ECI available prior to publication.
Most recently, OEWS implemented another change in method that could influence wage estimates. In previous surveys, workers who reported their wages to be in the highest wage interval had their wage set at the lowest wage in that range. This wage was used when calculating the mean or percentile wage estimates. So a dentist who makes $90 an hour would be in the top wage range of $80.00 or more and would have been assigned a wage of $80.00. OEWS now sets the wage for the highest interval at the mean wage that workers in that interval would be expected to make. This mean wage is estimated from data collected by the National Compensation Survey. This method has been used since the release of the 2002 estimates.
Procedural changes in OEWS would also affect comparability of data over time. In 1999, OEWS implemented the Standard Occupation Classification (SOC) and introduced new boundaries of the wage intervals on the survey form. New boundaries were again introduced to the
survey in 2005.
When the OEWS survey switched from the old OEWS occupational classification system to the new SOC in 1999, it was not a one-to-one change. Some OEWS occupations split into more than one new SOC and other were merged together. To accommodate this change, some occupational wage estimates (where the prior occupation description was consistent with the SOC) used three years of survey data, while others (where the SOC occupation did not match to a previous OEWS survey description) used only one. In 2000, all occupations were estimated with just two years of data. In 2001, additional changes were made to the occupational classification structure, including collecting all detailed SOC occupations that were in scope.
The most relevant procedural change that affects the wage estimates is about the value of the highest wage range. In 1999, the OEWS survey raised the low-end of the highest wage range from $60 to $70. In 2005, the low-end of the highest wage range went from $70 to $80. Since the OEWS program pools surveys from three years, changes like these will still affect the wage estimates for three years.
Finally, OEWS has implemented other changes, like semi-annual collection and estimation and the conversion to the industry coding using NAICS. These also have an effect on comparability of the data.
See the "Occupational Employment and Wages Technical Note" on the Bureau of Labor Statistics Web site for additional technical information about the OEWS survey.
In conclusion, the use of OEWS data for time series analysis is not encouraged. If you plan to make such comparisons, you should understand that changes in both the survey methodology and procedures over the last several years influence the comparability of the wages over time.