Measuring the distribution of wages in the United States from 1996 through 2010 using the Occupational Employment Survey
Using the OES data, we replicate many of the wage variance trends that other authors have found using the CPS data. We show that most of the growth in wage inequality for the period of our study has occurred within the private sector, within particular industry groups such as professional and business services, and within occupational groups such as healthcare. We show that occupations explain a growing fraction of all wage variance, and, in an examination not possible with the CPS data, we find that within the private sector, wage differences among establishments explain far more of the level and the trend of wage variance than do wage differences among occupations.
Using the OES data to make comparisons over time
The Occupational Employment Statistics survey is designed to measure occupational employment and wages in the United States by geography and industry and is the only such survey of its size and scope. Since 1997, the OES has covered all workers in the United States except for agricultural workers, private household workers, and unincorporated self-employed workers without employees. Every year, approximately 400,000 private and local government establishments are asked to report the number of employees in each occupation within specific wage intervals. At the same time, complete employment data are obtained from the federal and state governments.13 An example of a portion of a survey form is given in the following exhibit.
For large establishments, the survey form lists 50 to 225 detailed occupations (occupations preprinted on the survey form are selected based on the industry and the size of the establishment). Small establishments receive a blank survey form and write in descriptions of the work done by their employees. These employer-provided descriptions are coded into occupations by staff in state labor agencies (as part of the OES federal–state partnership). Wage intervals on the OES survey form are given in both hourly and annual nominal dollars.14 To calculate average wages, the OES program obtains the mean of each wage interval every year from the National Compensation Survey (NCS). These mean wages are then assigned to all employees in that wage interval.15 Approximately 25 percent of jobs in the OES data are imputed (with an establishment nonresponse rate of approximately 20 percent).
The OES program underwent a large number of improvements between 1996 and 2002, many affecting the comparability of data over time. The first year of a truly national OES survey was 1996, when wage data were collected from 400,000 establishments in every industry and every state. In 1997, the program began including establishments with less than 5 employees in the sample. Improvements to the sampling and weighting methodologies took place in 1998. In 1999, changes in the width of the wage intervals were instituted, making them more uniform in natural logarithm (wage) terms, as well as the addition of a 12th wage category. Nominal wage intervals printed on the form were shifted again in 2005 and 2009. In addition, 1999 was the first year that the occupations used in the OES were those of the Standard Occupational Classification system. The year 2001 was the last in which data were collected in October, November, and December; beginning with November 2002, data have been collected from 200,000 establishments each November and 200,000 establishments each May. The 2002 sample was the first designed using the North American Industry Classification System (NAICS) in place of the older Standard Industrial Classification (SIC) system. Each of these survey changes is designed to improve the quality of OES data but has the downside of creating difficulty in comparing estimates from year to year.
13 Data provided to OES by federal and state governments are not broken out into individual establishments.
14 Annual wage rates listed on OES survey forms are 2,080 times the hourly wage rates.
15 For 1996 to 2001, the lower bound of the top-most (open-ended) interval was used as the wage for workers in this interval. From 2002 forward, the mean wage from the NCS for workers in the top-most interval has been used. For 1996–2001, we multiply the wage in the top-most interval by 1.4.