Data sources

The analysis in this report is based on the combination of job skills and preparation data from the U.S. Department of Labor’s Occupational Information Network (O*NET) and occupational employment and wage data from the Current Population Survey (CPS).

Occupational Information Network (O*NET): The O*NET database provides a variety of information related to the requirements of more than 950 occupations. The occupations are classified according to a coding scheme that is consistent with the 2010 Standard Occupational Classification. Among other things, O*NET includes information on 35 specific skills representing attributes of workers related to work performance (critical thinking or service orientation, for example). Each skill is rated on a scale of one to five measuring its importance to job performance, from not important to extremely important. This report focuses on the analysis of the importance ratings of the 35 job skills in the O*NET data.32 The ratings are based on information generated by trained job analysts.

Much of the analysis in this report used the most recent version of the O*NET database available at the time (Version 23, released August 2018). This file contains importance ratings of skills for 967 occupations (at the eight-digit level of classification). One occupation – mathematical technicians – was dropped because it was last evaluated prior to 2002. The importance ratings for the remaining 966 occupations were developed by analysts on a rolling basis from 2010 to 2018, with about 10% or slightly more of occupations rated in any given year.

Some of the analysis in this report used one of the first releases of O*NET (Version 5.1, released November 2003) with the goal of evaluating changes in skill ratings within occupations. This file contained importance ratings for 901 occupations, after the omission of mathematical technicians. The importance ratings in the 2003 O*NET data were developed by analysts using data collected mostly from the late 1970s through the early 1990s for the Dictionary of Occupational Titles (DOT). For that reason, changes in skills ratings from 2003 to 2018 are referred to as changes from circa 1990 to 2018. However, because of revisions to the methodology between the DOT and the O*NET the estimated changes in skill ratings within occupations should be treated with caution.

Current Population Survey (CPS): Conducted jointly by the U.S. Census Bureau and the Bureau of Labor Statistics, the CPS is a monthly survey of approximately 55,000 households and is the source of the nation’s official statistics on unemployment. The CPS sample covers the civilian, non-institutionalized population. In this report, 12 monthly CPS files in each year were combined to generate annual estimates of occupational employment in 1980 and 2018. Wages are estimated from the annual outgoing rotation group (ORG) files which consist of the sample of workers from whom wage information was collected. Some of the CPS microdata files used in this report are the Integrated Public Use Microdata Series (IPUMS-CPS) provided by the University of Minnesota.33

Determining job skills

For the analysis in this report, we group the 35 skills rated in the O*NET into five major families of job skills – social, fundamental, analytical, managerial and mechanical (see table below). The grouping is similar to the O*NET classification of skill categories. In general terms, social skills refer to interpersonal skills, fundamental skills lay the foundation for acquiring other skills, analytical skills capture scientific and technological prowess, managerial skills pertain to the management of people, things and finances, and mechanical skills describe the ability to work with and to control machinery or equipment.

As noted, the importance of each detailed skill element to an occupation is given a numerical rating on a scale of one (not important) to five (extremely important) in the O*NET data. The midpoint, a rating of three, indicates the skill is important for the occupation. A simple average of the ratings of detailed skills is used to represent the importance of a skill group to an occupation. For example, the 2018 O*NET importance rating for each of the seven social skills for chief executives (occupation code 11-1011.00) is as follows: 4.12 for monitoring (element ID 2.A.2.d), 4.25 for social perceptiveness (element ID 2.B.1.a), 4.25 for coordination (element ID 2.b.1.b), 4.12 for persuasion (element ID 2.B.1.c), 4.12 for negotiation (element ID 2.B.1.d), 3.12 for instructing (element ID 2.B.1.e) and 3.12 for service orientation (element ID 2.B.1.f). The average of these seven scores – 3.87 – is taken as the measure of the importance of social skills for chief executives. A similar process of averaging the importance ratings of detailed skill elements is used to determine the importance ratings of fundamental (4.08), analytical (3.03), managerial (4.06) and mechanical skills (1.31) for chief executives. The end result of this method is an average numerical rating for the importance of social, fundamental, analytical, managerial and mechanical skills in each of the 966 occupations retained from the 2018 O*NET and the 901 occupations retained from the 2003 O*NET.

O*NET skill elements that represent social, fundamental, analytical, managerial and mechanical skills

Linking the 2003 and 2018 O*NET data files

To determine changes in skill ratings within occupations it was necessary to match the occupations present in the 2003 O*NET data with the occupations present in the 2018 data. Other than the difference in the numbers of occupations – 901 in 2003 and 966 in 2018 – the occupation codes and definitions also vary somewhat due to revisions in occupational classifications. These differences were resolved partly by use of crosswalks provided by O*NET and partly by manual inspection of occupation codes and definitions. An example of a match is that “pressure vessel inspectors,” a 2003 occupation with code 13-1041.05, was linked to “construction and building inspectors,” a 2018 occupation with code 47-4011.00.

Additionally, some occupations from 2003 were combined to match to a single occupation in 2018, and vice versa. For example, three occupations from 2003, “employment, recruitment and placement specialists” (code 13-1071.00), “employment interviewers, private and public employment service” (code 13-1071.01) and “personnel recruiters” (code 13-1071.02) were matched to one occupation in 2018, “human resource specialists” (code 13-1071.00).

The final step in the process of matching the 2003 and 2018 O*NET files was to aggregate occupations from the eight-digit level of detail, e.g. code 13-1071.01, to the six-digit level of detail, e.g. code 13-1071. The result was a dataset consisting of 670 occupations at the six-digit level matched from 2003 to 2018. In addition to facilitating the matching of occupations over time, as in the example above, the aggregation was also done for consistency with the analysis based on the matching of O*NET data to CPS data (see below).

Matching O*NET and CPS data

Because O*NET does not contain employment or wage information for occupations it is necessary to match the skills data to CPS data. Although both the 2018 O*NET and the CPS use the 2010 standard occupational classification there is one key difference: O*NET lists more than 950 occupations coded at the eight-digit level, the finest detail possible, whereas the CPS lists fewer than 500 occupations coded at the four-digit level. In other words, an occupation listed in the CPS typically encompasses more than one occupation listed in O*NET. Thus, occupational data in O*NET must be aggregated to match up to the CPS data. This was done in three steps, as detailed below:

Step 1: The job skills and preparation ratings for eight-digit occupations in O*NET were aggregated to the six-digit level. For example, “financial managers,” a six-digit occupation, are broken apart into two eight-digit occupations in O*NET: “treasurers and controllers” and “financial managers, branch or department.” The job skills and preparation ratings for these two eight-digit occupations in O*NET were averaged to estimate the ratings for financial managers. This process was repeated as necessary, and the end result was a set of numerical ratings on job skills and preparation for 772 six-digit occupations in 2018.

Step 2: The ratings for six-digit occupations were further aggregated to the four-digit level using an occupational crosswalk from the Bureau of Labor Statistics. For example, “marketing and sales managers,” a four-digit occupation, consists of the following two six-digit occupations: “marketing managers” and “sales managers.” In this step of the aggregation process, the job skills and preparation ratings for marketing managers and sales managers are averaged using the employment in each occupation as the weight. If employment data were not available, as was the case for a few occupations, simple averages of the ratings for six-digit occupations were used to estimate ratings for the broader four-digit occupations. The result of this process was average jobs skills and preparation ratings for some 480 four-digit occupations that could be matched to the CPS in 2018.

Step 3: Because occupational classifications are frequently revised, an additional step was necessary to match the job skills and preparation ratings to a harmonized occupation coding scheme that could be used to trace employment and wage trends going back in time. The scheme in the IPUMS-CPS data (OCC2010) provides a consistent, long-term classification of occupations based on the 2010 standard occupational classification. Because of some inconsistencies between the latest CPS occupational codes and the harmonized occupation coding in OCC2010, additional aggregation and recoding was needed to maximize the number of occupations with valid skill ratings. For example, job skills and preparation ratings for “advertising and promotions managers,” “marketing and sales managers,” and “public relations managers” – three distinct four-digit occupations in the current CPS – were averaged using employment weights to estimate the ratings for “managers in marketing, advertising, and public relations” – a single occupation in the time-consistent OCC2010 classification.

The final datasets with job skills, employment and wage data from O*NET and the CPS includes 431 occupations for 2018 and 266 occupations for 1980. These datasets provide virtually complete coverage of U.S. employment. The 431 occupations within the 2018 dataset employed 155.3 million workers, close to the official government estimate of total U.S. employment of 155.8 million. The 266 occupations in the 1980 dataset employed 98.5 million workers, compared with the official estimate of total employment of 99.3 million.

Grouping occupations by the importance of a skill

The analysis in this report divides occupations into four skill tiers based on the importance rating of each of the five skill groups. Occupations are ranked by their importance rating on, say, social skills, and the top 25% of occupations (the highest quartile) are listed as “most important” users of social skills. The second and third quartile of occupations are defined to be “more important” and “less important” users of social skills, respectively. Finally, the bottom 25% of occupations (the lowest quartile) are listed as “least important” users of social skills. This process is repeated for each of the five major skill groups. The importance ratings that define the boundaries for each of the four tiers for the various skill types are shown in the table below.

 

Importance ratings that determine whether an occupation is a ‘least,’ ‘less,’ ‘more’ or ‘most’ important user of a skill, by type of skill

The relative importance of a skill

In this report, the changing mix of skills within a job is assessed by looking at the relative importance of each skill within the job. The relative importance of a skill is its rating divided by the sum of the ratings for all five skills. Suppose that the importance ratings for the major skill groups within a job are as follows: social – 3.1; fundamental – 3.2; analytical – 2.5; managerial – 2.4; and mechanical – 1.8. The sum of the ratings for the five skills is 13 and the relative importance of social skills in this job 0.24 (or 3.1/13), the relative importance of analytical skills is 0.19 (or 2.5/13), and so on. If all skills have the same importance rating, say, 3.0, then the relative importance of each skill is 0.2.

If the importance rating of, say, social skills increases by more than the ratings of other skills in an occupation, then the relative importance of social skills will increase, and the job will be judged to have become a more intensive user of social skills. For example, suppose that the importance ratings for the five skills in the same job at a later point in time are as follows: social – 3.5; fundamental – 3.6; analytical – 2.6; managerial – 2.7; and mechanical – 1.8. Then the relative importance of social skills is estimated to have increased to 0.25 and the relative importance of analytical skills to have decreased to 0.18, even though the ratings of both skills have increased over time.

Hourly wages

Wage estimates pertain to a worker’s main job. Workers paid by the hour report hourly wages. For workers who are not paid by the hour, the hourly wage is calculated as weekly earnings divided by the usual numbers of hours worked in a week. The CPS collects data on wages from outgoing rotation groups only, which represent one-quarter of the monthly sample. Self-employed workers are excluded from this subsample. Wages are adjusted for inflation with the Consumer Price Index Research Series (CPI-U-RS).

Using regression analysis to determine how skills, education and other factors affect the gender wage gap

How much a worker earns on her job is related to her skills, education, experience and other qualifications, and to the nature of the job, such as occupation and industry. In this report, a regression analysis is used to determine the relationship between the hourly wages of women and men and the following characteristics for which data were available: job skills, education level, occupation, industry, union membership, part-time work, age, race and ethnicity, geographic region, and metropolitan area status. Regressions are estimated separately for women and men following an approach in common use (see table below for the results).

The results from the regressions are used to estimate the earnings of women, relative to men’s earnings, assuming women and men had the same characteristics on average – that is, if they worked in jobs with the same skill requirements, had the same education levels, were equally likely to be union members, and so on. (This widely used technique is known as the Blinder-Oaxaca decomposition.) The regression analysis can also be used to assess how a difference between women and men with respect to any given characteristic, say the average rating on nonmechanical skills, may contribute to the gender wage gap.

Among the key findings, the regression analysis shows that wages are estimated to increase by about 22% for women and 24% for men with a one-point increase in the importance of nonmechanical skills in 2018 on average (the importance ratings for social, fundamental, analytical and managerial skills are combined to represent nonmechanical skills). On the other hand, wages decreased by about 4% for women and 2% for men with a one-point increase in the importance of mechanical skills. The gender differences in the returns to nonmechanical and mechanical skills are among the “unexplained” differences in the earnings of women and men.

Women held a slight edge over men on nonmechanical skills in 2018: The average rating on nonmechanical skills for women was 2.75 compared with 2.69 for men, with the difference arising from how women and men were distributed across occupations. Women also benefited from a lower rating on mechanical skills compared with men, an average of 1.51 versus 1.94. Gender differences in skill endowment are among the “explained” differences in the earnings of women and men. Examples of other differences include the greater share of women (40%) than men (35%) who have a bachelor’s degree or higher level of education and the greater share of women (18%) than men (7%) who work in administrative support occupations.