angle-up angle-right angle-down angle-left close user menu open menu closed search globe bars phone store

Skills and Earnings in the Full-Time Labor Market
Neeta Fogg, Paul Harrington, and Ishwar Khatiwada

A Word about the Data

PIAAC was designed to measure adult proficiencies in three key information processing skills—literacy, numeracy, and problem-solving in technology rich environments. The Survey of Adult Skills (PIAAC) was administered in 33 nations during the 2011-2014 period. PIAAC data for the United States that are used in this study are limited to the household portion that was conducted during 2011-2012 among a sample of the 16- to 65-year-old noninstitutional population as well as a supplemental survey conducted during 2013-2014 that was targeted to sample teens and young adults, unemployed adults, and older workers. The combined samples, also referred to as the enhanced U.S. household sample data, include about 8,700 persons between the ages of 16 and 74 who lived in households at the time of the survey.

The PIAAC survey instrument was composed of a background questionnaire as well as cognitive assessments of literacy and numeracy of respondents. PIAAC defines literacy as "understanding, evaluating, using and engaging with written text..." and numeracy as "the ability to access, use, interpret and communicate mathematical information and ideas..." 15 The PIAAC data collection process limits the time and related burdens required of respondents by only administering a fraction of the proficiency tests to individual adult participants in the survey. PIAAC survey respondents were not administered every skill proficiency question. Instead, 10 plausible values (PVs) for literacy and numeracy test scores are provided in the PIAAC data file. PVs are a statistical means to replicate a probable score distribution that summarizes how well each respondent answered a small subset of the assessment items and how well other respondents from a similar background performed on the rest of the assessment item pool. These plausible values are estimated using item response theory models. 16

According to the PIAAC technical documentation, in addition to the estimation of survey errors from the complex sampling design of PIAAC, one should also estimate the measurement errors presented in the proficiency assessments whenever the 10 plausible scores are used in the analysis. The measurement error accounts for variations in these 10 plausible values. All of the PIAAC proficiency measures in this report—for both descriptive and regression-based estimates and associated measurement errors—are estimated using 10 plausible values.

We begin the report with a descriptive analysis of the monthly earnings of full-time employed 25- to 54-year-old workers in 2012-2014, which are part of the enhanced U.S. household sample data, by their literacy proficiencies, educational attainment, and a wide array of demographic traits. Mean monthly earnings are examined for several demographic and educational subgroups of workers, and for each subgroup we present a separate analysis of their mean monthly earnings by the level of their literacy and numeracy proficiencies. The central focus of the analysis in this report is the connection between earnings and the human capital of workers measured by their educational attainment and literacy and numeracy proficiencies.

The descriptive analysis is followed by multivariate regression analysis of the monthly earnings of 25- to 54-year-old full-time employed workers. Using human capital earnings functions, these regression models estimate the independent effect of literacy and numeracy proficiencies and the educational attainment of U.S. workers on their monthly earnings by statistically holding constant the effect on earnings of all the other independent variables included in the regression model. Earnings are found to vary by education, skills, paid work experience, intensity of employment (hours per week), economic sector of employment, region, and demographic traits of workers. By including these worker traits as explanatory variables, these multivariate regression models are designed to estimate the independent effect of each of these variables on the monthly earnings of workers.


15 Measures of problem solving in a technology-rich environment were also included in the competency assessment portion of the PIAAC study in the United States.

16 For review of this topic, see Kentaro Yamamoto, Lale Khorramdel, and Matthias von Davier, "Scaling PIAAC Cognitive Data," Chapter 17, Technical Report of the Survey of Adult Skills (PIAAC), 2nd Edition (Paris: OECD Publishing, 2016).