Desktop version

Home arrow Education arrow The Dynamics of Opportunity in America

The Findings of Logistic Probability Models to Predict Labor Underutilization among Workers in 2013–2014

The above findings on the labor market problems of adults have primarily focused on variations in these problems across educational attainment and family income groups with a few separate breakouts of key findings for gender and race-ethnic groups. To illustrate the independent effects of other demographic variables on the

Table 7.5 Predicated probabilities for selected individuals 16 and older of being an underutilized member of the nation's labor force in 2013–2014 (in %)

Characteristics of individual

Probability (%)


16to 24-year-old, Black, male, native born, high school dropout, family income under $20,000



16to 24-year-old, White, male, native born, high school graduate, family income under $20,000



25to 34-year-old, White, male, native born, high school graduate,

family income $20,000-$40,000



35to 44-year-old, White, male, native born, some college, family income $40,000-$75,000



45to 54-year-old, White, male, native born, associate's degree, family income $75,000-$100,000



55to 64-year-old, White, male, native born, bachelor's or higher degree, family income $150,000 and over




underutilization rates of workers in 2013–2014, we have estimated a set of logistic probability models of their underutilization status over this 2-year period (for a description of this process and full detail about the logistic probability regression model, see Appendix 7C, including Table 7C.2).

The findings of the logistic probability regression model of the underutilized status of workers in 2013–2014 can be used to predict the probability of a given labor force participant with specific demographic and socioeconomic traits being underutilized at the time of the CPS household surveys in 2013–2014. The predicted probabilities of being underutilized in the labor market of six male individuals with very different demographic and socioeconomic backgrounds are presented in Table 7.5 (the specific formula used to generate these probability estimates is explained in Appendix 7D). [1]

The first individual was a young (16to 24-year-old) Black, native born male who was a high school dropout and lived in a low-income household (annual income under $20,000). His predicted probability of being underutilized in the labor market was an extraordinarily high 66.7 %. If this individual had been White and had a high school diploma and lived in a low-income family, his predicted probability of being underutilized was also quite high at 45.5 %. As the age of the respondent and family income increased, the predicted probability of being underutilized declined. A 25to 34-year-old White, male high school graduate from a low-middle-income family ($20,000–$40,000) had a 14 % probability of being underutilized.

If the respondent's age rose to 35–44, his education increased to 13–15 years with no formal degree, and his family income increased to the $40,000–75,000 range, then his probability of being underutilized declined to 8.2 %. A native born 55to 64-year-old male with a bachelor's or higher degree who lived in an affluent family ($150,000 or higher) had only a 4.5 % probability of being underutilized.

The findings of the above analyses are quite clear. Young, poorly educated adults from low-income families faced underutilization rates of historic proportions. They encountered Depression-era unemployment and other labor market problems in 2013–2014. Even young high school graduates from low-middle-income families faced high rates of labor underutilization. In contrast, older males (45–64) with a bachelor's or higher degree and above average incomes experienced very low labor underutilization rates that would have to be considered the equivalent of super full employment in the labor market. America's labor markets have become extremely stratified by age, education, and family income since 2000. Gaps in labor underutilization rates between the top and bottom of the distribution exceeded 60 percentage points, representing more than 15 times difference in relative terms.

  • [1] The estimated impact of gender on the probability of being underutilized was quite small (<1 percentage point), thus, we have limited our analysis to males only though the results for women would be quite similar.
Found a mistake? Please highlight the word and press Shift + Enter  
< Prev   CONTENTS   Next >

Related topics