Changes across cohorts in wage returns to schooling and early work experiences

Jared Ashworth, V Joseph Hotz, Arnaud Maurel, Tyler Ransom 14 April 2018



Since the early 1970s, the US labour market has undergone some dramatic changes. The wage premium for college graduates relative to high school graduates has increased from 30% to 55% (Valletta, forthcoming). Over the same period, the rate at which college students work while in school has risen steadily from 30% to 50% (Scott-Clayton 2012), causing college students to take longer to complete their degrees (Bound et al. 2012).

Do these facts present a bleak picture? The answer depends on whether in-school work is a productive use of time for college students. Previous studies (e.g. Stinebrickner and Stinebrickner 2003) have suggested that these increasing rates of in-school work may distract students from their studies, increasing their chances of dropping out of school and harming their later-life labour market prospects. And, to the extent that college students work while in school in part to cover their educational expenses, this may increase their time-to-degree and delay the start of their post-schooling work careers. Alternatively, in-school work may, in fact, represent a human capital investment by students that benefits, rather than harms, their careers. For example, working while in college may give a student a leg up in the labour market upon graduation because she already possesses basic skills for interacting in the workplace. In this light, increased time-to-degree may not be problematic if students use the time to accrue additional skills.

The increase in the college wage premium may be related to the increased propensity for students to work while in school. While a substantial share of the increase is likely due to labour demand factors (such as technological change), some of it may reflect the possibility that graduates today possess more skills than the graduates of the previous generation.

Parsing out the separate impacts of work and schooling experience on future wages is a challenging task that requires detailed data. Few datasets collect information on a youth’s employment and schooling activities, and then follow the same youth into adulthood. Possessing such longitudinal data is particularly useful in our case because it allows us to account for skills that are difficult to measure, but which are important determinants of in-school work decisions and future wages.

Data and trends in schooling and work experience over time

In a recent paper, we examine whether work experiences as a youth have positive effects on wages in adulthood (Ashworth et al. 2017). We also quantify the amount by which existing estimates of the returns to schooling are biased by ignoring unobserved skills and accumulated work experience by youth.

The data we use are from the 1979 and 1997 cohorts of the National Longitudinal Surveys of Youth (NLSY). These surveys follow representative cohorts of American youth from adolescence through middle age. The older cohort we studied were high-school students during the late seventies, while the younger cohort were high-school students during the late 1990s.1 We measure the schooling and work experiences of young men in each survey between age 16 and 29 and then examine their wages at age 29 to see how schooling, degree attainment, and work experience contribute to their adult wages. We break up work experience into five components: in-high-school, in-college, part-time, full-time, and military service.

Compared to youths from the older cohort, we find that those in the younger cohort spend about one more year in schooling, work about one more month while in high school and six more months while in college, and work about eight months less outside of school. As a result, students in more recent cohorts spend more time in school even though their total amount of accumulated work experience is about the same as those in earlier cohorts.

Returns to schooling and degree attainment

As mentioned above, we are interested in estimating the wage returns to years of schooling, high school diplomas, and bachelor’s degrees and how these returns have changed across recent cohorts. If we ignore the possibility that possessing natural ability and/or accumulating early work experience may impact one’s schooling choices and adult wages, we find that the return to both an additional year of schooling and a high school diploma went up slightly over the time period of our study. The return to a bachelor’s degree was hump-shaped: it rose sharply among the older cohort, then fell more recently for the younger cohort.

Results are changed when we account for changes in the accumulation of students’ work experiences, or in the levels of unobservable skills associated with each schooling level.  The returns to an additional year of schooling are significantly overstated when one does not account for those changes in compositions, as are the returns to a high school diploma or bachelor’s degree. For example, we measure a raw earnings premium to an additional year of schooling at 5%-7%, whereas this drops to less than 2% when accounting for the fact that individuals may have simultaneously accumulated work experience at the same time as schooling. The returns to high school and college degrees are also overstated – by about 60% - when we account for actual work experience and unobservable skills.

Returns to work experience

Also of interest in our study are the returns to the various work experiences. We find that, once accounting for selection, the returns to an additional year of working while in school – either during high school or college – are larger than to an additional year of schooling. Overall, returns to work experiences have generally been fairly stable over the time period we cover, although we document a slight increase in the returns to working while in high school, and working full-time (out of school).

Our findings underscore the importance of accounting for acquired skills when analysing the wage impacts of education. In particular, we show that controlling for work experience that was actually acquired results in sizable reductions in the contribution of completed schooling to adult wages.

The role of other skills

Our data allow us to measure years of completed schooling and work experience and to assess their returns for wages. But such measures don’t fully capture an individual’s skills and their consequences for wages. There are other skills – cognitive and non-cognitive ones – that we measure with less reliability. To assess their influence on wages and whether these skills and their returns have changed across cohorts, we use scores from components of the Armed Services Vocational Aptitude Battery (ASVAB) available in our data to provide a measure, albeit noisy measurement, of cognitive skills. We also allow for the influence of a second potential component of skills – one we think partially captures non-cognitive skills – on the schooling and work decisions, and wages, of the young men in our samples.

We find that the returns to both cognitive and non-cognitive skills have remained steady over time, with cognitive skills commanding a slightly higher return (15%-17% for a one-standard-deviation increase) relative to non-cognitive skills (11% for a one-standard-deviation increase).

Breaking down changes in the estimated returns to schooling and work experience into price effects (which quantify secular changes in returns to skills over this period) and composition effects (which quantify changes in the amount of labour market preparation and skills of students over this period), we find substantial composition effects, especially in terms of both cognitive and non-cognitive skills. Our results indicate that not all of the changes in the returns to schooling over our time period of interest are due to changes in labour market factors, but reflect instead changes in the level of labour market preparation.

Concluding remarks

While recent trends among college students have shown an increase in work while in college accompanied by an increase in time-to-degree completion, our analysis suggests that these trends need not be alarming. We find indeed that in-school work experience significantly contributes to adult wages, with the measured returns to schooling being lower once work experience and unobserved skills are accounted for. Taken together, our results suggest that work while in high school or college represents a human capital investment that has a sizable future payoff.


Ashworth J, V J Hotz, A Maurel, and T Ransom (2017), “Changes across Cohorts in Wage Returns to Schooling and Early Work Experiences,” NBER Working Paper No. 24160.

Bound J, M Lovenheim, S Turner (2012), “Increasing Time to Baccalaureate Degree in the United States,” Education Finance and Policy, 7(4): 375-424

Scott-Clayton, J (2012), “What Explains Trends in Labor Supply Among U.S. Undergraduates?” National Tax Journal, 65(1): 181-210.

Stinebrickner, R and T Stinebrickner (2003), “Working during School and Academic Performance,” Journal of Labor Economics 21(2): 473-491.

Valletta, R (Forthcoming), “Recent Flattening in the Higher Education Wage Premium: Polarization, Skill Downgrading, or Both?” in Education, Skills, and Technical Change: Implications for Future U.S. GDP Growth, ed. Charles Hulten and Valerie Ramey. University of Chicago Press.


[1] With regards to generations, the oldest cohort are the youngest group of Baby Boomers and the younger cohort are the oldest group of Millennials.



Topics:  Education Labour markets

Tags:  returns to education, early work experience

Assistant Professor of Economics, Pepperdine University

Arts & Sciences Professor of Economics, Duke University

Assistant Professor in the Department of Economics, Duke University

Assistant Professor, University of Oklahoma


CEPR Policy Research