Literacy and numeracy of overeducated and undereducated workers: revisiting the allocation process in the labour market

ABSTRACT

According to a prominent hypothesis, the occurrence of educational mismatches is consistent with human capital theory since over- and undereducation are substitutes for heterogeneity in the abilities and skills among educational peers. Using German data¹ of literacy and numeracy test scores, I find evidence that compared to their correctly matched educational peers, overeducated (undereducated) workers have lower (better) numeracy and literacy. Controlling for former periods of educational mismatch or unemployment confirms these results. However, only a small proportion of the wage penalty (wage premium) associated with overeducation (undereducation) can be attributed to the wider consideration of human capital endowment.

KEYWORDS:

• Overeducation
• undereducation
• mismatch
• human capital
• wages

JEL CLASSIFICATION:

• I21
• J24
• J31
• I26

Acknowledgments

I gratefully acknowledge the helpful comments and suggestions by Mario Bossler and the participants of the 2015 Colloquium on Personnel Economics and the 2015 Annual Conference of the European Society for Population Economics. The usual disclaimers apply.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data used in this paper are not public. They are subject to the data usage guidelines of the Leibniz Institute for Educational Trajectories and can be accessed upon request from its research data centre (www.neps-data.de/Mainpage).

Notes

1. This paper uses data from the National Educational Panel Study (NEPS): Starting Cohort 6 – Adults, doi:10.5157/NEPS:SC6:3.0.1. From 2008 to 2013, NEPS data were collected as a part of the Framework Programme for the Promotion of Empirical Educational Research funded by the German Federal Ministry of Education and Research (BMBF). Since 2014, the NEPS survey is carried out by the Leibniz Institute for Educational Trajectories (LIfBi) at the University of Bamberg in cooperation with a nationwide network.

2. A further test of procedural metacognition depicts knowledge of and control over one’s cognitive system as measured by asking the participants to rate their performance after completing the competence tests. However, this concept is not used in the present analysis.

3. Compared to the WLE, plausible values are often regarded as superior in coping with measurement error in latent competence variables. Unfortunately, thus far, there are no plausible values available for the NEPS data. However, since the WLE scores of each skill area are based on a large subset of tests, I argue that the standard errors should be very close to those that would be derived using plausible values.

4. I also employ an alternative, more objective measure, which is derived from the 5-digit occupational classification KldB2010. Its fifth digit provides information regarding the task complexity of the job, which can be translated into the required education.

5. As information regarding the actual working hours refers to the current situation and, therefore, is non-randomly affected by potentially occupation-specific vacations, sickness leaves, and overtime work, I refrain from calculating and using hourly wages.

6. All participants willing to complete the tests were asked to complete a quick reading speed test. However, due to the time-consuming design, only approximately three of four of those additionally performed either the reading comprehension or math competency test, and only one of two completed all three tests. Please note that each column in Appendix Tables A2a and A2b covers the regression sample in one skill domain as displayed in Table 2 (mismatch) and Table 5 (wages), respectively, and the observations of the men and women are combined.

7. For the calculation of the years spent in each state of (un)employment, information is utilized on a monthly level, hence resulting in more accurate measures than those obtained by exploiting years only.

8. I chose to not use the final schooling grades as the main measure in the analyses in this paper mainly because this variable only depicts a very crude measure of cognitive ability. In addition, this information is only available for a subsample of the surveyed individuals.

9. Please consider that schooling grades in Germany are constructed differently than those in other countries, with a low grade being superior to a high grade. The grades range from 1 (‘excellent’) to 6 (‘inadequate’).

10. Although the wage effects of overeducation and undereducation are not the core of this paper, I conducted a sensitivity check including both as categorical variables. These estimations further confirm the findings reported in other studies showing that the degree of educational mismatch is important. The higher the degree of overeducation (undereducation), the stronger the wage penalty (premium). See appendix Table A5.