Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019

The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 This paper studies the evolution of three higher education wage differentials from 1996 to 2019 in Germany. We distinguish between degrees from academic universities, degrees from universities of applied sciences, and the mas- ter craftsman\craftswoman certificate. The educational reference category is a standard degree within the German vocational education and training system. Based on samples of male and female workers from the Socio-Economic Panel Study (SOEP), regression methods show that all three educational wage differentials in 2019 exceeded the ones in 1996. However, workers graduating from universities experienced an inverse u-shape pattern with a maximum of about 0.5 log points around 2012. Since then, their wage differential decreased by nearly ten percent (about 0.045 log points). Although the decrease is not statistically significant at conventional levels, we think that nearly ten per - cent can be regarded as economically meaningful. We argue that this pattern is related to university expansion and changes in graduates’ subject-choice composition during that expansion. The paper concludes with a discussion of possible alternative explanations. Keywords: Educational Wage Differentials, Gender Gaps, Higher Education, Returns to Education JEL Classification: J31, J16, I23, I26 1 Introduction may foster innovation and trade, boosting investment The expansion of university education fostered a dynamic into new capital-intensive automation technologies like change in the workforce’s educational composition, and artificial intelligence. This paper studies whether relative which has received attention from policymakers and wages of high-skilled individuals increased or decreased researchers (e.g., Authoring Group NRoE 2018; Araki during the recent expansion of university education in 2020; Goldin and Katz 2008; Horowitz 2018). The debate Germany. revolves around whether the increase of highly edu- The literature often focuses on two educational cat - cated individuals may, presumably with some lag, lead egories, college graduates and others, while the result- to stronger competition among university graduates ing wage differential is referred to as the skill (or college and put pressure on their relative wages. Alternatively, wage) premium. According to Goldin and Katz (2008), highly educated individuals may experience even larger the skill premium in the United States increased from wage differentials (and vice versa for low-skilled indi - 1980 to 2005, which they explain as resulting from a tech- viduals). The increase in university-educated individuals nology-driven growth in the demand for college gradu- ates and non-routine tasks (see also Lindley and Machin 2016, among others). More recent evidence suggests a *Correspondence: ordemann@dzhw.eu stagnating skill premium in the US after 2010 (Valletta German Centre for Higher Education Research and Science Studies (DZHW ), 2018) and a moderate decrease in college wage premium Lange Laube 12, 30159 Hannover, Germany in European countries between 2005 and 2015 (Green Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 17 Page 2 of 12 J. Ordemann , F. Pfeiffer and Henseke 2021). Reinhold and Thomsen (2017) study discusses the new empirical findings on the evolution of the development of daily wages from a cohort of young educational wage differentials. Sect.  5 discusses changes employees based on administrative data in Germany in the subject composition of study choices among grad- until 2010. Differentiating between three skills categories, uates and the evolution of wage differentials. Finally, they find that high-skilled employees, compared to mid - Sect.  6 concludes with a discussion on further factors dle and low-skilled employees, experienced a rising skill that may have contributed to our main findings. premium. The German higher education system presumably 2 The evolution of the educational composition is more differentiated than the Anglo-Saxon one. Tak - of the population, aged 30 to 55 ing the perspective of one higher education premium 2.1 Education and wages ignores the fact that there are at least three specific and Economic reasoning suggests that when young adults well-defined higher educational degree categories in Ger - invest in higher education, they compare costs and many (Authoring Group NRoE 2018): Degrees from returns over the life cycle and consider their own educa- academic universities (referred to as universities, U, in tional and occupational preferences (e.g., Backes-Gellner what follows), degrees from universities of applied sci- et  al. 2021; Flossmann and Pohlmeier 2006; Pfeiffer and ences (UAS), and the master craftsman\craftswoman cer- Stichnoth 2015; Westphal et al. 2022). Wage differentials tificate (MC). MC is the highest post-secondary degree emerge in order to compensate for these investments (for outside the university system in Germany. It builds on a other reasons for wage differentials, such as amenities, fourth qualification-type obtained via the dual vocational skills, effort, risk, or social interactions, see e.g., Anger apprenticeship system (named vocational education and Heineck 2010; Gebel and Heineck 2019; Krueger and training, VET). These degrees vary significantly in and Schkade 2008). Educational wage differentials sig - academic content and length of study (see Sect.  2). Our nal investment opportunities and differences in the effort study contributes to the international literature in a novel needed to acquire specific degrees. In a thought experi - way by looking at the evolution of these three higher edu- ment where wage differentials would be zero, the eco - cational (gross) wage differentials compared to a VET nomic incentives for educational investments would be degree using data from the Socio-Economic Panel Study low or absent. (SOEP) from 1996 to 2019. In addition, we consider the Since wages and educational wage differentials result major studied while at a university of applied sciences from several economy-wide and individual-specific fac - or university to highlight recent changes in the student tors, identifying specific factors strong enough to change composition of these majors and their potential relation their trajectory can be challenging. The part of the wage to educational wage differentials. attributed to the level of education depends on the com- Similar to the literature, we find that in Germany, in the petencies attained in formal educational institutions. period from 1996 to 2019, educational wage differentials In addition, individuals select themselves into these increased despite the higher education expansion and the institutions depending on their perceived abilities and higher educated workforce participation rates. We fur- socio-economic background (e.g., Becker and Hecken ther this field of literature by demonstrating specific evo - 2008; Hillmert and Jacob 2003; Müller and Pollak 2007), lutionary patterns for each educational degree—a pattern highlighting the role of preferences for academic educa- that evolves differently for women and men. While all tion (see Kamhöfer et al. 2019) as well as otherwise often three educational wage differentials increased compared unobserved barriers to and benefits from educational to a VET degree, workers graduating from universities pathways. experienced a halt starting around 2012, when it reached The overall amount of educational investment in soci - its highest level so far (about 0.5 log points) and after- ety, or the wage levels for a given competence profile are ward a decrease of about 0.045 log points. Although the driven by factors that the investing individual, as a rule, decrease is not statistically significant at the conventional cannot control. Thus, educational wage differentials level of 95%, we think it is economically meaningful. We depend on the amount and quality of educational invest- argue that this pattern is related to the expansion of uni- ments, on competencies that are not certified or are versity education. Furthermore, relatively more students hard to certify, and on factors that determine the over- graduated in arts and social sciences, subjects for which all supply of and demand for these competencies in the wage differentials are lower than for other majors. economy. The article proceeds as follows. Sect.  2 introduces the chosen educational categories and highlights the expan- 2.2 Educational categories used in the study sion of university education. Sect.  3 describes the data, The German education system has traditionally been its operationalization, and our research approach. Sect. 4 stratified and separated between occupational and The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 3 of 12 17 academic pillars (Authoring Group NRoE 2018; Baethge acquisition of the MC certificate then lasts an additional 2007). There are two types of academic educational insti - 2 to 3 years. tutions available: universities and universities of applied sciences. Both types of institutions vary in their aca- 2.3 Dynamic upskilling in the population aged 30 to 55 demic orientation and the subjects offered. Universities Germany experienced a significant expansion in univer - of applied sciences focus on a practically relevant set of sity education in the new millennium, which changed the qualifications predominantly in economics (as a rule, educational composition of the German population (e.g., business economics), social work, and engineering. They Authoring Group NRoE 2018). Figure  1 illustrates the often have strong ties to the local economy. Universities evolution of the educational composition in the popula- are broader in their portfolio and offer subjects encom - tion aged 30 to 55, separately for women and men, based passing the entire academic spectrum from the arts, on samples taken from SOEP. economics and business economics, social and natu- Note that the scales used for VET (left side) and the ral sciences, and sports to law, medicine, and veterinary three higher educational degrees (right side) differ. Over medicine. the 24  years considered here, highly educated young In Germany, matriculation at universities of applied people steadily entered the observed age group whereas sciences and universities requires an entrance qualifica - older and less educated people left it. As a result, the tion, which demands successful graduation from upper share of individuals with a U (UAS) degree increased secondary schooling (typically after 12 or 13  years of from 10.8 (10.8) percent in 1996 to 17.8 (11.9) percent overall schooling). Since 2009, both offer two degrees, in 2019. The share of individuals with a VET degree bachelor’s and master’s degree,  gradually replacing tra- decreased from 60.0 percent in 1996 to 50.5 percent in ditional degrees such as the “Diplom”. In addition, uni- 2019. Upskilling among women is a driver of the change versities offer the “Staatsexamen” for teaching, medicine, in the educational composition from 1996 to 2019. The and law, which is, by and large, comparable to a mas- share of women with a U degree increased by 7.8% points ter’s degree. A bachelor’s degree requires an investment and that of men by 6.0% points. In this period, the share of at least three years and a master’s degree of at least of women with UAS degrees remained approximately the two additional years. In addition to choosing the type same (−0.0% points) while the share of men increased by of degree and higher education institution, students can 2.1% points. Summing up, in 2019, approximately 28.0 also choose between a wealth of different subjects (from percent of women and 30.6 percent of men in the 30 to among more than 17,000 different courses, Authoring 55 age bracket held either a UAS or a U degree, compared Group NRoE 2018). to 20.8 percent for women and 22.5 percent for men in There is a third avenue to achieving a tertiary educa - 1996. More women (from 3.5 to 5.6 percent) and fewer tional qualification in Germany: The MC certificate is men (from 12.8 to 11.1 percent) with an MC certificate part of the vocational pillar, the central qualification sys - entered this age bracket. tem besides the academic pillar. It is specific to a craft such as a hairstylist or mechatronic technician and is less 3 The empirical approach for assessing academic in its learning contents. It enables certificate educational wage differentials holders to open their own firms in their respective craft Data. Our empirical analyses of the evolution of wage and opens up supervising positions for them. An MC differentials are based on samples from the German certificate builds on the already acquired qualification SOEP (Goebel et  al. 2019; SOEP-Core 2021). The SOEP of a VET degree, which typically lasts 3 to 4  years. The is a representative longitudinal panel study of German households. It concentrates on multiple topics ranging from employment, well-being, health, working hours, and earnings to daily life. The study began in 1984 and currently includes 36 waves. We use individual-level data Higher education has gradually opened to individuals with vocational train- spanning 24 years from the years 1996 to 2019. Our esti- ing and work experience or an MC certificate who do not otherwise possess a higher education entry qualification. The overall share of these students mation sample is unbalanced and restricted to 259,555 remains below four percent of all students, and less than two percent of all observations of 35,890 employed individuals, 18,455 of alumni (Brändle and Ordemann 2020). We do not investigate the pathways whom are women. The samples include employees and into higher education or this subpopulation. According to Ordemann (2019) they have similar monetary but slightly lower non-monetary labour market the self-employed. Although the wage determining pro- returns then other alumni. Arguably, the content of the educational degrees cesses may differ between the two groups, economic may have differed at the beginning of the observation period from those com - arguments suggest that they are related. While the self- mon in (West) Germany, and this may have influenced the evolution of edu - cational wage differentials. In our previous version of this paper (Ordemann employed have to generate their wages from residual and Pfeiffer 2021), we also performed the analysis separately for East and West profits, employees receive a fixed wage bargained ex-ante Germany. 17 Page 4 of 12 J. Ordemann , F. Pfeiffer Fig. 1 The educational composition in the population 1996–2019 ( Women/Men; in %). Note: Individuals aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. The shares of the four educational categories do not reach 100% because the fifth category, no degree, is excluded in this figure. Their average shares (total) are 11.8% for 1996 and 13.1% for 2019. Source: SOEP v36, authors’ own calculations (e.g., Pfeiffer and Pohlmeier 1992). If a risk-adjusted wage entries into and exits from young adulthood and retire- in self-employment differs from an employee’s wage, ment respectively. workers can become employees and vice versa. The investigation starts in 1996 for two reasons. The We concentrate on female and male prime-age workers first reason is that, according to Gebel and Pfeiffer aged 30 to 55. In this age group, as a rule, individuals are (2010), 1996 was the year in which estimates of the members of the workforce, although participation rates returns to education reached their minimum value in are still higher for the better educated. A difference that the period 1984 to 2006 in West Germany. The period is higher among women compared to men. In our SOEP of strong educational expansion after World War II samples, the share of working women increased from exerted downward pressure on wages for skilled work- 67.6% in 1996 to 85.5% in 2019 and from 91.2 to 92.6% ers, and the estimated returns to education were (mod- among men. More investment in education increases the erately) decreasing from 1984 onward. However, after opportunity cost of not working. Therefore, individu - 1996 estimated returns to education started to increase als with a higher educational degree tend to show higher once again. participation rates compared to VET (for women in The second reason is that German reunification in Germany, also compare Westphal et  al. 2022). However, 1990 may have influenced the German wage structure, the employment participation rates also increased for especially during the years immediately following reuni- women with a VET degree and women with no degree. fication (e.g., Gernandt and Pfeiffer 2007, 2009). Thus by While the participation rates of women increased, men 1996, 6  years after reunification, a relevant part of the still display higher participation rates. Participation rates specific impact of reunification on the educational wage among men exceeded the ones for women by 7.1% points differentials should already have taken place. in 2019, compared to 23.6 in 1996, which is a result in Variables. The dependent variable of our analyses is the part of the upskilling among women (for more details, natural logarithm of gross earnings per hours worked. It see Ordemann and Pfeiffer 2021). Nevertheless, we think is obtained separately for each year by the trimmed real concentrating on the age group of 30 to 55 year old work- gross monthly income reported in the previous month. ers could be helpful in lowering potential estimation The reported income is divided by the factor of 4.33 biases associated with the endogeneity of labour market times the trimmed actual working hours at the end of the sample selection. In addition, the obtained wage was The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 5 of 12 17 trimmed before transforming it into the natural loga- models, we report the partial coefficients for the highest rithm. The trimming of all variables is performed on the educational degrees for each year from 1996 to 2019. one percent level; all incomes are depreciated to 2015. VET is used as the reference group in our analyses. These Additional payments, such as holiday pay are excluded. partial coefficients may not indicate the causal economic The explanatory variable in focus is the highest edu - effect of the educational investments needed to gain the cational degree. We use degrees provided by the SOEP, respective educational degree. which reflect the unique characteristics of the German Estimates of educational wage differences and differ - education system: no (post-secondary) degree (or not yet entials over 24 years may exhibit, to some degree, erratic completed), apprenticeship or vocational training, master patterns from year to year. There is no explicit theory craftsmanship, and university (of applied sciences). supporting the notion that educational wage differences Some adjustments to the variables provided by the and differentials should not display such a pattern. Nev - SOEP are made. First, we add cooperative education and ertheless, we cannot exclude the possibility that part civil servant training to the category of master crafts- of this pattern found in our analyses is the result of the manship. Cooperative education combines vocational various samples retrieved over such a long time period. training with academic study but is still bound to the firm For instance, the number of observations in our estima- with which students have a work contract, who shape tion samples varied from 3681 to 7990. To get rid of such the curricula of the cooperative education institution. types of randomness to some degree for our subsequent Second, higher education degrees obtained in a foreign analyses, we use Epanechnikov kernel-weighted local- country were added to the category of universities of mean polynomial smoothing for the figures produced applied sciences to reflect the diversity of higher educa - from the estimates. tion from all over the world in this educational category. We focus on these three higher education categories, 4 The evolution of educational wage differences although each of them may have further heterogeneities. and differentials, 1996 to 2019 While it will not be possible with the SOEP data to ana- 4.1 The evolution of educational wage differences lyse the variety of study subjects described in Sect.  2.2, In the years under investigation, 1996 to 2019, the aver- we will group and examine seven majors: arts, law, eco- age real wages in our samples doubled (Additional file  1: nomics, social sciences, medicine, natural sciences, and Table  S2). On average, they grew annually by 3.33% engineering. among women and 3.32% among men (Additional file  1: We control for the individual potential work experience Table  S3). This significant growth is, at least to some subdivided into percentiles, sex, migration background, extent, the result of the stable performance of the Ger- partner, employment of the partner, children in the man economy (e.g., Burda and Seele 2017, 2020; Dust- household, city vs. country living, West vs. East Germany, mann et  al. 2014). The wage growth rates vary between and for the sample the respondent initially belonged to. the educational categories. Workers with a degree from Additional file  1: Table S1 in the contains descriptive sta- UAS experienced above average growth rates (especially tistics of all variables. women, at 3.92%, and to a lesser extent men, at 3.53%), Method. We start with the average wage differences and workers with no degree below average growth rates of the three higher educational degrees compared to a (2.78% for women, 2.47% for men). Women with an VET degree. Subsequently, we estimate adjusted educa- MC certificate experienced below average growth rates tional wage differentials. Based on OLS wage regression (3.17%), men above (3.51%). Women with a U degree experienced below average growth rates while men with a U degree experience an average growth rate. Despite the significant decrease in the share of workers with a VET For reasons of robustness, we performed additional regressions which degree, their wages also grew below the average (women: include three additional dummy variables, one for cooperative education, one for civil servant training, and one for higher education degrees obtained in 3.05%, and men 3.07%). a foreign country, and found no differences for the adjusted U differentials. The growth rates vary between the age groups of There are some moderate differences for UAS and MC which are discussed in younger (30 to 39  years old) and older (40 to 55  years Sect. 4.2 below. old) workers (Additional file  1: Table  S3). They are, Further estimates were calculated separately for younger (30–39) and older workers (40–55), based on samples of workers aged 25 to 65, and on average, higher for the samples of younger women separately for workers from Eastern and Western Germany. Additional with a tertiary degree compared to the samples of older checks restricted the sample to employees only, without the self-employed. Furthermore, we estimated the wage differentials separating VET gradu- ates into those who attained an Abitur as formal entrance certification into Footnote higher education and those who had a lower secondary degree. The regres - 3 (continued) sion tables for these additional findings as well as the number of observa- tions for each year and the adjusted R of our main analyses are available in Ordemann and Pfeiffer (2021). 17 Page 6 of 12 J. Ordemann , F. Pfeiffer Fig. 2 Smoothed educational wage differences, 1996 to 2019 ( Women/Men; in ln, 95%- CI). Note: Employed individuals aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. Differences in the ln of real wages compared to VET. The average mean differences of no degree to VET for men are −0.13 and for women −0.17. The average real wages can be found in Additional file 1: Table S2. Source: SOEP v36, authors’ own calculations women with the same degree, which reflects the pro - increase only until 2008, and decline from 2014 onward. cess of upskilling among young women in particular. As a result, average wage differences for U and UAS con - Among men, younger workers experienced moderately verge towards the end of the observation period in 2019. lower wage growth, except for those with a VET and MC degree. Here, younger men experience a stronger wage 4.2 The evolution of educational wage differentials growth than older men. The evolution of the adjusted three higher educational Figure  2 displays the smoothed average differences wage differentials for U and UAS degrees, as well as for of the natural logarithm of the educational wage of the MC in comparison to a VET qualification is shown in three highest educational degrees relative to the VET Fig.  3 for the period 1996 to 2019. The adjusted educa - degree from 1996 to 2019. All three categories of higher tional wage differentials display, by and large, a similar education show an upward trend in wage differences evolutionary pattern as the educational wage differences compared to the VET degree. Women (on the left side) in Fig.  2 above. However, the adjusted wage differentials with a U degree have higher wage differences compared for men with a U degree are slightly higher than the mean to men (right side), although the gap narrows toward the wage differences. The adjusted wage differentials for end of the observation period. In 2019, the average wage women and men workers with a U degree are decreas- differentials are 0.45 log points for women and 0.42 log ing after 2012. The decrease amounts to about 0.045 points for men. For UAS, the differences among women log points. Although the decrease is not statistically sig- are significantly lower compared to the ones among men. nificant, given the overlapping confidence intervals, it is In addition, the differences for a UAS degree are more economically meaningful. Compared to the highest esti- similar to those of an MC degree for women and more mated wage differentials for U workers so far, which was similar to a U degree for men. about 0.5 log points around 2012, it is nearly ten percent The three educational wage differences among women lower in 2019. increase until 2012 and then stagnate (MC, UAS) or The adjusted wage differentials were higher for women decline (U). Among men, the pattern differs slightly. The at the beginning of the observation period while they are average wage differences increased steadily for MC and of a similar magnitude among both men and women at UAC after 2000. Growth slows down after 2014. How - the end. The convergence of the adjusted educational ever, for male workers graduating from U, the differences wage differentials for the group of female and male The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 7 of 12 17 Fig. 3 Adjusted smoothed educational wage differentials, 1996 to 2019 ( Women/Men; in ln, 95% CI). Note: Employed individuals, aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. OLS estimates with ln real wages (educational reference category: VET ). Individuals without any degree earn significantly less (for women on average, 19.6%, for men 11.8% without a clear time trend. The wage differentials, including 95% CI based on robust errors, can be found in Additional file 1: Table S4. Source: SOEP v36, authors’ own calculations workers with a U degree may have several causes. One Since the increase of the wage differentials of MC out - cause, presumably, is the significant expansion of uni - performs those of UAS, the difference between the two versity education among women after 2000. Given an is lower in 2019 compared to 1996. For workers with a increasing number of highly educated women entering UAS degree, the estimated wage differentials are higher employment, it may have been no longer necessary for for men compared to women. For workers with an MC firms to increase monetary incentives for the employ - certificate, the estimated wage differentials are higher ment participation of women compared to men. Such for women compared to men. Among women, the wage an explanation assumes that women and men with a U differentials between UAS and MC workers do not sta - degree compete in comparable economic segments and tistically differ in the observation period. However, the are substitutes at this aggregate level. There is some evi - estimated coefficients are always higher for UAS com - dence to support this idea from Francesconi and Parey pared to MC. (2018), who find that there is no gender wage gap at the As a robustness check, we performed additional regres- beginning of the career. A second cause may result from sions in which a degree from cooperative education, civil a change in the composition of subject choice as subjects servant training, and a higher education degree obtained differ in their content, prestige, and expected wages. We in a foreign country were added in the form of dummy further investigate this potential cause in more detail in variables in the estimation equation instead of including the next section. them in the categories of the MC or UAS degrees. The In addition, Fig.  3 indicates a stronger increase for comparison reveals that there are virtually no differences UAS and MC compared to U throughout the observa- in the estimated coefficients for the adjusted U differen - tion period for men. Thus the wage differentials between tial. In contrast, the UAS differentials are, on average, U and UAS, which were relatively high and significantly 0.01 log points higher over all 24 coefficients for both different around 2012 converge towards the end of the women and men. The adjusted wage differentials for MC, observation period. While on average the adjusted differ - averaged over all 24 estimates, turned out to be 0.02 log ential for U is still higher compared to UAS, the confi - points lower for women and 0.01 log points higher for dence intervals overlap in 2019. men. These later findings are interesting on their own and The adjusted wage differentials for men with a UAS may even deserve additional research to better capture degree are always significantly higher compared to MC. the diversity of higher education degrees in Germany. 17 Page 8 of 12 J. Ordemann , F. Pfeiffer Table 1 Share of first degrees in study majors from UAS and U, 1993 and 2011 (in %). Type UAS U Sex Women Men Women Men Year 1993 2011 1993 2011 1993 2011 1993 2011 Arts – – – – 20.7 26.5 7.6 11.7 Law – – – – 7.6 4.4 7.5 4.6 Economics 40.9 41.5 25.3 27.6 11.3 11.0 15.1 15.9 Social Sc 21.5 24.5 4.1 5.5 15.3 21.8 5.9 12.5 Medicine – – – – 12.0 6.6 11.9 5.2 Natural Sc – – – – 15.9 15.5 16.8 18.6 Engineering 20.5 19.1 64.1 58.6 6.4 6.1 28.4 27.0 The notation reads as follows: UAS university of applied sciences, U university, Sc sciences. The numbers in columns do not add to 100 percent because not all majors have been included; teachers are included in the group of social sciences Source: DZHW ICE (Federal Statistical Office, Main Reports, 3301); authors’ own calculations However, they are quantitatively not significant enough The findings also reveal that U graduates earn higher to modify our main findings and conclusions. wages compared to UAS graduates in general and in particular when they studied the same major. For exam- 5 Changes in the composition of subjects studied ple, the adjusted wage differential for economists with and educational wage differentials a U degree was 0.60 (0.49) for women (men) in 2012 The returns to university education are heterogeneous and 0.43 (0.42) for women (men) with a UAS degree. and empirically vary between subjects (e.g., Francesconi According to our interpretation, this difference mirrors and Parey 2018; Klein 2016). Table 1 groups the distribu- the higher investment costs since time-to-graduation at tion of subjects in the seven most prominent academic a university lasts 5 to 6  years, on average; In contrast, majors (arts, law, economics, social sciences, medicine, it lasts 3 to 4  years at a university of applied sciences natural science, and engineering) separately for women (Authoring Group NRoE 2020). and men and for U and UAS, and comparing 1993 and The adjusted subject-specific wage differentials are 2011. There appear to be some relevant changes over relatively stable over time, especially among engineer- time. For all graduates, there is an increase of 12.3% ing and law. However, the wage differentials for arts, points in the share of arts and social sciences among social and also for natural sciences decreased, although women and of 10.7% points among men. In comparison, not without variations among women and men. For the share of students graduating in law and medicine, instance, the wage differentials among women graduat - both high-wage subjects, decreased. The share for natural ing at U decreased from 0.42 in 2012 to 0.36 in 2019 for sciences, engineering, and economics remained almost arts (minus 0.06 log points) and from 0.46 to 0.36 for constant. social sciences (minus 0.10 log points). Since the share To assess the possible role these changes may exert on of arts and social sciences increased by 12.3% points educational wage differentials, our analysis proceeds in among women and 10.7 among men between 1993 and two steps: We examine the subject-specific wage differ - 2011 (Table 1), the decrease in these adjusted wage dif- entials before we turn to the strength of their relation- ferentials presumably contributed to the stagnation of ship with the expansion of study majors. Table 2 presents the overall U adjusted wage differential. the adjusted educational wage differentials by major for The graduation growth rates are interpreted as prox - females (part a) and males (part b). Graduates in medi- ies for the expansionary effect that the increasing num - cine, law, economics, engineering, and natural sciences ber of graduates may have had on wages (see Additional experience higher wage differentials relative to employ - file  1: Table S5 in the Online Appendix). We analyse the ees with a VET degree, and compared to arts and social period between 2002 and 2008, when the significant sciences graduates. A straightforward calculation reveals expansion in first degree attainment took place (Orde - an average difference of around 0.18 log points in 2012 mann and Pfeiffer 2021). As universities offer the most for women and 0.28 log points for men between the two encompassing subject portfolio, we concentrate on the poles of wage differentials. changes for university graduates. During this period the graduation growth rates vary between 0.97 among The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 9 of 12 17 Table 2 Educational wage differentials by study majors, four selected years (in ln) (a) Women 2012 2014 2016 2019 UAS U UAS U UAS U UAS U Arts – 0.42*** – 0.35*** – 0.31*** – 0.36*** Law – 0.59*** – 0.56*** – 0.58*** – 0.57*** Economics 0.43*** 0.60*** 0.41*** 0.59*** 0.42*** 0.61*** 0.38*** 0.53*** Social Sc 0.27*** 0.46*** 0.34*** 0.34*** 0.25*** 0.32*** 0.21*** 0.36*** Medicine – 0.83*** – 0.82*** – 0.78*** – 0.96*** Natural Sc – 0.62*** – 0.59*** – 0.47*** – 0.52*** Engineering 0.39*** 0.48*** 0.34*** 0.46*** 0.32*** 0.44*** 0.31*** 0.53*** # obs./R 2,900 0.22 2,409 0.23 2,045 0.22 1,990 0.23 (b) Men 2012 2014 2016 2019 UAS U UAS U UAS U UAS U Arts – 0.24*** – 0.33*** – 0.19* – 0.26*** Law – 0.64*** – 0.68*** – 0.66*** – 0.74*** Economics 0.42*** 0.49*** 0.35*** 0.41*** 0.39*** 0.42*** 0.41*** 0.52*** a a Social Sc 0.21*** 0.37*** 0.07ns 0.34*** 0.28*** 0.34*** 0.10 0.35*** Medicine – 0.71*** – 0.71*** – 0.78*** – 0.71*** Natural Sc – 0.56*** – 0.57*** – 0.55*** – 0.48*** Engineering 0.48*** 0.54*** 0.48*** 0.55*** 0.46*** 0.50*** 0.51*** 0.58*** # obs./R 2,877 0.32 2,383 0.32 1,960 0.30 1,883 0.33 Employed individuals aged 30 to 55. The notation reads as follows: UAS university of applied sciences, U university, Sc sciences. OLS estimates with ln real wages (educational reference category: VET ). Statistical significance level of the estimates: * for p < .05, ** for p < 0.01, and *** for p < .001. Findings for other subjects, such as agricultural studies and fine arts are not included in the table. The estimated coefficient for the subject in the year was implausible. We therefore report the coefficient from the previous/following year’s estimates Source: SOEP v36; authors’ own calculations rates of women graduates have been higher compared to women for arts and −0.36 among men for law. We those of men. expect a delayed effect of the expansion on the labour Figure  4 summarises the resulting strength of the rela- market. Therefore, we investigate how the wage differ - tionship between the expansion in study majors (gradu- entials changed between 2012 and 2019. The growth ate growth rates) and subject-specific educational wage rates in subject-specific wage differentials range for differentials. Overall, the figure suggests that there is a women from −0.10 for the social sciences to 0.13 for negative relationship between the two growth rates for medicine, and for men from −0.08 in natural sciences university graduates. On average, a ten percent increase to 0.10 in law (see Additional file 1 : Table S5). in graduates from a specific subject is associated with a For instance, the graduate growth rate in arts stud- roughly 0.01 log point reduction in the adjusted sub- ied at university was 0.97 for women, and 0.54 for men, ject-specific wage differential relative to VET. This rela - whereas the growth in wage differentials was −0.06 tionship should not be interpreted as a “law”. It is not for women and 0.02 for men. Similarly, the gradu- irrespective of time, subject choice, and economic con- ate growth rate in economics studied at university was ditions. The relationship illustrated in the figure depends 0.90 for women, and 0.33 for men, whereas the growth on the specific conditions and socio-economic circum - in wage differentials was −0.07 for women and 0.03 for stances at the time when the educational expansion men. Majors that expanded most in terms of graduates started, i.e., the existing stock of graduates, the strength between 2002 and 2008 are also, on average, subject to of the expansion, the rate of retirement among lower- the strongest decreases in wage differentials. However, skilled workers, as well as the economic circumstances there are two that act as outliers in this relationship: engi- when the graduates begin their careers (e.g., Goldin and neering and medicine retained higher wage differentials Katz 2008). among women compared to men, although the growth 17 Page 10 of 12 J. Ordemann , F. Pfeiffer Fig. 4 The change of subject-specific wage differentials and graduate growth rates. Note: The notation reads as follows: sc sciences. Illustration based on findings presented in Additional file 1: Table S5. 6 Summary and open questions wage majors such as the arts and social sciences during This study investigates the evolution of educational wage the university expansion post-2000. According to our differentials for three categories of higher education interpretation, there is some initial evidence that the halt compared to a VET degree in Germany during the period in the increase of the adjusted U wage differentials, the of expansion of university education. It has been docu- expansion of university education and the changes in the mented that in the German context, the three categories composition of majors studied during this expansion are of tertiary education matter for the evolution of wage dif- related—a conclusion that is underlined by international ferentials. OLS findings demonstrate that the adjusted evidence regarding the importance of subject choice for wage differentials for university graduates increased until wage differentials (e.g., Machin and Puhani 2003; Michel - around 2012 to about 0.5 log points before they stagnate more and Sassler 2016). for some years and decline after 2015. In 2019, the wage A number of open questions remain. After 2012, the differential is assessed as being about 0.045 log point Bologna Process, with its introduction of first- and sec - lower compared to 2012, which nearly amounts to a ten ond-cycle degrees may have influenced the evolution of percent decrease. the adjusted educational wage differentials. If the share of The adjusted wage differentials for men with a U and workers with a bachelor’s degree from a university in our a UAS degree converged. The adjusted wage differentials samples is increasing this may also have contributed to for men with a UAS degree increased during the obser- the stagnation of the adjusted U wage differentials. Due vation period as well, although the speed of the increase to data restrictions, we are not able to assess this pos- slowed down after 2015. The adjusted wage differentials sibility. In our estimation samples, the case numbers of for men with an MC qualification increased steadily, workers with a UAS bachelor’s degree doubled between although they always remain significantly lower com - 2014 and 2019 (from 56 to 110). Since these workers pared to UAS. Among women, the wage differentials for have lower wage differentials compared to workers with U are significantly higher compared to UAS throughout a Diploma or a master’s degree, this development may the observation period, while the wage differentials for have contributed to the stagnation of the increase in UAS UAS increase only slightly, and for MC stagnate towards adjusted wage differentials after 2015. However, most the end of the observation period. The study documents bachelor graduates from universities seem to go on to an above average growth in graduates of lower average The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 11 of 12 17 attain a master’s degree (Authoring Group NRoE 2018). u-shape pattern, such that wages decline when working In our samples, there are no workers with a U Bachelor’s hours per week exceed specific benchmarks such as the degree. average working hours. If working hours exceed such a In the last few decades, more students with a lower benchmark, productivity and wages may start to decline. preference for education at a university may have Such a relationship may have restricted the expansion of entered universities and this type of self-selection may working hours in times of increasing wage differentials have, in turn, contributed to the stagnation of U wage and even caused working hours to decline. The decrease differentials (e.g., Carneiro and Lee 2011; Kroher et  al. in working hours may have contributed to the stagnation 2021; Ordemann 2021). However, the incentive to enter of the adjusted U wage differentials after 2014. It may university may also have been fostered by expectations indicate, for instance, a higher leisure preference result- about a higher future demand for graduates driven by ing in a decrease in the demand for goods and the stagna- technological change. According to Dauth et al. (2021), tion of the U wage differentials. continued automation in Germany positively affected Although the adjusted wage differentials for U are still incumbent high-skilled workers, decreased the demand higher in 2019 compared to 1996, one may ask whether for workers with vocational education and increased a period of decreasing wage differentials is ahead and the incentives for talented young adults to enter univer- how it might develop. Our empirical approach should sity instead of vocational education. Since we have nei- be useful for thinking about this question. For West ther information on university preference parameters Germany, earlier findings by Gebel and Pfeiffer (2010) nor expectations about future labour market prospects, suggest that average returns to education between 1984 it is left for further research to look deeper into these and 2006 reached a minimum in 1996. The current types of explanation. study suggests that the adjusted U wage differential for Another hypothesis posits that the composition of university graduates was highest around 2012 and in workers with a VET qualification may have changed with 2019 was still above the values in 1996. However, the respect to formal secondary education between 1996 and labour market momentum of university expansion is 2019. While more school leavers graduated with the for- ongoing and there may be room for a further decline mal certificate (“Abitur”) needed to enter higher educa - in the future. It remains a task for future research to tion, the wish to attain a similar position to their parents assess the further evolution of educational wage differ - may have diverted many of them into VET (Konietzka entials and their underlying forces. and Hensel 2017). In this process, the human capital of workers with a VET qualification may have increased Supplementary Information over time, contributing to the stagnation of the wage dif- The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12651- 022- 00323-6. ferentials compared to U workers. Indeed, further regres- sion analysis confirmed that VET workers with an Abitur Additional file 1: Table S1. Descriptive statistics from the (pooled) earned significantly higher wages than those with a VET estimation samples, 1996 to 2019 ( Total/Women/Men). Table S2. Average qualification but without an Abitur in almost all years. real wages by educational degrees, 1996 to 2019 ( Women/Men; Mean). Table S3. Average real wages and annual wage growth by educational However, this estimated wage premium decreased in the degrees ( Women/Men; in EUR and in %). Table S4. Educational wage years 2015/19 compared to the years 2008/12 according differentials, 1996 to 2019 ( Women/Men; in ln, [CI]). Table S5. Dynamics of to our regressions, for women from 0.174 to 0.142 and for degrees and subject-specific wage differentials (in thousands/in %). men from 0.211 to 0.162. This finding does not seem to support the change-in-VET-hypothesis, although it is not Acknowledgements We would like to thank Michael Gebel, Sarah McNamara, Matthias Parey, a formal falsification. Future research on the relevance of Frauke Peter, Stephan Thomsen, and Thomas Zwick as well as the editor and these considerations and the influence of possible other the anonymous reviewers for many insightful and very helpful comments. All changes to the contents of VET could be helpful. remaining errors are our own. Ordemann and Pfeiffer (2021) analysed the gender spe - Author contributions cific participation rates and working hours, based on the All authors read and approved the final manuscript. same samples from the SOEP used in the present study. Funding They report increasing participation rates between 1996 Not applicable. and 2019. Women’s participation rates were still lower than men’s in 2019, although the gap narrowed. Quite Availability of data and materials The dataset analysed during the current study is available at the Research surprisingly, despite upskilling, the average hours worked Data Center of the Socio-Economic Panel, https:// doi. org/ 10. 5684/ soep. core. in the samples of workers aged 30 to 55 decreased from v36. The additional statistical information is based on data of Destatis and is 39.4 h in 1996 to 37.1 h in 2019. According to Bick et al. available at https:// icela nd. dzhw. eu. (2019), working hours and wages display an inverted 17 Page 12 of 12 J. Ordemann , F. Pfeiffer Gernandt, J., Pfeiffer, F.: Wage convergence and inequality after unification: Declarations (East) Germany in transition. In: Kanbur, R., Svejnar, J. (eds.) Labor Market and Development, pp. 387–404. Routledge, London (2009) Ethics approval and consent to participate Goebel, J., Grabka, M.M., Liebig, S., Kroh, M., Richter, D., Schröder, C., Schupp, Not applicable. J.: The German Socio-Economic Panel (SOEP). J. Econ. Stat. 239, 345–360 (2019) Competing interests Goldin, G., Katz, L.F.: The race between education and technology. Havard The authors declare that they have no competing interests. University Press, London and Massachusetts (2008) Green, F., Henseke, G.: Europe’s evolving graduate labour markets: supply, Author details demand, underemployment and pay. J. Lab. Mar. Res. 55, 1–13 (2021) German Centre for Higher Education Research and Science Studies (DZHW ), Hillmert, S., Jacob, M.: Social inequality in higher education is vocational train- Lange Laube 12, 30159 Hannover, Germany. ZEW Leibniz Centre for Euro- ing a pathway leading to or away from university. Europ. Soc. Rev. (2003). pean Economic Research, L7, 1, 68161 Mannheim, Germany. https:// doi. org/ 10. 1093/ esr/ 19.3. 319 Horowitz, J.: Relative education and the advantage of a college degree. Am. Received: 7 December 2021 Accepted: 5 October 2022 Soc. Rev. 83, 771–801 (2018) Kamhöfer, D., Schmitz, H., Westphal, M.: Heterogeneity in marginal non-mon- etary returns to higher education. J. Europ. Econ. Assoc. 17(1), 205–244 (2019) Klein, M.: The association between graduate’s field of study and occupational References attainment in West Germany, 1980–2008. J. Lab. Mar. Res. 49, 43–58 Anger, S., Heineck, G.: Cognitive abilities and earnings – first evidence for (2016) Germany. Appl. Econ. Let. 17, 699–702 (2010). https:// doi. org/ 10. 1080/ Konietzka, D., Hensel, T.: Berufliche Erstausbildung im Lebensverlauf: 13504 85080 22978 55 Grundlagen und empirische Befunde. In: Becker, R. (ed.) Lehrbuch der Araki, S.: Educational expansion, skill diffusion, and the economic value of Bildungssoziologie, pp. 281–309. Springer, Wiesbaden (2017) credentials and skills. Am. Soc. Rev. 85(2), 128–175 (2020) Kroher, M., Leuze, K., Thomsen S. L., Trunzer J.: Did the “Bologna Process” Authoring Group NRoE: Bildung in Deutschland 2018. Ein indikatorengestütz- achieve its goals? 20 years of empirical evidence on student enrolment, ter Bericht mit einer Analyse zu Wirkungen und Erträgen von Bildung. study success and labour market outcomes. IZA DP No. 14757, Bonn wbv Verlag, Bielefeld (2018). (2021) Authoring Group NRoE: Bildung in Deutschland 2020. Ein indikatorengestütz- Krueger, D., Schkade, A.B.: Sorting in the labor market: do gregarious workers ter Bericht mit einer Analyse zu Bildung in einer digitalisierten Welt. wbv flock to interactive jobs? J. Hum. Res. 43(4), 859–883 (2008) Verlag, Bielefeld (2020). Lindley, J., Machin, S.: The rising postgraduate wage premium. Economica 83, Backes-Gellner, U., Herz, H., Kosfeld, M., Oswald, Y.: Do preferences and biases 281–306 (2016) predict life outcomes? Evidence from education and labor market entry Machin, S., Puhani, P.A.: Subject of degree and the gender wage differential: decisions Econ. Rev. Europ. 134, 103709 (2021) evidence from the UK and Germany. Econ. Let. 79, 393–400 (2003) Baethge, M.: Das deutsche Bildungs-Schisma: Welche Probleme ein vorin- Michelmore, K., Sassler, S.: Explaining the gender wage gap in stem: does field dustrielles Bildungssystem in einer nachindustriellen Gesellschaft hat. sex composition matter? R. S. F. J. Soc. Sci. 2, 194–215 (2016) In: Lemmermöhle, D., Hasselhorn, M. (eds.) Bildung-Lernen, pp. 93–116. Müller, W., Pollak, R.: Weshalb gibt es so wenige Arbeiterkinder in Deutschlands Wallstein, Göttingen (2007) Universitäten? In: Becker, R., Lauterbach, W. (eds.) Bildung als Privileg, pp. Becker, R., Hecken, A.E.: Warum werden Arbeiterkinder vom Studium an 303–342. Springer, Wiesbaden (2007) Universitäten abgelenkt? Eine empirische Überprüfung der „Ablenkung- Ordemann, J.: Studium ohne Abitur. Bildungserträge nichttraditioneller Hochs- sthese“ von Müller und Pollak (2007) und Hillmert und Jacob (2003). Köl. chulabsolventen im Vergleich. Springer, Wiesbaden (2019) Z. Soz. 60(1), 7–33 (2013) Ordemann, J., Pfeiffer, F.: The evolution of educational wage differentials for Bick, A., Brüggemann, B., Fuchs-Schündeln, N.: Hours worked in Europe and women and men, from 1996 to 2019. ZEW Disc Pap 21–066, 1–55 (2021) the US: new data, new answers. Scan. J. Econ. 121, 1381–1416 (2019) Ordemann, J.: Academic Pay Gap 2015. A Snapshot of the within difference of Brändle, T., Ordemann, J.: Same same but different? Non-traditional students higher education graduates income (unpublished manuscript) (2021) and alumni in Germany. Stud. Paed. 25, 35–50 (2020) Pfeiffer, F., Pohlmeier, W.: Income, uncertainty and the probability of self- Burda, M.C., Seele, S.: Das deutsche Arbeitsmarktwunder: Eine Bilanz. Persp. employment. Rech. Econ. Louv. 58, 265–281 (1992) Wpol. 18, 179–204 (2017) Pfeiffer, F., Stichnoth, H.: Fiskalische und individuelle Bildungsrenditen - Burda, M.C., Seele, S.: Reevaluating the German labor market miracle. Ger. aktuelle Befunde für Deutschland. Persp. Wpol. 16, 393–411 (2015) Econ. Rev. 21, 139–179 (2020) Reinhold, M., Thomsen, S.L.: The changing situation of labor market entrants in Carneiro, P., Lee, S.: Trends in quality-adjusted skill premia in the United States, Germany. J. L. M. R. 50, 161–174 (2017) 1960–2000. Am. Econ. Rev. 101, 2309–2349 (2011) SOEP-Core v36: Socio-Economic Panel (SOEP), data for years 1984–2019, EU Dauth, W., Findeisen, S., Südekum, J., Wösser, N.: The adjustment of labor Edition. https:// doi. org/ 10. 5684/ soep. core. v36eu (2021). markets to robots. J. Europ. Econ. Ass. (2021). https:// doi. org/ 10. 1093/ Valletta, R.G.: Recent flattening in the highereducation wage premium: polari- jeea/ jvab0 12 zation, skill downgrading, or both? In: Education, Skills, and Technical Dustmann, C., Fitzenberger, B., Schönberg, U., Spitz-Oener, A.: From sick man Change: implications for Future US GDP Growth, pp. 313–342. Nat. Bur. of Europe to economic superstar: Germany’s resurgent economy. J. Econ. Econ. Res, Massachusetts (2018) Persp. 28, 167–188 (2014) Westphal, M., Kamhöfer, D.A., Schmitz, H.: Marginal college wage premiums Flossmann, A., Pohlmeier, W.: Causal returns to education: a survey on empiri- under selection into employment. Econ. J. (2022). https:// doi. org/ 10. cal evidence for Germany. J. Econ. Stat. 226, 6–23 (2006) 1093/ ej/ ueac0 21 Francesconi, M., Parey, M.: Early gender gaps among university graduates. Europ. Econ. Rev. 109, 63–82 (2018) Gebel, M., Pfeiffer, F.: Educational expansion and its heterogeneous returns for Publisher’s Note wage workers. J. Appl. Soc. Sci. S. 130, 19–42 (2010) Springer Nature remains neutral with regard to jurisdictional claims in pub- Gebel, M., Heineck, G.: Returns to education in the life course. In: Becker, R. (ed.) lished maps and institutional affiliations. Research Handbook on the Sociology of Education, pp. 454–475. Edward Elgar Publishing, Cheltenham, UK, Northhampton, MA (2019) Gernandt, J., Pfeiffer, F.: Rising wage inequality in germany. J. Econ. Stat. 227, 358–380 (2007) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal for Labour Market Research Springer Journals

The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019

Loading next page...
 
/lp/springer-journals/the-evolution-of-educational-wage-differentials-for-women-and-men-in-ksawYLU0KO
Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
ISSN
1614-3485
eISSN
2510-5027
DOI
10.1186/s12651-022-00323-6
Publisher site
See Article on Publisher Site

Abstract

This paper studies the evolution of three higher education wage differentials from 1996 to 2019 in Germany. We distinguish between degrees from academic universities, degrees from universities of applied sciences, and the mas- ter craftsman\craftswoman certificate. The educational reference category is a standard degree within the German vocational education and training system. Based on samples of male and female workers from the Socio-Economic Panel Study (SOEP), regression methods show that all three educational wage differentials in 2019 exceeded the ones in 1996. However, workers graduating from universities experienced an inverse u-shape pattern with a maximum of about 0.5 log points around 2012. Since then, their wage differential decreased by nearly ten percent (about 0.045 log points). Although the decrease is not statistically significant at conventional levels, we think that nearly ten per - cent can be regarded as economically meaningful. We argue that this pattern is related to university expansion and changes in graduates’ subject-choice composition during that expansion. The paper concludes with a discussion of possible alternative explanations. Keywords: Educational Wage Differentials, Gender Gaps, Higher Education, Returns to Education JEL Classification: J31, J16, I23, I26 1 Introduction may foster innovation and trade, boosting investment The expansion of university education fostered a dynamic into new capital-intensive automation technologies like change in the workforce’s educational composition, and artificial intelligence. This paper studies whether relative which has received attention from policymakers and wages of high-skilled individuals increased or decreased researchers (e.g., Authoring Group NRoE 2018; Araki during the recent expansion of university education in 2020; Goldin and Katz 2008; Horowitz 2018). The debate Germany. revolves around whether the increase of highly edu- The literature often focuses on two educational cat - cated individuals may, presumably with some lag, lead egories, college graduates and others, while the result- to stronger competition among university graduates ing wage differential is referred to as the skill (or college and put pressure on their relative wages. Alternatively, wage) premium. According to Goldin and Katz (2008), highly educated individuals may experience even larger the skill premium in the United States increased from wage differentials (and vice versa for low-skilled indi - 1980 to 2005, which they explain as resulting from a tech- viduals). The increase in university-educated individuals nology-driven growth in the demand for college gradu- ates and non-routine tasks (see also Lindley and Machin 2016, among others). More recent evidence suggests a *Correspondence: ordemann@dzhw.eu stagnating skill premium in the US after 2010 (Valletta German Centre for Higher Education Research and Science Studies (DZHW ), 2018) and a moderate decrease in college wage premium Lange Laube 12, 30159 Hannover, Germany in European countries between 2005 and 2015 (Green Full list of author information is available at the end of the article © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. 17 Page 2 of 12 J. Ordemann , F. Pfeiffer and Henseke 2021). Reinhold and Thomsen (2017) study discusses the new empirical findings on the evolution of the development of daily wages from a cohort of young educational wage differentials. Sect.  5 discusses changes employees based on administrative data in Germany in the subject composition of study choices among grad- until 2010. Differentiating between three skills categories, uates and the evolution of wage differentials. Finally, they find that high-skilled employees, compared to mid - Sect.  6 concludes with a discussion on further factors dle and low-skilled employees, experienced a rising skill that may have contributed to our main findings. premium. The German higher education system presumably 2 The evolution of the educational composition is more differentiated than the Anglo-Saxon one. Tak - of the population, aged 30 to 55 ing the perspective of one higher education premium 2.1 Education and wages ignores the fact that there are at least three specific and Economic reasoning suggests that when young adults well-defined higher educational degree categories in Ger - invest in higher education, they compare costs and many (Authoring Group NRoE 2018): Degrees from returns over the life cycle and consider their own educa- academic universities (referred to as universities, U, in tional and occupational preferences (e.g., Backes-Gellner what follows), degrees from universities of applied sci- et  al. 2021; Flossmann and Pohlmeier 2006; Pfeiffer and ences (UAS), and the master craftsman\craftswoman cer- Stichnoth 2015; Westphal et al. 2022). Wage differentials tificate (MC). MC is the highest post-secondary degree emerge in order to compensate for these investments (for outside the university system in Germany. It builds on a other reasons for wage differentials, such as amenities, fourth qualification-type obtained via the dual vocational skills, effort, risk, or social interactions, see e.g., Anger apprenticeship system (named vocational education and Heineck 2010; Gebel and Heineck 2019; Krueger and training, VET). These degrees vary significantly in and Schkade 2008). Educational wage differentials sig - academic content and length of study (see Sect.  2). Our nal investment opportunities and differences in the effort study contributes to the international literature in a novel needed to acquire specific degrees. In a thought experi - way by looking at the evolution of these three higher edu- ment where wage differentials would be zero, the eco - cational (gross) wage differentials compared to a VET nomic incentives for educational investments would be degree using data from the Socio-Economic Panel Study low or absent. (SOEP) from 1996 to 2019. In addition, we consider the Since wages and educational wage differentials result major studied while at a university of applied sciences from several economy-wide and individual-specific fac - or university to highlight recent changes in the student tors, identifying specific factors strong enough to change composition of these majors and their potential relation their trajectory can be challenging. The part of the wage to educational wage differentials. attributed to the level of education depends on the com- Similar to the literature, we find that in Germany, in the petencies attained in formal educational institutions. period from 1996 to 2019, educational wage differentials In addition, individuals select themselves into these increased despite the higher education expansion and the institutions depending on their perceived abilities and higher educated workforce participation rates. We fur- socio-economic background (e.g., Becker and Hecken ther this field of literature by demonstrating specific evo - 2008; Hillmert and Jacob 2003; Müller and Pollak 2007), lutionary patterns for each educational degree—a pattern highlighting the role of preferences for academic educa- that evolves differently for women and men. While all tion (see Kamhöfer et al. 2019) as well as otherwise often three educational wage differentials increased compared unobserved barriers to and benefits from educational to a VET degree, workers graduating from universities pathways. experienced a halt starting around 2012, when it reached The overall amount of educational investment in soci - its highest level so far (about 0.5 log points) and after- ety, or the wage levels for a given competence profile are ward a decrease of about 0.045 log points. Although the driven by factors that the investing individual, as a rule, decrease is not statistically significant at the conventional cannot control. Thus, educational wage differentials level of 95%, we think it is economically meaningful. We depend on the amount and quality of educational invest- argue that this pattern is related to the expansion of uni- ments, on competencies that are not certified or are versity education. Furthermore, relatively more students hard to certify, and on factors that determine the over- graduated in arts and social sciences, subjects for which all supply of and demand for these competencies in the wage differentials are lower than for other majors. economy. The article proceeds as follows. Sect.  2 introduces the chosen educational categories and highlights the expan- 2.2 Educational categories used in the study sion of university education. Sect.  3 describes the data, The German education system has traditionally been its operationalization, and our research approach. Sect. 4 stratified and separated between occupational and The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 3 of 12 17 academic pillars (Authoring Group NRoE 2018; Baethge acquisition of the MC certificate then lasts an additional 2007). There are two types of academic educational insti - 2 to 3 years. tutions available: universities and universities of applied sciences. Both types of institutions vary in their aca- 2.3 Dynamic upskilling in the population aged 30 to 55 demic orientation and the subjects offered. Universities Germany experienced a significant expansion in univer - of applied sciences focus on a practically relevant set of sity education in the new millennium, which changed the qualifications predominantly in economics (as a rule, educational composition of the German population (e.g., business economics), social work, and engineering. They Authoring Group NRoE 2018). Figure  1 illustrates the often have strong ties to the local economy. Universities evolution of the educational composition in the popula- are broader in their portfolio and offer subjects encom - tion aged 30 to 55, separately for women and men, based passing the entire academic spectrum from the arts, on samples taken from SOEP. economics and business economics, social and natu- Note that the scales used for VET (left side) and the ral sciences, and sports to law, medicine, and veterinary three higher educational degrees (right side) differ. Over medicine. the 24  years considered here, highly educated young In Germany, matriculation at universities of applied people steadily entered the observed age group whereas sciences and universities requires an entrance qualifica - older and less educated people left it. As a result, the tion, which demands successful graduation from upper share of individuals with a U (UAS) degree increased secondary schooling (typically after 12 or 13  years of from 10.8 (10.8) percent in 1996 to 17.8 (11.9) percent overall schooling). Since 2009, both offer two degrees, in 2019. The share of individuals with a VET degree bachelor’s and master’s degree,  gradually replacing tra- decreased from 60.0 percent in 1996 to 50.5 percent in ditional degrees such as the “Diplom”. In addition, uni- 2019. Upskilling among women is a driver of the change versities offer the “Staatsexamen” for teaching, medicine, in the educational composition from 1996 to 2019. The and law, which is, by and large, comparable to a mas- share of women with a U degree increased by 7.8% points ter’s degree. A bachelor’s degree requires an investment and that of men by 6.0% points. In this period, the share of at least three years and a master’s degree of at least of women with UAS degrees remained approximately the two additional years. In addition to choosing the type same (−0.0% points) while the share of men increased by of degree and higher education institution, students can 2.1% points. Summing up, in 2019, approximately 28.0 also choose between a wealth of different subjects (from percent of women and 30.6 percent of men in the 30 to among more than 17,000 different courses, Authoring 55 age bracket held either a UAS or a U degree, compared Group NRoE 2018). to 20.8 percent for women and 22.5 percent for men in There is a third avenue to achieving a tertiary educa - 1996. More women (from 3.5 to 5.6 percent) and fewer tional qualification in Germany: The MC certificate is men (from 12.8 to 11.1 percent) with an MC certificate part of the vocational pillar, the central qualification sys - entered this age bracket. tem besides the academic pillar. It is specific to a craft such as a hairstylist or mechatronic technician and is less 3 The empirical approach for assessing academic in its learning contents. It enables certificate educational wage differentials holders to open their own firms in their respective craft Data. Our empirical analyses of the evolution of wage and opens up supervising positions for them. An MC differentials are based on samples from the German certificate builds on the already acquired qualification SOEP (Goebel et  al. 2019; SOEP-Core 2021). The SOEP of a VET degree, which typically lasts 3 to 4  years. The is a representative longitudinal panel study of German households. It concentrates on multiple topics ranging from employment, well-being, health, working hours, and earnings to daily life. The study began in 1984 and currently includes 36 waves. We use individual-level data Higher education has gradually opened to individuals with vocational train- spanning 24 years from the years 1996 to 2019. Our esti- ing and work experience or an MC certificate who do not otherwise possess a higher education entry qualification. The overall share of these students mation sample is unbalanced and restricted to 259,555 remains below four percent of all students, and less than two percent of all observations of 35,890 employed individuals, 18,455 of alumni (Brändle and Ordemann 2020). We do not investigate the pathways whom are women. The samples include employees and into higher education or this subpopulation. According to Ordemann (2019) they have similar monetary but slightly lower non-monetary labour market the self-employed. Although the wage determining pro- returns then other alumni. Arguably, the content of the educational degrees cesses may differ between the two groups, economic may have differed at the beginning of the observation period from those com - arguments suggest that they are related. While the self- mon in (West) Germany, and this may have influenced the evolution of edu - cational wage differentials. In our previous version of this paper (Ordemann employed have to generate their wages from residual and Pfeiffer 2021), we also performed the analysis separately for East and West profits, employees receive a fixed wage bargained ex-ante Germany. 17 Page 4 of 12 J. Ordemann , F. Pfeiffer Fig. 1 The educational composition in the population 1996–2019 ( Women/Men; in %). Note: Individuals aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. The shares of the four educational categories do not reach 100% because the fifth category, no degree, is excluded in this figure. Their average shares (total) are 11.8% for 1996 and 13.1% for 2019. Source: SOEP v36, authors’ own calculations (e.g., Pfeiffer and Pohlmeier 1992). If a risk-adjusted wage entries into and exits from young adulthood and retire- in self-employment differs from an employee’s wage, ment respectively. workers can become employees and vice versa. The investigation starts in 1996 for two reasons. The We concentrate on female and male prime-age workers first reason is that, according to Gebel and Pfeiffer aged 30 to 55. In this age group, as a rule, individuals are (2010), 1996 was the year in which estimates of the members of the workforce, although participation rates returns to education reached their minimum value in are still higher for the better educated. A difference that the period 1984 to 2006 in West Germany. The period is higher among women compared to men. In our SOEP of strong educational expansion after World War II samples, the share of working women increased from exerted downward pressure on wages for skilled work- 67.6% in 1996 to 85.5% in 2019 and from 91.2 to 92.6% ers, and the estimated returns to education were (mod- among men. More investment in education increases the erately) decreasing from 1984 onward. However, after opportunity cost of not working. Therefore, individu - 1996 estimated returns to education started to increase als with a higher educational degree tend to show higher once again. participation rates compared to VET (for women in The second reason is that German reunification in Germany, also compare Westphal et  al. 2022). However, 1990 may have influenced the German wage structure, the employment participation rates also increased for especially during the years immediately following reuni- women with a VET degree and women with no degree. fication (e.g., Gernandt and Pfeiffer 2007, 2009). Thus by While the participation rates of women increased, men 1996, 6  years after reunification, a relevant part of the still display higher participation rates. Participation rates specific impact of reunification on the educational wage among men exceeded the ones for women by 7.1% points differentials should already have taken place. in 2019, compared to 23.6 in 1996, which is a result in Variables. The dependent variable of our analyses is the part of the upskilling among women (for more details, natural logarithm of gross earnings per hours worked. It see Ordemann and Pfeiffer 2021). Nevertheless, we think is obtained separately for each year by the trimmed real concentrating on the age group of 30 to 55 year old work- gross monthly income reported in the previous month. ers could be helpful in lowering potential estimation The reported income is divided by the factor of 4.33 biases associated with the endogeneity of labour market times the trimmed actual working hours at the end of the sample selection. In addition, the obtained wage was The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 5 of 12 17 trimmed before transforming it into the natural loga- models, we report the partial coefficients for the highest rithm. The trimming of all variables is performed on the educational degrees for each year from 1996 to 2019. one percent level; all incomes are depreciated to 2015. VET is used as the reference group in our analyses. These Additional payments, such as holiday pay are excluded. partial coefficients may not indicate the causal economic The explanatory variable in focus is the highest edu - effect of the educational investments needed to gain the cational degree. We use degrees provided by the SOEP, respective educational degree. which reflect the unique characteristics of the German Estimates of educational wage differences and differ - education system: no (post-secondary) degree (or not yet entials over 24 years may exhibit, to some degree, erratic completed), apprenticeship or vocational training, master patterns from year to year. There is no explicit theory craftsmanship, and university (of applied sciences). supporting the notion that educational wage differences Some adjustments to the variables provided by the and differentials should not display such a pattern. Nev - SOEP are made. First, we add cooperative education and ertheless, we cannot exclude the possibility that part civil servant training to the category of master crafts- of this pattern found in our analyses is the result of the manship. Cooperative education combines vocational various samples retrieved over such a long time period. training with academic study but is still bound to the firm For instance, the number of observations in our estima- with which students have a work contract, who shape tion samples varied from 3681 to 7990. To get rid of such the curricula of the cooperative education institution. types of randomness to some degree for our subsequent Second, higher education degrees obtained in a foreign analyses, we use Epanechnikov kernel-weighted local- country were added to the category of universities of mean polynomial smoothing for the figures produced applied sciences to reflect the diversity of higher educa - from the estimates. tion from all over the world in this educational category. We focus on these three higher education categories, 4 The evolution of educational wage differences although each of them may have further heterogeneities. and differentials, 1996 to 2019 While it will not be possible with the SOEP data to ana- 4.1 The evolution of educational wage differences lyse the variety of study subjects described in Sect.  2.2, In the years under investigation, 1996 to 2019, the aver- we will group and examine seven majors: arts, law, eco- age real wages in our samples doubled (Additional file  1: nomics, social sciences, medicine, natural sciences, and Table  S2). On average, they grew annually by 3.33% engineering. among women and 3.32% among men (Additional file  1: We control for the individual potential work experience Table  S3). This significant growth is, at least to some subdivided into percentiles, sex, migration background, extent, the result of the stable performance of the Ger- partner, employment of the partner, children in the man economy (e.g., Burda and Seele 2017, 2020; Dust- household, city vs. country living, West vs. East Germany, mann et  al. 2014). The wage growth rates vary between and for the sample the respondent initially belonged to. the educational categories. Workers with a degree from Additional file  1: Table S1 in the contains descriptive sta- UAS experienced above average growth rates (especially tistics of all variables. women, at 3.92%, and to a lesser extent men, at 3.53%), Method. We start with the average wage differences and workers with no degree below average growth rates of the three higher educational degrees compared to a (2.78% for women, 2.47% for men). Women with an VET degree. Subsequently, we estimate adjusted educa- MC certificate experienced below average growth rates tional wage differentials. Based on OLS wage regression (3.17%), men above (3.51%). Women with a U degree experienced below average growth rates while men with a U degree experience an average growth rate. Despite the significant decrease in the share of workers with a VET For reasons of robustness, we performed additional regressions which degree, their wages also grew below the average (women: include three additional dummy variables, one for cooperative education, one for civil servant training, and one for higher education degrees obtained in 3.05%, and men 3.07%). a foreign country, and found no differences for the adjusted U differentials. The growth rates vary between the age groups of There are some moderate differences for UAS and MC which are discussed in younger (30 to 39  years old) and older (40 to 55  years Sect. 4.2 below. old) workers (Additional file  1: Table  S3). They are, Further estimates were calculated separately for younger (30–39) and older workers (40–55), based on samples of workers aged 25 to 65, and on average, higher for the samples of younger women separately for workers from Eastern and Western Germany. Additional with a tertiary degree compared to the samples of older checks restricted the sample to employees only, without the self-employed. Furthermore, we estimated the wage differentials separating VET gradu- ates into those who attained an Abitur as formal entrance certification into Footnote higher education and those who had a lower secondary degree. The regres - 3 (continued) sion tables for these additional findings as well as the number of observa- tions for each year and the adjusted R of our main analyses are available in Ordemann and Pfeiffer (2021). 17 Page 6 of 12 J. Ordemann , F. Pfeiffer Fig. 2 Smoothed educational wage differences, 1996 to 2019 ( Women/Men; in ln, 95%- CI). Note: Employed individuals aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. Differences in the ln of real wages compared to VET. The average mean differences of no degree to VET for men are −0.13 and for women −0.17. The average real wages can be found in Additional file 1: Table S2. Source: SOEP v36, authors’ own calculations women with the same degree, which reflects the pro - increase only until 2008, and decline from 2014 onward. cess of upskilling among young women in particular. As a result, average wage differences for U and UAS con - Among men, younger workers experienced moderately verge towards the end of the observation period in 2019. lower wage growth, except for those with a VET and MC degree. Here, younger men experience a stronger wage 4.2 The evolution of educational wage differentials growth than older men. The evolution of the adjusted three higher educational Figure  2 displays the smoothed average differences wage differentials for U and UAS degrees, as well as for of the natural logarithm of the educational wage of the MC in comparison to a VET qualification is shown in three highest educational degrees relative to the VET Fig.  3 for the period 1996 to 2019. The adjusted educa - degree from 1996 to 2019. All three categories of higher tional wage differentials display, by and large, a similar education show an upward trend in wage differences evolutionary pattern as the educational wage differences compared to the VET degree. Women (on the left side) in Fig.  2 above. However, the adjusted wage differentials with a U degree have higher wage differences compared for men with a U degree are slightly higher than the mean to men (right side), although the gap narrows toward the wage differences. The adjusted wage differentials for end of the observation period. In 2019, the average wage women and men workers with a U degree are decreas- differentials are 0.45 log points for women and 0.42 log ing after 2012. The decrease amounts to about 0.045 points for men. For UAS, the differences among women log points. Although the decrease is not statistically sig- are significantly lower compared to the ones among men. nificant, given the overlapping confidence intervals, it is In addition, the differences for a UAS degree are more economically meaningful. Compared to the highest esti- similar to those of an MC degree for women and more mated wage differentials for U workers so far, which was similar to a U degree for men. about 0.5 log points around 2012, it is nearly ten percent The three educational wage differences among women lower in 2019. increase until 2012 and then stagnate (MC, UAS) or The adjusted wage differentials were higher for women decline (U). Among men, the pattern differs slightly. The at the beginning of the observation period while they are average wage differences increased steadily for MC and of a similar magnitude among both men and women at UAC after 2000. Growth slows down after 2014. How - the end. The convergence of the adjusted educational ever, for male workers graduating from U, the differences wage differentials for the group of female and male The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 7 of 12 17 Fig. 3 Adjusted smoothed educational wage differentials, 1996 to 2019 ( Women/Men; in ln, 95% CI). Note: Employed individuals, aged 30 to 55. The notation reads as follows: VET vocational education and training, MC master craftsmen/craftswomen, UAS university of applied sciences, and U university. OLS estimates with ln real wages (educational reference category: VET ). Individuals without any degree earn significantly less (for women on average, 19.6%, for men 11.8% without a clear time trend. The wage differentials, including 95% CI based on robust errors, can be found in Additional file 1: Table S4. Source: SOEP v36, authors’ own calculations workers with a U degree may have several causes. One Since the increase of the wage differentials of MC out - cause, presumably, is the significant expansion of uni - performs those of UAS, the difference between the two versity education among women after 2000. Given an is lower in 2019 compared to 1996. For workers with a increasing number of highly educated women entering UAS degree, the estimated wage differentials are higher employment, it may have been no longer necessary for for men compared to women. For workers with an MC firms to increase monetary incentives for the employ - certificate, the estimated wage differentials are higher ment participation of women compared to men. Such for women compared to men. Among women, the wage an explanation assumes that women and men with a U differentials between UAS and MC workers do not sta - degree compete in comparable economic segments and tistically differ in the observation period. However, the are substitutes at this aggregate level. There is some evi - estimated coefficients are always higher for UAS com - dence to support this idea from Francesconi and Parey pared to MC. (2018), who find that there is no gender wage gap at the As a robustness check, we performed additional regres- beginning of the career. A second cause may result from sions in which a degree from cooperative education, civil a change in the composition of subject choice as subjects servant training, and a higher education degree obtained differ in their content, prestige, and expected wages. We in a foreign country were added in the form of dummy further investigate this potential cause in more detail in variables in the estimation equation instead of including the next section. them in the categories of the MC or UAS degrees. The In addition, Fig.  3 indicates a stronger increase for comparison reveals that there are virtually no differences UAS and MC compared to U throughout the observa- in the estimated coefficients for the adjusted U differen - tion period for men. Thus the wage differentials between tial. In contrast, the UAS differentials are, on average, U and UAS, which were relatively high and significantly 0.01 log points higher over all 24 coefficients for both different around 2012 converge towards the end of the women and men. The adjusted wage differentials for MC, observation period. While on average the adjusted differ - averaged over all 24 estimates, turned out to be 0.02 log ential for U is still higher compared to UAS, the confi - points lower for women and 0.01 log points higher for dence intervals overlap in 2019. men. These later findings are interesting on their own and The adjusted wage differentials for men with a UAS may even deserve additional research to better capture degree are always significantly higher compared to MC. the diversity of higher education degrees in Germany. 17 Page 8 of 12 J. Ordemann , F. Pfeiffer Table 1 Share of first degrees in study majors from UAS and U, 1993 and 2011 (in %). Type UAS U Sex Women Men Women Men Year 1993 2011 1993 2011 1993 2011 1993 2011 Arts – – – – 20.7 26.5 7.6 11.7 Law – – – – 7.6 4.4 7.5 4.6 Economics 40.9 41.5 25.3 27.6 11.3 11.0 15.1 15.9 Social Sc 21.5 24.5 4.1 5.5 15.3 21.8 5.9 12.5 Medicine – – – – 12.0 6.6 11.9 5.2 Natural Sc – – – – 15.9 15.5 16.8 18.6 Engineering 20.5 19.1 64.1 58.6 6.4 6.1 28.4 27.0 The notation reads as follows: UAS university of applied sciences, U university, Sc sciences. The numbers in columns do not add to 100 percent because not all majors have been included; teachers are included in the group of social sciences Source: DZHW ICE (Federal Statistical Office, Main Reports, 3301); authors’ own calculations However, they are quantitatively not significant enough The findings also reveal that U graduates earn higher to modify our main findings and conclusions. wages compared to UAS graduates in general and in particular when they studied the same major. For exam- 5 Changes in the composition of subjects studied ple, the adjusted wage differential for economists with and educational wage differentials a U degree was 0.60 (0.49) for women (men) in 2012 The returns to university education are heterogeneous and 0.43 (0.42) for women (men) with a UAS degree. and empirically vary between subjects (e.g., Francesconi According to our interpretation, this difference mirrors and Parey 2018; Klein 2016). Table 1 groups the distribu- the higher investment costs since time-to-graduation at tion of subjects in the seven most prominent academic a university lasts 5 to 6  years, on average; In contrast, majors (arts, law, economics, social sciences, medicine, it lasts 3 to 4  years at a university of applied sciences natural science, and engineering) separately for women (Authoring Group NRoE 2020). and men and for U and UAS, and comparing 1993 and The adjusted subject-specific wage differentials are 2011. There appear to be some relevant changes over relatively stable over time, especially among engineer- time. For all graduates, there is an increase of 12.3% ing and law. However, the wage differentials for arts, points in the share of arts and social sciences among social and also for natural sciences decreased, although women and of 10.7% points among men. In comparison, not without variations among women and men. For the share of students graduating in law and medicine, instance, the wage differentials among women graduat - both high-wage subjects, decreased. The share for natural ing at U decreased from 0.42 in 2012 to 0.36 in 2019 for sciences, engineering, and economics remained almost arts (minus 0.06 log points) and from 0.46 to 0.36 for constant. social sciences (minus 0.10 log points). Since the share To assess the possible role these changes may exert on of arts and social sciences increased by 12.3% points educational wage differentials, our analysis proceeds in among women and 10.7 among men between 1993 and two steps: We examine the subject-specific wage differ - 2011 (Table 1), the decrease in these adjusted wage dif- entials before we turn to the strength of their relation- ferentials presumably contributed to the stagnation of ship with the expansion of study majors. Table 2 presents the overall U adjusted wage differential. the adjusted educational wage differentials by major for The graduation growth rates are interpreted as prox - females (part a) and males (part b). Graduates in medi- ies for the expansionary effect that the increasing num - cine, law, economics, engineering, and natural sciences ber of graduates may have had on wages (see Additional experience higher wage differentials relative to employ - file  1: Table S5 in the Online Appendix). We analyse the ees with a VET degree, and compared to arts and social period between 2002 and 2008, when the significant sciences graduates. A straightforward calculation reveals expansion in first degree attainment took place (Orde - an average difference of around 0.18 log points in 2012 mann and Pfeiffer 2021). As universities offer the most for women and 0.28 log points for men between the two encompassing subject portfolio, we concentrate on the poles of wage differentials. changes for university graduates. During this period the graduation growth rates vary between 0.97 among The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 9 of 12 17 Table 2 Educational wage differentials by study majors, four selected years (in ln) (a) Women 2012 2014 2016 2019 UAS U UAS U UAS U UAS U Arts – 0.42*** – 0.35*** – 0.31*** – 0.36*** Law – 0.59*** – 0.56*** – 0.58*** – 0.57*** Economics 0.43*** 0.60*** 0.41*** 0.59*** 0.42*** 0.61*** 0.38*** 0.53*** Social Sc 0.27*** 0.46*** 0.34*** 0.34*** 0.25*** 0.32*** 0.21*** 0.36*** Medicine – 0.83*** – 0.82*** – 0.78*** – 0.96*** Natural Sc – 0.62*** – 0.59*** – 0.47*** – 0.52*** Engineering 0.39*** 0.48*** 0.34*** 0.46*** 0.32*** 0.44*** 0.31*** 0.53*** # obs./R 2,900 0.22 2,409 0.23 2,045 0.22 1,990 0.23 (b) Men 2012 2014 2016 2019 UAS U UAS U UAS U UAS U Arts – 0.24*** – 0.33*** – 0.19* – 0.26*** Law – 0.64*** – 0.68*** – 0.66*** – 0.74*** Economics 0.42*** 0.49*** 0.35*** 0.41*** 0.39*** 0.42*** 0.41*** 0.52*** a a Social Sc 0.21*** 0.37*** 0.07ns 0.34*** 0.28*** 0.34*** 0.10 0.35*** Medicine – 0.71*** – 0.71*** – 0.78*** – 0.71*** Natural Sc – 0.56*** – 0.57*** – 0.55*** – 0.48*** Engineering 0.48*** 0.54*** 0.48*** 0.55*** 0.46*** 0.50*** 0.51*** 0.58*** # obs./R 2,877 0.32 2,383 0.32 1,960 0.30 1,883 0.33 Employed individuals aged 30 to 55. The notation reads as follows: UAS university of applied sciences, U university, Sc sciences. OLS estimates with ln real wages (educational reference category: VET ). Statistical significance level of the estimates: * for p < .05, ** for p < 0.01, and *** for p < .001. Findings for other subjects, such as agricultural studies and fine arts are not included in the table. The estimated coefficient for the subject in the year was implausible. We therefore report the coefficient from the previous/following year’s estimates Source: SOEP v36; authors’ own calculations rates of women graduates have been higher compared to women for arts and −0.36 among men for law. We those of men. expect a delayed effect of the expansion on the labour Figure  4 summarises the resulting strength of the rela- market. Therefore, we investigate how the wage differ - tionship between the expansion in study majors (gradu- entials changed between 2012 and 2019. The growth ate growth rates) and subject-specific educational wage rates in subject-specific wage differentials range for differentials. Overall, the figure suggests that there is a women from −0.10 for the social sciences to 0.13 for negative relationship between the two growth rates for medicine, and for men from −0.08 in natural sciences university graduates. On average, a ten percent increase to 0.10 in law (see Additional file 1 : Table S5). in graduates from a specific subject is associated with a For instance, the graduate growth rate in arts stud- roughly 0.01 log point reduction in the adjusted sub- ied at university was 0.97 for women, and 0.54 for men, ject-specific wage differential relative to VET. This rela - whereas the growth in wage differentials was −0.06 tionship should not be interpreted as a “law”. It is not for women and 0.02 for men. Similarly, the gradu- irrespective of time, subject choice, and economic con- ate growth rate in economics studied at university was ditions. The relationship illustrated in the figure depends 0.90 for women, and 0.33 for men, whereas the growth on the specific conditions and socio-economic circum - in wage differentials was −0.07 for women and 0.03 for stances at the time when the educational expansion men. Majors that expanded most in terms of graduates started, i.e., the existing stock of graduates, the strength between 2002 and 2008 are also, on average, subject to of the expansion, the rate of retirement among lower- the strongest decreases in wage differentials. However, skilled workers, as well as the economic circumstances there are two that act as outliers in this relationship: engi- when the graduates begin their careers (e.g., Goldin and neering and medicine retained higher wage differentials Katz 2008). among women compared to men, although the growth 17 Page 10 of 12 J. Ordemann , F. Pfeiffer Fig. 4 The change of subject-specific wage differentials and graduate growth rates. Note: The notation reads as follows: sc sciences. Illustration based on findings presented in Additional file 1: Table S5. 6 Summary and open questions wage majors such as the arts and social sciences during This study investigates the evolution of educational wage the university expansion post-2000. According to our differentials for three categories of higher education interpretation, there is some initial evidence that the halt compared to a VET degree in Germany during the period in the increase of the adjusted U wage differentials, the of expansion of university education. It has been docu- expansion of university education and the changes in the mented that in the German context, the three categories composition of majors studied during this expansion are of tertiary education matter for the evolution of wage dif- related—a conclusion that is underlined by international ferentials. OLS findings demonstrate that the adjusted evidence regarding the importance of subject choice for wage differentials for university graduates increased until wage differentials (e.g., Machin and Puhani 2003; Michel - around 2012 to about 0.5 log points before they stagnate more and Sassler 2016). for some years and decline after 2015. In 2019, the wage A number of open questions remain. After 2012, the differential is assessed as being about 0.045 log point Bologna Process, with its introduction of first- and sec - lower compared to 2012, which nearly amounts to a ten ond-cycle degrees may have influenced the evolution of percent decrease. the adjusted educational wage differentials. If the share of The adjusted wage differentials for men with a U and workers with a bachelor’s degree from a university in our a UAS degree converged. The adjusted wage differentials samples is increasing this may also have contributed to for men with a UAS degree increased during the obser- the stagnation of the adjusted U wage differentials. Due vation period as well, although the speed of the increase to data restrictions, we are not able to assess this pos- slowed down after 2015. The adjusted wage differentials sibility. In our estimation samples, the case numbers of for men with an MC qualification increased steadily, workers with a UAS bachelor’s degree doubled between although they always remain significantly lower com - 2014 and 2019 (from 56 to 110). Since these workers pared to UAS. Among women, the wage differentials for have lower wage differentials compared to workers with U are significantly higher compared to UAS throughout a Diploma or a master’s degree, this development may the observation period, while the wage differentials for have contributed to the stagnation of the increase in UAS UAS increase only slightly, and for MC stagnate towards adjusted wage differentials after 2015. However, most the end of the observation period. The study documents bachelor graduates from universities seem to go on to an above average growth in graduates of lower average The evolution of educational wage differentials for women and men in Germany, from 1996 to 2019 Page 11 of 12 17 attain a master’s degree (Authoring Group NRoE 2018). u-shape pattern, such that wages decline when working In our samples, there are no workers with a U Bachelor’s hours per week exceed specific benchmarks such as the degree. average working hours. If working hours exceed such a In the last few decades, more students with a lower benchmark, productivity and wages may start to decline. preference for education at a university may have Such a relationship may have restricted the expansion of entered universities and this type of self-selection may working hours in times of increasing wage differentials have, in turn, contributed to the stagnation of U wage and even caused working hours to decline. The decrease differentials (e.g., Carneiro and Lee 2011; Kroher et  al. in working hours may have contributed to the stagnation 2021; Ordemann 2021). However, the incentive to enter of the adjusted U wage differentials after 2014. It may university may also have been fostered by expectations indicate, for instance, a higher leisure preference result- about a higher future demand for graduates driven by ing in a decrease in the demand for goods and the stagna- technological change. According to Dauth et al. (2021), tion of the U wage differentials. continued automation in Germany positively affected Although the adjusted wage differentials for U are still incumbent high-skilled workers, decreased the demand higher in 2019 compared to 1996, one may ask whether for workers with vocational education and increased a period of decreasing wage differentials is ahead and the incentives for talented young adults to enter univer- how it might develop. Our empirical approach should sity instead of vocational education. Since we have nei- be useful for thinking about this question. For West ther information on university preference parameters Germany, earlier findings by Gebel and Pfeiffer (2010) nor expectations about future labour market prospects, suggest that average returns to education between 1984 it is left for further research to look deeper into these and 2006 reached a minimum in 1996. The current types of explanation. study suggests that the adjusted U wage differential for Another hypothesis posits that the composition of university graduates was highest around 2012 and in workers with a VET qualification may have changed with 2019 was still above the values in 1996. However, the respect to formal secondary education between 1996 and labour market momentum of university expansion is 2019. While more school leavers graduated with the for- ongoing and there may be room for a further decline mal certificate (“Abitur”) needed to enter higher educa - in the future. It remains a task for future research to tion, the wish to attain a similar position to their parents assess the further evolution of educational wage differ - may have diverted many of them into VET (Konietzka entials and their underlying forces. and Hensel 2017). In this process, the human capital of workers with a VET qualification may have increased Supplementary Information over time, contributing to the stagnation of the wage dif- The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12651- 022- 00323-6. ferentials compared to U workers. Indeed, further regres- sion analysis confirmed that VET workers with an Abitur Additional file 1: Table S1. Descriptive statistics from the (pooled) earned significantly higher wages than those with a VET estimation samples, 1996 to 2019 ( Total/Women/Men). Table S2. Average qualification but without an Abitur in almost all years. real wages by educational degrees, 1996 to 2019 ( Women/Men; Mean). Table S3. Average real wages and annual wage growth by educational However, this estimated wage premium decreased in the degrees ( Women/Men; in EUR and in %). Table S4. Educational wage years 2015/19 compared to the years 2008/12 according differentials, 1996 to 2019 ( Women/Men; in ln, [CI]). Table S5. Dynamics of to our regressions, for women from 0.174 to 0.142 and for degrees and subject-specific wage differentials (in thousands/in %). men from 0.211 to 0.162. This finding does not seem to support the change-in-VET-hypothesis, although it is not Acknowledgements We would like to thank Michael Gebel, Sarah McNamara, Matthias Parey, a formal falsification. Future research on the relevance of Frauke Peter, Stephan Thomsen, and Thomas Zwick as well as the editor and these considerations and the influence of possible other the anonymous reviewers for many insightful and very helpful comments. All changes to the contents of VET could be helpful. remaining errors are our own. Ordemann and Pfeiffer (2021) analysed the gender spe - Author contributions cific participation rates and working hours, based on the All authors read and approved the final manuscript. same samples from the SOEP used in the present study. Funding They report increasing participation rates between 1996 Not applicable. and 2019. Women’s participation rates were still lower than men’s in 2019, although the gap narrowed. Quite Availability of data and materials The dataset analysed during the current study is available at the Research surprisingly, despite upskilling, the average hours worked Data Center of the Socio-Economic Panel, https:// doi. org/ 10. 5684/ soep. core. in the samples of workers aged 30 to 55 decreased from v36. The additional statistical information is based on data of Destatis and is 39.4 h in 1996 to 37.1 h in 2019. According to Bick et al. available at https:// icela nd. dzhw. eu. (2019), working hours and wages display an inverted 17 Page 12 of 12 J. Ordemann , F. Pfeiffer Gernandt, J., Pfeiffer, F.: Wage convergence and inequality after unification: Declarations (East) Germany in transition. In: Kanbur, R., Svejnar, J. (eds.) Labor Market and Development, pp. 387–404. Routledge, London (2009) Ethics approval and consent to participate Goebel, J., Grabka, M.M., Liebig, S., Kroh, M., Richter, D., Schröder, C., Schupp, Not applicable. J.: The German Socio-Economic Panel (SOEP). J. Econ. Stat. 239, 345–360 (2019) Competing interests Goldin, G., Katz, L.F.: The race between education and technology. Havard The authors declare that they have no competing interests. University Press, London and Massachusetts (2008) Green, F., Henseke, G.: Europe’s evolving graduate labour markets: supply, Author details demand, underemployment and pay. J. Lab. Mar. Res. 55, 1–13 (2021) German Centre for Higher Education Research and Science Studies (DZHW ), Hillmert, S., Jacob, M.: Social inequality in higher education is vocational train- Lange Laube 12, 30159 Hannover, Germany. ZEW Leibniz Centre for Euro- ing a pathway leading to or away from university. Europ. Soc. Rev. (2003). pean Economic Research, L7, 1, 68161 Mannheim, Germany. https:// doi. org/ 10. 1093/ esr/ 19.3. 319 Horowitz, J.: Relative education and the advantage of a college degree. Am. Received: 7 December 2021 Accepted: 5 October 2022 Soc. Rev. 83, 771–801 (2018) Kamhöfer, D., Schmitz, H., Westphal, M.: Heterogeneity in marginal non-mon- etary returns to higher education. J. Europ. Econ. Assoc. 17(1), 205–244 (2019) Klein, M.: The association between graduate’s field of study and occupational References attainment in West Germany, 1980–2008. J. Lab. Mar. Res. 49, 43–58 Anger, S., Heineck, G.: Cognitive abilities and earnings – first evidence for (2016) Germany. Appl. Econ. Let. 17, 699–702 (2010). https:// doi. org/ 10. 1080/ Konietzka, D., Hensel, T.: Berufliche Erstausbildung im Lebensverlauf: 13504 85080 22978 55 Grundlagen und empirische Befunde. In: Becker, R. (ed.) Lehrbuch der Araki, S.: Educational expansion, skill diffusion, and the economic value of Bildungssoziologie, pp. 281–309. Springer, Wiesbaden (2017) credentials and skills. Am. Soc. Rev. 85(2), 128–175 (2020) Kroher, M., Leuze, K., Thomsen S. L., Trunzer J.: Did the “Bologna Process” Authoring Group NRoE: Bildung in Deutschland 2018. Ein indikatorengestütz- achieve its goals? 20 years of empirical evidence on student enrolment, ter Bericht mit einer Analyse zu Wirkungen und Erträgen von Bildung. study success and labour market outcomes. IZA DP No. 14757, Bonn wbv Verlag, Bielefeld (2018). (2021) Authoring Group NRoE: Bildung in Deutschland 2020. Ein indikatorengestütz- Krueger, D., Schkade, A.B.: Sorting in the labor market: do gregarious workers ter Bericht mit einer Analyse zu Bildung in einer digitalisierten Welt. wbv flock to interactive jobs? J. Hum. Res. 43(4), 859–883 (2008) Verlag, Bielefeld (2020). Lindley, J., Machin, S.: The rising postgraduate wage premium. Economica 83, Backes-Gellner, U., Herz, H., Kosfeld, M., Oswald, Y.: Do preferences and biases 281–306 (2016) predict life outcomes? Evidence from education and labor market entry Machin, S., Puhani, P.A.: Subject of degree and the gender wage differential: decisions Econ. Rev. Europ. 134, 103709 (2021) evidence from the UK and Germany. Econ. Let. 79, 393–400 (2003) Baethge, M.: Das deutsche Bildungs-Schisma: Welche Probleme ein vorin- Michelmore, K., Sassler, S.: Explaining the gender wage gap in stem: does field dustrielles Bildungssystem in einer nachindustriellen Gesellschaft hat. sex composition matter? R. S. F. J. Soc. Sci. 2, 194–215 (2016) In: Lemmermöhle, D., Hasselhorn, M. (eds.) Bildung-Lernen, pp. 93–116. Müller, W., Pollak, R.: Weshalb gibt es so wenige Arbeiterkinder in Deutschlands Wallstein, Göttingen (2007) Universitäten? In: Becker, R., Lauterbach, W. (eds.) Bildung als Privileg, pp. Becker, R., Hecken, A.E.: Warum werden Arbeiterkinder vom Studium an 303–342. Springer, Wiesbaden (2007) Universitäten abgelenkt? Eine empirische Überprüfung der „Ablenkung- Ordemann, J.: Studium ohne Abitur. Bildungserträge nichttraditioneller Hochs- sthese“ von Müller und Pollak (2007) und Hillmert und Jacob (2003). Köl. chulabsolventen im Vergleich. Springer, Wiesbaden (2019) Z. Soz. 60(1), 7–33 (2013) Ordemann, J., Pfeiffer, F.: The evolution of educational wage differentials for Bick, A., Brüggemann, B., Fuchs-Schündeln, N.: Hours worked in Europe and women and men, from 1996 to 2019. ZEW Disc Pap 21–066, 1–55 (2021) the US: new data, new answers. Scan. J. Econ. 121, 1381–1416 (2019) Ordemann, J.: Academic Pay Gap 2015. A Snapshot of the within difference of Brändle, T., Ordemann, J.: Same same but different? Non-traditional students higher education graduates income (unpublished manuscript) (2021) and alumni in Germany. Stud. Paed. 25, 35–50 (2020) Pfeiffer, F., Pohlmeier, W.: Income, uncertainty and the probability of self- Burda, M.C., Seele, S.: Das deutsche Arbeitsmarktwunder: Eine Bilanz. Persp. employment. Rech. Econ. Louv. 58, 265–281 (1992) Wpol. 18, 179–204 (2017) Pfeiffer, F., Stichnoth, H.: Fiskalische und individuelle Bildungsrenditen - Burda, M.C., Seele, S.: Reevaluating the German labor market miracle. Ger. aktuelle Befunde für Deutschland. Persp. Wpol. 16, 393–411 (2015) Econ. Rev. 21, 139–179 (2020) Reinhold, M., Thomsen, S.L.: The changing situation of labor market entrants in Carneiro, P., Lee, S.: Trends in quality-adjusted skill premia in the United States, Germany. J. L. M. R. 50, 161–174 (2017) 1960–2000. Am. Econ. Rev. 101, 2309–2349 (2011) SOEP-Core v36: Socio-Economic Panel (SOEP), data for years 1984–2019, EU Dauth, W., Findeisen, S., Südekum, J., Wösser, N.: The adjustment of labor Edition. https:// doi. org/ 10. 5684/ soep. core. v36eu (2021). markets to robots. J. Europ. Econ. Ass. (2021). https:// doi. org/ 10. 1093/ Valletta, R.G.: Recent flattening in the highereducation wage premium: polari- jeea/ jvab0 12 zation, skill downgrading, or both? In: Education, Skills, and Technical Dustmann, C., Fitzenberger, B., Schönberg, U., Spitz-Oener, A.: From sick man Change: implications for Future US GDP Growth, pp. 313–342. Nat. Bur. of Europe to economic superstar: Germany’s resurgent economy. J. Econ. Econ. Res, Massachusetts (2018) Persp. 28, 167–188 (2014) Westphal, M., Kamhöfer, D.A., Schmitz, H.: Marginal college wage premiums Flossmann, A., Pohlmeier, W.: Causal returns to education: a survey on empiri- under selection into employment. Econ. J. (2022). https:// doi. org/ 10. cal evidence for Germany. J. Econ. Stat. 226, 6–23 (2006) 1093/ ej/ ueac0 21 Francesconi, M., Parey, M.: Early gender gaps among university graduates. Europ. Econ. Rev. 109, 63–82 (2018) Gebel, M., Pfeiffer, F.: Educational expansion and its heterogeneous returns for Publisher’s Note wage workers. J. Appl. Soc. Sci. S. 130, 19–42 (2010) Springer Nature remains neutral with regard to jurisdictional claims in pub- Gebel, M., Heineck, G.: Returns to education in the life course. In: Becker, R. (ed.) lished maps and institutional affiliations. Research Handbook on the Sociology of Education, pp. 454–475. Edward Elgar Publishing, Cheltenham, UK, Northhampton, MA (2019) Gernandt, J., Pfeiffer, F.: Rising wage inequality in germany. J. Econ. Stat. 227, 358–380 (2007)

Journal

Journal for Labour Market ResearchSpringer Journals

Published: Dec 1, 2022

Keywords: Educational Wage Differentials; Gender Gaps; Higher Education; Returns to Education; J31; J16; I23; I26

References