Search Results

Almost all papers on the motivation of German firms to train apprentices take as a stylised fact that training firms have to incur net costs. This notion stems practically from one source only - a series of descriptive cross section analyses of the costs and benefits of apprenticeship training conducted by the Federal Institute for Vocational Education and Training (BIBB). While an influential theoretical literature tries to motivate why firms should accept net training costs, the stylised fact itself has not been further tested. This paper therefore presents a complementary validation of the net cost hypothesis by simulating the decision personnel managers have to take when they want to replace unskilled or semiskilled employees [with] apprentices. It is the first causal assessment of the impact different occupational groups of apprentices have on gross enterprise profits. We use multivariate panel estimation techniques because we can hereby tackle several sources of estimation bias and reversed causality. Our paper shows that it is necessary to discriminate between different groups of occupations when assessing the costs and benefits of apprenticeship training. In particular, we find that the share of apprentices in trade, commercial, craft and construction occupations has a positive impact on contemporary gross profits and the apprentices are substitutes for unskilled or semi-skilled workers. In contrast, an increase in the share of apprentices in the manufacturing occupations reduces contemporary gross profits. This means that enterprises offering apprenticeships in manufacturing occupations do not cover their training costs during the apprenticeship period. The apprenticeship training rather is a human capital investment. This paper demonstrates the efficiency of the German apprenticeship system: it allows companies to provide general and occupation-specific skills in highly specified occupations such as manufacturing. Otherwise, it offers cost-neutral apprenticeships in occupations where skills are more general and the mobility is higher, such as commercial or trade occupations.

Published abstract reprinted by permission of the copyright owner. Show less

This working paper examines how returns to education in the West German labour market have evolved over the last 20 years. This period saw the entry of graduates from the period of educational expansion in the 1960s and 1970s to the labour market and an upgrading of the skill structure. The authors apply two estimation methods, Wooldridge’s (2004) approach that relies on conditional mean independence and Garen’s (1984) control function approach that requires an exclusion restriction, to address the issues of ‘endogeneity of schooling and its heterogeneous returns’. For the population of workers from the German Socio-Economic Panel (GSOEP), findings reveal that both approaches produce estimates of average returns to education that decrease until the late 1990s and increase significantly afterwards. During the period observed, the gender gap in returns to education appears to vanish. Also, analysis reveals that the ‘baby boomer’ cohort has the lowest average return to education in young ages although this effect disappears when they become older. Show less

This paper shows that earnings losses after unemployment increase with age. First, older employees start out with relatively high earnings in comparison to employees without employment interruptions several years before the non-employment spell. This earnings advantage turns into a strong earnings disadvantage shortly before the non-employment spell. Younger unemployed have a relatively stable and small earnings disadvantage before non-employment. Second, while the younger employees quickly enjoy earnings higher than those without employment interruptions after the non-employment spell, earnings for older employees are lower even six years after the unemployment spell. If those with non-employment spells re-enter the labour market at the same employer, the earnings impact is the more positive the younger the employee. This paper uses representative administrative spell data for 1993-2001 that allow us to take into account the precise length of all non-employment spells and calculate the exact dates before and after the spells.

Published abstract reprinted by permission of the copyright owner. Show less

This study uses a new approach for analysing changes in the gender pay gap that is based on direct measures of job tasks and gives a comprehensive characterisation of how work for men and women has changed in recent decades. Drawing on data from West Germany, the authors find that women have seen relative increases in non-routine analytic tasks and non-routine interactive tasks, which are associated with higher skill levels. The most significant difference between the genders is the pronounced relative decline in routine task inputs among women with little change for men. These relative task changes explain a substantial fraction of the closing of the gender wage gap. The authors argue that their evidence suggests that these task changes are driven, at least in part, by technological change. They also demonstrate that the task changes are related to the recent polarisation of employment between low and high skilled occupations that was observed during the 1990s. Show less

While there is a broad literature on the general wage effect of training, little is known about the effects of different training forms and about the effects for heterogeneous training participants. This study therefore adds two aspects to the literature on earnings effects of training. First, the earnings effect of training is calculated for different 'types' of employees, i.e. discriminating between qualification level, experience, job tenure, and other attributes. Second, we distinguish between the earnings impact of external and internal training. For our analysis, we use the 'Qualification and Career survey', a rich German data set with information on 0.1 per cent of all individuals employed in Germany in 1998/1999. We use a one-step full-information maximum likelihood and a two stage least squares estimation to regress the impact of training participation on earnings correcting for the endogeneity of training participation. By using a broad list of employee and employer characteristics, we try to avoid omitted variable bias.

We find the training earnings markup to be positively correlated with qualification and experience. The analysis of internal and external training reveals that this result is driven by external training only. Internal training does not have a significant earnings effect. The correction for selection into training leads to an increase in the training coefficients and a decrease of its significance.

Published abstract. Show less

This paper analyzes the economic and social consequences of job loss which contribute to exclusion from society based on German linked survey and administrative data. To study the causal relationship between unemployment and multiple dimensions of social marginalization, I combine inverse propensity score weighting with a difference-in-differences approach. The results suggest that job loss has particularly detrimental effects on the subjective perception of social integration, life satisfaction, the access to economic resources and mental health. Moreover, this paper shows that becoming unemployed hinders the fulfillment of psychosocial needs that are typically associated with working, such as social status and higher self-efficacy. The effects of job loss are long-lasting, growing more profound the longer the duration of unemployment and persisting following reemployment. Looking at effect heterogeneity, I find that having a partner and being highly educated reduces the negative effects of job loss.

Published abstract. Show less

Online labor markets experienced a rapid growth in recent years. They allow for long-distance transactions and offer workers access to a potentially 'global' pool of labor demand. As such, they bear the potential to act as a substitute for shrinking local income opportunities. Using detailed U.S. data from a large online labor platform for microtasks, we study how local unemployment affects participation and work intensity online. We find that, at the extensive margin, an increase in commuting zone level unemployment is associated with more individuals joining the platform and becoming active in fulfilling tasks. At the intensive margin, our results show that with higher unemployment rates, online labor supply becomes more elastic. These results are driven by a decrease of the reservation wage during standard working hours. Finally, the effects are transient and do not translate to a permanent increase in platform participation by incumbent users. Our findings highlight that many workers consider online labor markets as a substitute to offline work for generating income, especially in periods of low local labor demand. However, the evidence also suggests that, despite their potential to attract workers, online markets for microtasks are currently not viable as a long run alternative for most workers.

Published abstract. Show less

Using personnel records from a single large German firm in the financial industry, this paper provides detailed evidence on the effect of age and the supervisor’s gender on gender differences in workplace training, holding constant various work-place characteristics. [The authors] implement an age-specific decomposition of the incidence and the duration of training into three terms: an age-specific coefficients effect, an age-specific characteristics effect, and an age composition effect. [The] results show that the gender training gap changes with age. Females obtain less training during the early career, and their training occurs at higher age. The timing of the gender training gap seems to be driven by diverging career paths associated with employment interruptions. However, [the authors] find no evidence for catching-up effects after parental leave. A decomposition of the training gap including supervisor fixed effects reveals that supervisors do not treat male and female employees differently. Supervisors assign more training to all employees if they themselves participate more in training.

Published abstract. Show less

Studies on the underlying mechanisms of social mobility commonly find that half of the intergenerational earnings persistence remains unexplained. Focusing on the phenomenon of overqualification, this study examines a transmission channel that might operate beyond the mechanisms previously analysed. [The author] explores how the family background of university graduates affects the probability to hold a job that does not require a tertiary degree, i.e. to be overqualified. Potential pathways of the family background effects are discussed and proxy variables for the mediating factors cognitive skills, study characteristics, social capital, financial capital, and aspiration are incorporated into the empirical analysis. Graduates from families with a high socioeconomic status are found to be less likely to be overqualified. The unconditional social overqualification gap amounts to 7.4 percentage points. Non-linear Blinder-Oaxaca decompositions show that roughly 60 per cent of the social overqualification gap can be attributed to group differences in observable characteristics. Differences in cognitive skills, study characteristics, and social capital are found to be important mediators of the family background effects.

Published abstract. Show less

Using data from the German Socio-Economic Panel for the years 2000 to 2006 we analyze the determinants and labor market effects of educational dropout. In addition to classical variables like family background and occupation, we examine noncognitive skills and Internet use. Noncognitive skills and Internet availability at home are negatively associated with the probability of becoming an educational dropout. The wage gap between dropouts and those with completed school and professional education vanishes for males once we control for additional characteristics such as occupations, professional Internet use and noncognitive skills. For females it is reduced to four per cent.

Published abstract reprinted by permission of the copyright owner. Show less