Abstract
Employment re-entry opportunities decrease with age. For middle-aged welfare benefit recipients, employment obstacles connected to age exacerbate further disadvantages connected to welfare receipt. At the same time, there is considerable diversity in middle-aged welfare benefit recipients’ long-term employment trajectories, which has thus far received little attention. Policies aim to increase labour market participation at higher ages. To this end, it is important to understand specific difficulties and to be realistic when formulating goals for people with very diverse types of employment histories. Using large-scale register data, this paper’s focus is on a cohort aged 45–54 in August 2012 in Germany. Sequence analysis aids in identifying characteristics relevant to employment histories over the past 19 years, from January 1993 to July 2012. Subsequent employment outcomes over the time span September 2012 to December 2018 are investigated, differentiating between jobs of different quality, and effects of training programmes on these outcomes are analysed using entropy balancing methods. Findings are that middle-aged welfare recipients’ employment biographies are very diverse, ranging from very little employment experience, over long histories of intermittent employment, to long continuous employment histories. Employment history attributes significantly affect employment prospects. The analyses further show that it is not too late to invest in skills, independent of employment history type.
Key messages
Current policies aim to increase labour market participation at older ages.
Older benefit recipients’ employment biographies and connected employment obstacles are very diverse.
Their experiences range from very little, over intermittent, to long continuous employment.
Skill training shows positive effects, independent of employment history.
Introduction
Finding new employment after a period of unemployment can be challenging for labour market participants of an advanced age. Yet, due to population ageing in many countries, retaining older workers and improving their employment opportunities has become a policy goal (Dietz and Walwei, 2011). Policies previously installed to encourage early retirement have been reformed to increase labour market participation at older ages. Welfare benefit recipients are particularly affected by policies aiming to activate people formerly not participating in the labour market.
However, little is known about the diversity of middle-aged benefit recipients’ labour market careers. Knowledge on whether their employment histories are characterised by job-cycling, uninsured employment, or rather long-term unemployment, and how this relates to employment prospects, is essential for understanding what types of policies are needed. Focusing only on increasing employment without increasing its quality could perpetuate difficulties inherent in the employment careers of many middle-aged benefit recipients.
Active labour market programmes, such as training programmes and workfare, are central policies for welfare recipients in Germany. Some programme types target welfare recipients with relatively greater, some those with lesser employment obstacles. Yet, knowledge is limited on how well these programmes actually benefit members of their target groups, whether older welfare recipients can still profit, and which programmes address specific difficulties stemming from characteristics of their long-term employment biographies. This paper adds a long-term perspective to studying middle-aged benefit recipients’ labour market histories. The focus is on a cohort aged 45–54 in August 2012, who were not employed and received the flat-rate means-tested Unemployment Benefit II, a benefit for people not or no longer eligible for Unemployment Insurance. Sequence analysis is combined with multidimensional scaling to identify attributes relevant to employment histories over the past 19 years, separately for men and women in eastern and western Germany. The analyses make use of the earliest available administrative data after German unification. The second part of the analyses investigate how employment history attributes affect prospects of obtaining jobs of different quality in the years following the sampling time point, up to the end of 2018. These analyses additionally study whether employment effects of participating in short- and medium-term training programmes and workfare vary with attributes of sample members’ employment histories.
The analyses in this article differentiate subgroups of sample members by region and gender. In eastern Germany in the early 1990s, large-scale unemployment and employment instability followed German unification (Diewald et al, 2002; Simonson et al, 2015). The cohort members in the present study were in their late 20s and early 30s at that time, thus at an early phase of their careers. Their experiences can be expected to have long-term repercussions for their employment outcomes throughout their life courses. To account for differences in experiences, this article therefore conducts separate analyses for eastern and western Germany.
Far more mothers than fathers take parental leave in Germany (Geisler and Kreyenfeld, 2011; Keller and Kahle, 2018). Yet, employment interruptions can lead to loss of human capital and firm-specific skills, and employers may discriminate against mothers with long employment breaks. Thus, childcare related employment interruptions can have long-term consequences for women’s employment careers and earnings (Ziefle, 2004; Stewart, 2014; Schmelzer et al, 2015), and unemployed mothers have been shown to have lower re-employment opportunities than childless women (Frodermann and Müller, 2019). However, employment rates were much higher in East than West Germany for mothers who had their children before German unification (Trappe et al, 2015), particularly due to a better childcare infrastructure in East Germany. Children of the sample members of the present study were born mostly in the 1980s and 1990s, at a time when the childcare infrastructure in eastern and western Germany was still vastly different. To account for differences in employment histories between men and women in eastern and western Germany resulting from these structural differences, this article conducts separate analyses for these groups.
Employment obstacles of older jobseekers: theoretical considerations and literature review
Long-term unemployment
Employment rates tend to decrease in the years before reaching legal retirement age. In part, this is due to workers exiting employment into early retirement, disability pension or unemployment (Dietz and Walwei, 2011; Wuebbeke, 2011; Boockmann et al, 2012; Visser et al, 2016). Lower rates of unemployment-to-employment transitions among older compared to younger jobseekers further contribute to comparatively low levels of employment of those nearing retirement age (Tisch, 2015; Beste and Trappmann, 2016; Hägglund and Bächmann, 2017; Nagore García and van Soest, 2017; Dengler et al, 2021). Lower employment entry rates are reflected in longer unemployment durations among older people (Chan and Stevens, 2001) and longer durations of benefit receipt (SDFLA, 2020).
Long-term unemployment itself further reduces employment opportunities. Re-employment rates tend to decline with the length of the unemployment spell, known as duration dependence (Abraham et al, 2019). Hohmeyer and Lietzmann (2020) find evidence for duration dependence both of welfare receipt and unemployment specifically for welfare benefit recipients in Germany, and Carpentier et al (2017) similarly find duration dependence for exits from welfare to employment for Belgium.
Skill depreciation is one factor contributing to duration dependence of exit rates from unemployment (Mincer and Ofek, 1982; Gregory and Jukes, 2001). Moreover, given technological change, skills older jobseekers were originally trained for may have become obsolete (Dietz and Walwei, 2011). At long unemployment durations, discouragement can amplify difficulties of regaining employment (Henkens et al, 1996; Nivorozhkin and Nivorozhkin, 2020), and can decrease older jobseekers’ job search intensity (Zacher, 2013; de Coen et al, 2015). Wuebbeke (2011) finds that older welfare recipients tended to choose early retirement not so much due to a preference for more leisure, but due to discouragement and a perceived lack of adequate support by public employment agencies.
A further factor leading to lower job-entry rates at longer unemployment durations is the growing selectivity of people who remain unemployed for long durations of time. People with characteristics connected to better employment opportunities, such as higher levels of education, employment experience or motivation, leave unemployment at higher rates. Thus, at longer unemployment durations, people lacking such characteristics are over-represented. In addition, the long-term unemployed may be confronted with stigma and discrimination by potential employers who, for instance, assume a lack of motivation or skills, even when this is not the case (Eriksson and Rooth, 2014; Farber et al, 2017; Hohmeyer and Lietzmann, 2020).
The present article investigates the extent to which middle-aged benefit recipients are affected by long-term unemployment, how unemployment duration relates to prospective employment outcomes, and whether there is effect heterogeneity of employment programmes by unemployment duration.
Employment instability and job-cycling
A further factor potentially affecting the employment opportunities of older jobseekers is a long history of cycling between work and welfare. For several countries, studies have found evidence of cycling between low-paid employment and unemployment for a part of the labour force (Uhlendorff, 2006; Stewart, 2007; Cortis et al, 2013; Bruckmeier and Hohmeyer, 2018), or find such evidence only for women but not for men (Buddelmeyer et al, 2010). Moreover, experiences of unemployment are shown to foster future unemployment (Arulampalam et al, 2000; Jarosch, 2021).
Reasons formerly unemployed people take up low-quality, unstable employment may include human capital depreciation or a lowering of reservation wages over time spent in unemployment (Arulampalam et al, 2000). Moreover, employers may use applicants’ prior experiences of unemployment (Arulampalam et al, 2000) or prior job quality to screen workers, such that those formerly working in low-quality jobs primarily have access to similar jobs (Fok et al, 2015). Frequent job changes can also lead to loss of job-specific human capital, and could be interpreted as a signal of low ability by potential employers (Gagliarducci, 2005). Gagliarducci (2005) shows that repeated intermittent spells of temporary employment are indeed related to lower opportunities of obtaining secure employment.
Evidence for Germany shows that the jobs taken up by former welfare recipients tend to be very short, with nearly half lasting less than six months (Bruckmeier and Hohmeyer, 2018). Jobs available to former welfare benefit recipients tend to be low-paid and are often located in sectors with only seasonal or short-term employment options, such as the temporary agency, agriculture, construction or transport sectors. Furthermore, welfare recipients who take up employment in small firms, as well as in jobs requiring low levels of education, have shorter job durations (Bruckmeier and Hohmeyer, 2018; Dengler et al, 2021).
The present analyses identify to what extent unstable employment is prevalent in the employment histories of welfare recipients. Analyses on employment outcomes study how people’s past job-spell durations are related to their prospective employment opportunities, and whether employment programme participation interacts with past job-spell durations in affecting employment outcomes.
Institutional setting
The European employment strategy, incorporated into the Lisbon strategy and Europe 2020 strategy, encourages higher employment rates throughout the life course to improve the European Union (EU)’s competitiveness, foster economic growth and reduce poverty (Cantillon, 2011). In this context and in response to population ageing, many EU countries have reformed policies in an effort to increase labour force participation at older ages (Annesley, 2007). Indeed, employment rates at older ages have risen throughout Europe in recent years (Loichinger and Skirbekk, 2016; Loichinger and Prskawetz, 2017) as well as specifically in Germany (Dietz and Walwei, 2011).
In Germany, one reform in this context has been to gradually abolish early retirement options for long-term unemployed people aged over 60, beginning in the 1990s (Mika and Baumann, 2008). Moreover, the Hartz Reforms in the first half of the 2000s in Germany involve far-reaching activation policies, such as greater benefit conditionality for welfare recipients and stronger use of active labour market programmes, with the goal of encouraging employment.
The most frequent active labour market programmes for welfare benefit recipients are short training, further vocational training and workfare. Workfare programmes in Germany, also known as One-Euro-Jobs, are a form of job creation programme, providing public or non-profit sector work that is required to be of public interest, to not replace regular jobs, nor distort competition. Participants receive one to two euros an hour in addition to their regular welfare benefits. Workfare programmes target benefit recipients with low employment opportunities, with the intentions of improving their employability and social integration, and providing a first step into the labour market (Federal Employment Agency, 2022). Workfare is most often in the areas of infrastructure improvement or environmental preservation, and the average planned duration in 2021 was 4.6 months (SDFLA, 2022a).
Short training includes application training or other short training programmes and capability tests, and may not run for more than two months (eight weeks) (Social Code III, §45).
Further vocational training in contrast has no fixed duration. Courses that ended at the beginning of 2022 ran for an average of 4.3 months. Further vocational training provides occupation-specific skills. The most frequent types of further vocational training in the first months of 2022 were courses for acquiring motor vehicle driving licences, training in the areas of social work and childcare, security occupations and clerical work (SDFLA, 2022b). Guidelines for case workers state that further vocational training should especially target people without a vocational degree, those living in regions subject to industrial restructuring, as well as long-term unemployed people. Before granting the measure, case workers are also required to assess whether training is necessary for labour market integration, and whether welfare recipients are likely to successfully complete a course (sufficient motivation and endurance, and no obstacles preventing successful participation; Federal Employment Agency, 2021).
Workfare jobs are often located in sectors with low labour demand, which can limit the utility of work experience gained there (Harrer and Stockinger, 2021). Nonetheless, workfare might provide a first step into the labour market, for instance by providing an opportunity to become accustomed to a regular work schedule, and may therefore have positive effects on subsequent regular employment for people with low employment experience. For participants with better employment prospects to begin with, on the other hand, such very basic work experience can be expected to have little or no effect on regular employment opportunities.
Short training might be expected to have moderate positive employment effects, as participation can augment general skills important for finding a job. Further vocational training should have larger effects, as it provides more specific occupational skills and allows more time to invest in these skills. Both types of courses should be beneficial for people with little employment experience, as they can provide up-to-date skills helpful for re-entering employment, and participation can signal commitment to potential employers. People with high employment experience can likewise profit, for instance if they need to retrain or are looking for more skilled jobs.
Previous research has found active labour market programmes such as short training, further vocational training and workfare to vary in their ability to improve employment opportunities, with the greatest positive effects for further vocational training and lowest effects for workfare (Bernhard, 2016; Harrer et al, 2020; Harrer and Stockinger, 2021). Dengler (2019) finds that positive job quality effects are particularly evident for further vocational training. Studying effects of short training programmes for older welfare recipients, Romeu Gordo and Wolff (2011) find small positive employment effects. While Hohmeyer and Wolff (2012) find no positive employment effects of workfare for welfare recipients under the age of 25, they find small positive effects for those aged over 25, without systematic variation between the age groups 25–35, 36–50 and 51–62. Studying welfare recipients who began further vocational training in 2005, Bernhard (2016) finds somewhat larger long-term employment effects for those aged over than under 25, but no large differences between those aged 25–44 and 45–57.
This article contributes evidence on heterogeneity of effects of active labour market programmes for older welfare recipients with diverse employment biographies.
Data and method
The analyses combine sequence analysis, multidimensional scaling and entropy balancing methods. The data is large-scale administrative data from multiple sources that is available for scientific analysis as the Integrated Employment Biographies (IEB), Unemployment Benefit II History data set (LHG), Jobseekers’ vitas (WGH) and Jobseeker History (ASU) data sets. Employment histories in the IEB are derived from notifications sent by employers to health and pension insurance funds. Data in the LHG and ASU on episodes of benefit receipt and on household composition is based on job centre and employment office records. Finally, the WGH includes self-reported information on jobseekers’ vita that they provided for the job centres and employment offices, for instance on periods of self-employment.
The sample includes people aged 45–54 who were receiving flat-rate means-tested Unemployment Benefit II – a welfare benefit for people who have run out of or are not eligible for Unemployment Insurance, but are capable of employment – and were not employed on 31 August 2012. The administrative data sources only record employment spells that took place in Germany. As the sample members’ long-term employment biographies are a central focus of the study, the sample thus excludes people who are likely not to have been living in Germany during the entire past 19 years. These were identified as people who named an episode in a different country in their employment-office vita. Furthermore, people who were naturalised or have a foreign nationality and additionally have no record (and for those with a partner no record for their partner either) in any administrative data source dating back at least to the first quarter (that is, the first five years) of the 19-year observation period were also excluded. Altogether, 10.8% of the original sample was excluded due to these reasons. The resulting total sample size was 183,098 for men in western Germany, 142,335 for women in western Germany, 95,764 for men in eastern Germany and 73,578 for women in eastern Germany.
Sequence analysis of sample members’ employment biographies (1993–2012)
Sequence analysis demands high computing capacity, which increases quadratically with the number of observations and sequence length (Brzinsky-Fay et al, 2006). Thus, it was not feasible to run the sequence analyses using the full samples. Moreover, sequence index plots for very large samples tend to become unintelligible, as the individual rows become increasingly blurred for large sample sizes. Therefore, four random samples consisting of 12,500 people each were drawn from the initial full administrative sample for the four subgroups of men and women, living in eastern and western Germany, respectively.
For the sequence analysis, monthly employment histories were prepared for a 19-year period, from January 1993 to July 2012, the month before the sampling time point. Reliable data for eastern Germany is available starting in 1993, which was therefore chosen as the beginning of the observation period for the employment histories. In each month of their employment histories, sample members were classified as being either not employed, employed in a minijob, employed in a socially insured job or self-employed. Jobs paying up to €450 (€400 between 2003 and 2013) a month are not socially insured in Germany, and are commonly referred to as ‘minijobs’. Minijobs are recorded in the data starting in 2003. Thus, these types of jobs unfortunately do not appear in the employment histories in the first half of the observation period. Nevertheless, to exploit the available information as best possible, the study takes account of minijobs in the second half of the observation period.
The program SADI: Sequence Analysis Tools for Stata (Halpin, 2014) is used to perform optimal matching analysis. Optimal matching analysis calculates a distance matrix that quantifies the dissimilarity between each pair of sequences. The more operations necessary to transform one sequence into another, and the more costly the specific operations used, the more dissimilar the two sequences are conceived to be.
In a next step, the pairwise distance matrix resulting from optimal matching analysis was used to attempt cluster analysis, using the Ward clustering algorithm. Index plots (Brzinsky-Fay et al, 2006) for four clusters for each subgroup are shown in Figures A1–A4, using random samples of 500 people per cluster for better readability of the index plots. However, the quality of the clustering solutions proved to be low. Average silhouette widths (Halpin, 2016), which indicate how similar sequences are to other sequences in their own cluster compared to other clusters, were small for any chosen number of clusters (Table A1). Thus, it appears the sequences cannot well be grouped into distinct clusters. A reason for this may be that the sequences are long and highly varied; almost no two sequences are exactly alike (with the exception of sequences of people with no employment experience at all). Several authors discuss difficulties of grouping and graphing highly complex sequences (Piccarreta, 2012; Lengyel and Botta‐Dukát, 2019). Moreover, Halpin and Chan (1998) point out that cluster analysis is relatively indeterminant, in that results are dependent on the specific linkage mechanism and specific sample used.
Halpin and Chan (1998) propose multidimensional scaling (MDS) for exploratory descriptive analysis of sequences, and Piccarreta and Lior (2010) suggest sorting sequences in index plots by first dimension MDS distances to better visualise the main features of the sequences. MDS is a factorial technique that projects sequences into a low-dimensional space, such that distances in this space reflect as best possible the original distances between all pairs of sequences from optimal matching analysis (Piccarreta and Lior, 2010).
For the present analysis, MDS index plots provide a method for sorting sequences without grouping them. This better reflects the gradual nature of differences between sequences than would be possible using cluster analysis.
MDS is applied to the distance matrices created using optimal matching analysis (Halpin, 2014), using the utility sqmdsadd in the sq-package (Kohler et al, 2020). For this purpose, as a first step, optimal matching analysis was rerun using smaller samples of 500 people per subgroup of women and men in eastern and western Germany, as MDS requires high computing capacity. MDS is used for exploratory analysis only; the subsequent analyses (described later) use the full samples from administrative data.
Next, following Piccarreta and Lior (2010), sequences are sorted by the first two MDS dimension distances for visualisation, using index plots (Brzinsky-Fay et al, 2006), as shown in Figure 1 in the next section.
Employment histories of welfare benefit recipients in August 2012
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Notes: Sorted by first two multidimensional scaling distances. Random sample of 500 sequences for each subsample.Employment histories of welfare benefit recipients in August 2012
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Notes: Sorted by first two multidimensional scaling distances. Random sample of 500 sequences for each subsample.Employment histories of welfare benefit recipients in August 2012
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Notes: Sorted by first two multidimensional scaling distances. Random sample of 500 sequences for each subsample.This method of exploratory analysis aids in identifying characteristics relevant for distinguishing sequences. To investigate how strongly such characteristics (for example, total job experience, proportion of time employed in minijobs, average length of job spells) reflect the first two MDS dimension distances, sequence characteristics are used as explanatory variables in regression analyses for these distances. Table A3 shows R2 values of 0.96–0.99 for models with the first MDS dimension distance as a dependent variable. Thus, the identified sequence characteristics seem to reflect first dimension MDS distances quite well. For second dimension MDS distances, R2 values are smaller, but at 0.37 and 0.47 still indicate that the sequence characteristics explain a large proportion of variation of second dimension MDS distances (Table A4).
Entropy balancing to evaluate employment programme effects on employment outcomes (2012–18), after the sampling time point
Finally, the paper evaluates effects of the three most frequent types of employment policy programmes for welfare recipients in Germany. These programmes are short training programmes, further vocational training and workfare programmes. The analyses account for interaction effects of employment history characteristics and programme participation, to provide insight as to which benefit recipients can best profit from which programme types. The paper studies effects of programme participations that took place in September to November 2012, the first three months after the sampling time point. Long-term effects of these programme participations on several indicators of employment duration and employment quality are evaluated over the time span September 2012–December 2018.
The programme evaluations use the full samples of eastern German women, eastern German men, western German women and western German men as described at the beginning of the data and method section.
The programme evaluations apply entropy balancing methods (Hainmueller, 2012), using the Stata utility ebalance (Hainmueller and Xu, 2013) to obtain weights to adjust the covariate distribution of the control group to that of the participant group, with respect to the means and variances (first two moment constraints) of a large set of covariates. These include individual socio-demographic variables, labour market history indicators, household characteristics, partner characteristics, as well as regional characteristics, and are listed in detail in Table A5 in the appendix. The weighted sample is then used to calculate Average Treatment Effects on the Treated for the investigated employment programmes, using linear regression analysis. Entropy balancing is run separately for the four subgroups of men and women in eastern and western Germany, and for each of the three programme types, thus for 12 separate analyses.
The employment outcomes investigated are total months in regular insured employment, total months in uninsured minijobs, as well as the probability of obtaining at least one job lasting for at least two years in the period September 2012–December 2018. These measures provide indicators of overall employment integration as well as of employment quality. The measures were chosen as they reflect aspects of sample members’ employment histories, which will be discussed in more detail in the following. Employment histories not only vary as to total employment experience, but also with respect to the proportion of uninsured minijob employment versus insured regular employment, as well as the duration of individual employment spells. Using indicators of employment integration and quality as outcomes helps in investigating whether employment history attributes are perpetuated in prospective employment outcomes, and whether participation in active labour market programmes alters outcomes.
Results
Sequence analysis and multidimensional scaling
The empirical analyses set out with a sequence analysis (optimal matching analysis) of the 19-year employment histories of four random samples consisting of 500 people each, for western German men, western German women, eastern German men and eastern German women, respectively. Multidimensional scaling (MDS) is applied to the distance matrix resulting from optimal matching analysis.
Following Piccarreta and Lior (2010), Figure 1 shows index plots of the employment histories of the members of these four samples, from January 1993 to July 2012, sorted by the first two dimensions of MDS. Each row represents one person’s employment history. Each month of each person’s employment history is colour-coded to represent a given employment status.
The sorted employment histories shown in Figure 1 provide a chance to visualise welfare recipients’ employment histories for exploratory analysis. The index plots sorted by the first dimension of MDS indicate that in each of the four subsamples, a part of the sample members have no employment spells at all or have only infrequent short employment spells. Others appear to have frequent short employment spells throughout their careers. Some, finally, were employed for most of the observed period. Employment histories are also observed to differ by the proportion of time spent in minijobs and self-employment, as well as the duration of time since the last job.
The second dimension of MDS (Figure 1) especially highlights heterogeneity with respect to the phase during which people’s employment experience was most concentrated. While for some people, the ‘centre of gravity’ of employment was early in their careers, with recent job spells being shorter and more infrequent, for others, employment was concentrated more strongly in recent years. To reflect this dimension, a variable giving the duration since the median month of a person’s months in employment was generated.
Differences between sample members’ individual employment histories appear gradual (Figure 1). Thus, for this sample, MDS seems better suited for exploratory analysis than cluster analysis (Figures A1–A4 and Table A1). As described in the data and method section, cluster analysis did not provide distinct clusters for this sample.
Tables A3 and A4 in the appendix show that the employment history characteristics already described account for variation in the first and second dimension MDS distances quite well (see data and method section). Tables A3 and A4 also indicate that the measure ‘duration since median of months with a job’, is more relevant for the second MDS dimension than for the first.
Table 1 provides descriptive statistics for the sequence characteristics for the full samples. The proportion of the sample that has never been employed in the observation period ranges from approximately 3% for eastern German men to approximately 15% for western German women. Among those who have ever had a job, average job experience over the 235-month (19-year) observation period ranges from 65 months for eastern German women to 87 months for western German men. A quarter of each subsample was employed for a total of three years or less during this time, while a further quarter was employed for nearly half or more of the observation period. The proportion of employment that was in uninsured minijobs ranges from 9% for eastern German men to 34% for western German women. The proportion of self-employment, on the other hand, ranges from 5% for eastern German women to 10% for western German men. The average duration of individual job spells reaches from 18 months for eastern German women to 24 months for western German men. For a quarter of each sample, average job duration is seven months or less. The average duration since the last job varies between 59 months for western German women and 67 months for eastern German women. The duration since the median of months with a job (‘centre of gravity of employment’) ranges from 154 months for western German men and women to 164 months for eastern German men. Variation in this dimension is highest for western German women.
Descriptives for characteristics of employment history sequences, January 1993–July 2012 (235 months). Full samples
| Percentile | |||||||
|---|---|---|---|---|---|---|---|
| Mean | 10 | 25 | 50 | 75 | 90 | N | |
| Women, east | |||||||
| Ever had a job (%) | 90.1 | 73,578 | |||||
| Among those who ever had a job | |||||||
| Duration (months) with a job | 64.5 | 11 | 23 | 49 | 96 | 144 | 66,328 |
| Proportion of time in minijobs of total months with a job (%) | 20.7 | 0.0 | 0.0 | 3.6 | 33.3 | 70.9 | 66,328 |
| Proportion of time in self-employment of total months with a job (%) | 4.8 | 0.0 | 0.0 | 0.0 | 0.0 | 1.2 | 66,328 |
| Average job-spell duration (months) | 18.1 | 4.4 | 7.0 | 11.0 | 19.2 | 37.3 | 66,328 |
| Duration since last job (months) | 67.3 | 5 | 17 | 49 | 107 | 155 | 66,328 |
| Duration since median of months with a job (months) | 158.0 | 89 | 128 | 164 | 196 | 220 | 66,328 |
| Men, east | |||||||
| Ever had a job (%) | 96.6 | 95,764 | |||||
| Among those who ever had a job | |||||||
| Duration (months) with a job | 79.4 | 17 | 36 | 70 | 115 | 158 | 92,470 |
| Proportion of time in minijobs of total months with a job (%) | 9.2 | 0.0 | 0.0 | 0.0 | 10.0 | 32.0 | 92,470 |
| Proportion of time in self-employment of total months with a job (%) | 8.3 | 0.0 | 0.0 | 0.0 | 0.0 | 36.9 | 92,470 |
| Average job-spell duration (months) | 19.9 | 4.5 | 6.9 | 11.3 | 21.0 | 42.0 | 92,470 |
| Duration since last job (months) | 66.1 | 6 | 18 | 48 | 104 | 151 | 92,470 |
| Duration since median of months with a job (months) | 163.6 | 109 | 141 | 169 | 194 | 210 | 92,470 |
| Women, west | |||||||
| Ever had a job (%) | 84.5 | 142,335 | |||||
| Among those who ever had a job | |||||||
| Duration (months) with a job | 70.5 | 9 | 25 | 60 | 107 | 150 | 120,319 |
| Proportion of time in minijobs of total months with a job (%) | 34.2 | 0.0 | 0.0 | 16.3 | 67.5 | 100.0 | 120,319 |
| Proportion of time in self-employment of total months with a job (%) | 5.7 | 0.0 | 0.0 | 0.0 | 0.0 | 13.3 | 120,319 |
| Average job-spell duration (months) | 20.4 | 3.8 | 6.8 | 12.4 | 23.6 | 44.0 | 120,319 |
| Duration since last job (months) | 58.8 | 4 | 14 | 42 | 94 | 137 | 120,319 |
| Duration since median of months with a job (months) | 153.6 | 71 | 115 | 152 | 206 | 235 | 120,319 |
| Men, west | |||||||
| Ever had a job (%) | 94.7 | 183,098 | |||||
| Among those who ever had a job | |||||||
| Duration (months) with a job | 86.8 | 15 | 36 | 79 | 131 | 173 | 173,414 |
| Proportion of time in minijobs of total months with a job (%) | 10.7 | 0.0 | 0.0 | 0.0 | 11.5 | 37.1 | 173,414 |
| Proportion of time in self-employment of total months with a job (%) | 10.4 | 0.0 | 0.0 | 0.0 | 0.0 | 48.2 | 173,414 |
| Average job-spell duration (months) | 24.1 | 4.1 | 7.0 | 13.0 | 27.0 | 54.3 | 173,414 |
| Duration since last job (months) | 62.6 | 5 | 16 | 45 | 100 | 143 | 173,414 |
| Duration since median of months with a job (months) | 154.2 | 101 | 129 | 155 | 185 | 210 | 173,414 |
Thus, employment histories of middle-aged welfare recipients are quite varied. A significant proportion has very little employment experience, and for many, employment has been unstable. Nonetheless, a part of the sample has comparatively high employment experience, particularly among western German men.
Association between characteristics of employment histories and socio-demographic variables are shown in Tables A6–A9. These tables, for instance, show that having children is negatively related to employment experience for women in eastern and western Germany. Having a partner is also negatively related to employment experience for women in western Germany, while this relationship is not as clear-cut in eastern Germany. For men, for the most part, having a partner and children appear to be positively associated with employment experience. Both employment experience and average job-spell duration is higher for those with vocational degrees. While people with German citizenship have less employment experience on average, they have longer average job spells (with the exception of eastern German women). Conversely, this indicates more employment experience but more unstable jobs among those without German citizenship.
Effects of participating in workfare, short training and further vocational training on subsequent employment outcomes
Table 2 shows main effects of participating in programmes that began in September to November 2012 on subsequent employment outcomes up until December 2018, on the basis of linear regression and linear probability models using the samples of programme participants and non-participants, weighted via entropy balancing weights. Findings are that participating in workfare raises months in regular employment by 0.8–0.85 months in western Germany, and has no significant effects in eastern Germany. Workfare also raises western German women’s minijob employment duration by 1.08 months, and their probability of having at least one job lasting at least two years by 2.57 percentage points. Effects of short training programmes are somewhat larger, raising time in regular employment by 2.44 to 3 months, time in minijobs by 0.29 to 0.81 months, and the probability of having at least one job lasting at least two years by 3.73 to 4.66 percentage points. Further vocational training has the largest effects of all three programmes on regular employment, at 7.7 to 8.35 months. It also raises the probability of having at least one job lasting at least two years by 7.72 to 11.63 percentage points. Further analyses (not shown here, available on request) show that the further vocational training disproportionately raises the probability of more stable compared to less stable employment, to a greater extent than short training and especially than workfare. However, further vocational training has no significant effects on months in minijobs for three of the four subgroups, and a significantly negative effect for western German women (−0.6 months).
Linear regressions for balanced samples. Main effects of programme participation in September–November 2012 on specified outcomes in the time period from programme start to December 2018. ATTs and baseline outcomes for balanced control group
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| total time in regular jobs (months) | total time in minijobs (months) | probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| ATTs | ||||||||
| workfare | ||||||||
| women, east | –0.059 | 0.222 | 0.620 | 2,899 | 66,076 | |||
| men, east | 0.083 | –0.054 | –0.221 | 4,113 | 86,125 | |||
| women, west | 0.852 | ** | 1.083 | *** | 2.572 | *** | 2,175 | 130,282 |
| men, west | 0.806 | *** | 0.073 | 0.989 | * | 4,019 | 167,000 | |
| short training | ||||||||
| women, east | 2.630 | *** | 0.416 | 4.657 | *** | 2,654 | 66,186 | |
| men, east | 2.442 | *** | 0.488 | ** | 3.924 | *** | 3,407 | 86,625 |
| women, west | 2.998 | *** | 0.805 | *** | 4.317 | *** | 4,747 | 127,325 |
| men, west | 2.709 | *** | 0.288 | ** | 3.734 | *** | 7,346 | 163,322 |
| further vocational training | ||||||||
| women, east | 8.348 | *** | –0.394 | 11.631 | *** | 903 | 68,058 | |
| men, east | 7.714 | *** | –0.440 | 10.080 | *** | 1,220 | 88,988 | |
| women, west | 7.700 | *** | –0.597 | * | 10.471 | *** | 1,497 | 130,909 |
| men, west | 7.531 | *** | –0.211 | 7.722 | *** | 2,005 | 169,088 | |
| Baseline (balanced control group) | ||||||||
| workfare | ||||||||
| women, east | 7.798 | *** | 6.230 | *** | 16.800 | *** | 2,899 | 66,076 |
| men, east | 6.519 | *** | 4.033 | *** | 12.013 | *** | 4,113 | 86,125 |
| women, west | 6.671 | *** | 5.892 | *** | 15.313 | *** | 2,175 | 130,282 |
| men, west | 6.164 | *** | 3.800 | *** | 11.825 | *** | 4,019 | 167,000 |
| short training | ||||||||
| women, east | 9.815 | *** | 6.485 | *** | 19.646 | *** | 2,654 | 66,186 |
| men, east | 8.984 | *** | 4.177 | *** | 15.155 | *** | 3,407 | 86,625 |
| women, west | 9.083 | *** | 6.739 | *** | 19.256 | *** | 4,747 | 127,325 |
| men, west | 8.704 | *** | 4.450 | *** | 15.242 | *** | 7,346 | 163,322 |
| further vocational training | ||||||||
| women, east | 13.468 | *** | 6.572 | *** | 24.028 | *** | 903 | 68,058 |
| men, east | 11.730 | *** | 4.147 | *** | 18.363 | *** | 1,220 | 88,988 |
| women, west | 12.021 | *** | 7.254 | *** | 22.796 | *** | 1,497 | 130,909 |
| men, west | 10.730 | *** | 4.490 | *** | 17.615 | *** | 2,005 | 169,088 |
Note: *p < 0.1; **p < 0.05; ***p < 0.01
The lower half of Table 2 shows the baseline levels of the outcome variables for the balanced control group (the estimates for the constants in the models). These figures indicate that participants in workfare would have been in regular employment for 6.16 to 7.8 months on average had they not participated in the programme. Participants in short training would have been in regular employment for 8.7 to 9.82 months, and participants in further vocational training for 10.73 to 13.47 months. Thus, it appears that benefit recipients with better employment opportunities to begin with are represented more strongly in further vocational training than in short training and especially than in workfare. Descriptive statistics (available on request) further show that employment experience is greatest among participants in further vocational training, followed by short training, and lowest among participants in workfare.
Tables 3–14 show interaction effects of programme participation with the employment history attributes that were discussed on the basis of the index plots in Figure 1 and the descriptive statistics in Table 1. Table 3 shows workfare effects for eastern German women. The table contains both the general effects of the employment history attributes as well as their interactions with workfare participation. Table 3 shows that each month of employment experience increases subsequent regular insured employment by 0.07 and minijob employment by 0.03 months on average. To allow for discontinuity in the effect of employment experience, having had no employment experience at all is entered as a separate variable. Having never been employed during the retrospective observation period has an additional negative effect of −5.63 months on regular employment and −3.97 months on minijob employment. The proportion of employment experience in minijobs has a negative effect on employment prospects in regular employment and a positive effect on minijob prospects, while the proportion of self-employment has a negative effect on prospective minijob employment. Opposite to what might be expected, the average duration of individual job spells has negative effects on all three employment outcomes. Thus, people with a history of short job spells (at a given level of overall employment experience) appear to have better employment prospects. This may indicate that willingness to accept unstable employment increases employment probabilities given the job options available to welfare benefit recipients. As would be expected, the duration since the last job has a negative effect on all three employment outcomes, as does the duration since the median of months with a job on regular employment prospects.
Linear regression for balanced sample. Interaction of workfare participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 2,899 | 66,076 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −5.629 | *** | −3.974 | *** | −9.081 | *** | ||
| Duration (months) with a job | 0.068 | *** | 0.026 | *** | 0.133 | *** | ||
| Proportion of time in minijobs of total months with a job (in %) | −1.768 | *** | 4.423 | *** | 2.712 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | 0.538 | −2.141 | *** | −3.819 | *** | |||
| Average job-spell duration (months) | −0.066 | *** | −0.031 | *** | −0.097 | *** | ||
| Duration since last job (months) | −0.042 | *** | −0.024 | *** | −0.065 | *** | ||
| Duration since median of months with a job (months) | −0.011 | *** | 0.006 | *** | −0.007 | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | −0.018 | 0.693 | 3.981 | |||||
| Interaction treatment / never had a job | 1.972 | −0.974 | 0.814 | |||||
| Interaction treatment / job duration | −0.022 | * | −0.021 | ** | −0.085 | *** | ||
| Interaction treatment / minijob proportion | 0.836 | −1.406 | 0.557 | |||||
| Interaction treatment / self-employment proportion | 0.546 | 1.267 | 2.164 | |||||
| Interaction treatment / average job-spell duration | 0.004 | 0.064 | *** | 0.131 | ** | |||
| Interaction treatment / duration since last job | 0.006 | −0.005 | −0.014 | |||||
| Interaction treatment / duration since median of months with a job (months) | 0.001 | 0.002 | −0.002 | |||||
| Constant | 10.867 | *** | 5.570 | *** | 17.654 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of workfare participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German men
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 4,113 | 86,125 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −2.691 | *** | −2.552 | *** | −4.278 | *** | ||
| Duration (months) with a job | 0.063 | *** | 0.020 | *** | 0.109 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −1.366 | *** | 5.419 | *** | 4.472 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | 0.393 | −1.371 | *** | −3.484 | *** | |||
| Average job-spell duration (months) | −0.055 | *** | −0.021 | *** | −0.070 | *** | ||
| Duration since last job (months) | −0.033 | *** | −0.015 | *** | −0.046 | *** | ||
| Duration since median of months with a job (months) | −0.018 | *** | 0.005 | *** | −0.022 | *** | ||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 1.619 | −0.829 | 1.741 | |||||
| Interaction treatment / never had a job | 3.775 | ** | 3.002 | ** | 7.019 | ** | ||
| Interaction treatment / job duration | −0.003 | 0.005 | −0.007 | |||||
| Interaction treatment / minijob proportion | 1.500 | 0.675 | 4.952 | |||||
| Interaction treatment / self-employment proportion | 3.340 | −0.440 | 3.891 | |||||
| Interaction treatment / average job-spell duration | 0.003 | 0.001 | 0.000 | |||||
| Interaction treatment / duration since last job | 0.015 | *** | 0.008 | ** | 0.033 | *** | ||
| Interaction treatment / duration since median of months with a job (months) | −0.017 | ** | −0.002 | −0.026 | ||||
| Constant | 8.971 | *** | 2.945 | *** | 12.878 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of workfare participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Western German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| total time in regular jobs (months) | total time in minijobs (months) | probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 2,175 | 130,282 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −5.159 | *** | −2.624 | *** | −6.981 | *** | ||
| Duration (months) with a job | 0.046 | *** | 0.024 | *** | 0.085 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −1.903 | *** | 3.207 | *** | 2.514 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −2.104 | *** | −1.265 | *** | −5.552 | *** | ||
| Average job-spell duration (months) | −0.028 | *** | −0.025 | *** | −0.013 | |||
| Duration since last job (months) | −0.038 | *** | −0.024 | *** | −0.060 | *** | ||
| Duration since median of months with a job (months) | −0.011 | *** | −0.002 | −0.014 | *** | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | −0.062 | −2.213 | ** | −2.960 | ||||
| Interaction treatment / never had a job | 4.996 | *** | 4.254 | ** | 13.892 | *** | ||
| Interaction treatment / job duration | 0.006 | 0.016 | 0.043 | |||||
| Interaction treatment / minijob proportion | 2.709 | ** | 2.895 | ** | 8.878 | *** | ||
| Interaction treatment / self-employment proportion | 1.598 | 2.485 | 1.951 | |||||
| Interaction treatment / average job-spell duration | −0.005 | −0.014 | −0.041 | |||||
| Interaction treatment / duration since last job | 0.018 | ** | 0.027 | *** | 0.058 | *** | ||
| Interaction treatment / duration since median of months with a job (months) | −0.011 | −0.002 | −0.023 | |||||
| Constant | 10.104 | *** | 6.283 | *** | 17.595 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of workfare participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Western German men.
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 4,019 | 167,000 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −3.095 | *** | −2.077 | *** | −4.094 | *** | ||
| Duration (months) with a job | 0.045 | *** | 0.021 | *** | 0.083 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −2.288 | *** | 4.541 | *** | 2.548 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −1.119 | *** | −0.675 | *** | −3.898 | *** | ||
| Average job-spell duration (months) | −0.032 | *** | −0.023 | *** | −0.039 | *** | ||
| Duration since last job (months) | −0.033 | *** | −0.015 | *** | −0.047 | *** | ||
| Duration since median of months with a job (months) | −0.019 | *** | −0.001 | −0.024 | *** | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 0.620 | −1.049 | −0.326 | |||||
| Interaction treatment / never had a job | −0.631 | 2.353 | ** | 2.402 | ||||
| Interaction treatment / job duration | −0.005 | 0.012 | ** | 0.009 | ||||
| Interaction treatment / minijob proportion | −1.082 | −0.544 | −0.495 | |||||
| Interaction treatment / self-employment proportion | 0.526 | 0.938 | 0.641 | |||||
| Interaction treatment / average job-spell duration | 0.014 | −0.007 | −0.005 | |||||
| Interaction treatment / duration since last job | 0.009 | 0.014 | *** | 0.021 | ||||
| Interaction treatment / duration since median of months with a job (months) | −0.001 | −0.004 | −0.004 | |||||
| Constant | 9.338 | *** | 3.586 | *** | 13.788 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of short-training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 2,654 | 66,186 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −5.776 | *** | −4.214 | *** | −9.671 | *** | ||
| Duration (months) with a job | 0.075 | *** | 0.017 | *** | 0.124 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −1.523 | *** | 4.303 | *** | 3.118 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −1.880 | *** | −2.271 | *** | −7.511 | *** | ||
| Average job-spell duration (months) | −0.063 | *** | −0.018 | *** | −0.076 | *** | ||
| Duration since last job (months) | −0.046 | *** | −0.025 | *** | −0.074 | *** | ||
| Duration since median of months with a job (months) | −0.016 | *** | 0.006 | *** | −0.016 | *** | ||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 3.646 | ** | −0.589 | 4.487 | ||||
| Interaction treatment / never had a job | 3.076 | * | 0.004 | 3.871 | ||||
| Interaction treatment / job duration | 0.036 | *** | −0.003 | 0.053 | * | |||
| Interaction treatment / minijob proportion | −1.588 | 1.888 | −1.492 | |||||
| Interaction treatment / self-employment proportion | 1.784 | 1.402 | 10.637 | |||||
| Interaction treatment / average job-spell duration | −0.054 | ** | −0.006 | −0.107 | ** | |||
| Interaction treatment / duration since last job | 0.010 | −0.003 | 0.009 | |||||
| Interaction treatment / duration since median of months with a job (months) | −0.017 | * | 0.006 | −0.012 | ||||
| Constant | 12.824 | *** | 6.013 | *** | 21.391 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of short-training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German men
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 3,407 | 86,625 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −3.731 | *** | −2.431 | *** | −5.620 | *** | ||
| Duration (months) with a job | 0.068 | *** | 0.017 | *** | 0.117 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −1.551 | *** | 6.491 | *** | 5.274 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −1.767 | *** | −1.283 | *** | −6.167 | *** | ||
| Average job-spell duration (months) | −0.055 | *** | −0.017 | *** | −0.062 | *** | ||
| Duration since last job (months) | −0.044 | *** | −0.016 | *** | −0.059 | *** | ||
| Duration since median of months with a job (months) | −0.025 | *** | 0.007 | *** | −0.021 | *** | ||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 3.404 | ** | −0.491 | 3.565 | ||||
| Interaction treatment / never had a job | 1.735 | −0.594 | 5.781 | |||||
| Interaction treatment / job duration | 0.019 | * | 0.001 | 0.027 | ||||
| Interaction treatment / minijob proportion | −0.389 | 2.021 | 0.989 | |||||
| Interaction treatment / self-employment proportion | −0.222 | 0.254 | −1.724 | |||||
| Interaction treatment / average job-spell duration | −0.025 | 0.009 | −0.015 | |||||
| Interaction treatment / duration since last job | 0.001 | 0.006 | 0.011 | |||||
| Interaction treatment / duration since median of months with a job (months) | −0.012 | 0.001 | −0.013 | |||||
| Constant | 11.661 | *** | 2.535 | *** | 14.293 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of short-training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme star to December 2018. ATTs. Western German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 4,747 | 127,325 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −6.270 | *** | −2.976 | *** | −8.721 | *** | ||
| Duration (months) with a job | 0.052 | *** | 0.020 | *** | 0.085 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −2.354 | *** | 3.845 | *** | 3.048 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −2.606 | *** | −1.465 | *** | −7.650 | *** | ||
| Average job-spell duration (months) | −0.036 | *** | −0.026 | *** | −0.029 | *** | ||
| Duration since last job (months) | −0.048 | *** | −0.026 | *** | −0.075 | *** | ||
| Duration since median of months with a job (months) | −0.015 | *** | −0.003 | * | −0.021 | *** | ||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 2.591 | ** | 0.284 | 1.274 | ||||
| Interaction treatment / never had a job | 0.467 | 2.815 | ** | 6.837 | ** | |||
| Interaction treatment / job duration | 0.010 | 0.000 | 0.016 | |||||
| Interaction treatment / minijob proportion | −0.327 | 1.287 | 2.020 | |||||
| Interaction treatment / self-employment proportion | 2.848 | −0.991 | 4.677 | |||||
| Interaction treatment / average job-spell duration | −0.013 | 0.020 | 0.049 | |||||
| Interaction treatment / duration since last job | 0.002 | 0.007 | 0.021 | |||||
| Interaction treatment / duration since median of months with a job (months) | −0.001 | −0.005 | −0.009 | |||||
| Constant | 12.305 | *** | 6.871 | *** | 21.339 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of short-training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Western German men
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 7,346 | 163,322 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −3.403 | *** | −1.930 | *** | −3.232 | *** | ||
| Duration (months) with a job | 0.057 | *** | 0.021 | *** | 0.098 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −3.578 | *** | 6.185 | *** | 3.218 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −2.931 | *** | −0.660 | *** | −5.676 | *** | ||
| Average job-spell duration (months) | −0.036 | *** | −0.022 | *** | −0.037 | *** | ||
| Duration since last job (months) | −0.043 | *** | −0.016 | *** | −0.052 | *** | ||
| Duration since median of months with a job (months) | −0.027 | *** | −0.001 | −0.037 | *** | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 2.948 | *** | 0.120 | 3.338 | ||||
| Interaction treatment / never had a job | 0.012 | −0.210 | −2.063 | |||||
| Interaction treatment / job duration | 0.006 | −0.004 | −0.005 | |||||
| Interaction treatment / minijob proportion | 2.849 | ** | 0.445 | 5.465 | ** | |||
| Interaction treatment / self-employment proportion | −0.497 | 0.216 | −0.967 | |||||
| Interaction treatment / average job-spell duration | 0.003 | 0.002 | 0.024 | |||||
| Interaction treatment / duration since last job | −0.002 | 0.002 | −0.008 | |||||
| Interaction treatment / duration since median of months with a job (months) | −0.006 | 0.002 | 0.002 | |||||
| Constant | 11.589 | *** | 3.651 | *** | 16.088 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of further vocational training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 903 | 68,058 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −7.601 | *** | −3.610 | *** | −11.121 | *** | ||
| Duration (months) with a job | 0.078 | *** | 0.015 | *** | 0.124 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −3.418 | *** | 6.081 | *** | 2.222 | * | ||
| Proportion of time in self-employment of total months with a job (%) | −4.543 | *** | −2.297 | *** | −9.863 | *** | ||
| Average job-spell duration (months) | −0.064 | *** | −0.019 | *** | −0.078 | *** | ||
| Duration since last job (months) | −0.062 | *** | −0.024 | *** | −0.092 | *** | ||
| Duration since median of months with a job (months) | −0.021 | *** | 0.008 | *** | −0.018 | ** | ||
| workfare participation: treatment effects | ||||||||
| Baseline effect | 3.932 | 2.396 | 10.364 | * | ||||
| Interaction treatment / never had a job | 1.572 | −1.207 | 2.683 | |||||
| Interaction treatment / job duration | 0.021 | −0.028 | ** | 0.027 | ||||
| Interaction treatment / minijob proportion | 3.580 | −2.728 | 1.774 | |||||
| Interaction treatment / self-employment proportion | 0.060 | 1.544 | −2.315 | |||||
| Interaction treatment / average job-spell duration | −0.017 | 0.007 | 0.018 | |||||
| Interaction treatment / duration since last job | 0.023 | −0.004 | 0.077 | * | ||||
| Interaction treatment / duration since median of months with a job (months) | 0.010 | −0.002 | −0.031 | |||||
| Constant | 16.282 | *** | 4.957 | *** | 24.127 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of further vocational training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Eastern German men
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Ttotal time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 1,220 | 88,988 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −4.364 | *** | −2.350 | *** | −5.310 | *** | ||
| Duration (months) with a job | 0.072 | *** | 0.018 | *** | 0.128 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −2.714 | *** | 7.067 | *** | 5.789 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −3.719 | *** | −1.086 | *** | −8.618 | *** | ||
| Average job-spell duration (months) | −0.059 | *** | −0.013 | *** | −0.059 | *** | ||
| Duration since last job (months) | −0.050 | *** | −0.015 | *** | −0.059 | *** | ||
| Duration since median of months with a job (months) | −0.033 | *** | 0.010 | *** | −0.031 | *** | ||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | −0.532 | 1.072 | 6.652 | |||||
| Interaction treatment / never had a job | 2.538 | 0.271 | 1.949 | |||||
| Interaction treatment / job duration | 0.067 | *** | −0.024 | *** | 0.073 | ** | ||
| Interaction treatment / minijob proportion | 3.906 | −4.151 | * | −3.380 | ||||
| Interaction treatment / self-employment proportion | 0.680 | 0.111 | −3.353 | |||||
| Interaction treatment / average job-spell duration | −0.030 | 0.018 | −0.058 | |||||
| Interaction treatment / duration since last job | 0.025 | −0.004 | 0.031 | |||||
| Interaction treatment / duration since median of months with a job (months) | 0.009 | 0.005 | −0.017 | |||||
| Constant | 14.875 | *** | 1.628 | *** | 16.603 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of further vocational training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Western German women
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| women, west | 1,497 | 130,909 | ||||||
| Employment history attributes | ||||||||
| Never had a job | −7.640 | *** | −2.588 | *** | −8.450 | *** | ||
| Duration (months) with a job | 0.057 | *** | 0.021 | *** | 0.092 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −2.938 | *** | 4.663 | *** | 3.579 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −3.393 | *** | −1.309 | *** | −7.623 | *** | ||
| Average job-spell duration (months) | −0.036 | *** | −0.026 | *** | −0.019 | |||
| Duration since last job (months) | −0.059 | *** | −0.027 | *** | −0.081 | *** | ||
| Duration since median of months with a job (months) | −0.019 | *** | −0.001 | −0.025 | *** | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 3.117 | 2.238 | 13.647 | *** | ||||
| Interaction treatment / never had a job | 6.032 | * | 5.588 | *** | 13.509 | ** | ||
| Interaction treatment / job duration | 0.028 | −0.015 | −0.005 | |||||
| Interaction treatment / minijob proportion | 5.142 | ** | 2.033 | 9.288 | ** | |||
| Interaction treatment / self-employment proportion | −0.179 | 1.415 | −6.169 | |||||
| Interaction treatment / average job-spell duration | −0.010 | 0.001 | −0.005 | |||||
| Interaction treatment / duration since last job | 0.037 | ** | 0.034 | *** | 0.114 | *** | ||
| Interaction treatment / duration since median of months with a job (months) | −0.003 | −0.027 | *** | −0.067 | *** | |||
| Constant | 15.115 | *** | 6.418 | *** | 23.329 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
Linear regression for balanced sample. Interaction of Further Vocational Training participation in September–November 2012 with employment history attributes on specified outcomes in the time period from programme start to December 2018. ATTs. Western German men
| Outcomes | ||||||||
|---|---|---|---|---|---|---|---|---|
| Total time in regular jobs (months) | Total time in minijobs (months) | Probability of having at least one continuous job lasting at least 2 years in one firm (percentage points) | N (treated) | N (control group) | ||||
| 2,005 | 169,088 | |||||||
| Employment history attributes | ||||||||
| Never had a job | −4.090 | *** | −1.945 | *** | −3.774 | *** | ||
| Duration (months) with a job | 0.061 | *** | 0.020 | *** | 0.101 | *** | ||
| Proportion of time in minijobs of total months with a job (%) | −3.865 | *** | 6.384 | *** | 2.862 | *** | ||
| Proportion of time in self-employment of total months with a job (%) | −3.814 | *** | −0.774 | *** | −6.963 | *** | ||
| Average job-spell duration (months) | −0.039 | *** | −0.020 | *** | −0.035 | *** | ||
| Duration since last job (months) | −0.054 | *** | −0.017 | *** | −0.065 | *** | ||
| Duration since median of months with a job (months) | −0.033 | *** | 0.000 | −0.042 | *** | |||
| Workfare participation: treatment effects | ||||||||
| Baseline effect | 7.409 | *** | −0.089 | 5.972 | ||||
| Interaction treatment / never had a job | 4.109 | 2.903 | * | 15.032 | ** | |||
| Interaction treatment / job duration | 0.024 | * | −0.004 | 0.039 | ||||
| Interaction treatment / minijob proportion | 1.511 | 3.368 | 9.578 | * | ||||
| interaction treatment / self-employment proportion | −1.415 | 2.471 | ** | −0.118 | ||||
| Interaction treatment / average job-spell duration | 0.005 | 0.008 | 0.030 | |||||
| Interaction treatment / duration since last job | −0.008 | 0.008 | 0.018 | |||||
| Interaction treatment / duration since median of months with a job (months) | −0.014 | −0.007 | −0.032 | |||||
| Constant | 13.773 | *** | 3.279 | *** | 18.130 | *** | ||
Note: *p < 0.1; **p < 0.05; ***p < 0.01
These effects of employment history attributes are very similar for all four subgroups of eastern German women, eastern German men, western German women and western German men, for all samples of programme participants and weighted controls (Tables 3–14).
The lower parts of Tables 3–6 show interaction effects of these employment history attributes with workfare participation. In the four subgroups of women and men in eastern and western Germany, respectively, workfare effects tend to be more positive for people with no employment experience or little employment experience, and for those with long durations since their last job. Among western German women, those with higher proportions of minijob experience also have more positive workfare effects. Thus, workfare indeed appears more effective among its target group of people with low employment resources. For instance, for people with no employment experience, the total workfare effect on regular employment amounts to between −0.11 months (western German men) and +5.39 months (eastern German men). In comparison, the workfare effect on regular employment for an example of a person with comparatively beneficial employment history attributes (118 months of employment experience, no minijob or self-employment experience, an average job-spell duration of 12 months, 18 months duration since the last job, and 120 months duration since the median month of employment) ranges between −2.3 months (eastern German women) and +0.18 months (western German men).
Tables 7–10 give results for interactions of employment history attributes with participation in short training. Overall, not many interaction effects are statistically significant. Short training in some cases has significantly more positive effects for people with more employment experience. However, the opposite holds for effects on regular employment among eastern German women and on minijob employment among western German women. Among western German men, short training especially seems to enable regular employment for those with high proportions of minijob experience. Using the same examples as for workfare, for a person with no employment experience, the total short-training effect on regular employment ranges from +2.96 months (western German men) to +6.72 months (eastern German women), and that for a person with comparably favourable employment history attributes (as described earlier) amounts to between +2.88 months (western German men) and +5.4 months (eastern German women).
Tables 11–14 show findings for further vocational training. In eastern Germany, further vocational training has more positive effects on regular employment and less positive effects on minijob employment for those with more employment experience. In western Germany, people with low employment resources especially appear to benefit from further vocational training, particularly among western German women. Using the same examples as before, for a person with no employment experience, the total further vocational training effect on regular employment ranges from +2 months for eastern German men to +11.52 months for western German men, and that for a person with comparably favourable employment history attributes (as described earlier) from +6.5 months (western German women) to +8.51 months (eastern German men).
Conclusion
When seeking employment, middle-aged welfare recipients face disadvantages connected to age, on top of disadvantages experienced by all welfare recipients. Despite these difficulties, policies have shifted in the direction of encouraging labour market participation at higher ages. This article investigated the long-term employment biographies of a cohort of welfare recipients aged 45–54 in 2012, and related their employment biographies to their prospective employment outcomes. Gaining knowledge on the nature of middle-aged welfare recipients’ employment biographies can help to better understand their diversity and specific employment obstacles. Findings on active labour market policies indicate which policies are most effective for benefit recipients with varying employment biographies, and whether middle-aged benefit recipients can still profit from upskilling programmes for improving their employment prospects during the remainder of their labour market careers.
The analyses show that employment experiences over the past 19 years range from no employment at all, over infrequent to more frequent short employment spells, to longer continuous employment. Employment histories also vary by amounts of self-employment and uninsured minijobs, and by the phase in which employment experience was most concentrated. Individual job spells were on average longest among western German men and shortest among eastern German women. Minijob work experience was especially prevalent among western German women.
Overall employment experience was quite low. On average, female sample members were employed for around one quarter, and male sample members for around one third, of the retrospective observation period. Moreover, employment was quite unstable for many sample members: approximately a quarter of the sample had job spells lasting for seven months or less on average.
These employment history attributes influence prospective employment outcomes of middle-aged benefit recipients. Employment experience has positive effects on finding new employment, while the duration since the last job has negative effects. Moreover, some people have recently had a job, but have only sparse employment spells in recent years, which also negatively affects employment outcomes. The proportion of employment experience in minijobs positively affects minijob employment, and negatively affects regular employment. Somewhat surprisingly, a history of short employment spells, at a given level of overall employment experience, positively affects future employment prospects. Possibly, willingness to accept unstable employment is beneficial for obtaining a job given the types of jobs available to welfare benefit recipients.
Analyses of the impacts of short training, further vocational training and workfare showed that overall, further vocational training had the largest positive effects on regular insured employment and on the probability of obtaining at least one job lasting for at least two years. Further vocational training had no significant positive effects on uninsured minijob employment. Thus, further vocational training especially appears to raise prospects of good-quality employment. Short training programmes had somewhat smaller, but likewise positive employment effects. Workfare programmes had no significant overall employment effects for eastern Germany and very small effects for western Germany.
Interactions with employment history attributes showed that positive workfare effects were largely restricted to people with very little employment experience or long durations since their last job. However, among western German men, workfare effects remained close to zero even for those with no employment experience. Positive employment effects of further vocational training were larger than for workfare, both for people with favourable and for those with less favourable employment history attributes. Eastern German men are an exception, with comparatively low further vocational training effects among those with little employment experience. Short training showed medium-level positive employment effects that did not differ much by employment history attributes.
Overall, the results on programme effects show that it is not too late for middle-aged welfare recipients to profit from upskilling programmes. Although they may have only one or two decades left before retirement, training programmes substantially improve their overall opportunities of finding employment. Further vocational training especially improves opportunities of finding good-quality employment.
Further vocational training is more strongly directed towards people for whom training is thought ‘likely to be successful’, while workfare is directed towards people with low employment resources, and short training is a quite broadly applied measure. Analyses show that employment experience is lower among participants in workfare than among those in short training and especially further vocational training. However, people with less favourable or more favourable employment history attributes profit from further vocational training and short training. As further vocational training improves prospects of obtaining insured and stable employment, a recommendation would be to further encourage participation by people with low employment experience, sparse recent employment experience or employment experience predominantly in uninsured minijobs.
Middle-aged benefit recipients with a history of short employment spells (at a given level of overall employment experience) have better employment prospects overall, presumably because they are willing to accept all types of employment. In this context, a further benefit of further vocational training can be to help prevent continued job-cycling by enabling stable employment.
Acknowledgements
I would like to thank Sarah Bernhard, Timo Fleckenstein, Katrin Hohmeyer, Zein Kasrin, Joachim Wolff, the two anonymous reviewers, the participants of the RC19 session on Labour Markets, the ESA RN17 session on precarious work, the WORK 2021 Conference and all the members of the IAB working group on long-term unemployment for very helpful comments and advice, as well as Laurenz Dischinger for excellent research assistance.
Conflict of interest
The author declares that there is no conflict of interest.
References
Abraham, K.G., Haltiwanger, J., Sandusky, K. and Spletzer, J. (2019) The consequences of long-term unemployment: evidence from linked survey and administrative data, ILR Review, 72(2): 266–99. doi: 10.1177/0019793918797624
Annesley, C. (2007) Lisbon and social Europe: towards a European ‘adult worker model’ welfare system, Journal of European Social Policy, 17(3): 195–205. doi: 10.1177/0958928707078363
Arulampalam, W., Booth, A. and Taylor, M. (2000) Unemployment persistence, Oxford Economic Papers, 52(1): 24–50. doi: 10.1093/oep/52.1.24
Bernhard, S. (2016) Berufliche Weiterbildung von Arbeitslosengeld-II-Empfängern: Langfristige Wirkungsanalysen, Sozialer Fortschritt, 65(7): 153–61. doi: 10.3790/sfo.65.7.153
Beste, J. and Trappmann, M. (2016) Erwerbsbedingte Abgänge aus der Grundsicherung: der Abbau von Hemmnissen macht‘s möglich, IAB-Kurzbericht, no. 21/2016.
Boockmann, B., Fries, J. and Göbel, C. (2012) Specific Measures for Older Employees and Late Career Employment, ZEW Discussion Paper, No. 12-059.
Bruckmeier, K. and Hohmeyer, K. (2018) Nachhaltige Integration bleibt schwierig, IAB-Kurzbericht, no. 2/2018.
Brzinsky-Fay, C., Kohler, U. and Luniak, M. (2006) Sequence analysis with Stata, Stata Journal, 6(4): 435–60. doi: 10.1177/1536867X0600600401
Buddelmeyer, H., Lee, W.S. and Wooden, M. (2010) Paid employment and unemployment dynamics in Australia, Economic Record, 86(272): 28–48. doi: 10.1111/j.1475-4932.2009.00595.x
Cantillon, B. (2011) The paradox of the social investment state: growth, employment and poverty in the Lisbon era, Journal of European Social Policy, 21(5): 432–49. doi: 10.1177/0958928711418856
Carpentier, S., Neels, K. and Van den Bosch, K. (2017) Exit from and re-entry into social assistance benefit in Belgium among people with migration background and the native-born, International Journal of Social Welfare, 26(4): 366–83. doi: 10.1111/ijsw.12270
Chan, S. and Stevens, A.H. (2001) Job loss and employment patterns of older workers, Journal of Labor Economics, 19(2): 484–521. doi: 10.1086/319568
Cortis, N., Bullen, J. and Hamilton, M. (2013) Sustaining transitions from welfare to work: the perceptions of employers and employment service providers, Australian Journal of Social Issues, 48(3): 363–84. doi: 10.1002/j.1839-4655.2013.tb00287.x
de Coen, A., Forrier, A., de Cuyper, N. and Sels, L. (2015) Job seekers’ search intensity and wage flexibility: does age matter?, Ageing & Society, 35(2): 346–66. doi: 10.1017/S0144686X13000731
Dengler, K. (2019) Effectiveness of active labour market programmes on the job quality of welfare recipients in Germany, Journal of Social Policy, 48(4): 807–38. doi: 10.1017/S0047279419000114
Dengler, K., Hohmeyer, K. and Zabel, C. (2021) Welfare recipients’ transition into employment and employment stability in Germany, LABOUR, 35(4): 450–84. doi: 10.1111/labr.12207
Dietz, M. and Walwei, U. (2011) Germany – no country for old workers?, Zeitschrift für Arbeitsmarkt Forschung / Journal for Labour Market Research, 44(4): 363–76.
Diewald, M., Solga, H. and Goedicke, A. (2002) Back to labour markets – who got ahead in post-communist societies after 1989? The case of East Germany, European Societies, 4(1): 27–52. doi: 10.1080/14616690220130982
Eriksson, S. and Rooth, D.O. (2014) Do employers use unemployment as a sorting criterion when hiring? Evidence from a field experiment, American Economic Review, 104(3): 1014–39. doi: 10.1257/aer.104.3.1014
Farber, H.S., Silverman, D. and von Wachter, T.M. (2017) Factors determining callbacks to job applications by the unemployed: an audit study, RSF: Russell Sage Foundation Journal of the Social Sciences, 3(3): 168–201. doi: 10.7758/rsf.2017.3.3.08
Federal Employment Agency (2021) Fachliche Weisungen Förderung der beruflichen Weiterbildung nach §16 Abs. 1 SGB II i. V. m. §§81ff. SGB III. 8 February 2021.
Federal Employment Agency (2022) Fachliche Weisungen Zweites Buch Sozialgesetzbuch – SGB II §16d SGB II. Arbeitsgelegenheiten. Stand 1 January 2022. https://www.arbeitsagentur.de/datei/dok_ba045619.pdf
Fok, Y.K., Scutella, R. and Wilkins, R. (2015) The low-pay no-pay cycle: are there systematic differences across demographic groups?, Oxford Bulletin of Economics and Statistics, 77(6): 872–96. doi: 10.1111/obes.12084
Frodermann, C. and Müller, D. (2019) Establishment closures in Germany: the motherhood penalty at job search durations, European Sociological Review, 35(6): 845–59. doi: 10.1093/esr/jcz043
Gagliarducci, S. (2005) The dynamics of repeated temporary jobs, Labour Economics, 12(4): 429–48. doi: 10.1016/j.labeco.2005.05.001
Geisler, E. and Kreyenfeld, M. (2011) Against all odds: fathers’ use of parental leave in Germany, Journal of European Social Policy, 21(1): 88–99. doi: 10.1177/0958928710385732
Gregory, M. and Jukes, R. (2001) Unemployment and subsequent earnings: estimating scarring among British men 1984–94, The Economic Journal, 111(475): 607–25. doi: 10.1111/1468-0297.00665
Hägglund, A. and Bächmann, A.C. (2017) Fast lane or down the drain? Does the occupation held prior to unemployment shape the transition back to work?, Research in Social Stratification and Mobility, 49: 32–46.
Hainmueller, J. (2012) Entropy balancing for causal effects: a multivariate reweighting method to produce balanced samples in observational studies, Political Analysis, 20(1): 25–46. doi: 10.1093/pan/mpr025
Hainmueller, J. and Xu, Y. (2013) ebalance: a stata package for entropy balancing, Journal of Statistical Software, 54(7): 1–18. doi: 10.18637/jss.v054.i07
Halpin, B. (2014) SADI: Sequence Analysis Tools for Stata, Working Paper, Department of Sociology, University of Limerick WP2014-03, http://www.ul.ie/sociology/pubs/wp2014-03.pdf.
Halpin, B. (2016) SILHOUETTE: Stata Module to Calculate and Graph Silhouette Width for Cluster Analysis, Statistical Software Components S458165, Boston College Department of Economics, revised 27 June 2016, https://econpapers.repec.org/software/bocbocode/s458165.htm.
Halpin, B. and Chan, T.W. (1998) Class careers as sequences: an optimal matching analysis of work-life histories, European Sociological Review, 14(2): 111–30. doi: 10.1093/oxfordjournals.esr.a018230
Harrer, T. and Stockinger, B. (2021) First step and last resort: One-Euro-Jobs after the reform, Journal of Social Policy, 51(2): 412–34. doi: 10.1017/S0047279421000313
Harrer, T., Moczall, A. and Wolff, J. (2020) Free, free, set them free? Are programmes effective that allow job centres considerable freedom to choose the exact design?, International Journal of Social Welfare, 29(2): 154–67. doi: 10.1111/ijsw.12405
Henkens, K., Sprengers, M. and Tazelaar, F. (1996) Unemployment and the older worker in the Netherlands: re-entry into the labour force or resignation, Ageing & Society, 16(5): 561–78. doi: 10.1017/S0144686X00020237
Hohmeyer, K. and Lietzmann, T. (2020) Persistence of welfare receipt and unemployment in Germany: determinants and duration dependence, Journal of Social Policy, 49(2): 299–322. doi: 10.1017/S0047279419000242
Hohmeyer, K. and Wolff, J. (2012) A fistful of euros: is the German one-euro job workfare scheme effective for participants?, International Journal of Social Welfare, 21(2): 174–85. doi: 10.1111/j.1468-2397.2011.00830.x
Jarosch, G. (2021) Searching for Job Security and the Consequences of Job Loss, NBER Working Paper Series No. 28481.
Keller, M. and Kahle, I. (2018) Realisierte Erwerbstätigkeit von Müttern und Vätern zur Vereinbarkeit von Familie und beruf, Statistisches Bundesamt (Destatis), 3: 55–72.
Kohler, U., Luniak, M. and Brzinsky-Fay, C. (2020) SQ: Stata module for sequence analysis. https://econpapers.repec.org/software/bocbocode/s456755.htm
Lengyel, A. and Botta‐Dukát, Z. (2019) Silhouette width using generalized mean: a flexible method for assessing clustering efficiency, Ecology and Evolution, 9(3): 13231–43. doi: 10.1002/ece3.5774
Loichinger, E. and Prskawetz, A. (2017) Changes in economic activity: the role of age and education, Demographic Research, 36: 1185–208. doi: 10.4054/DemRes.2017.36.40
Loichinger, E. and Skirbekk, V. (2016) International variation in ageing and economic dependency: a cohort perspective, Comparative Population Studies, 41(2), doi: 10.12765/CPoS-2016^=^#4.
Mika, T. and Baumann, J. (2008) Soziale Konsequenzen der Abschaffung des Vorruhestandes für langzeitarbeitslose, WSI Mitteilungen, 11/12: 605–11.
Mincer, J. and Ofek, H. (1982) Interrupted work careers: depreciation and restoration of human capital, Journal of Human Resources, 17(1): 3–24. doi: 10.2307/145520
Nagore García, A. and van Soest, A. (2017) Unemployment exits before and during the crisis, Labour, 31(4): 337–68.
Nivorozhkin, A. and Nivorozhkin, E. (2020) Job search, transition to employment and discouragement among older unemployed welfare recipients in Germany, Social Policy & Administration, 55(4): 747–65.
Piccarreta, R. (2012) Graphical and smoothing techniques for sequence analysis, Sociological Methods & Research, 41(2): 362–80. doi: 10.1177/0049124112452394
Piccarreta, R. and Lior, O. (2010) Exploring sequences: a graphical tool based on multi-dimensional scaling, Journal of the Royal Statistical Society: Series A (Statistics in Society), 173(1): 165–84. doi: 10.1111/j.1467-985X.2009.00606.x
Romeu Gordo, L. and Wolff, J. (2011) Creating employment or keeping them busy? An evaluation of training programs for older workers in Germany, Journal of Aging & Social Policy, 23(2): 198–218.
Schmelzer, P., Kurz, K. and Schulze, K. (2015) Einkommensnachteile von Müttern im Vergleich zu kinderlosen Frauen in Deutschland, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 67(4): 737–62.
SDFLA (Statistics Department of the Federal Labour Agency) (2020) Tabellen, Langzeitleistungsbezieher, November 2020, Nuremberg: Bundesagentur für Arbeit. https://statistik.arbeitsagentur.de/SiteGlobals/Forms/Suche/Einzelheftsuche_Formular.html?topic_f=lzb-lzb-strukturen
SDFLA (Statistics Department of the Federal Labour Agency) (2022a) Arbeitsgelegenheiten, May 2022, Nuremberg: Bundesagentur für Arbeit. https://statistik.arbeitsagentur.de/SiteGlobals/Forms/Suche/Einzelheftsuche_Formular.html?nn=1524036&topic_f=agh
SDFLA (Statistics Department of the Federal Labour Agency) (2022b) Förderung der Beruflichen Weiterbildung, Nuremberg: Bundesagentur für Arbeit. https://statistik.arbeitsagentur.de/SiteGlobals/Forms/Suche/Einzelheftsuche_Formular.html?nn=1524036&topic_f=teilnehmer-massnahmen-fbw
Simonson, J., Romeu Gordo, L. and Kelle, N. (2015) Separate paths, same direction? De-standardization of male employment biographies in East and West Germany, Current Sociology, 63(3): 387–410. doi: 10.1177/0011392115572380
Stewart, K. (2014) Employment trajectories and later employment outcomes for mothers in the British Household Panel Survey: an analysis by skill level, Journal of Social Policy, 43(1): 87–108. doi: 10.1017/S004727941300055X
Stewart, M.B. (2007) The interrelated dynamics of unemployment and low-wage employment, Journal of Applied Econometrics, 22(3): 511–31. doi: 10.1002/jae.922
Tisch, A. (2015) The employability of older job-seekers: evidence from Germany, Journal of the Economics of Ageing, 6: 102–12. doi: 10.1016/j.jeoa.2014.07.001
Trappe, H., Pollmann-Schult, M. and Schmitt, C. (2015) The rise and decline of the male breadwinner model: institutional underpinnings and future expectations, European Sociological Review, 31(2): 230–42. doi: 10.1093/esr/jcv015
Uhlendorff, A. (2006) From No Pay to Low Pay and Back Again? A Multi-State Model of Low Pay Dynamics, IZA Discussion Paper, No. 2482.
Visser, M., Gesthuizen, M., Kraaykamp, G. and Wolbers, M. (2016) Inequality among older workers in the Netherlands: a life course and social stratification perspective on early retirement, European Sociological Review, 32(3): 370–82. doi: 10.1093/esr/jcw013
Wuebbeke, C. (2011) The limitations of activation policies: unemployment at the end of working life, Ageing & Society, 31(6): 977–1002. doi: 10.1017/S0144686X10000929
Zacher, H. (2013) Older job seekers’ job search intensity: the interplay of proactive personality, age and occupational future time perspective, Ageing & Society, 33(7): 1139–66. doi: 10.1017/S0144686X12000451
Ziefle, A. (2004) Die individuellen kosten des Erziehungsurlaubs. eine empirische analyse der kurz- und längerfristigen folgen für den karriereverlauf von Frauen, Kölner Zeitschrift für Soziologie und Sozialpsychologie, 56(2): 213–31.
Appendix
Employment history clusters for eastern German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for eastern German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for eastern German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for eastern German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for eastern German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for eastern German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German women
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Employment history clusters for western German men
Citation: Longitudinal and Life Course Studies 14, 3; 10.1332/175795921X16643819920960
Mean silhouette width for 2 to 8 cluster solutions
| Women, east | Men, east | Women, west | Men, west | |
|---|---|---|---|---|
| 2 clusters | 0.36 | 0.38 | 0.34 | 0.45 |
| 3 clusters | 0.33 | 0.30 | 0.33 | 0.24 |
| 4 clusters | 0.28 | 0.09 | 0.34 | 0.21 |
| 5 clusters | 0.28 | 0.07 | 0.18 | 0.19 |
| 6 clusters | 0.11 | 0.06 | 0.17 | 0.19 |
| 7 clusters | 0.12 | 0.07 | 0.18 | 0.18 |
| 8 clusters | 0.12 | 0.07 | 0.17 | 0.18 |
Multidimensional scaling, eigenvalues
| Dimension | Eigenvalue | Abs(eigenvalue), % | (eigenvalue)^2, % | |
|---|---|---|---|---|
| Women, east | 1 | 4891261 | 41.0 | 94.8 |
| 2 | 834227 | 7.0 | 2.8 | |
| 3 | 328906 | 2.8 | 0.4 | |
| 4 | 241009 | 2.0 | 0.2 | |
| 5 | 217886 | 1.8 | 0.2 | |
| Men, east | 1 | 5664714 | 41.1 | 95.2 |
| 2 | 930310 | 6.7 | 2.6 | |
| 3 | 376091 | 2.7 | 0.4 | |
| 4 | 301165 | 2.2 | 0.3 | |
| 5 | 248510 | 1.8 | 0.2 | |
| Women, west | 1 | 5695174 | 42.2 | 94.4 |
| 2 | 1032625 | 7.7 | 3.1 | |
| 3 | 425499 | 3.2 | 0.5 | |
| 4 | 260198 | 1.9 | 0.2 | |
| 5 | 257594 | 1.9 | 0.2 | |
| Men, west | 1 | 7326300 | 44.5 | 96.1 |
| 2 | 1035932 | 6.3 | 1.9 | |
| 3 | 499966 | 3.0 | 0.5 | |
| 4 | 371181 | 2.3 | 0.3 | |
| 5 | 276170 | 1.7 | 0.1 |
Regression results for first dimension distances of multidimensional scaling
| Women, east | Men, east | Women, west | Men, west | |||||
|---|---|---|---|---|---|---|---|---|
| Ever had a job (%) | −0.05 | −0.01 | −0.04 | 0.05 | ||||
| Duration (months) with a job | −1.74 | *** | −1.84 | *** | −1.72 | *** | −1.91 | *** |
| Proportion of time in minijobs of total months with a job (%) | 0.46 | *** | 0.71 | *** | 0.52 | *** | 0.56 | *** |
| Proportion of time in self-employment of total months with a job (%) | 0.10 | ** | 0.11 | *** | 0.07 | 0.20 | *** | |
| Average job spell duration (months) | −0.22 | *** | −0.17 | *** | −0.22 | *** | −0.11 | *** |
| Duration since last job (months) | 0.00 | −0.02 | −0.04 | −0.06 | *** | |||
| Duration since median of months with a job (months) | −0.04 | ** | 0.00 | −0.04 | * | 0.00 | ||
| Constant | 103.84 | *** | 140.40 | ** | 106.80 | *** | 151.45 | *** |
| R2 | 0.97 | 0.99 | 0.96 | 0.99 |
Note: Random sample of 500 sequences for each subsample.
Regression results for second dimension distances of multidimensional scaling
| Women, east | Men, east | Women, west | Men, west | |||||
|---|---|---|---|---|---|---|---|---|
| Ever had a job (%) | −0.58 | *** | −0.60 | *** | 0.60 | *** | −0.59 | *** |
| Duration (months) with a job | 0.06 | −0.02 | −0.12 | ** | 0.03 | |||
| Proportion of time in minijobs of total months with a job (%) | 0.55 | *** | 0.56 | *** | −0.64 | *** | 0.36 | *** |
| Proportion of time in self-employment of total months with a job (%) | 0.34 | *** | 0.27 | *** | −0.39 | *** | 0.03 | |
| Average job spell duration (months) | −0.20 | ** | −0.24 | *** | 0.17 | ** | −0.21 | *** |
| Duration since last job (months) | 0.05 | 0.02 | 0.01 | −0.04 | ||||
| Duration since median of months with a job (months) | −0.47 | *** | −0.68 | *** | 0.43 | *** | −0.64 | *** |
| Constant | 114.80 | *** | 169.14 | *** | −98.36 | *** | 156.81 | *** |
| R2 | 0.37 | 0.47 | 0.43 | 0.46 |
Note: Random sample of 500 sequences for each subsample.
Variables used for entropy balancing
| Individual socio-demographic variables | - age - disability - citizenship - vocational degree |
| Labour market history indicators | - ever employed in retrospective observation period - total employment experience - proportion of time in minijobs of employment experience - proportion of time self-employed of employment experience - average duration of job spells - duration since the last job - duration since median of months with a job - 1-digit ISCO of the last occupation - earnings in last job - participation in job creation schemes in last two years - participation in job subsidies in last two years - participation in further vocational training in last two years - participation in retraining in last two years - participation in short training in last two years - participation in firm-based training in last two years - participation in workfare in last two years - past duration of means-tested unemployment benefit (Unemployment Benefit II) receipt - present short-term availability for employment - past duration of unavailability due to health issues during unemployment - past duration with reduced capability of employment |
| Household characteristics | - partnership status (no partner and never married; no partner and ever married; currently cohabiting; currently married) - children in the household (yes/no) - age of children living in the household (0–25) - proxy for adult children who have already left the household |
| Partner characteristics | - disability - citizenship - vocational degree - current capability of employment - retirement - duration since last job - last job minijob, insured full-time, insured part-time, firm-based apprenticeship - earnings in the last job - 1-digit ISCO of the last occupation |
| Regional characteristics | - classification of the regional labour market structure - regional unemployment rate - regional long-term unemployment rate - regional male- and female-specific Unemployment Benefit II rates - regional vacancy-to-unemployment ratio - job centre-specific inflow rates into various employment programmes |
Association between characteristics of employment histories and socio-demographic variables. Women, east
| Among people who have ever had a job | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Probability ever had a job (%) | Duration (months) with a job | Proportion of time in minijobs of total months with a job (%) | Proportion of time in self-employment of total months with a job (%) | Average job duration (months) | Duration since last job (months) | Duration since median of months with a job (months) | ||||||||
| Age | −0.28 | *** | 0.31 | *** | −0.18 | *** | −0.03 | 0.31 | *** | 0.95 | *** | 0.87 | *** | |
| Partnership status (reference: no partner, never married) | ||||||||||||||
| No partner in household, ever married | −0.08 | 6.05 | *** | 2.52 | *** | 1.38 | *** | 1.46 | *** | −7.04 | *** | −4.31 | *** | |
| Cohabiting | −1.39 | *** | −2.59 | *** | 1.02 | ** | −0.51 | * | 0.28 | 2.93 | *** | 0.02 | ||
| Married | −4.26 | *** | 4.18 | *** | 3.57 | *** | 0.22 | 3.24 | *** | −2.59 | *** | −1.49 | *** | |
| Children (reference: no children) | ||||||||||||||
| Children aged <18 in household | −6.27 | *** | −22.55 | *** | 3.74 | *** | 0.70 | −8.49 | *** | 14.75 | *** | 4.35 | *** | |
| Children aged 18–24 only in household | −3.96 | *** | −28.09 | *** | 8.78 | *** | −2.39 | *** | −11.36 | *** | −1.47 | −6.17 | *** | |
| All children have already moved out of household (proxy) | 1.51 | *** | −19.12 | *** | 5.47 | *** | −2.37 | *** | −10.68 | *** | −0.28 | 0.35 | ||
| Age of the youngest child (reference: age <5 years) | ||||||||||||||
| Youngest child aged 6–9 | −1.80 | * | −7.28 | *** | −0.03 | −1.37 | ** | −2.75 | *** | −0.89 | 3.59 | * | ||
| Youngest child aged 10–14 | −0.07 | −10.35 | *** | 2.97 | *** | −2.05 | *** | −3.73 | *** | −7.25 | *** | 3.99 | ** | |
| Youngest child aged 15–17 | 0.84 | −10.61 | *** | 6.58 | *** | −1.56 | *** | −4.07 | *** | −14.33 | *** | −7.88 | *** | |
| With German citizenship | 1.37 | *** | −7.41 | *** | −1.58 | *** | −3.12 | *** | −1.49 | *** | 14.28 | *** | 16.47 | *** |
| With disability | 0.71 | * | 3.11 | *** | −2.84 | *** | −2.07 | *** | 0.70 | ** | 6.77 | *** | 1.97 | *** |
| Vocational degree (reference: no vocational degree) | ||||||||||||||
| Vocational degree | 15.14 | *** | 18.31 | *** | −2.60 | *** | −0.30 | 1.40 | *** | −22.00 | *** | −6.87 | *** | |
| University or technical college degree | 14.70 | *** | 29.17 | *** | −8.38 | *** | 9.00 | *** | 4.03 | *** | −33.94 | *** | −23.45 | *** |
| Constant | 79.02 | *** | 70.07 | *** | 18.02 | *** | 8.60 | *** | 25.15 | *** | 74.37 | *** | 150.42 | *** |
| N | 73,516 | 66,270 | 66,270 | 66,270 | 66,270 | 66,270 | 66,270 | |||||||
| * p<.1; **p<.05; ***p<.01 | ||||||||||||||
Association between characteristics of employment histories and socio-demographic variables. Men, east
| Among people who ever had a job | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Probability ever had a job (%) | Duration (months) with a job | Proportion of time in minijobs of total months with a job (%) | Proportion of time in self-employment of total months with a job (%) | Average job duration (months) | Duration since last job (months) | Duration since median of months with a job (months) | ||||||||
| Age | −0.09 | *** | 0.01 | −0.16 | *** | −0.13 | *** | 0.29 | *** | 1.15 | *** | 0.66 | *** | |
| Partnership status (reference: no partner, never married) | ||||||||||||||
| No partner in household, ever married | 0.82 | *** | 12.66 | *** | 0.25 | 3.18 | *** | 3.83 | *** | −6.20 | *** | −1.88 | *** | |
| Cohabiting | 0.50 | ** | 8.14 | *** | 0.94 | *** | 0.13 | 1.70 | *** | −6.34 | *** | −1.09 | ** | |
| Married | −0.04 | 19.04 | *** | 0.78 | *** | 0.15 | 6.08 | *** | −9.61 | *** | −4.10 | *** | ||
| Children (reference: no children) | ||||||||||||||
| Children aged <18 in household | 0.81 | ** | 1.87 | * | 1.87 | *** | 4.59 | *** | −3.29 | *** | −9.37 | *** | −9.42 | *** |
| Children aged 18–24 only in household | 1.01 | *** | −0.79 | 0.90 | ** | −0.91 | ** | −2.95 | *** | −5.18 | *** | −0.42 | ||
| All children have already moved out of household (proxy) | 1.70 | *** | 1.11 | ** | 1.33 | *** | −1.34 | *** | −5.41 | *** | −6.52 | *** | 1.07 | *** |
| Age of the youngest child (reference: age <5 years) | ||||||||||||||
| Youngest child aged 6–9 | −0.89 | * | −2.18 | −0.30 | −0.78 | 0.77 | 2.46 | 5.16 | *** | |||||
| Youngest child aged 10–14 | −0.13 | −1.92 | −0.61 | −3.52 | *** | 0.15 | 3.85 | *** | 7.28 | *** | ||||
| Youngest child aged 15–17 | −0.43 | 0.22 | −0.48 | −3.78 | *** | 0.74 | 1.37 | 7.94 | *** | |||||
| With German citizenship | −1.89 | *** | −3.42 | *** | −3.06 | *** | −5.80 | *** | 0.75 | * | 19.25 | *** | 15.57 | *** |
| With disability | −0.31 | −1.17 | ** | −0.65 | *** | −2.63 | *** | 0.73 | ** | 10.21 | *** | 2.11 | *** | |
| Vocational degree (reference: no vocational degree) | ||||||||||||||
| Vocational degree | 7.13 | *** | 22.46 | *** | −1.93 | *** | −2.62 | *** | 1.21 | *** | −22.09 | *** | −1.21 | *** |
| University or technical college degree | 6.24 | *** | 31.91 | *** | −3.04 | *** | 11.27 | *** | 8.51 | *** | −33.12 | *** | −21.54 | *** |
| Constant | 91.44 | *** | 54.23 | *** | 12.98 | *** | 14.78 | *** | 17.18 | *** | 73.87 | *** | 152.48 | *** |
| N | 95,692 | 92,399 | 92,399 | 92,399 | 92,399 | 92,399 | 92,399 | |||||||
| * p<.1; **p<.05; ***p<.01 | ||||||||||||||
Association between characteristics of employment histories and socio-demographic variables. Women, west
| Among people who ever had a job | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Probability ever had a job (%) | Duration (months) with a job | Proportion of time in minijobs of total months with a job (%) | Proportion of time in self-employment of total months with a job (%) | Average job duration (months) | Duration since last job (months) | Duration since median of months with a job (months) | ||||||||
| Age | −0.54 | *** | 0.40 | *** | −0.08 | ** | 0.06 | *** | 0.53 | *** | 0.69 | *** | 0.68 | *** |
| Partnership status (reference: no partner, never married) | ||||||||||||||
| No partner in household, ever married | 1.98 | *** | −0.65 | * | 9.46 | *** | 0.43 | *** | −1.86 | *** | −8.63 | *** | −5.40 | *** |
| Cohabiting | −1.97 | *** | −4.81 | *** | 9.13 | *** | −0.28 | −1.87 | *** | −3.84 | *** | −3.28 | *** | |
| Married | −7.83 | *** | −3.93 | *** | 19.62 | *** | 0.13 | 1.08 | *** | −4.14 | *** | 3.87 | *** | |
| Children (reference: no children) | ||||||||||||||
| Children aged <18 in household | −7.55 | *** | −20.31 | *** | 7.41 | *** | −0.51 | −6.24 | *** | 8.78 | *** | 4.06 | *** | |
| Children aged 18–24 only in household | −6.18 | *** | −24.19 | *** | 18.42 | *** | −1.44 | *** | −8.05 | *** | −8.42 | *** | −11.62 | *** |
| All children have already moved out of household (proxy) | −0.31 | −17.99 | *** | 9.78 | *** | −1.66 | *** | −8.97 | *** | −7.04 | *** | −10.63 | *** | |
| Age of the youngest child (reference: age <5 years) | ||||||||||||||
| Youngest child aged 6–9 | −2.03 | *** | −6.52 | *** | 0.45 | −0.89 | ** | −2.56 | *** | 0.25 | 8.26 | *** | ||
| Youngest child aged 10–14 | −1.89 | *** | −9.02 | *** | 6.17 | *** | −0.55 | −3.72 | *** | −6.57 | *** | 8.08 | *** | |
| Youngest child aged 15–17 | 1.12 | −7.14 | *** | 11.07 | *** | −0.29 | −3.62 | *** | −13.77 | *** | −4.23 | *** | ||
| With German citizenship | −3.18 | *** | −4.98 | *** | 3.96 | *** | −0.22 | 0.32 | 2.59 | *** | 2.94 | *** | ||
| With disability | −0.07 | 0.73 | −3.17 | *** | −2.22 | *** | −0.19 | 7.95 | *** | 3.69 | *** | |||
| Vocational degree (reference: no vocational degree) | ||||||||||||||
| Vocational degree | 15.31 | *** | 24.74 | *** | −10.65 | *** | 1.61 | *** | 2.69 | *** | −19.17 | *** | −14.32 | *** |
| University or technical college degree | 13.53 | *** | 17.62 | *** | −14.08 | *** | 7.64 | *** | −0.12 | −21.04 | *** | −24.79 | *** | |
| Constant | 83.46 | *** | 77.79 | *** | 19.26 | *** | 5.63 | *** | 25.83 | *** | 75.23 | *** | 164.03 | *** |
| N | 142,283 | 120,277 | 120,277 | 120,277 | 120,277 | 120,277 | 120,277 | |||||||
| * p<.1; **p<.05; ***p<.01 | ||||||||||||||
Association between characteristics of employment histories and socio-demographic variables. Men, west
| Among people who ever had a job | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Probability ever had a job (%) | Duration (months) with a job | Proportion of time in minijobs of total months with a job (%) | Proportion of time in self-employment of total months with a job (%) | Average job duration (months) | Duration since last job (months) | Duration since median of months with a job (months) | ||||||||
| Age | −0.32 | *** | −0.07 | 0.03 | * | 0.23 | *** | 0.58 | *** | 1.17 | *** | 0.63 | *** | |
| Partnership status (reference: no partner, never married) | ||||||||||||||
| No partner in household, ever married | 2.84 | *** | 17.30 | *** | −0.89 | *** | 2.78 | *** | 4.18 | *** | −11.49 | *** | −3.48 | *** |
| Cohabiting | 1.37 | *** | 8.31 | *** | 0.32 | 2.13 | *** | 1.05 | *** | −8.58 | *** | −2.86 | *** | |
| Married | 2.66 | *** | 21.91 | *** | −0.66 | *** | 0.01 | 6.65 | *** | −14.59 | *** | −8.49 | *** | |
| Children (reference: no children) | ||||||||||||||
| Children aged <18 in household | 1.64 | *** | 0.12 | 1.11 | *** | 3.58 | *** | −5.15 | *** | −10.07 | *** | −7.93 | *** | |
| Children aged 18–24 only in household | 0.69 | ** | 1.02 | 0.84 | *** | 0.14 | −0.39 | −3.63 | *** | −1.09 | * | |||
| All children have already moved out of household (proxy) | 2.15 | *** | −3.98 | *** | 1.50 | *** | −0.51 | *** | −7.32 | *** | −6.09 | *** | −0.94 | *** |
| Age of the youngest child (reference: age <5 years) | ||||||||||||||
| Youngest child aged 6–9 | −0.16 | −1.19 | 0.36 | −1.59 | *** | 1.29 | ** | 3.65 | *** | 3.88 | *** | |||
| Youngest child aged 10–14 | −0.62 | * | 1.40 | * | −0.20 | −1.91 | *** | 2.87 | *** | 4.14 | *** | 5.50 | *** | |
| Youngest child aged 15–17 | −0.62 | * | 2.81 | *** | −0.37 | −2.76 | *** | 2.97 | *** | 3.19 | *** | 6.93 | *** | |
| With German citizenship | −3.17 | *** | −1.45 | *** | −0.60 | *** | 0.23 | 3.12 | *** | 10.97 | *** | 5.87 | *** | |
| With disability | −0.55 | *** | −0.22 | −0.46 | *** | −4.19 | *** | 0.60 | ** | 8.32 | *** | 4.56 | *** | |
| Vocational degree (reference: no vocational degree) | ||||||||||||||
| Vocational degree | 6.50 | *** | 25.79 | *** | −2.47 | *** | 1.03 | *** | 2.31 | *** | −22.63 | *** | −7.36 | *** |
| University or technical college degree | 5.87 | *** | 23.06 | *** | −2.04 | *** | 9.28 | *** | 3.33 | *** | −26.16 | *** | −20.91 | *** |
| Constant | 90.99 | *** | 60.79 | *** | 12.78 | *** | 8.33 | *** | 19.12 | *** | 77.90 | *** | 158.48 | *** |
| N | 183,029 | 173,348 | 173,348 | 173,348 | 173,348 | 173,348 | 173,348 | |||||||
| * p<.1; **p<.05; ***p<.01 | ||||||||||||||
