Differences in the Patterns of in-work Poverty in Germany and the UK

In-work poverty has become an important issue across post-industrial countries and is widely viewed as a by-product of processes of globalisation and changing social- and labour market policies (Lohmann and Andreß 2008; Brady et al.2010; Fraser et al.2011). In this paper, we argue that such policies have been implemented both in Germany and the UK, but affect different socio-economic groups in both countries.

We focus on Germany and the UK, as these countries are at the centre of interest in recent comparative policy research (Clasen 2005; Seeleib-Kaiser and Fleckenstein 2007). This research identified similarities in the strategies of both countries for adapting to macro-economic challenges, but also found remaining institutional core differences (Clasen et al.2011). Consequently, the development of social and labour market policies in Germany and the UK is often discussed as partial policy convergence (Roder 2003; Mohr 2008). However, most of the relevant comparative literature focuses on policy analysis, not on (individual level) policy outcomes. The present study therefore seeks to contribute to the literature by linking institutional patterns in Germany and the UK with micro-level outcomes; in this case, with in-work poverty.

We argue that centralised bargaining arrangements and strict employment protection legislation (EPL) in Germany give both core workers and older employees protective rights and thereby channel the effects of labour market deregulation to the periphery of the labour market. By contrast, a context with open employment relationships, as in the UK, is assumed to shift economic risks to individuals with low human capital resources. In addition, we assume that the more dominant male breadwinner orientation of family policy in Germany causes increased economic dependency of women, resulting in a more precarious situation for single-female households.

Respective hypotheses on country-specific differences in the determinants of in-work poverty are tested on the basis of the German Socio-Economic Panel Study (SOEP) and the British Household Panel Study (BHPS).

This section provides an overview of the institutional labour market contexts in Germany and the UK. The two labour markets are classified principally on the basis of macro-indicators from large-scale comparative projects. We focus on the degrees of decommodification and regulation, strictness of employment protection, bargaining centralisation and family policy orientation. Therefore, this study refers to measures that are treated as relevant in the aforementioned policy discourse and established as key macro-variables for explaining economic micro-level outcomes (Lohmann 2009; Baranowska and Gebel 2010; Brady et al.2010). Table 1 compares Germany and the UK based on these measures. In all dimensions, Germany and the UK are compared with reference to the distribution within the group of EU-15 countries. Reported are standardised continuous scores. As rule of thumb, we regard cross-country differences of more than one standard deviation as substantial.

The average net replacement rate during unemployment (OECD 2010) is used as indicator for the degree of decommodification (Lohmann 2009). This measure refers to the overall generosity of the welfare state with regard to unemployment benefits. Both Germany (0.3) and the UK (0.1) perform close to the EU average in this dimension. Thus, the degree of decommodification is moderate in both countries when compared to the other EU-15 countries. This assessment is supported by a modernised version of Esping-Andersen's (1990) decommodification index for unemployment (Scruggs and Allan 2006; Scruggs 2005), based on policy measures from 2002.

The similarity of indicators for the UK and Germany contradicts an established view on the respective benefit systems: traditionally, the German system of unemployment insurance grants unlimited earnings-related benefits and is focused on status preservation. The UK benefit system, in contrast, reflects a liberal workfare paradigm and is based on means-tested allowances that are designed to provide a minimum standard of living (Clasen 2005). Obviously, the standardised indicators provided above reflect recent changes in the unemployment benefit system in Germany.

While the Hartz IV reform of 2005 is usually characterised as a paradigmatic shift from the conservative German welfare state towards a more liberal model (Mohr 2008; Palier and Thelen 2010), earlier reform measures (like the 2001 Job-Aqtiv legislation) had already started to erode the principle of status maintenance and employed many measures focused on a means-tested provision of welfare. Accordingly, the similarity of standardised indicators is in line with policy analyses that find a convergence in unemployment benefit policies in Germany and the UK (Seeleib-Kaiser and Fleckenstein 2007; Mohr 2008). In conclusion, we can cautiously assume that the level of decommodification is no longer a core difference in the institutional setting between Germany and the UK.¹

The strictness of regulation of temporary employment (OECD 2009) is used to describe the degree of regulation. This indicator refers to the maximum number of successive fixed-term and temporary work contracts, the maximum cumulated duration of such contracts and other restrictions on atypical employment (OECD 2004). Taken together, its items reflect a wide range of measures typically associated with politically administered labour market deregulation.

While the figures in Table 1 still exhibit a lower level of labour market regulation in the UK (–1.6) than in Germany (–0.8), the two countries are separated by less than one standard deviation. These similar levels of indicators are in line with findings of recent processes of asymmetric labour policy convergence (Seeleib-Kaiser and Fleckenstein 2007; Clasen et al.2011): several policy measures implemented between 1997 and 2005 loosened restrictions on atypical employment in Germany. Like previously in the UK, instruments were implemented to encourage greater flexibility in marginal employment, temporary and agency work and subsidised self-employment (Palier and Thelen 2010). Thus, regarding available indicators and findings from policy analysis, the degree of regulation is no longer a core difference in the institutional configurations of Germany and the UK.

Minimum-wage settlements, as another relevant dimension of labour market regulation, are not regarded in the OECD-measure. Nevertheless, country-specific differences in minimum wage policies exist: while there is a national minimum (hourly) wage in the UK, no mandatory legislative regulations are in place in Germany (Clasen 2005). However, due to collective agreements, a variety of industry-specific wage standards do exist in Germany, covering approximately 60% of employees (Visser 2004). An indicator on minimum wages provided by the World Bank (2013) covers legislative regulations as well as collective agreements. This indicator outlines similar levels of minimum wages (measured as ratio to the value of gross domestic product (GDP) added per worker in 2008) in Germany (0.2) and the UK (0.3, see standardised scores in Table 1) and consequently is in line with our general assessment in this policy area.

To measure differences in the system of Employment protection legislation (EPL), this study refers to regulations governing the dismissal of employees on regular contracts (OECD 2009). This indicator reflects several aspects of labour market legislation, including the strictness of settlements regarding notification procedures, the length of notice periods, severance pay arrangements, the generosity of definitions of justified dismissals and the length of trial periods (OECD 2004).

Of all of the institutional dimensions discussed in this paper, EPL is the one in which Germany and the UK differ most widely; while Germany performs at the upper bound of the standard deviation around the mean of EU-15 countries (0.8), the UK is found at the lower end of the distribution (–1.6). This difference reflects major disparities in the definition of notice periods, specifically for employees with a high level of seniority (Ebbinghaus and Eichhorst 2009). Another core difference in the two countries' employment protection laws lies in their definition of ‘justified’ dismissals, which are substantially more generous to employers in the UK than in Germany (OECD 2004). In conclusion, the majority of workers in Germany are better protected from job loss, status decrease and shifts into atypical types of employment via EPL than workers in the UK.

The degree of centralisation is used as indicator of the configuration of the bargaining system (Lohmann 2009). This concept is strongly related to the authority held by the trade unions, the competitive situation between trade unions and the options for concentrated and coordinated action (Visser 2004). The indicator is taken from the ICTWSS Database (Visser 2009).

Germany and the UK exhibit considerable differences in the configuration of their bargaining systems. While the degree of centralisation in Germany (0.4) ranks above the EU-15 average, the UK (–1.1) is located at the lower end of the standard interval around the mean. This indicator reveals that in Germany the trade unions are organised on a sectoral basis and are highly concentrated, whereas in the UK bargaining authorities are company-based and fragmented (Visser 2004). It also mirrors differences in other key features of the collective bargaining system: although union density is similar in both countries (approximately 30%, OECD 2004), the degree of inclusion, which measures the share of employees whose working conditions are negotiated by unions, is much higher in Germany (61%) than it is in the UK (36%, Visser 2004). Thus, available indicators reveal that employee interests regarding labour conditions and labour outcomes are subject to a higher level of coordination through the collective bargaining system in Germany than in the UK.

For the assessment of country-specific family policies, this paper adopts the typology of Korpi (2000). This classification reflects the traditional view on the countries' family policy orientation: the UK employs a market model, while Germany provides explicit family support. The male breadwinner model traditionally prevalent in both countries (Lewis and Ostner 1994) therefore results from different policies (Hank et al.2004; Daly 2011).

However, Germany and the UK have progressed towards a dual-earner support policy model in the last decade by promoting an expansion of the public childcare infrastructure (Daly 2011; Jüttner et al.2011). Still, Germany and the UK have undergone neither paradigmatic changes nor a convergence in family policies, as indicated most significantly by the stable configuration of the transfer systems (Warth 2011). Relevant measures in this regard are (1) the separate taxation system in the UK (which favours dual-earner couples, as incomes of a second earner are taxed at a low level), (2) dependent insurance regulations in Germany, where married partners of employed persons are automatically covered by their health insurance and (3) fixed child allowances and comparatively generous parental leave arrangements in Germany (Daly 2011). Thus, country-specific disparities in family policies have remained, and still reveal a core difference in the institutional settings of Germany and the UK.

Several newer OECD-measures on the degree of defamilisation support this assessment: the average effective tax rate of a second full-time earner on the household level in 2008 is at 50%in Germany and only at 40% in the UK, if the spouse is employed full time (OECD 2012). Additionally, public expenditure on childcare and also formal care enrolment is significantly greater in the UK than in Germany (OECD 2012), indicating that family and work are more reconcilable in the UK. However, recent comparative studies suggest that childcare provision in the UK is less stable, less reliable and stronger based on user fees than in Germany (Evers et al.2005; Zagel 2013). Therefore, we handle the aspect of childcare provision with caution when developing hypotheses.

To summarise the above findings on institutional disparities: major differences persist between Germany and the UK in the openness of employment relationships, as reflected in indicators on ELP and bargaining centralisation. In the institutional aspects of decommodification and regulation, the classic regime boundaries are blurred. Significant differences in family policy still exist: with some exceptions, Germany still displays many of the attributes of a conservative welfare state, while in the UK, market-oriented family policies are dominant. In a next step, the regime clusters identified at the macro level are employed to develop hypotheses on differences in country-specific determinants of working poverty at the individual level.

The previous section has portrayed both Germany and the UK as countries in which modest degrees of decommodification implicitly define relatively weak criteria for acceptable working conditions. Therefore, both countries offer considerable incentives for active labour market participation, even on the basis of low paid or atypical employment. Additionally, both regimes also directly support these employment types by providing a variety of options to employers in arranging employment relationships besides regular contracts. Since atypical jobs are highly associated with poverty (Goerne 2011; van Lancker 2011), in-work poverty risks consequently are produced in both countries.

This section deals with the question of how these in-work poverty risks are channelled to certain socio-economic groups. We expect institutional differences in general employment protection (EPL) and bargaining centralisation to play a major role in this matter. Strong EPL and centralised bargaining are established as key features of regimes with closed employment relationships (Regini 2000). Accordingly, Germany is often classified as having a labour market with closed employment relationships, while the UK – with weak EPL and a decentralised bargaining structure – is regarded as a prototype of a context with open employment relationships (Mills and Blossfeld 2003). According to Sörensen (1983), regimes with closed employment relationships tend to associate insiders' position on the labour market with a barrier and extract them from the competitive market. Hence, Germany's centralised bargaining structure and strong EPL are predicted to protect specifically the core-workers from the forces of commodification and deregulation (Mills and Blossfeld 2003).

While the impact of EPL in this matter is evident, the implications of centralised bargaining require a more detailed explanation. In economic theory, centralised bargaining is considered a key factor in determining the influence (or, rather, success) of unions (Blau and Kahn 1996). Unions, at the same time, have established as insiders' interest organisations (Lindbeck and Snower 1986). Accordingly, the degree of centralisation largely defines the ability of unions to protect insiders from the impact of welfare-state retrenchment and labour market deregulation. A similar perspective regards a high aggregation level of bargaining as an effective means to coordinate the collective interests of core-workers: according to Weber (1956), the extent to which an economic or social group accumulates protection from competitive market forces depends on the groups' ability to organise its collective interests. In this light, a collective (centralised) representation of core-workers' interests accentuates disparities in the distribution of protection between core-workers and persons in (transitory) positions at the periphery of the labour market.

Therefore, in line with the aforementioned references discussing Germany as a prototype of a regime with closed employment relationships, we conclude that insiders' positions on the labour market in Germany appear to be shielded from competitive market forces. Consequently, the impacts of commodification and deregulation are shifted to the outsiders on the periphery of the German labour market. In the UK – due to the country's highly decentralised bargaining system and marginal employment protection – institutionalised protection for core-workers is absent. Therefore, entrants to the labour market from the educational system (H1) and re-entrants from unemployment (H2) are exposed to a higher relative poverty risk in Germany than in the UK.

A further consequence of centralised and highly coordinated bargaining is the comparatively low degree of variation in wages, which is attributable to trade union bargaining aimed at compressing the distribution of labour incomes (Blau and Kahn 1996). Therefore, skills (which largely determine the earning potential on the labour market, see Becker 1993) shall have a greater impact on individual economic risks in a decentralised bargaining context as the UK. Additionally, human capital resources can be assumed to be crucial in the UK as employment relationships are open and market mechanisms are not restricted by EPL (DiPrete et al.1997). Thus, employees with a low level of education are assumed to face a greater relative in-work poverty risk in the UK than in Germany (H3).

In addition to wage compression, wage distributions bearing the mark of centralised bargaining are also characterised by strong references to age and seniority (Oswald 1985). Thus, the higher the degree of centralisation, the more senior positions will be protected. Thus, older employees in Germany are shielded from market forces – not only through specific employment protections but also through above-average wages, as stipulated in collective agreements. As a result, older employees in the UK are expected to face a greater relative in-work poverty risk than their German counterparts (H4).

Finally, the overview of institutional measures in the previous section has revealed a stronger family policy focus on the male breadwinner model in Germany: Germany provides explicit incentives for female partners not to work full time in the labour market. By contrast, many policies in the UK explicitly encourage labour market participation of both partners. As long as partners form an economic unit, this has no impact on country-specific gender disparities in (working) poverty. However, if new households are formed after events like separation or divorce, economic dependence becomes relevant with regard to the poverty risk (Vandecasteele 2009): in Germany, women's week labour market attachment during marriage will rarely allow for her to secure the previously achieved economic standard after a separation. Consequently, the gender specific difference in the working poverty risk is assumed to be greater in Germany than in the UK (H5).

The current body of literature does provide tentative evidence confirming several of the hypotheses formulated above. Lohmann and Marx (2008) demonstrate – in a descriptive cross-country comparison based on data from 2001 – that workers under the age of 30 and employed women in Germany are disproportionately affected by poverty compared to those in other countries. In the UK, by contrast, the workers who appear to be most affected by in-work poverty are those older than 50. This is confirmed by a multivariate analysis of Goerne (2011), showing that the relative risk of older workers in the UK is higher compared with other European countries. So far, however, there has been no systematic test of the differences in determinants of in-work poverty risks between Germany and the UK. This study aims to fill this gap by examining the hypotheses regarding cross-country differences within a multivariate framework. The following section outlines the methodological approach in detail.

Our empirical analyses are based on data from the SOEP (Wagner et al.2007) and the BHPS (Taylor et al.2010). In order to maximise comparability, we use waves 2003–2006 for Germany and waves 2002–2005 for the UK. This restriction was necessary due to data limitations: Only these years provide similarly measured variables on job status and comparable job history data for both countries. Cross-sectional weights are applied in all of our analyses to correct the under- or over-representation of socio-demographic groups. The population is comprised of the labour force in private households in Germany and the UK. Individuals are considered to be employed when they are between the ages of 17 and 64 and work a minimum of one hour per week (for robustness-checks with stricter definitions, see note ² ). However, civil service and military personnel, as well as students, apprentices, retired persons, and registered unemployed persons are excluded. The sample contains 74,156 observations, of which 58,397 (34,791 from Germany and 23,606 from the UK) are available for the multivariate analysis. These observations are distributed among 20,669 respondents (12,320 in Germany and 8349 in the UK) who were interviewed multiple times.

Poverty is measured on the basis of the recent monthly household income.³ We use the modified OECD equivalence scale to regard differences in households' sizes and demand. Persons are identified as poor when they live in a household whose equivalence-scaled income amounts to less than 60% of the country- and year-specific median (OECD 2004).

According to the hypotheses postulated above, transitions into employment constitute the most important independent factors in the present analysis. Re-entrants to the labour market are currently in a stage of their employment history that was preceded by unemployment. These persons are considered as re-entrants for up to two years after their transition back into the labour market. Individuals are defined as entrants to the labour market if they started working in their first job within the three years prior to the date of the survey. Periods of vocational training are, at least in Germany, not regarded as part of the entrance phase.

Occupational history is measured as calendar data in the SOEP and in spell format in the BHPS. For the purposes of data processing and distribution, however, the relevant information is extracted from the original data-sets and provided to data users in the form of event-history data (on a monthly basis). Entrance and re-entrance, as defined above, are operationalised on the basis of these event-history data-sets. Education is measured on the basis of the internationally comparable, ordinally scaled Casmin classification. Casmin levels are combined into three categories: ‘high’ (Casmin 6–9), ‘mean’ (Casmin 4–5) and ‘low’ (Casmin 1–3). Persons over the age of 55 are classified as older workers.

While this study focuses on individual-level variables, we acknowledge that correlated household characteristics are important determinants of in-work poverty, too (Gleicher and Stevans 2005; Brady et al.2010). Thus, two dichotomous household-level variables are integrated in the multivariate analysis as controls: the first indicator measures whether children (under age 16) are living in the household, while the second differentiates between single-earner and multiple-earner households. Additionally, the horizontal (sectors) and vertical segregation (job status) of the labour market is controlled for at the individual level.

Atypical types of employment are identified as relevant determinants of in-work poverty (van Lancker 2011) and constitute an important channel of deregulation, as discussed in the theoretical part. Therefore, we integrate measures on type of employment (standard, fixed-term, agency work, part time and self-employment) in an additional model.

Country-specific poverty rates for groups of employed people are outlined in the descriptive analysis. A formal, statistical validation of the hypotheses and a control of the effects of possible confounders and mediators are undertaken within a logistic regression framework. The differences between Germany and the UK in the poverty risks of specific groups are specified through the integration of interaction variables. A robust variance estimation for cluster-correlated data is used to account for statistical problem associated with repeated measurements on the individual level (Wooldridge 2002). Several robustness-checks with alternative specifications and estimations are performed.⁴ The substantive size of coefficients is illustrated by plots on predicted risk ratios (RR).⁵

Table 2 shows group-specific working poverty rates in Germany and the UK. In addition, group-related deviations from country-specific overall rates are displayed (for an overview on confidence intervals, see Table A1 in the appendix, ⁰⁰⁰⁷).

Particularly noteworthy is the finding that, in the UK, the share of working poor in entry-level positions is only marginally larger than the poverty rate among all employed, while in Germany the poverty risk among entry-level workers is more than one-third above the average rate. Additionally, in Germany the poverty rate among re-entrants is more than two times the rate for all employed persons while in the UK the risk levels for both groups differ by about 70%.

Table 2 also provides evidence on the disproportionality in the country-specific poverty risks of older workers. In the UK, the poverty rate for this group exceeds the country-specific overall rate by about 50%, whereas in Germany, the risk of poverty among older workers is approximately 2% lower than that of all employed persons. In addition, Table 2 shows that disparities based on differences between educational degrees are more pronounced in Germany than in the UK: In Germany, the in-work poverty risk of a person with a higher education degree is almost 70% below the average, whereas in the UK, the gap in the in-work poverty rate between all workers and those with higher education degrees is only 33%. Furthermore, the relative difference in poverty risk between employed men and women in Germany (58%) is substantially larger than in the UK (8%).

Table 3 presents the results of a series of logistic regression analyses. Model 1 reports the country-specific in-work poverty risks associated with this study's key characteristics. Reported are coefficients on the logged odds of having a household income beneath the poverty threshold. Males under age 55 with a lower education degree who have not recently (re-)entered the labour market constitute the model's reference category. These individuals face a significantly higher risk of poverty in the UK than in Germany.

In line with H1 and H2, the positive signs of both country-specific interaction terms in Section B point to a significantly larger poverty gap between (re-)entrants and other employees in Germany than in the UK. Country-specific differences in the effects of socio-demographic predictors (Section C) are also confirmed by the regression analysis: the coefficient referring to older workers in the UK carries a positive sign, whereas the coefficient of the country-interaction variable shows a significant negative effect and overlaps the main coefficient, which is consistent to H4. Women bear a greater in-work poverty risk than men in both countries. However, in the UK, the differences between female and male employees are not significant. In addition, in line with H5, the gender-specific disparities in the risk are significantly greater in Germany.

In line with expectations, we find that having a higher level of education reduces the risk of in-work poverty in both countries (Section D). However, the significant negative country-specific interactions indicate that the effect of education in preventing poverty risks is stronger in Germany than in the UK – which contradicts H3.

Model 2 reveals that all of the effects addressed are robust if household composition and occupational structure are controlled for.⁶ In order to illustrate the substantive size of (net) country-specific interactions, we computed RR with confidence intervals based on coefficients and standard errors of interactions in Model 2. Covariates are held constant at their means. Thus, Figure 1 shows relative, country-specific differences in the poverty-risk of entrants, re-entrants and older employees to that of a core-worker, controlling for household composition, education, gender and occupational status.⁷

Holding covariates constant, an entrant's and a re-entrant's poverty risks are predicted to be about two times as high than a core-worker's risk in Germany. In the UK, the entrant's risk is predicted to be only about 20% higher than the core worker's. For re-entrants, the RR to core-workers is also substantively smaller in the UK (45%). Older employees are predicted to be only slightly more affected by poverty than core-workers in Germany, while older employees in the UK are predicted to have a 65% higher poverty-risk than the reference group of insiders in the UK.

In the theoretical part, we argued that impacts of deregulation are shifted to the periphery of the labour market in Germany and consequently produce disproportional in-work poverty risks for outsiders. As key feature of deregulation, policies promoting atypical jobs have been discussed. Similarly, family policy in the UK has been introduced as more supportive in favour of female standard employment. Consistent with these explanations, country-specific interactions of entrants, re-entrants and women substantially erode and are non-significant, if type of employment is controlled for in Model 3. Obviously, atypical types of employment are one relevant channel mediating higher in-work poverty risks of outsiders and women in Germany.

Our study adds to a body of literature that studies the (partial) convergence of labour and social policies in the UK and Germany (Clasen 2005; Seeleib-Kaiser and Fleckenstein 2007; Mohr 2008; Clasen et al. 2011), as it suggests that this process goes along with a precarisation of entry-level positions in Germany and of older employees in the UK.

More specifically, the empirical analysis reveals that, in Germany, both entrants (H1) and re-entrants (H2) face a relatively greater risk of in-work poverty than those in the UK. Additionally, as underlying mechanism, a stronger concentration of atypical employment at entry-level positions in Germany is identified. These empirical findings are consistent with our explanatory model, which (1) introduces atypical employment as important channel of in-work poverty; (2) discusses atypical employment as a consequence of deregulation and commodification; and (3) finally assumes that a context with closed employment relationships, as Germany, tends to shift the impacts of such policies to the periphery of the labour market. Consequently, this study suggests that higher in-work poverty risks of outsiders in Germany are a result of a remaining institutional core difference between the countries: in Germany, closed employment relationships (produced by a system of generous employment protection paired with centralised bargaining) make entrants and re-entrants specifically vulnerable to processes of deregulation and activation. In the UK, by contrast, similar activating policies in the more open employment system are associated with increased poverty risks for older employees (H4).

We also addressed country differences in gender disparities. As predicted by H5, the analysis reveals that women are less affected by in-work poverty in the UK than in Germany. Additionally, it reveals a stronger concentration of German women in non-standard types of employment as one relevant underlying mechanism. These findings are in line with our explanatory model, which assumes that a weaker male breadwinner notation in the transfer system stronger supports standard employment and economic independency of women in the UK. However, our empirical findings leave room for alternative (or, rather, additional) explanations: as Daly (2011) finds, activating policies directed at single mothers are strongly focused on – often precarious – atypical employment in Germany. This mechanism might lead to a bigger shift from non-work poverty to working poverty of single mothers in Germany than in the UK, resulting in the observed country-specific gender disparities in working poverty. This interpretation is supported by empirical results from Zagel (2013), who finds that single mothers working part time is more common in Germany, while periods of inactivity are more prevalent among single mothers in the UK.

The hypothesis regarding the country-specific effect of education (H3) is not verified in the multivariate analysis: contrary to our expectations, higher levels of education were found to protect individuals from in-work poverty to a greater extent in Germany than in the UK. Country-specific differences in the configuration of educational systems may explain this result: the educational system in the UK is characterised by a low degree of stratification and standardisation (Müller and Gangl 2003). Thus, the signalling role of educational certificates in UK may be weaker than in Germany.

A final issue is the extent to which the study's findings can be generalised. Hypotheses developed based on our theoretical assumptions were mainly confirmed, yet the question of the general (or rather, external) validity of the theory remains open. A multi-country comparison would provide evidence of the macro-/micro-interactions investigated here, and could be a means of explicitly verifying the underlying theoretical assumptions of this study.

This manuscript benefited from the elaborated review of two anonymous reviewers and the editor. The author is also grateful for the constructive feedback received from Hans-Jürgen Andreß, Daniel Clegg, Karsten Hank, Henning Lohmann, and Jake H. Rosenfeld.

Supplemental data for this article can be accessed here.

Marco Giesselmann is a research associate at the Socio-Economic Panel Study (SOEP) at the DIW Berlin. He is also assistant professor for Social Science Research Methods at the University of Bielefeld. His research is devoted primarily to the impact of labour market reform on individual living conditions. His publications include works on poverty and low-wage work, on the economic consequences of critical life events, and on the methodology of longitudinal data analysis. He has published on these issues in the European Sociological Review, European Societies, Zeitschrift für Soziologie and Kölner Zeitschrift für Soziologie and Sozialpsychologie.