Hidden in Plain Sight? Does Stricter Employment Protection Legislation Lead to an Increased Risk of Hidden Unemployment in Later Life?

Abstract

This article examines the influence of national employment protection legislation (EPL) on the likelihood of older workers in Europe being forced into retirement. Data are drawn from 4 waves of the Survey of Health, Ageing and Retirement in Europe (SHARE) covering the period from 2004 to 2013. The sample is restricted to those who were aged between 50 and 80 and exited from paid work during the study period (N = 3,446). EPL was measured using the OECD indicators of employment protection concerning regulations for individual dismissals. Exits from work were defined as forced or unforced based on the respondent’s description of the reason for leaving work. Our cross-national study shows considerable variety in the prevalence of forced career exit across 13 European countries. Furthermore, the results show that career exit through retirement is less likely to have been forced as compared to career exit through non-retirement routes. However, the results also show that with every unit increase in the EPL index, the probability of forced career exit through retirement becomes more likely. Apparently in countries with high levels of employment protection, retirement is a more attractive route to lay off older workers than in countries with low EPL. By forcing older adults to leave their jobs through retirement, these employers are shedding workers who would have preferred to continue their working lives.

Research on retirement is burgeoning, driven by population ageing and increased government interest in extending working lives (OECD, 2015). Yet, despite concerted efforts by governments throughout the world to encourage people to remain in work until later in life, there are still concerns that many older workers are being forced out of work (Brand, 2015; Business in the Community, 2015; OECD, 2013a). Forced exits from work are defined as those that result from external structural forces, such a plant closure, redundancy, poor health or the end of a temporary contract. To date, however, most of the existing studies on forced career exit concentrate on the individual risk factors of being pushed out of the labor force and often focus on the subjective experience of career exit. However, more recently research emphasized the importance of looking at macro-social factors that force older adults to leave their careers (Ebbinghaus & Radl, 2015). In line with this argument, we explore whether the route through which older people are forced out of work, that is, via retirement or non-retirement routes, changes depending on strictness of employment protection legislation (EPL) in different European countries. In so doing our paper seeks to make 4 contributions to the literature; (a) to use a cross-national comparative research design to examine the prevalence and predictors of forced career exit across countries in Europe, (b) to add to the growing literature on the role of macro-social factors in late life career exit, (c) to broaden our focus on forced exit from work in later life by moving beyond looking at individual routes and considering forced exit across all routes, (d) to add to the debates on the relative roles of structure and agency in late life career exit. Our research question is: Does the strictness of EPL affect the risk being forced out of work through either retirement or non-retirement routes in Europe?

This is an important question because forcing older workers out of work can have a detrimental impact on the economy, through the loss taxes and increased social security payouts, on the organization, through the loss of human capital, and on the older worker, through loss of income and impacts on well-being. Moreover, involuntary career exit in later life could undermine efforts by European governments to extend working lives. The UK government estimates that increasing the average age of labor market exit by 1 year could raise real GDP by around 1%, via tax revenues and reduced pension spending (Department for Work and Pensions, 2017). Analyses also shows that companies that shed older workers could lose significant competitive advantage in terms of recruiting and retaining productive workers. In particular those companies that push older workers out of work risk losing the experience and knowledge of workers gained from many years of employment, increased levels of poor morale and premature exit among staff who feel they have no options for phasing their retirement and being liable to age discrimination claims and associated costs (Department for Work and Pensions, 2013a, 2013b, 2015). Leaving work can also have serious financial consequences for individuals resulting from a loss in income, employment-related fringe benefits and, potentially, accumulated job-specific skills and experience (Brand, 2015). Thus, while it is generally accepted that older workers are less likely to lose their jobs (Bernard & Galarneau, 2010), when they are laid off, they often take longer to find another job and, on average, suffer substantial, persistent loss of earnings (Chan & Huff Stevens, 2001). Being forced out of work also carries health risks. Involuntary job loss in later life has been shown to be associated with an increased risk of poorer general health (Gallo, Bradley, Siegel, & Kasl, 2000) and poor psychological health (Brand, Levy, & Gallo, 2008; Hyde, Hanson, Chungkham, Leineweber, & Westerlund, 2015). Given the strength of evidence for the negative effects of being forced out of work in later life across a wide range of factors it is crucial that we gain a better understanding of the determinants of forced career exit.

To date, however, the majority of research on the determinants of forced career exit comes from single country studies, predominantly the United States, the UK, and the Scandinavian countries (van der Heide, van Rijn, Robroek, Burdorf, & Proper, 2013). While such single country studies are useful for developing a more in-depth understanding of the country-specific context through which older workers are forced out of work, the results in these country-specific contexts are hard to generalize to other contexts because policies on work and retirement vary so widely across countries. In addition, they are limited in what they can tell us about the relative impact of different policy regimes on forced exit. From both a research and a policy perspective it is crucial to understand how macro-social and institutional forces may influence forced career exit transitions to better assess the consequences of certain policies and policy reform. To redress this, we employ a cross-national comparative research design using data from a harmonized study, in which the question formats and data collection protocols are the same in each country, to explore and better understand differences in the prevalence and predictors of forced career exit across countries in Europe.

A large share of the existing literature on career exit of older adults focuses on individual level predictors while broader macrosocial factors have been largely ignored. An exception is the previous research by Ebbinghaus and Radl (2015) which showed that a higher statutory pension age and stricter hiring and firing regulation increased the likelihood that retirement was forced. However, while this study has been crucial in highlighting the need to include macrosocial factors it was based on cross-sectional data and only looked at whether retirement was forced or not. Our study advances their work by using longitudinal data, allowing us to directly control for pre-exit factors, and by looking at the extent to which the strictness of EPL affects whether older workers are forced out of work via retirement or non-retirement routes. As such, by looking at the forced/unforced nature of exit across all routes of labor market exit, we seek to address a somewhat different question, whether the way in which older people are forced out of work changes depending on strictness of EPL in different countries.

We also seek to make a theoretical contribution to this field of research by drawing on the wider sociological debates about the relationships between structure and agency. We define structure, or structural forces, as the recurrent patterned arrangements which influence or constrain individual choices or behavior, while, conversely agency is conceptualized as the capacity of individuals to act independently and to make their own free choices (Barker, 2005, p. 448). From this perspective, we distinguish two structural levels. First, we conceive of forced exits from work as an indicator of the effects of external structural forces, such a plant closure, while non-forced exits from work are seen as an exercise of individual agency. Second, by looking at the impact that EPL has on the likelihood of being forced out of work via certain routes we introduce another “structural” level into our analyses that operates at the macro-social policy level. By looking at the combination of these factors we aim to see how structure(s) and agency interact and have an impact on individuals.

Furthermore, in this study we aim to go beyond the traditional focus on retirement by taking a more holistic view on career exit in later life. We take a broad view to look at the risk of forced exit regardless of the route of exit (i.e., non-retirement or retirement route). The increased complexity of the timing and reasons for career exit in later life calls for this more comprehensive approach which focuses on the degree of choice in the transition from employment rather than on the various routes out of work. Moreover, by separating out the degree of choice or compulsion in the exit decision from the route of exit this then allows us to (a) look at how the risk of forced career exit varies across route and (b) explore the ways in which EPL impacts on this relationship. Before we explain our theoretical approach and develop our hypothesis, we start with explaining this conceptual contribution in more detail in the next section.

CONCEPTUALIZING FORCED VERSUS UNFORCED LATE CAREER EXIT

Traditionally research on the causes and consequences of employment exit in later life has tended to focus exclusively on one particular route, for example, retirement, disability pension or unemployment. Even where the distinction has been made between forced and unforced exit the focus has still been within, rather than across, specific routes and has predominantly focused on the retirement route, for example, un/forced retirement (Dorn & Sousa-Poza, 2010; Szinovacz & Davey, 2005; Van Solinge & Henkens, 2007). However, in this study we argue that if we are to capture the complexity of the labor market transitions in later life we need to go beyond this traditional approach and beyond the focus on retirement. Instead, as is shown in Figure 1, in our analyses we look at forced versus unforced exit regardless of route. In so doing we follow researchers such as Hyde and colleagues (2015) and Mandemakers and Monden (2013) who have argued that the distinction between retirement and other forms of employment exit in later life is becoming increasingly blurred.

In the past retirement was rightly seen as a special form of employment exit as it coincided with labor market exit and the uptake of pension income. Indeed, in many countries receipt of a pension income used to be predicated on withdrawal from the labor market. In this sense retirement, unlike unemployment, was seen as nonreversible. In technical terms it was an absorbing state. However, this is no longer the case and increasing numbers of people are returning to work following retirement either through encore careers (Quinn, 2010), un-retirement (Larsen & Pedersen, 2013) or bridge jobs (Dingemans & Henkens, 2014; Dingemans, Henkens, & Solinge, 2015). So, if retirement is no longer necessarily associated with detachment from the labor market then, in terms of its labor market position, it is no longer distinct from other nonwork states such as unemployment. Conversely, if retirement no longer represents the cessation of working life, unemployment in later life can often lead to permanent exit from the labor force (Tatsiramos, 2010). This is because older workers have lower rates of reemployment than younger workers (Chan & Huff Stevens, 2001) which can lead to them becoming discouraged and “elegantly withdrawing” from the labor market (Fournier, Zimmermann, & Gauthier, 2011). Additionally, many people are forced into early retirement through poor health and disability.

Furthermore, in light of reforms to retirement and pension policies, retirement as such may no longer be the most prevalent barrier to continuing work careers. While previous early retirement arrangements were important instruments for employers to lay off their older workforces, the closing down of such arrangements may result in the increased prevalence of unemployment in later life. The same may be true for older adults with ill health who have seen their opportunities to retire early disappear. Indeed, data from Finland show that changes in the eligibility age thresholds for unemployment and part-time pension schemes and the tightening of the medical criteria for disability pension eligibility have jointly raised the average age at which older workers leave employment. Crucially the study found that this increase was mainly due to a sharp drop in disability pension enrollment (Kyyrä, 2015).

In line with these arguments, and evidence from elsewhere (Calvo, Sarkisian, & Tamborini, 2013; Henkens, van Solinge, & Gallo, 2008), we argue that it is not the form of job loss per se but whether this was forced or unforced. Thus, to broaden the concept of forced exit, it is important to include all forms of forced career exits. For example, unemployment can be unforced if someone chooses to leave a job to look for another job or takes time out from work. On the other hand retirement is sometimes forced if it is the result of a plant closure or redundancy (Dorn & Sousa-Poza, 2010). This more holistic view is needed in order to better understand the totality of the ways in which older workers are being forced out of work. Thus, following these arguments, this study focuses on all forms of forced career exit in later life, regardless of the specific pathway out of employment, to get a full picture of the total amount of forced exit.

EPL AND FORCED CAREER EXIT

Previous research has investigated the impact of macro-social factors on labor market exit in later life. For instance, some studies, conducted during the earlier period of falling rates of labor market participation, demonstrated that generous social security and pension systems provided strong incentives for people to choose to retire early (Berkel & Börsch-Supan, 2004; Börsch-Supan, 2000; Gruber & Wise, 2009; Guillemard, 2003). Currently there appears to be growing evidence that the removal of these early exit pathways by most governments has resulted in an increase in labor force participation rates in later life (Euwals, Van Vuren, & van Vuuren, 2011; Kyyrä, 2015). These studies demonstrate that such policies play a key role in the employment position of older workers.

However, despite the evidence that shows how important they are, pension systems are still only one part of the policy mix that can impact on labor force participation in later life. Indeed, one could argue that by focusing largely on the effects of pensions and retirement reform in extending working lives there is a risk that we overlook other forms of labor market policies that prevent older workers from being forced out of work. To redress this, we look at the role that EPL, defined as the rules that govern the hiring and firing of workers, has on the likelihood of forced career exit across 13 different European countries.

The OECD (2006) and the European Commission (2006) have identified EPL as an important determinant of the employment situation of older workers. However, this relationship is not as straightforward as it might first appear. On the one hand EPL is generally seen as a social good as it protects workers from arbitrary dismissal or discriminatory practices by their employers and ensures that firms also absorb some of the social costs of labor turnover (OECD, 2013b). This could be particularly advantageous for older workers because, as Bennett and Möhring (2015) note, although EPL should be “age blind,” in many countries it is implicitly biased towards older workers as the level of legal protection depends on tenure. As older workers have had greater opportunities to establish longer tenure and, in most cases, higher pay, this would incur higher severance payments if they were made redundant. This would therefore make it more expensive for organizations to dismiss these workers (Chéron, Hairault, & Langot, 2011). From this perspective, stricter EPL has a positive impact on older workers by increasing job security and reducing the likelihood of involuntary labor market exit. This is supported by evidence that countries with lower EPL have not only higher dismissal rates but also greater rates of voluntary career exits (Gielen & Tatsiramos, 2012).

However, EPL can also have some negative or contradictory effects (Engelhardt, 2012; Skedinger, 2010). Bennett (2016) notes that may remain hidden by only looking at the direct effect of EPL on forced career exit. A few studies have looked at the impact of EPL on career exit in later life. However, the results are ambiguous. Bennett and Möhring (2015) found that stricter EPL was associated with a lower likelihood of early retirement for men who had been in regular employment for most of their lives. Conversely, Ebbinghaus and Radl (2015) found that stricter regulation was associated with an increased likelihood of being forced out of work. What these seemingly contradictory findings suggest is that EPL can impact differently on the likelihood of being forced out of work depending on the route by which a person leaves work. As Saint-Paul (2009) notes a high EPL increases the employability of older adults in the years running up to retirement age, as high firing costs may lead employers to decide to not fire older workers with decreasing productivity but wait until they are eligible for retirement. This may result in a lower critical age at which employers prefer retirement over unemployment when EPL is high. This then leads us to our central hypothesis: we expect that stricter EPL leads to a greater likelihood that those who are forced out of work in later life will leave work through retirement rather than through non-retirement routes. If this hypothesis is confirmed it will show that stricter EPL does not unambiguously protect the employment of older workers, as it may create a greater risk of hidden unemployment in later life.

METHOD

In the current article, we use data from the Survey of Health, Aging and Retirement in Europe (SHARE). This is a cross-national and longitudinal research project that has collected data on adults aged 50 years and older, in several European countries since 2004 (Börsch-Supan et al., 2013). A clear advantage of SHARE’s specific focus on older adults is that it offers large enough samples per country to investigate our research questions. After the first data collection in 2004 (Wave 1), data has been collected in 2006 (Wave 2), 2008 (Wave 3), 2011 (Wave 4), and 2013 (Wave 5). Respondents from Wave 1 were followed in the next waves as far as possible and refresher samples were added to the country samples in later waves. At Wave 1 the average household response rate for all participating countries was 61.6%. However, this ranged from 81.0% in France to 38.8% in Switzerland (SHARE-ERIC, 2016).

Our analytical sample is comprised of adults aged between 50 and 80 as we are interested in career exits. Data from the third wave of the SHARE project is excluded because of the different format of this wave focusing on the specific life histories of respondents. Only respondents from countries which had contributed data from at least two waves (not including Wave 3) were included. Hence our final sample included respondents from the following 13 countries: Austria, Germany, Sweden, Netherlands, Spain, Italy, France, Denmark, Switzerland, Belgium, Czech Republic, Poland, and Estonia. We made use of the panel character of the data and selected respondents in two steps. In the first step, we selected individuals who reported being employed at their first wave of participation in the survey. For some this was at the start of data collection in 2004, for others this was at a later wave as they were part of one of the refresher samples. Within this group, a further selection was made by including only the respondents who moved from being employed to being retired, unemployed or disabled in the subsequent waves in the panel. As a result, the data are structured as a kind of staggered cohort design, including respondents from whom data were taken in Wave 1–Wave 2, Wave 2–Wave 4, or Wave 4–Wave 5. In total, about one in five of the cases had to be excluded due to missing values on the dependent variable. Among the unemployed and some of the retirees, forced or unforced career exit could not be classified, because respondents reported “other reasons” for their career exit. We will address these issues in the discussion section later. The final sample size is 3,446. In total, 47% of the sample is women and the mean age is 62.

Measures

Dependent variable

The dependent variable in the study is whether career exit was forced (1) or unforced (0). The construction of this variable is illustrated in Table 1. It is derived from questions on: (a) labor market position, (b) reasons becoming retired, and (c) reasons for becoming unemployed. For those who retired or became unemployed whether they were forced to leave their career job or not was determined by asking them reasons why they left their last job. As can be seen by Table 1, although these reasons were specific to each route there was still considerable overlap between them. Hence respondents who reported being retired were asked the reasons they retired. Similarly, unemployed respondents were asked how they became unemployed. Following Radl (2013) and Ebbinghaus and Radl (2015) responses were categorized as forced, for example, made redundant or own ill health, or unforced, for example, because they resigned or by mutual agreement. In the case of retirement, respondents were allowed to code all categories that apply to them, instead of giving the main reasons like in the case of unemployment. Around 10% of retirees gave multiple responses. In cases where a forced and a non-forced reason was given we privileged the forced reason. While we are aware that leaving work is a potentially complex phenomenon with multiple motivations we had to make a decision when faced with respondents who gave both negative and positive reasons for leaving. Rather than eliminating them from the analyses we decided to privilege the negative reason. It was felt that if a respondent gave both positive and negative reasons for leaving work then it is possible that the positive reason might be a post-hoc rationalization designed to mask the fact that they had been forced out of work. Finally, those who reported leaving work due to disability were categorized as experiencing a forced career exit.

Independent variables

The main independent variable at the individual level is whether career exit happened through a retirement route or a non-retirement route (unemployment or disability). The classification of retirement and non-retirement routes was based on the respondent’s self-reported labor market status, for example, retired, unemployed or sick/disabled, in the wave after which they had left work.

At the country level, our main variable is the strictness of EPL. We use the measure of EPL from the OECD that is concerned with the individual dismissal rules for regular employment contracts. This index includes eight indicators of procedural inconveniences, notice periods and severance pay, and the difficulty of dismissal. The indicators are compiled using the OECD Secretariat’s own reading of statutory laws, collective bargaining agreements and case law as well as contributions from officials from OECD member countries and advice from country experts. The data refer to the year 2009 for which the EPL-scores have been revised by the OECD in 2013 (OECD, 2013A higher score on the EPL measure, which ranges from 0 to 6, refers to stricter EPL. In our data, EPL ranges from 1.60 in Switzerland to 3.05 in Czech Republic.). While it is possible to take the value of EPL at T2 for the respondents (with the exception of Estonia for whom the data are only available from 2007 onwards) the relatively small sample sizes per country per year limited us from taking such an approach. However, the values for EPL are very stable for each country over the study period (see  Appendix A).

Controls

In the analyses, we control for individual level factors that are known as important indicators for forced career exit (Van Solinge & Henkens, 2007). All these measures are based on data from the wave before employment exit. First, we account for gender by including a dummy variable in which women are coded as 1. Second, age is taken into account by including the centered measure. Third, educational level is included as a measure of socioeconomic status (SES). The SHARE team constructed ISCED classifications to get a comparable measure for education across the different educational systems in the countries (for detailed information, see www.share-project.org). Fourth, we took into account characteristics of the career job, which is here defined as the job prior to the forced or unforced exit from the labor force. We controlled for the job being full-time or part-time and for job satisfaction. Part-time employment was operationalized as working less than 32 hrs a week (based on total worked hours, not contracted hours). Job satisfaction was measured by asking: “All things considered I am satisfied with my job. Would you say you strongly agree, agree, disagree or strongly disagree.” We constructed a dummy variable in which respondents who (strongly) agreed were coded as 1. We ran alternative analyses with job satisfaction included as a continuous variable to see if this had any effect on the overall results (analyses not shown). But as it did not we retained the dummy measure for simplicity of interpretation. Finally, we included dummies for the waves of measurement as control for time-effects.

Analytical Framework

There are several analytical approaches for using cross-national datasets like SHARE. In the comparative social science literature, the use of multilevel models is increasingly popular. In such multilevel models, also referred to as mixed models, fixed regression parameters are estimated, while the variance that is left unexplained is captured in individual-level and country-level random effects (Bryan & Jenkins, 2013). An advantage of this way of modeling the hierarchical structure in the data is that the direct impact of country-level predictors on the individual-level dependent variable can be tested. A drawback of this approach, however, is that the models have only a small number of degrees of freedom if the number of macro-level units is small (N < 30), potentially biasing the results (Bryan & Jenkins, 2013).

A valuable alternative in this case is the application of a country fixed effects approach (Möhring, 2012). In country fixed effects models country-level heterogeneity is controlled for by including N_country – 1 dummy-variables for the countries. A result of this approach is that it is not possible to estimate the direct effects of country-level factors. However, it is possible to test moderator effects of country-level predictors on individual-level relationships (Bennett & Möhring, 2015; Möhring, 2012); the so-called cross-level interaction effects. In the current paper, we are particularly interested in how the strictness of EPL (country level) channels older adults in forced career exit via non-retirement or retirement routes (individual level). Therefore, the country fixed effects approach is well suited to test whether such a cross-level interaction effect exists.

Recently the use of logistic regression models has come under serious criticism as a method of estimation. Mood (2010), recommends the use of a linear probability model instead. However, we have decided to use a logistic approach as linear probability models can be problematic in cases where the values of the dependent variable are close to either 0 or 1. This is the case for Sweden and Switzerland in our sample and, therefore, a logit model is preferred. Nevertheless, we conducted additional analyses using linear probability models to see if this had an impact on our findings. These are discussed in the subsection on sensitivity analyses.

RESULTS

Descriptive Analyses

The sample is described in Table 2. In total, one in three older workers in the panel experienced forced career exit during the study period. However, there is substantial variation in the percentage of workers who report that they were forced to leave work across countries. Figure 2 shows that in a number of countries, such as Estonia, Spain, and Poland, the rates of forced career exit are relatively high. Almost one in two older workers who leave work in Spain and Estonia report that they were forced to do so. Conversely the percentage is much lower in countries like as Sweden, Switzerland, and Austria where around one in five older workers is forced to leave their career jobs. Additional analysis by do not show any consistent gendered pattern the rates of forced career exit across these countries as a whole ( Appendix B)

Table 2 also shows that in about three-quarters of cases people left work via retirement and one-quarter left through non-retirement routes (16% via unemployment and 10% via disability). It is not surprising that retirement was the dominant way out, because the age-range on which the sample was selected includes the traditional retirement ages across the European countries under study. However, as is shown in Figure 3, the share of forced exit via retirement and non-retirement routes differs across the countries. This scatterplot shows that there is no clear relationship between forced career exit through non-retirement or through retirement. Estonia (EE) is the extreme case with high percentages of both forced career exit through pre-retirement and retirement routes. Poland (PL), in contrast, has high levels of forced career exit through non-retirement routes, but forced retirement is scarce. In countries such as Austria (AU), Germany (DE), and the Netherlands (NL) is forced exit through non-retirement routes is relatively low compared to the other countries, but especially in Austria, forced retirement is relatively common. Regarding the control variables, Table 2 reports that about one in four respondents had a high level of education as compared to about one in three who followed the lowest educational levels. A large majority of respondents was (very) satisfied with their career job and one in four worked in part-time jobs before career exit. Finally, Table 2 shows that the majority of respondents experienced the transition from career employment to nonemployment in between Wave 4 and Wave 5. This may have to do with the relatively large sample sizes in Wave 4 and Wave 5 and the additional countries included in the sample.

Multivariate Analyses

Table 3 shows the results of the country fixed effects logit models predicting forced career exit. In Model A, the main variable retirement status is added together with the control variables. We found a negative coefficient for retirement status in predicting forced career exit. More specifically, this means that career exit through retirement is less likely to have been forced as compared to career exit through non-retirement routes. This is the logical result of the traditional roles of these exit routes. For example older workers who reach retirement age most often voluntarily decide to leave the labor force, while unemployment and disability are much more likely to be forced. Still, our operationalization of the main variables has shown that, on the one hand, unemployment was not always forced while, on the other hand, retirement was not always voluntary (Table 2). However, overall, Model A shows that career exit is less often forced when it goes through retirement as compared to non-retirement routes.

Model A in Table 3 also includes some other important covariates that are also associated with the likelihood of forced career exit. First, educational level is negatively related to forced career exit, which means that those with high levels of education are less likely to be forced to leave their careers as compared to their lower educated counterparts. Second, age is an important predictor of forced career exit. Older age is also associated with a lower likelihood of being forced to leave the career job. Gender and the characteristics of the career job (i.e., job satisfaction and part-time vs. full-time employment) were not found to be statistically significantly associated with the likelihood of forced career exit.

Our central hypothesis in this study is tested in Model B of Table 3. Our expectation was that the relationship between the forced nature of exit and the route of exit differs across countries based on the strictness of EPL in countries. Specifically, we hypothesized that in countries with stronger EPL, the likelihood of being forced to leave career employment via retirement instead of non-retirement is higher as compared to low EPL countries. As the negative coefficient for retirement status shows in Model A of Table 3, the likelihood of forced career exit through retirement is lower than forced career exit through non-retirement routes. We would expect it to stay negative after including the interaction with EPL in Model B because, as we said retirement is by definition much more likely to be voluntary as compared to non-retirement. However, confirmation of our hypothesis would mean that the difference between the likelihood of being forced through retirement or non-retirement decreases in the case of a strong EPL context. The results in Model B point into this direction. Again, and as expected, we find a negative coefficient for retirement status. Furthermore, we find a positive coefficient for the interaction between retirement status and EPL. With every unit increase in the EPL index, the effect of career exit through retirement becomes a little less negative, and thus a little more likely.

To better grasp the findings, this result is illustrated in  Appendix B. We calculated the predicted probabilities of having experienced forced career exit through retirement in different EPL scenarios. The x-axis covers the full range of EPL in our sample (1.6 to 3.05). Keeping all other factors constant,  Appendix B shows that the likelihood of experiencing forced career exit through retirement is about seven to ten percent in low EPL-countries. In the high EPL countries, the likelihood to be forced to leave your career through retirement is much higher, about 20% (Figure 4).

Sensitivity Analyses

We have performed extra sensitivity checks to test the robustness of our results. First, we have replicated Model B in Table 3 using a jackknife approach (Abdi & Williams, 2010), meaning that one country at the time is omitted from the analysis. The results of this procedure are presented in  Appendix C. The overall jackknife coefficient which is calculated from the separate coefficient estimations, is very close to the coefficient we found in our main model (0.862 in Table 3, Model B vs. 0.864 in the jackknife procedure). Second, in response to Mood’s (2010) article on the use of logistic regression, we have replicated Model B in Table 3 using a linear probability approach (results not shown). The results show that the direction and significance of the main coefficients of interest do not change. Finally, we have checked to what extent the differences of time between the waves, that is, the gap in the series created by the fact that the third wave of SHARE did not collect the data we required for the analyses, impacted upon our results. We tested whether the difference in time between waves had an effect in the models by replacing the year dummies for a dummy which measured whether there was a 2- or 4-year gap in between the waves (analyses not shown). However, this did not have a significant effect or impact on the relationship between EPL, retirement and whether labor market exit was forced or not.

DISCUSSION

Driven by population aging and aging workforces, many Western governments are reforming their work and pension systems to encourage older adults to extend their working lives. However, some older adults are forced out of work and unable to extend their participation in the labor force. Since forced career exit has been shown to have negative consequences for employability and health in later life, our aim has been to develop a greater understanding of the structural forces that push people out of work. While many studies have examined individual-level factors that are associated with forced exit from work in later life, in this study we have advanced the literature by taking a macro-social, cross-national comparative approach.

Theoretical and Practical Implications

This cross-national study shows considerable variation in the prevalence of forced career exit across 13 European countries. The results show that in most of the countries in the study, the majority of older adults did not feel that they were forced to leave work. This was particularly so in Sweden and Switzerland where the prevalence of forced career exit was rather low. Still, even in these countries about one in five older adults reported that structural factors had forced them to leave the labor force. By contrast, in countries such as Estonia, Spain, and Poland, around half of the older adults experienced forced career exit. While career exit via non-retirement routes was more likely to be forced, a substantial share of the transitions to retirement were also forced. These findings are in line with previous research that shows that retirement is not always a matter of choice (Ekerdt, 2010; Van Solinge & Henkens, 2007).

In order to understand whether macro-level factors impacted on being forced out of work in later life we examined whether the likelihood of forced career exit through retirement instead of non-retirement is higher when EPL in the country is stricter. The argument underpinning this hypothesis was that laying off older workers prior to retirement is much more expensive where EPL is strict due to higher severance pay-outs. In this situation retirement then becomes an attractive alternative because it is a more legitimized route of exit in later life and does not incur the same level of costs to the employer. The results of our study provide support for this hypothesis and show that in countries with higher levels of EPL, the likelihood of forced career exit through retirement is higher than in countries with lower levels of EPL. This would suggest that in high EPL countries, retirement is a more attractive route to lay off older workers than in low EPL countries. Thus, it is not that EPL necessarily reduces the likelihood of being forced out of work in later life. It could be that it merely alters the way in which older workers are forced out of work. So rather than being forced out via unemployment, stricter EPL, which would result in higher firing costs, means that those older workers who are forced out are more likely to be forced out via retirement. What these findings suggest is that in countries with stricter EPL, policy makers need to ensure that there is also sufficiently strong legislation to prevent employers forcing older workers out through retirement.

We also sought to contribute to the theoretical debates about the changing balance between structure and agency in late life working. Hence, despite a growing policy discourse that older workers now have greater agency in their decision to remain in work or leave work in later life our analyses show that structural factors continue to play a role in the dynamics of labor market participation in later life. The descriptive analyses show that, in line with other research in this area, a significant proportion of older workers across Europe felt that they had been forced out of work. This was particularly high in Spain, Poland, and Estonia. These figures alone should caution us from taking claims, either by policy makers or academic researchers, that older workers are now in control of their decision to work or not in later life. However, by looking at the interaction between macro-social structural factors, that is, EPL, and the route of labor market exit on the likelihood of being forced out of work we were able to extend the theoretical contribution of this paper to look at the role of “structure in structures.” While previous research that has looked at macro-level factors has generally included them as additional independent variables we have sought to look at how macro-social structural factors shape micro-level structural factors. What the results of the interaction between EPL and labor market exit route show is that macro-level structural factors change the way micro-level structural factors operate across different labor market exit routes. Hence these analyses suggest that (a) we ought to include macro-social structural factors in our analyses, but, (b) we cannot assume that they will have an equivalent effect across all routes or groups.

Limitations

This study is not without limitations. First, the measurement of forced and unforced career exit is not completely unambiguous. As Ebbinghaus and Radl (2015) show in their study there are differences between in/voluntary retirement and forced versus unforced exits from work in later life. As we have noted earlier, our decision to focus on forced versus unforced exit from work was informed by our theoretical framework, looking at structure versus agency, and methodological rationale to limit the risk of misreporting the reason for leaving work due to recall bias. However, this does mean that our measure does not necessarily reveal how people perceive their career exit. For instance, redundancy may be perceived as a voluntary decision when people are offered generous redundancy arrangements. From our perspective, redundancy still selects some older workers to leave the labor force who would otherwise have successfully continued their work careers, therefore the use of these more objective measure was particularly suitable for our attempt to investigate structure versus agency. However, it would be good if future research sought to replicate our analyses using a measure of perceived voluntariness of career exit.

Another limitation is the number of missing values on our dependent variable. Partly, the missing values are explained by the fact that some respondents selected the category “other reasons,” which we could not classify as forced or unforced. This was particularly the case for unemployment. We checked to what extent the missing values were evenly distributed across countries. The analyses reveal that the level of missingness was particularly high in Poland and Italy. To examine whether this had any effect on our results we conducted sensitivity analyses, excluding Poland and Italy (analyses not shown). The results did not change substantively, which suggests that the selective distribution of missing values across countries (and across the EPL regimes) does not impact on our results.

The study design also created some issues for our analyses. As noted in the methods section we were unable to use date from the third wave of the SHARE study as this did not collect the information we required (instead it focused on recording lifecourse information). Because of this the spacing between the waves differs in our study. It is plausible that the impact of our independent variables was weaker for respondents who left work between Waves 2 and 4, and that by including them in the analyses we have underestimated the true strength of this effect. However, as noted in the above section on sensitivity analyses, this did not have a significant effect. Also, because SHARE did not collect occupational title after Wave 1 we were unable to use socio-economic class as a possible measure for socio-economic position (SEP). This was unfortunate as previous research has shown that the timing of retirement is highly stratified by class (Radl, 2013). However, as Dubé (2004) notes there are a number of reasons why lower educated older workers might be more likely to take involuntary retirement following a job loss due to economic conditions or poor health. Because of their lower educational attainment such workers are more likely than more highly educated workers to have a prolonged job search period and be much more likely not to find re-employment due to their poorer job prospects. As such they may prefer to “elegantly withdraw” from the labor market via retirement.

CONCLUSIONS

Overall our analyses point to two main conclusions. First, that despite the official discourse that policy reforms are, in part, designed to give us greater choice and agency in our decision to leave the labor market there are still significant numbers of older worker who encounter structural forces, such as plant closure, redundancy, and dismissals, that exclude them from the labor market. Although the experience of such organizational forces may not always be perceived as an involuntary transition out of work by older workers, these events limit the room for individual agency of older workers to freely decide their work status (Ebbinghaus & Radl, 2015). There is a danger that, as both policy makers and researchers turn their attention to the factors that can lead to extended working lives, this group of older workers get overlooked or even blamed for their inability to work for longer. Ensuring that these workers are able to continue in work would have a net positive effect on labor market participation rates in later life. Second, however, we show that although employment policies play an important role in protecting older workers the presence of EPL alone is not sufficient to prevent older workers from being forced out of work. Moreover, simply tightening those regulations will not reduce the chance that some older works will be forced out of work as employers who wish to do so will find alternative ways of achieving this. As previous research has shown, employers are not particularly willing to retain older workers (Mulders, Henkens, & Schippers, 2015; Oude Mulders, Henkens, Liu, Schippers, & Wang, 2016), despite changes in legislation that should make older workers more attractive. By forcing older adults to leave their jobs through retirement, these employers are shedding workers who would have actually preferred to continue their working lives. What this shows is that national policies to protect older workers from being pushed out of work will only work if they are in concert with employers’ policies to retain older workers.

REFERENCES

Appendix A.

Appendix B.

Appendix C.

Author notes