Determinants of (in-)voluntary retirement: A systematic literature review

Introduction

‘Retirement is one of the main life course transitions in late adult life. How retirees experience this transition (as voluntary vs forced) has strong implications for adaptation to retirement and wellbeing in retirement’ (Van Solinge and Henkens, 2007: 300).

An important context for how retirement is perceived and whether it is seen as voluntary or involuntary is the current economic situation of a country. When there is a scarcity of labour older workers are often seen as a pool of skilled and reliable employees. Examples are the United Kingdom during the Second World War and the economic boom in Germany after the Second World War (Phillipson, 1982; Hess et al., 2020). In times of recession older workers are, however, often pushed into early retirement to make space for younger workers and, thus, lower the unemployment rate (Townsend, 1981). After the oil crisis and increased competition from Asia in the 1970s, many European countries – in particular those with conservative welfare states – offered older workers different early retirement options, for example, via disability pensions or unemployment insurance (Ebbinghaus, 2006). This policy resulted in a culture of early retirement in which early retirement was the rule rather than the exception.

In the late 1990s and early 2000s, a policy shift from promoting early retirement to extending working lives took place. The reason for this was that policymakers and scientists had concerns that, because of demographic ageing, the shrinking number of contributors and growing number of beneficiaries in public pay-as-you-go pension systems might jeopardize the welfare states’ financial sustainability (Ebbinghaus, 2006; Guillemard and Rein, 1993). The policy shift was implemented, for example, by restricting early retirement options and introducing measures to improve the employability of older workers. Furthermore, statutory retirement ages, which can be seen as a reference point for what a suitable retirement age is, were raised (Ebbinghaus, 2006). These measures have been implemented in almost all OECD countries. Due to these reforms, the actual retirement age as well as the employment rate of older workers increased (OECD, 2019). Furthermore, the number of pensioners working in addition to receiving a pension is growing as well (Ebbinghaus, 2006), hinting at a destandardization of the life course with regard to a distinction between work and retirement (Kohli, 2007). Further potential reasons for this development are the general good development of the labour market in some countries, rising female labour market participation, as well as demographic trends (Hess et al., 2020).

In addition to the economic situation, shared societal notions on how retirement is defined are important for the voluntariness of retirement transitions – the culture of early retirement being one example. Earlier gerontological disengagement theory states that retirement is a possibility for older people to detach themselves from the labour market and withdraw from society, which is the preferred state for the older people as well as for the society (Achenbaum and Bengtson, 1994; Hochschild, 1975). Townsend (1981) argues that retirement has led to a structured dependency for older people and that ‘[r]etirement is in a real sense a euphemism for unemployment’ (Townsend, 1981: 10), into which older workers are pushed against their will. In contrast, one can see retirement also as a time of leisure and freedom, enjoyed by (most) retirees (Higgs and Gilleard, 2006).

Thus, it becomes clear that retirement and how it is perceived by the retiring person – voluntary or involuntary/forced – is affected by the societal contexts of the labour market and norms, and can have many reasons. These reasons can be economic or occupational but also of an individual nature. Depending on the reasons a person retires, the transition can be operationalized in, at least, voluntary or involuntary retirement (Hofäcker et al., 2016). The degree of voluntariness or choice is determined by external structural forces and individual agency. Structural forces are arrangements on the meso- and macro-level, while agency is seen as an individual and independent act of free choice (Hyde and Dingemans, 2017). However, agency – depending on the field of decision – is determined by social norms and individual circumstances, for example, education (Barker, 2005: 236). Accordingly, factors that allow individuals a high degree of agency are more likely to lead to voluntary retirement. In contrast, structural conditions or structural forces increase the likelihood of involuntary or forced retirement. Reasons that imply a high degree of agency (such as the desire to retire to have more leisure or to retire at the same time as a partner) are considered voluntary. Reasons that imply low or no agency (such as dismissal or ill health) are therefore to be considered involuntary. The conditions that influence the control are not the same for all employees.

Studies from Europe, North America and Japan show different degrees of distribution of involuntary retirement with a range from nearly 10% up to more than 50% (Dorn and Sousa-Poza, 2010; Ebbinghaus and Radl, 2015; Henkens et al., 2008; Hofäcker et al., 2016). These variations are probably due to the operationalization of voluntariness of retirement and depend especially on country and cohort context. Involuntary retirement has a negative effect on health (Bacharach et al., 2008; Rhee et al., 2016; Van der Heide et al., 2013; Van Solinge, 2007) as well as on wellbeing (Dingemans and Henkens, 2014; Hershey and Henkens, 2014; Radó and Boissonneault, 2018). In addition, several studies have shown the importance of retirement’s voluntariness to individual’s adjustment to retirement (Reitzes and Mutran, 2004; Wang et al., 2011).

Thus, it is of high societal relevance to investigate what determinants, drivers and contexts of the voluntariness of retirement, can be identified. In this study, we build a compilation of previous results on factors influencing (in-)voluntary retirement based on a systematic literature search. We explore studies in which retirement’s voluntariness is the dependent variable. Focusing on retirement’s voluntariness as a dependent variable will allow us to identify and classify the main drivers and determinants of involuntary retirement. Based on this, groups of older workers at risk of involuntary retirement can be categorized.

Methods

Results

The literature search generated 3102 results (Figure 1). After removing duplicates, we screened 2076 titles for relevance by title and abstract. An additional search via Google Scholar was performed afterwards. Following recommendations by Haddaway et al. (2015), we screened the first 250 results. After 250 titles, we did not encounter any further relevant results. Based on the title and abstract, we removed 2256 titles. For the full-text assessment, we selected 70 titles. Considering the selection criteria (see above), 14 titles met the criteria and were included in the final review.OPEN IN VIEWER

Most of the studies use data from several countries and mainly from Europe. ² Only five non-European countries were included in at least one study: Japan, New Zealand, Australia, Canada and the United States. This country selection is probably caused by limitations in the data, as mainly single country data sets in European countries or cross-country data sets with a focus on Europe were used. When interpreting the results, it must be acknowledged that these are limited to ‘modern’ western societies with developed welfare states and pension systems. Samples were restricted to older people after their retirement transition.

Voluntariness of retirement was operationalized in three ways. i) Retirees were asked directly if they had retired voluntarily or involuntarily (Denton et al., 2013; Dorn and Sousa-Poza, 2010; Szinovacz and Davey, 2005; Welsh et al., 2018). Van Solinge and Henkens (2007) also applied the direct approach but used a variable that ranges from 0 (voluntary retirement) to 3 (involuntary retirement). The two next ways of operationalizing the voluntariness of retirement employ an indirect approach. ii) The second type contrasts the preferred retirement timing against the actual retirement age (Ebbinghaus and Radl, 2015; Streiber and Kohli, 2017). Retirees were asked if they would have wanted to work longer (involuntary retirement) or retired when they wanted to (voluntary retirement). iii) The third type of operationalized voluntariness of retirement uses the main reason for retirement, which is then categorized into voluntary and involuntary retirement (Hofäcker et al., 2016; Hyde and Dingemans, 2017; Madero-Cabib and Kaeser 2016; Radl, 2013b; Radl and Himmelreicher, 2015; Sheppard and Wallace, 2018). Being made redundant was, for example, considered involuntary retirement, while to enjoy life was seen as voluntary retirement. An interesting adaption of this is carried out by Radl (2013a) who sorts the reasons for retirement into three categories: voluntary, conventional and involuntary retirement.

Voluntariness of retirement was used as a dependent variable in all the included studies in different types of regressions including logistic (Dorn and Sousa-Poza, 2010; Hofäcker et al., 2016; Hyde and Dingemans, 2017; Madero-Cabib and Kaeser 2016; Streiber and Kohli, 2017; Szinovacz and Davey, 2005), ordered logistic (Van Solinge and Henkens, 2007) and multilevel (Ebbinghaus and Radl, 2015) and cox regression (Welsh et al., 2018) as well as Chi-square test (Sheppard and Wallace, 2018) and event history models (Radl, 2013a, 2013b; Radl and Himmelreicher, 2015)

The independent variables in these models, which are potential determinants of the voluntariness of retirement, will be discussed in more detail. We follow Hofäcker (2010), who identifies three levels of determinants for retirement transitions: i) The micro- or individual level includes determinants that are related to the person retiring, for example, their health status and the level of education. ii) The meso- or company level includes determinants that are related to the company in which the retiring person is working, for example, the sector or the sizes of the company. iii) The macro- or country level includes determinants at the national or welfare state level, for example, the economic development or situation of the labour market.

Discussion

Involuntary retirement can lead to poor health (Bacharach et al., 2008; Rhee et al., 2016; Van der Heide et al., 2013; Van Solinge, 2007) and affects retiree’s wellbeing (Dingemans and Henkens, 2014; Hershey and Henkens, 2014; Radó and Boissonneault, 2018). It is, hence, of high relevance to explore what factors lead to voluntary and what factors lead to involuntary retirement. To summarize and systematize previous research on determinants of retirements’ voluntariness, we conducted a systematic review of the literature.

All included studies were restricted to ‘modern’ western societies which clearly show the limited scope of previous research. Although some of the fastest ageing countries can be found in Asia (China, South Korea) no studies could be identified. Furthermore, most of the studies used cross-sectional data limiting causal interpretation as well as comparisons over time. Voluntariness of retirement was operationalized in most studies as a dichotomous variable, which is either involuntary or voluntary. One can ask if this reflects the complete range of how retirees perceive their retirement transition. ‘[V]oluntary versus involuntary retirement […] are more accurately conceived as continuous rather than dichotomous variables’ (Beehr, 1986: 34). Alternative operationalization, which might help capture the whole range of retirements’ voluntariness, used involuntary, voluntary and conventional retirement (Radl, 2013a).

The results of our study find several factors influencing the voluntariness of retirement transition on the micro- or individual level that are in line with ‘classical’ dimensions of social inequality. We identify vertical dimensions, like education or income, which affect the lives (including retirement transitions) in a negative way of persons in a less favourable position. In addition, there are horizontal dimensions such as age or gender that are additionally crucial to the retirement process. We assume that these factors interact and reinforce each other in the sense of a cumulative disadvantage (Dannefer, 2003). Besides these factors of social inequality, we also expect social norms to play a role, which are most dominant for age. Freedom of choice about the type and the timing of retirement is important for the voluntariness of retirement as well (Hyde et al., 2015). The company contexts or meso-level plays an important role in the voluntariness of retirement transitions. It seems that the workplace can either increase or restrict older workers’ agency, which then affects the voluntariness of retirement (Hyde and Dingemans, 2017). The connection between voluntariness of retirement and the company’s sector is less clear. Differences between countries and surveys in the definitions and operationalizations of the sector might be an explanation. A potential second explanation is that the association between sector and retirement transitions’ voluntariness is indeed ambivalent. On the macro-level that captures determinants of the labour market and the welfare state, contexts that increase older workers’ agency, again correlated with more voluntary retirement transitions. Such contexts are an affluent economy and a labour market with a high demand for workers. This supports older workers in taking command of their retirement.

The literature review allowed us to identify six gaps in the literature: i) More research on countries outside of Europe should be conducted. ii) More research should compare the voluntariness of retirement across retirement cohorts. This literature review has shown how important the institutional context is for how voluntarily a retiree perceives the retirement transition. This institutional context is changing (Ebbinghaus, 2006) and, hence, one would also expect changes in retirements’ voluntariness between retirement cohorts. iii) More research is necessary to investigate the role of the company and workplace level in retirements’ voluntariness. A first step would be the inclusion of more information on the retiree’s previous job in large national and international surveys. This should be complemented by company case studies. iv) More research should also focus on the country level and explore what configurations of the welfare state and the labour market are correlated with retirements’ voluntariness. Including more countries outside Europe would help to increase the variation on the country level. v) More research should consider potential interactions between the three levels. Hofäcker et al. (2016) show, for example, that the relationship between gender and retirement’s voluntariness does depend on the country context. This would also allow the issue of social inequality to be addressed with more precision. vi) More research should employ a more differentiated scale to retirement’s voluntariness, examples of which can be found in Radl (2013a) and Van Solinge and Henkens (2007). This would allow a more detailed analysis of the determinants.

When interpreting the findings of the systematic review at hand one must acknowledge two limitations: first, the language of the studies included was limited to English. This means that single country studies written in other languages might have been missed. Second, the studies do differ largely in the sample composition and statistical methods used and, thus, comparability is limited. Nevertheless, the study makes three main contributions. First, to our absolute best knowledge, it is the first attempt to systematically search for and evaluate research on the voluntariness of retirement transitions and its potential determinants. Second, this allows contexts to be detected in which older workers have a higher risk of involuntary retirement. Policymakers, employers, trade unions and other stakeholders can use this as a starting point to reduce the share of involuntary retirement transitions, and, as a result, increase pensioners’ health and wellbeing. Third, gaps in the literature regarding the determinants were identified. This can guide future research on the voluntariness of retirement transitions and its determinants.

Supplemental Material