Retirement transitions in the 21st century: a scoping review of the changing nature of retirement in Europe

The ways to transition from work-life to retirement are undergoing important transformations. The timing and pathways are changing, and many individuals are undergoing long periods of being in between working and retirement life. Yet, our cultural understandings of retirement tend to maintain a clear distinction between pre-and post-retirement life. Although the changes in retirement transitions are not new, the trend has accelerated in recent decades. We focus on what is known from the literature about the tendencies in alternative retirement transitions of healthy seniors in Europe since 2000. We review some conceptual and political transitions in the societal un-derstandings of retirement, followed by a scoping review in three sections: (1) later life employment transitions (bridge employment and employment after retirement) , (2) self-employment , and (3) unretirement . We conclude that


Introduction
There are important transformations taking place regarding the nature of retirement and the ways to transition from working life.This scoping review addresses these transformations and shows how the retirement transition has become a process rather than a point in time.Thus, a clear distinction between pre-and post-retirement life is increasingly obsolete.
Decreasing infant mortality rates at the beginning of the 20 th century and the application of new health technologies during the century prolonged life and created a surplus of labour.Thus, retirement was institutionalised, and older people were segregated from the labour market.When mandatory retirement and pension schemes became widespread in European welfare states during the 20 th century -which have since been a key factor in the consolidation and maintenance of such welfare states -retirement was a point in time.It was usually decided by chronological age and sometimes by years in the labour market.
Closely related to the industrialised society, retirement was at the end of an institutionalised life course organised temporally into rather fixed periods of childhood and education, adulthood and production, and finally old age and rest (Kohli 2007).Celebrated as a universal security system, retirement and pensions were also criticised for creating dependence and more sedentary lives among the older generations (Townsend 1981;Walker 1980).In the last decades of the 20 th century, this criticism was counteracted by a continued increase in life and health expectancy, which entailed long lives after retirement age.Those years were not used sedentarily.Instead, the nature of retirement was transformed by policies of active, healthy, successful and productive ageing combined with healthy generations of senior citizens, often engaged in many post-retirement activities.
In recent years, not only the increasing life and health expectancy has changed life post-retirement, but also the timing of retirement and the different pathways to it are evolving as European citizens increasingly postpone retirement and extend their working lives (European Commission 2018).Whilst some people work full-time beyond retirement age, most engage in different kinds of jobs with a transitional character (i.e.reduced time, reduced status, small-scale consulting, etc.).
Therefore, in this scoping review, we ask what is known from the existing literature about the current, different retirement transitions of healthy seniors in Europe in the 2000s.As retirement is changing, our understanding of this important transitional phase should change as well.Our aim is to summarise and disseminate research findings, which scoping reviews are ideal for (Arksey & O'Malley 2005).We believe that highlighting and disseminating the existing heterogeneity of retirement transitions plays an integral part in countering a stereotyped stance towards people approaching retirement age.
Recent review studies and conceptual contributions have focused on two main issues: how to characterise the myriad of emerging flexible retirement trajectories and how to explain this trend (Alcover 2017;Cahill et al. 2013;Earl & Taylor 2017;Hofäcker & Radl 2016;Sullivan & Al Ariss 2019).An illustrative example of the former can be found in the paper by Earl and Taylor (2017).They use the concept of "bridge employment (BE)" to signify a "broad range of late and post-career jobs including parttime work with the same employer that involves a reduction of working hours (also called phased retirement); a new or modified role with the same employer that involves a reduction of job complexity or physical demand (also called partial retirement); a job with a new employer in the same career field; a job with a new employer in a new career field; and self-employment (SE) … 'bridge jobs' share an ambiguity of being located beyond career but preceding retirement" (pp. 332-333).This is echoed by Alcover (2017: 247), who states that "The most common definition of BE refers to any kind of paid work (part-time, full-time, or SE) carried on after the end of an individual's professional career or full-time employment before complete withdrawal from the labour force or retirement.BE alternatives may therefore be considered modalities of retirement that prolong working life, allowing the term 'full retirement' to be used to refer to final withdrawal from the workforce".
According to these definitions, BE1 signifies any late career employment transitions before full retirement.In contrast, employment after retirement (EAR) emphasises continued employment or transition to new EAR age.The problem with both concepts is that inclusion in the study populations is unclear.For BE, it is difficult to say when you enter "post-career" as the job market is increasingly characterised by more flexible work forms, short-term contracts, etcetera.Furthermore, there is a gradual, sliding shift to new job functions or requirements for many employees, which means that it is difficult to say when you have entered BE.EAR may seem more well defined.However, the individualisation of pension schemes in many countries means that opportunities for retiring have become more flexible as opposed to the idea of a fixed retirement age for all.These ambiguities should be acknowledged, but they cannot be fully resolved as they reflect underlying empirical heterogeneity.A key implication for our review is that we must recognise a degree of empirical overlap between EAR and BE studies in our material.We will use the concept of "later life employment" (LLE) as the umbrella term but will refer to BE and EAR as subcategories in the results section, as these terms are used in the reviewed studies.
Building on this discussion, we take retirement to be the absence of formal work, and retirement transitions to be a process towards retirement with varying degrees of labour participation.This is in line with the dynamic perspective of retirement found in Cahill et al. (2013), who identified six periods through which older workers pass (or may pass) in their transition from full employment to permanent or definitive retirement.
A useful summary of the various explanations for the flexible retirement trend is found in the institutional rational-choice model developed by Hofäcker and Radl (2016: 8).This model suggests that retirement decisions at the micro-level (individual) are also influenced by several factors at the macro (societal) and meso (organisational) levels.A similar perspective of the interaction between micro-, macro-and meso-level factors is found in earlier contributions (e.g.Beehr et al. 2007).Hofäcker and Radl further identify four different factors or mechanisms that influence the transformation from institutional structures to individual choices.
"Push" and "pull" factors have traditionally been used in economic analyses, whilst the additional factors of "maintain" and "need" provide a more nuanced perspective, particularly suited for the current era and the emerging dynamics of flexible retirement patterns.Individual choices are clearly important explanatory factors for understanding retirement patterns.Yet, we agree that additional insights may be gained by supplementing this with an awareness of the meso-and macro-level factors that influence individual choice options.Such factors not only include the economic and policy contexts mentioned by Hofäcker and Radl but should also be extended to include cultural and normative pressures at societal and organisational levels (March & Olsen 2010).An illustration of such normative factors can be found in a study by Manfredi and Vickers (2009), which showed that many people feel pushed out of the labour marketby their employers, peers or themselves -when they reach a certain age.
Whilst these explanatory frameworks are very useful, it is essential to note that many of the studies in our review present associational relationships rather than causal explanations.We will summarise the study results whilst emphasising the need to remain aware of this weakness in the material.
The following scoping review should be seen as an introduction to a broad field with many approaches, explanations and complexities rather than a systematic review aiming to summarise results across a narrowly defined set of studies with similar scopes and methods.After the design and method section, we introduce the field by reviewing some conceptual and political transitions in the societal understandings of retirement.This is followed by the actual scoping review, which we have divided into three sections.(1) LLE includes BE and EAR, thus, consists of studies investigating people who transition into new work either full-time, part-time or in new transitional jobs in late career or after statutory retirement age.(2) SE consists of studies investigating people starting new enterprises around retirement age.(3) Unretirement covers studies investigating people who have exited the labour market only to enter again after a period of retirement.In the three review sections, we ask (1) who engages in such practices, (2) why do people engage in such practices, and (3) which structures and initiatives enable such practices.The three sections have emerged from the main themes in the empirical papers identified in our search.

Design and Methods
We have used the research databases PubMed and Scopus and included relevant studies published after 2000 and until May 2019 in those databases.In a second search, we updated with more recent studies published until June 2020.We have only included studies in English.As we have included studies investigating a minimum of one European country, studies comparing, for example, retirement practices in Germany, Japan and the US have also been included, but only the European findings are reported in the review.Numerous studies have focused on specific medical conditions that each have a significant impact on retirement transition.Thus, we chose to focus on studies that include only healthy seniors or do not specify the included subjects' health conditions.
Regarding the content and quality of the studies included, we have not excluded studies due to poor research designs, as this is not the role of a scoping review (Arksey & O'Malley 2005).In many cases, we include the research design in our discussion of their findings.This is important, for example, if a study includes people from 50 to 69 years of age (Haynes et al. 2014) because most of them are not in retirement age.It should also be noted that we have only been interested in studies that include people in or around retirement age.Therefore, studies that have asked future retirees about their prospective and speculative retirement/work practices have not been included (e.g.Matthijs Bal et al. 2012).
Initially, we found 142 relevant articles or chapters.After a screening of abstracts by the authors, the authors and student assistants went through all included abstracts to find duplicates and studies that should be excluded due to the above criteria.This left us with 100 studies.These studies were screened in full length by the authors, which left us with 72 studies.Initially, we also included a fourth category, BE.However, after analysing the nature of the studies in this category and those in the EAR category, we decided to combine BE and EAR into the LLE category, as there were too many overlaps in terms of what was actually being studied.
In the spring of 2020, a second search was conducted to update the sample of studies.Initially, 33 new relevant studies were included, which finally led to 15 new studies included in the sample.In total, we have included 87 studies divided into 67 studies investigating EAR, 15 studies investigating SE and five studies investigating unretirement (see Appendix 1).
Before presenting the results of this review, we will explore the shifting nature of retirement in the following section.As this exploration is primarily conceptual and policy-oriented, we have included studies within and outside Europe as well as policy reports from the European Union (EU).

Reforms and the shifting nature of retirement
In the past decade, the EU member states have undergone substantial pension reforms centred on "the introduction of higher pensionable ages, tighter eligibility conditions and reductions in early-retirement opportunities" (European Commission 2018).Central to these reforms is the postponement of retirement age, mirroring the increasing life expectancy (Moreira 2016) as well as policies and discourses of active ageing, which reframe old age as a life phase including societal participation (Lassen & Moreira 2014).
In the same period, the stability of retirement institutions across Europe has been questioned.In many countries, debates about the standards and designs of pension schemes and retirement rights have flourished, as models of flexible, gradual and differentiated retirement have been proposed and tested.As described recently by Phillipson (2019), retirement has become a contested institution as people in their 50s and 60s experience an increasingly fragmented nature of working life and engage in a range of precarious jobs.In the EU, part-time work is widespread among the 65+ age group, and 38% of the 65+ workforce were engaged in parttime work in 2011 (Eurofound 2014), and more than 50% of retirees would like to work at least some hours weekly.
An important measure for calculating the economic sustainability of European member states has been the old-age dependency ratio (e.g.European Commission 2012).However, this is currently being questioned, as the ratio's clear delineation between pre-and post-retirement is out of touch with current retirement practices.In current dependency ratios, 65+ is deemed out of the workforce, but in the EU, about one out of every ten persons between 65 and 69 years of age work (Eurofound 2016).
As numerous scholars have argued, retirement is a process that can take years, and it is often complex, with various pathways not necessarily entailing a complete withdrawal from the labour market (Cahill et al. 2013;Calvo et al. 2017;Fasbender et al. 2016;Leinonen et al. 2018;Tang & Burr 2015).The complexity of retirement pathways suggests that the traditional explanatory model of push-pull factors should be modified to include additional factors such as "need" and measures to "maintain" seniors (Hess et al. 2016).In this regard, it is relevant to note that one of the most important factors for post-retirement life satisfaction is the ability to choose one's retirement timing, instead of being forced to retire by sickness or unemployment (Fisher et al. 2016).
Whilst some identify working beyond retirement as a consequence of increasing precariousness, others embrace it as caused by increased health expectancy and a continued wish to engage and contribute.What should be clear is that there is rising uncertainty regarding the status of retirement and retirement timing (McDonald & Donahue 2011), as it becomes individualised.As Vickerstaff et al. have argued (2015), in the past retirement happened to us at a fixed age, whereas now it is something we do, and we do it in different ways.In a British qualitative study about senior workers, Phillipson et al. (2018) showed that after the end of mandatory retirement, senior workers are still implicitly expected to retire voluntarily in times of cutbacks and are implicitly expected to plan for their retirement.As such, the systems and organisations around workers appear to have retreated from the management of work-endings, thereby individualising retirement timing and creating uncertainties with limited support for taking good and timely retirement decisions.
A wealth of studies highlight the importance of retirement planning (e.g.Adams & Rau 2011;Elder & Rudolph 1999;Hershey et al. 2007;Taylor & Doverspike 2003).However, explored qualitatively, Moffatt and Heaven (2017) found that many workers were not able to plan, as they did not experience control and the ability to choose retirement timing.Unanticipated events (such as disease or unemployment) impeded them from acting according to plans.As such, retirement planning is a socially structured issue containing normative ideals about the ability to plan the many bumps of life.Retirement is uncertain to a degree.One US study found that white-collar boomers now perceive complete retirement as negative.Indeed, retirement is in flux (Kojola & Moen 2016).
Above, we have emphasised important shifts in the policies, practices and cultural understandings of retirement to frame the following review.We have also shown that whilst some attempts have been made to summarise aspects of the new retirement patterns across Europe (Cylus et al. 2018;Eurofound 2012Eurofound , 2014Eurofound , 2016;;Hofäcker & Unt 2013), more systematic information about the retirement transitions across Europe is needed.We will address these issues by looking further into the shifting nature of retirement through the studies investigating the heterogeneity of European retirement transitions.Although we have identified a considerable number of studies, there are clear limitations in converting this into a systematic analysis of differences across countries, regions or institutional designs, as there are simply too few studies to populate such subcategories.Therefore, our main contribution is to highlight several trends regarding retirement transitions as they unfold in the European countries covered by the available peer-reviewed studies.

Scoping 1: Later life employment
EAR can be defined as continued full-or part-time employment after the statutory retirement age.Across Europe, people tend to retire in their 60s (see Figure 1), although there are many signs that labour market participation, in general, is increasing beyond retirement age.Across Europe, people work longer than previously (European Commission 2018: 85; also Figure 2).Simultaneously, rates of sick leave are falling among the 65+ group, as documented, for example, in Sweden (Farrants et al. 2018), which could contribute further to this movement.Whilst we explore this tendency in the following, it should be noted that conventional retirement at the official retirement age remains the dominant pathway (Madero-Cabib et al. 2019).
Several studies use the concept of BE.As noted above, this concept is usually presented to signify a wide range of different types of parttime employment and dis/continuity.As such, it is challenging to define narrowly and has considerable empirical overlap with the EAR concept.Some people stay in the same job after pension age but at reduced hours.Others find new jobs with fewer hours, sometimes in entirely different fields than their previous jobs.Some people start businesses (which we will review in Section 2).Some people engage in one of the above but do not consider it part of their retirement transition.Others define BE as participation in paid work when you also receive a pension income (Dingemans et al. 2017).Thus, BE is not easily defined (Beehr & Bennett 2015), and in many instances, it corresponds to gradual, phased or partial retirement.The empirical perspective on BE is complicated because, in countries like the Netherlands, 75% of women and 25% of men work parttime throughout the life course.As such, BE can simply be a continuation of part-time work, perhaps with a gradual reduction in working hours.BE has been widely studied in the US, where it has, for example, been shown to have positive effects on mental health (e.g.Cahill et al. 2013), but it has been less studied in Europe.Much of the European research has been conducted in the Netherlands.We conclude that a uniform definition of BE is lacking.It often overlaps with EAR in terms of the empirical phenomena studied, although with the difference that BE is defined as a pre-retirement phenomenon in some studies.
LLE as a societal practice appears to be increasing, although the variability in definitions makes it harder to conclude across studies.One Dutch study found that 25% of retirees participate in BE after retirement (Dingemans et al. 2016).In another study using SHARE data from continental Europe, only 14.6% of the sample aged 55-70 engaged in BE when interviewed, making it less common than in the US (Brunello & Langella 2013).The growth in LLE makes it important to study the complex empirical reality of this phenomenon.
Who engages in later life employment and why?Whilst the majority of employees stay in the same working environment when continuing post-retirement work, a German study showed considerable heterogeneity in who engages in such activities (Burkert & Hochfellner 2017).Similarly, Swedish research points to differences in propensity to continue work due to a complex interplay between working life experiences, economy, health status and geographical location (Lundgren et al. 2018).In a cross-European study, clear differences in the likelihood of EAR were found, depending on whether the working retiree is employed in a highstrain or low-strain job (Dingemans & Henkens 2019a).
Whilst such complexity should be acknowledged, some common traits emerge from studies across Europe that generally point to four overall characteristics of people continuing to work: good health, high education, good working conditions and meaningful work.A European review that included 15 studies focusing on the importance of gender for EAR found that health is the most important factor for EAR across genders (Edge et al. 2017).Those who work at 65+ in the Netherlands are in better health than the retirees (Van der Zwaan et al. 2019;Wahrendorf et al. 2017).Another Dutch study based on qualitative data confirmed this observation of good health as the main predictor of EAR (Sewdas et al. 2017), whilst a longitudinal study from Scotland also pointed to the importance of mental health for EAR (Demou et al. 2017).
Across Europe, education is an important factor for the uptake of EAR.A study comparing Germany, the US and Korea found that in Germany, it is those with high education who engage in EAR (Cho et al. 2016).Also, in countries with low employment rates among the old, such as Poland, higher education increases the probability of EAR (Oleksiyenko & Życzyńska-Ciołek 2018).A qualitative study about EAR among academics in the UK, New Zealand and Australia found that this particular group experienced greater satisfaction levels when working after retirement age, which was caused by a decrease in administrative responsibilities (Hutchings et al. 2020).In Finland, a country with a flexible retirement age between 63 and 68, people with higher education retire later.However, people with lower education engage in more EAR, thereby decreasing the difference in actual working hours between the two groups (Leinonen et al. 2018).
Overall, we find the same tendencies in studies about BE and EAR regarding health and education.Another emerging insight from a growing number of studies is that barriers and options in the pension system are important, as greater flexibility promotes LLE.
One scoping review, including countries across the world, supported the conclusion that BE is primarily for those in good health (Carlstedt et al. 2018).This is also confirmed by a study across 16 European countries, which concluded that BE is most likely among those in good health, and high education increases the likelihood of BE.The same study also mentioned the importance of pension income and marital status.When people have a high pension income, their likelihood of BE decreases, and when they have experienced divorce or widowhood, their likelihood of BE increases (Dingemans et al. 2017).A British study showed not only that more men than women are engaged in BE but also that the strongest predictors are physical and cognitive abilities (Stafford et al. 2017), which confirms that health is strongly correlated with BE.
Working conditions throughout the life course are also important factors in EAR.A study including 14 European countries found that people with disadvantageous and discontinuous working lives were less likely to engage in EAR (Hoven et al. 2018).Furthermore, people in more privileged positions engaged in EAR, and their psychosocial working conditions when doing so were better than those reported by retirees reflecting on their last occupation (Wahrendorf et al. 2017).
Another cluster of studies points to meaningfulness in the job as an essential factor for EAR.Personal attitudes to work (Van der Zwaan et al. 2019), work engagement (De Wind et al. 2016), and high levels of personal and social meaning in work (Fasbender et al. 2016) are all associated with EAR.The importance of perceiving employment as meaningful is also confirmed in a Dutch study about BE (Veth et al. 2018).Other key factors include task autonomy (Alcover & Topa 2018;Müller et al. 2015) and flexible working conditions (Oude Mulders et al. 2013).Whilst such positive "pull" factors are important, it should also be kept in mind that some people engage in LLE due to economic necessity and/or a desire to gain financial security.This complexity of background factors is displayed in a German study about EAR (Burkert & Hochfellner 2017).
When scrutinising companies' HR policies, a Dutch study found that in many cases, there are no formal agreements on BE.However, the most valued workers are often offered BE through informal agreements (Oude Mulders et al. 2013).This finding suggests that BE is more easily available for the most privileged or specialised workers, who often do not need formal agreements.
Two of the key reasons for EAR that emerge across many studies are social relations and fulfilment in the workplace.Those who engage in EAR tend to have more friends than those who do not work (Haynes et al. 2014).Furthermore, people with jobs contributing to their identity and a sense of accomplishment are more likely to be employed after traditional retirement age (Hovbrandt et al. 2019).Another factor, although this has only been scarcely investigated, could be people's attitudes towards ageing.A German longitudinal study investigating psychological aspects of EAR showed that employees who perceive retirement as a social loss or working as personal growth are more likely to engage in EAR.In contrast, people who experience ageing as a gain of self-knowledge are not likely to continue working beyond the normal retirement age (Fasbender et al. 2016).
What are the consequences of engaging in later life employment?The reasons for engaging in LLE appear to be correlated with the effects in terms of self-reported wellbeing.In the UK, those who state that they engage in EAR for financial reasons score lower in quality of life than those who retired at the usual age.Conversely, those who continue employment for non-financial reasons score much higher (Di Gessa et al. 2018).Nevertheless, it is also essential to keep the broader societal and economic context in mind.A study of 16 European countries found that retirees with low incomes or who live in countries with low retirement income increase their life satisfaction through EAR; however, this is not evident for retirees with high income (Dingemans & Henkens 2019b).
Some Dutch studies point to BE as good for the quality of life, life satisfaction and mental wellbeing, particularly if the transition from career job to retirement was involuntary (Dingemans & Henkens 2014, 2015).However, a European study concluded that there is no difference in overall wellbeing between people who retire early or late, partially or fully (Sohieret al. 2020).
Which structures and initiatives enable employment after retirement?In recent years, there have been many initiatives across Europe to extend working lives.Whilst some of these have targeted retirement age in general, others have endeavoured to create more flexible working conditions around retirement, which has been shown as a precondition for increased EAR (Sewdas et al. 2017).Flexible working conditions can be organised through formal legislation and contracts, but informal arrangements also appear to play an important role (Cebulla et al. 2007).Key to creating such flexible working conditions is the removal of barriers for financially vulnerable groups who find it difficult to obtain employment (Dingemans & Henkens 2019b) as well as the maintenance of employment rights after retirement age (Lain 2012).Initiatives such as increasing the minimum retirement age and reducing employment protection rights have raised BE levels in Northern and Central Europe and increased permanence in fulltime employment in Southern Europe (Brunello & Langella 2013).Also, policies encouraging BE by allowing for additional pension income seem to work well in the Netherlands (Dingemans & Henkens 2014) and Norway (Furunes et al. 2015).A European macro-level study found that the higher the tendency to engage in part-time employment in general across the population, the lower the propensity to retire early, as many continue in what could be considered BE (Been & Van Vliet 2017).
Besides flexible conditions and structures, age management is highlighted by some studies as being essential in promoting EAR (Principi et al. 2020).A Swedish qualitative study highlighted the introduction of age management in the workplace as important for retaining older workers (Blomé et al. 2018).This has also been studied in another Swedish qualitative study.Through interviews with managers, it is suggested that workplace norms and cultural understandings are crucial, and the issue of transitioning to a less demanding role is often seen as a problem for the individual worker rather than a typical part of the "socio-temporal order of companies" (Krekula 2018).Another study of Swedish managers showed that managers' attitudes towards EAR determine whether they try to retain older workers (Nilsson 2018).One aspect of such age management strategies could be to provide older workers with an increased sense of control, as a British longitudinal study points to this as key for EAR (Carr et al. 2016).Furthermore, it has been suggested that guiding senior workers and reducing ageism in the workplace can also promote BE (Carlstedt et al. 2018).
Another cluster of studies focuses on the importance of working conditions throughout the life course for EAR.A study across 14 European countries revealed that work trajectories throughout the life course are important for EAR, and disadvantage accumulates during the life course, leading to early retirement (Hoven et al. 2018).Likewise, a study of eleven European countries found that acquiring skills and good working conditions through the life course can facilitate the ability to engage in EAR (Komp et al. 2010).In this regard, gender plays an important role, particularly in countries with caregiving policies requiring the mother to take care of the children, often leading to less stable work trajectories (Wildman 2020; Worts et al. 2016).
To sum up, good health, a high level of education, good working conditions and meaningful work contribute to an increased likelihood of engaging in LLE.Other factors such as a widespread culture of part-time work throughout the life course also play a role.People tend to engage in EAR when they experience a sense of meaning through work.EAR is reported to be good for wellbeing if it is not caused by financial need.These results should be tempered with the observation that most studies only observe associations, and there may be issues of bias and reverse causality.In any case, it appears that LLE can be promoted through various pathways.At the same time, a flexible infrastructure around retirement and good working conditions, in general, are likely to promote LLE.
Scoping 2: Self-employment SE in old age appears rather easy to define.However, a recent systematic review has pointed to several issues for empirical studies (Ratten 2019).First, when describing SE, should people who are entering old age as self-employed also be included, or only people who begin SE at older ages?Most of the studies included focus exclusively on newly started companies by older people, but some studies do not differ clearly between the two.Second, SE can vary substantially in terms of how the older entrepreneur envisions the company and whether it is a small hobby enterprise, an endeavour to make a living or to build a larger company.Third, the boundary between adults and older adults seems lower within this area than other areas of senior employment.As such, most studies included in this section involve people in their 50s.We have included these studies, as few studies on SE only including people 60+ have been conducted.
Whilst the EU encourages entrepreneurship in older ages (Eurofound 2014), and so-called grey entrepreneurship (Stirzaker & Galloway 2017) has been embraced in some European nations through support programs (e.g. the Prince's Initiative for Mature Enterprise [PRIME] in the UK), there are also some issues regarding precariousness and ageism forcing people out of the labour market, which we will highlight in the following sections.Whilst evidence shows that more SE people are working after 65 compared to wage earners (Wahrendorf et al. 2017), it is unclear whether this is due to poorer pension schemes or other factors (e.g. a different concept of work or stronger identification with own company compared to wage earners).A German study on SE people (not in retirement age) showed that they do not save for pensions sufficiently (Fachinger & Frankus 2017).Whilst this suggests that people might remain in SE for financial reasons, other studies point to more positive aspects of SE or suggest that some seniors simply continue what they are doing because they can (physically) and because no one is telling them not to.This may have positive mental and physical health implications, as shown in an Italian longitudinal study, where SE people were shown to live significantly longer than the rest of the population (Lallo & Raitano 2018).
Who engages in self-employment and why?The studies in our sample present a mixed picture regarding the issue of who engages in SE.Overall, it is difficult to claim that older entrepreneurs are generally well-off or in advantaged positions.Whilst a Dutch longitudinal study showed that it is primarily older people with high levels of finance and education starting SE and that they do so due to opportunity rather than necessity (Van Solinge 2014), another Dutch study revealed that SE is prominent and growing among older workers in the creative industries, but only the most successful strive at this.The rest get supplementary income from unskilled labour jobs (Hennekam 2015).A UK study showed that for individuals who have been made redundant at 50+, SE is a valuable option.However, although triggered by redundancy, the qualitative findings show that the study participants view their SE as a positive experience, enabling increased satisfaction and personal growth (Stirzaker & Galloway 2017).
In terms of how widespread SE is, a UK study showed that it is a commonplace, and it seems a valid option for many to extend their working lives (Small 2012).However, the total numbers are based on estimates and response biases.A cross-European study found that SE's relative importance is higher for the 50-69 cohort than for younger groups (Morris & Mallier 2003).Moreover, a study comparing SE patterns in Greece and the UK found that older people are more likely to be engaged in SE than younger generations.However, who these older people are in terms of educational and professional backgrounds differs from country to country and region to region.In terms of gender, more males than females are engaged in SE in general, but with age, SE increases for both genders (Morris & Mallier 2003).
The reasons for SE among older people are often studied as either push or pull; however, this does not appear to fully cover the complex pathways of SE (Stirzaker & Galloway 2017).For some people, redundancy is the event triggering SE (Small 2012;Stirzaker & Galloway 2017;Van Solinge 2014), but in other instances, redundancy can be a triggering event for positive experiences.Moreover, whilst redundancy may be a triggering event, the decision to become SE is not only caused by this event.For example, a study across the Rhine Valley found that whilst half of older people who engage in SE are pulled and the other half pushed; factors such as family traditions for SE matter in terms of the probability of SE (Harms et al. 2014).Another study pointed to higher age as increasing the likelihood of engaging in SE out of self-will (Kautonen et al. 2014).Finnish research using register data showed that people pushed into SE have lower education than those who were pulled, and that males and married people, in general, tend to be opportunity-driven.In contrast, women tend to be necessity-driven into SE (Tervo & Haapanen 2017).
In many cases, SE offers older people the opportunity to retire gradually (Morris & Mallier 2003).A UK-based study showed that older people, for the most part, engage in SE in the same field as their former employment (Small 2012), which adds to the findings from studies concluding that older people do not engage in SE to increase wealth but to use their skills and be their own boss (Stirzaker et al. 2019).SE can also be a response to ageism in the workplace (Stirzaker & Galloway 2017).Moreover, SE brings joy, meaning and purpose to some older people (Stirzaker et al. 2019).A qualitative study from the UK found that women aged 50+ draw on narratives of a new life stage with more freedom, self-knowledge and greater financial and personal security when they engage in SE.At the same time, they actively use this narrative to distance themselves from the category of older woman (Tomlinson & Colgan 2014).As such, the reasons for SE are many, and limiting the decision to either push versus pull or necessity-driven versus opportunity-driven does not account for the complex pathways towards SE.
Which structures and initiatives enable self-employment?Only a few studies have investigated the structures and policy initiatives that enable SE in older age.Overall, government support and favourable policies seem to work as pull factors for SE, whilst an ageist labour market and precariousness sometimes push older people into SE.Also, general patterns of SE within the population affect the SE levels of the older population.
Specific evidence on a policy initiative may be found in a UK study on the PRIME scheme, which supports people aged 50-70 considering SE.This initiative shows positive effects in terms of how many participants end up starting a business; however, this finding is biased because people contacting PRIME are already inclined to start up a business (Kautonen et al. 2008).This study also found that only 15% of the people contacting PRIME are 60+, but when contacting PRIME, this group is more likely than younger age groups to actually start a business.
In the Netherlands, policies favouring SE among the older population have increased the trend (Van Es & Van Vuuren 2011).Other factors such as an ageist labour market also drive older people to SE (Van Solinge 2014).Some countries operate benefit schemes that create disadvantages for people engaged in SE.This can be seen in a study from Finland, where older SE persons who have to close a business must become regular job seekers for a period before they can obtain incapacity benefits (Kautonen et al. 2008).
Whilst policies supporting or penalising SE in old age might influence the likelihood of SE, an EU study showed that if a country, in general, has a high level of SE, the 50-69 cohort are also more likely to be SE (Morris & Mallier 2003).Hence, whilst policies and support structures might prove beneficial to increase SE in old age, the general SE patterns and cultural norms within the population could prove just as important.
In summary, SE is widespread among older workers, and the likelihood of self-initiated SE increases with age.More males than females engage in SE in high age.However, it must be considered that most studies do not differentiate between SE throughout the life course and engagement in SE when approaching retirement.There are many pathways to SE, and redundancy is merely one among others.However, SE can become a positive way forward for people experiencing redundancy.

Scoping 3: Unretirement
Unretirement has mostly been studied in the US.For example, it has been shown that 26% of US retirees unretire (Maestas 2010).Whilst one explanation for this could be a financial shock or unexpected boredom after retirement, the study shows that unretirement is often expected prior to retirement, reflecting a complex process with many pathways to retirement.Although this phenomenon has not been widely studied in Europe, the few studies that exist suggest that the practice is widespread.
As with the other scoping categories, unretirement is not easily defined.Unretirement could both entail people ceasing paid work entirely and then, after some time, starting to work a few hours weekly, and people who gradually retire and start working full-time again (Platts & Glaser 2017).Also, the numbers suggesting it is a widespread practice could be caused by fluid boundaries between formal and informal work in practice.If a person retires from formal work but works informally for some hours a week at a small local business, and then formalises this work through a contract at some point, this would be considered unretirement in some research designs.Unretirement is distinguished from EAR and SE by a temporary break in labour participation, where the individual retires completely, followed by some kind of re-uptake of labour participation.Based on our empirical review, we find this phenomenon sufficiently unique to justify separate attention.
Who unretires and why?Similar to EAR, unretirement is mostly practised by the wealthy, the healthy and the well-educated.A study from Germany, Russia and the UK found that financial need was generally not the reason people unretired (Platts & Glaser 2017).The study found that 42% in Russia, 25% in the UK and 17% of retirees in Germany unretired.This may be compared to another study showing that only 6% in the UK and 2% in Italy unretire (Smeaton et al. 2018).The large difference for the UK is due to differences in the study designs and varying definitions of unretirement.In Sweden, the number seems to be low and varies between 6 and 14% depending on the definition (Pettersson 2014).In the Swedish study, the same association between unretirement and education was found, but it was also found that higher pensions decrease the probability of unretiring (Pettersson 2014).Another study not only confirmed the UK pattern but also showed that males tend to unretire more than females, and people with a mortgage when reaching retirement are more likely to unretire (Platts et al. 2019).Also, in the UK, if a person's partner is still engaged in paid work, that person is more likely to unretire (Kanabar 2015;Platts et al. 2019).Moreover, people who have recently retired have the highest probability of unretiring (Platts et al. 2019).
As suggested in the section above, unretirement is usually not caused by financial need (Pettersson 2014;Platts & Glaser 2017).Unretirement is not a way for people with few resources to obtain an improved financial situation during retirement.Rather, unretirement reinforces social differences, as it is usually the resourceful who unretire (Platts et al. 2019).In a comparative study of the UK, the US and Italy (Smeaton et al. 2018), the predictors in countries where unretirement is common (the UK and the US) are financial needs, high level of education, good health, no care responsibilities and being male.In Italy, where unretirement is less common, it is solely being male that is associated with a higher frequency of unretirement.
Across the studies, unretirement seems to be more a choice of lifestyle rather than driven by financial necessity (Kanabar 2015;Pettersson 2014), although financial necessity plays a larger role in some countries than others, with the US and the UK as the main examples (Kanabar 2015;Smeaton et al. 2018).However, the complex pathways towards unretirement have only been scarcely studied qualitatively.In a mixed-methods study that included qualitative interviews (Smeaton et al. 2018), the dichotomy between lifestyle choice and financial necessity is challenged.For example, whilst income level did not predict unretirement in the UK, the qualitative data showed that people often had mixed financial and social considerations when unretiring.Debt and children under the age of 30 (in need of expensive education) were part of the motivation to unretire.Hence, structural, social, cultural, intergenerational and financial considerations can be difficult to distinguish in unretirement practices.
Which structures and initiatives enable unretirement?As the research on unretirement in Europe has been sparse, it is difficult to determine which structures hinder or promote this phenomenon.Although the studies suggest that unretirement is often a lifestyle choice, in countries with high levels of pension adequacy, there is a decreased probability of unretirement (Pettersson 2014).Also, it has been suggested that the early retirement culture in Italy -in part due to a perceived moral duty to step aside for younger adults in countries with high unemploymentexplains low unretirement probability.This is combined with a cultural perception of retirement as a well-deserved period of resting that includes new and important social roles in terms of grandparenting (Smeaton et al. 2018).
Although such research findings seem somewhat speculative, they reveal a crucial gap in knowledge regarding unretirement as a social and cultural phenomenon.To sum up, more research is needed that analyses the retirement patterns leading to unretirement.With this scoping review of unretirement, we have shown that simple causal explanations (such as financial need or boredom with retirement) are insufficient.

Discussion and Conclusions
This scoping review has presented a number of trends and observations across European countries based on the available studies.This is useful for generating hypotheses to be explored in subsequent studies with a systematic inclusion of more countries, regions or institutional designs for comparisons.
The present study focuses on retirement from formal employment.Yet, it should be acknowledged that formal retirement is not the same as retiring from activities that are productive or useful for society, such as volunteering, paid work and caregiving (Mergenthaler et al. 2019).In many European countries, retirees are actively engaged in volunteering (Erlinghagen & Hank 2006).Likewise, grandparenting is widespread and has proven good for subjective wellbeing (Arpino et al. 2018), health (Di Gessa et al. 2016) and verbal fluency (Arpino & Bordone 2014), as well as contributing to the coherence of families and the productivity of the grandchildren's parents.Today, we see new gender roles also in this life phase, as grandfathers are increasingly involved in grandparenting (Mann & Leeson 2010).Also, many older persons engage in informal care for their spouses (Bertogg & Strauss 2018), and the people reaching retirement age are now increasingly seen to provide informal care for their parents, as the life course is stretched out (Dudová 2015).However, in this review, we only include studies dedicated to examining paid work.Nevertheless, as we have noted here, there are other retirement transitions that could also be classified as participatory, engaged and contributing.Indeed, such unpaid activities and associated identities may play a critical role in decision-making regarding paid work (Sullivan & Al Ariss 2019: 278).
This review has largely confirmed the relevance of macro-, meso-and micro-level factors influencing retirement decisions.It appears that national policies, general economic context, labour market conditions and a range of individual-level factors serve to push, pull, maintain and create needs that influence retirement transitions and engagement in LLE, SE and unretirement.Yet, this review has also indicated that retirement choice is embedded in a broader cultural and institutional context of values, norms and perceptions.Nevertheless, the interaction between such factors and traditional push, pull, maintain and need factors should be studied more extensively.
This review has also illustrated that the choice of LLE, SE and unretirement is more real for some senior citizens than for others.Pull and maintain appear to be particularly relevant for a highly skilled, healthy and well-connected labour force.In contrast, push and need seem to be the reality for many other groups.Whilst the first group may thrive under the turn to more flexible retirement schemes, the latter may experience this as a continuation of an increasingly precarious labour market situation.Some may not be able to find late career employment despite need (Earl & Taylor 2017: 332).Thus, the differences in choice architecture and opportunities for various employee groups should be studied in more detail across different institutional settings.
In any case, it is clear from our analysis that the tendency to create fixed, spatial, political and cognitive boundaries between life pre-and post-retirement does not fit well with the complexity in actual retirement practices that has emerged over the past decades.Indeed, drawing the line between retirement and post-retirement is complicated.
This mixture of push-pull factors and necessity-and opportunity-driven reasons is general for the three themes of LLE, SE and unretirement.However, it is important to note that many studies in our review present associational relationships rather than causal explanations.More studies and different study designs are needed to clearly establish causalities in the frameworks that have been put forward to understand LLE, SE and unretirement trends in different contexts.

Statement of funding
This manuscript has been prepared as part of the work in the Center for Healthy Aging at the University of Copenhagen and has been supported by the Nordea Foundation.The foundation did not play a role in the design, execution, analysis or writing of the study.

Declaration of contribution of authors
The authors have contributed equally to the review and the preparation of the manuscript.

Figure 1 .Figure 2 .
Figure 1.Age structure of the workforce, including formal careers, adjusted for full-time equivalence.Data from the European Social Survey including Austria, Belgium, Czechia, Denmark, Estonia, Finland, France, Germany, Hungary, Ireland, Israel, Lithuania, Netherlands, Norway, Poland, Portugal, Slovenia, Spain, Sweden, Switzerland and United Kingdom.The group "Paid work and fulltime carer" is unfortunately not visible in black and white graphics, but it is much smaller than the other groups and primarily exists in the age groups from 30 to 60.For details, please see the original source (Source:Cylus et al. 2018: 20)