Bridge employment quality and its impact on retirement adjustment: A structural equation model with SHARE panel data

Abstract

Bridge employment refers to the workforce participation pattern displayed by older workers between their partial retirement and their complete withdrawal from the workforce. Based on Shultz’s (2003) model of antecedents and consequences of bridge employment, this article proposes a set of hypotheses, using SHARE panel data (N = 634) from Wave 1 (2004) and Wave 2 (2006). These data are analyzed via structural equation modeling (SEM), testing both a direct effects model and a partial mediation model. Results show that bridge employment quality partially mediates the influences of T1 antecedents on T2 consequences. The implications of this study are discussed at both the theoretical and practical level.

Keywords

Bridge employment panel study retirement structural equation modeling

Participation in the labor market by workers over 50 years of age is currently a concern for industrialized countries. Older workers’ increased life expectancy and improved health conditions in developed countries leads many retirees to seek some kind of paid employment, known as bridge employment (hereafter, BE). Preliminary comparisons of international studies, such as SHARE (Survey of Health, Ageing and Retirement in Europe; Börsch-Supan and Jürges, 2005), HRS (Health and Retirement Study) from the USA, and ELSA (English Longitudinal Study of Ageing) from England, showed that the rate of employment among older workers (over 50) is much higher in the USA than in Europe, but the difference is more pronounced between 60 and 64 years. The data from the first phase of the SHARE study show that in Europe, the percentage of healthy participants between 50 and 60 years with bridge jobs ranges between 13.5% in Greece and 0.8% in France (Börsch-Supan et al., 2005).

BE refers to the diverse forms of older workers’ participation in the labor market that occurs between their retirement from the position they occupied for 10 or more years and their complete workforce withdrawal. BE can be either in the same occupation or different occupations, on a part-time, temporary or full-time basis, and either in salaried modalities or self-employment/entrepreneurship form, and often involves a combination of fewer hours with greater flexibility, less workload, responsibility or stress, and fewer physical demands (Beehr and Bennett, 2008). In other words, it is an employment carried out after retirement from a full-time post, but before quitting the labor market definitively. In this sense, for a job to be considered as BE, the worker must be receiving some kind of pension or economic remuneration depending on the previously held post, at the same time as he or she receives some monetary compensation or salary for the current position (Davis, 2003; Feldman, 1994). In countries like the USA or Canada, an increase in the frequency with which prematurely retired workers (before age 65) rejoin the workforce has been detected (Alley and Crimmins, 2007). Longitudinal data from the HRS indicate that approximately 15% of older North Americans who had a full-time professional career re-enter the workforce after having retired (Cahill et al., 2011).

BE is important to workers, organizations, and societies as a whole for many reasons. It is beneficial for maintaining the workforce and the productive activities of older workers from the baby boomer generations, avoiding the negative and dysfunctional effects in terms of the social security contribution and risks for the social protection system derived from the massive retirement of workers in a very short period of time (Alley and Crimmins, 2007; Pengcharoen and Shultz, 2010; Wang et al., 2008). BE contributes to the increase of quality of life, satisfaction with retirement, and adjustment to retirement (Dendinger et al., 2005; Topa et al., 2009). It strengthens autonomy, financial security, and economic well-being after retirement (Munnell and Sass, 2008; Topa et al., 2011). It reduces the experiences of discrimination due to older workers’ age, as it establishes agreements to flexibilize the occupational relation, benefiting aged workers and organizations, at the same time (Beehr and Bowling, 2002; Schalk, 2010). In addition, BE has the advantage of maintaining the continuity of life patterns and styles (that is, a balance between work activities and family, social, and leisure activities), avoiding the sudden interruption of activity which is typical of traditional retirement, promoting positive psychological results associated with personal and social identity, perceptions of personal accomplishment, role performance, entrepreneurial behaviors, etc. (Wang and Shultz, 2010). In the past decade, organizations have tended to consider BE strategies as an increasingly important part of human resources management policies, specifically designed for workers over 55 years, a mechanism to retain (and attract) highly qualified workers with experience (Rau and Adams, 2005).

Empirical studies have paid increasing attention to BE, exploring its antecedents and consequences, and motives for continuing to work among the aged population (Davis, 2003; Feldman, 1994; Feldman and Kim, 2003; Gobeski and Beehr, 2009; Kim and Feldman, 2000; Kooij et al., 2011; Šimová, 2010; Von Bonsdorff et al., 2009; Wang et al., 2008). Nevertheless, many of them have used convenience samples and North American samples. Some studies have used large national samples, such as the HRS (Weckerle and Shultz, 1999), and longitudinal designs (Wang et al., 2008), but were restricted to predicting the different forms of BE that the employee may have chosen – career BE or BE in a different field (Feldman, 1994; Shultz, 2003) – without considering its long-term impact on other outcomes. Various theoretical models of the antecedents and consequences of BE have been proposed and tested in empirical studies. Despite increasing interest in BE, to date there are no empirical longitudinal works with cross-national data about the long-term antecedents and consequences of BE. As a result, the objective of this study is twofold. First, to test a model of antecedents and consequences of BE, using panel data from SHARE (Wave 1 and Wave 2). Second, to extend the literature by including specific job-related psychosocial variables, such as BE quality, to predict their impact on employees’ well-being.

BE antecedents

Among others, Kim and Feldman (2000) and Shultz (2003) presented models of the antecedents of BE. The factors that directly affect the BE decision and experience could be grouped into individual and organizational factors. Among the first group, demographic status, working spouses, dependent children, health conditions, job-related skills, organizational tenure, psychological distress, role overload, type of pension and pension eligibility, and financial circumstances must be taken into account (see also Beehr and Bennett, 2007; Quinn, 1999). Organizational factors include attitudes toward older workers and flexible job options, among others. But the specific characteristic of Shultz’s (2003) model was the addition of specific job-related psychosocial variables, such as attitudes to work and to retirement, offering a more comprehensive model.

Different theories attempt to explain why people wish to continue their work activity with BE and the mechanisms by which this contributes to personal well-being. According to the continuity theory (Atchley, 1989), retirement has negative consequences, associated with the loss of previous roles, and people seek continuity between the moments before and after retirement, which is materialized in BE. The higher the role continuity, the fewer the adverse side effects of the loss sustained upon retiring. When we add the life span perspective, we can see the importance of the link between different life settings and, thereby, the relevance of demographic factors and personal dispositions toward retirement and its consequences. Regarding the individual factors that would affect BE, most of the studies support the fact that age (in a negative sense) and good health (in a positive sense) are antecedents of BE (see Kim and Feldman, 2000 among others) (H1a and H1b). A complete set of hypotheses for this study has been depicted in Figure 1.

Figure 1.

Hypothesized pattern of relationships for the partial mediation effects model.

Recent works have recommended extending the personal antecedents of BE to include personality variables, such as optimism or openness to change. Specifically, optimism has been proposed as a measure of positive expectations and has been shown to be a good predictor of desirable results, such as happiness or effective coping styles (Scheier and Carver, 1985). Therefore, we proposed optimism as a positive antecedent of BE (H2).

Regarding the specific job-related psychosocial variables aforementioned, some empirical studies had shown that attitudes toward work and retirement should influence perceptions of BE. For instance, Zaniboni et al. (2010) found that the anticipation of lost social integration upon retirement was positively related to the intention to take only part-time retirement. More specifically, people who were more satisfied with their job, with their current financial situation, and retirement were more likely to become involved in BE (Weckerle and Shultz, 1999). Similarly, some authors (Schmitt et al., 1979) found a small but significant relationship between the way people describe their jobs and their intention to retire early. Those with early retirement intentions described them as lower in autonomy, skill variety, and opportunity to interact with others.

With regard to the quality of life (QoL), it has been defined from a broader perspective as an indicator focused more on adult life characteristics than on the presence or absence of symptoms of disease, including various facets, such as autonomy, control, pleasure, and self-realization (Higgs et al., 2003). Based on this, it is reasonable to assume that both older people’s QoL and life satisfaction, understood as the extent to which their needs are satisfied, are antecedents of their real decision to engage in BE (H3a and H3b).

Job satisfaction should be defined as a positive and pleasant emotional state arising from workers’ perception of their work (Warr et al., 1979). Likewise, stress theories suggest that when one leaves a stressful role, such as adverse work conditions, one will experience relief associated with the change in the situation. In this sense, if full retirement can be conceptualized as withdrawal, it may be encouraged by low job satisfaction as a result of workers’ perception of their work as stressful and unpleasant. Specifically, Wang et al. (2008) showed that job satisfaction predicts involvement in career BE, while Gobeski and Beehr (2009) found that job satisfaction was associated with a high probability of career BE. In addition, those with high job satisfaction were more likely to work on BE (H4).

BE consequences

Specific models of BE have underlined life satisfaction and job satisfaction among the more notable results, to which other factors, such as physical and psychological health, have been subsequently added, and supported by the empirical studies. Therefore, we contend that BE will be positively related both to QoL and to life and job satisfaction (H5, H6, and H7).

BE characteristics

BE provides structure to daily life, offers opportunities to continue to perform rewarding activities, and provides social contacts, therefore its mediating role in the relation between antecedents and consequences seems to be theoretically supported. However, the positive influences of employment on personal well-being do not only occur in the quantitative dimension – number of hours dedicated to work – but especially as a function of its quality. Thus, Ulrich and Brott (2005) confirmed that BE flexibility is also perceived in the higher capacity to work on self-selected projects, working at one’s own pace, or working in a family setting or comfortable environment. In this sense, BE in its forms of self-employment is very attractive for a growing number of older workers. As has been noted (Giandrea et al., 2008), the choice between a wage-and-salary BE and a self-employment BE may be determined by the preference for a higher average income but less autonomy in the former versus greater freedom and leisure time but lower income in the latter.

There is a long tradition of research into the characteristics of job quality, understood as the balance between efforts invested and perceived rewards (Marmot et al., 2006; Siegrist, 1996; Siegrist et al., 2004). Empirical studies show that unbalanced situations generate stress and have negative effects on health. In the active population over 50 years of age, the predictive power of perceived reward on personal well-being and health has been revealed (Blanchet and Debrand, 2007; Debrand and Lengagne, 2007).

Specifically, BE conditions may be inadequate for the worker because positions change, and spaces are reduced or displaced. In addition, the allocation of tasks of less responsibility to those who perform BE jobs can lead the latter to consider themselves to be under-employed. Based on these findings, we conceptualized BE quality as the balance between efforts and rewards in four job dimensions: support, recognition, earnings, and advancement possibilities offered by work.

The framework of the relationships proposed in this study assumes that there are direct effects between antecedents and consequences, based on the preceding research. This is the case of the positive relations between optimism and results, between physical health and job satisfaction, and between QoL and life satisfaction. We also assume direct effects between QoL at T1, on the one hand, and QoL at T2 on the other.

Based on these rationales, we propose that BE quality will be a partial mediator in the relationship between BE antecedents and BE consequences (H8). This is not to imply that the direct effect of antecedents on consequences is always insignificant. Rather, our contention is that the direct effect of antecedents on consequences after controlling for BE quality is mostly marginal compared with its effects before controlling the influences of BE quality.

Method

Participants and procedure

The Survey of Health, Ageing and Retirement in Europe (SHARE) is a multidisciplinary and cross-national panel database of individual data on employment, health, socioeconomic status, and social and family networks of more than 55,000 persons aged 50 or over, at this moment, from 20 European countries. As De Lucca and Lipps stated (2005: 75), ‘SHARE is designed to be a genuine cross-national survey. The common interview mode, questionnaire design, effort devoted to the translation of the questionnaire, and, finally, the standardization of the fieldwork procedures across countries – including, wherever possible, a common electronic case management system – were the most important design tools used in SHARE to ensure strict cross-national comparability and high quality of the data.’

For these reasons, panel data were obtained from the SHARE study which included Wave 1, release 2.0.1 (2004) and Wave 2 (2006), with a total number of 28,296 and 19,309 respondents respectively, showing a global attrition rate of 68.23%. Eleven European countries participated in both W1 and W2 (Austria, Belgium, Denmark, France, Germany, Greece, Italy, the Netherlands, Spain, Sweden, and Switzerland) and the total number of respondents was 5403. The data permit the identification of labor-force participation in 2004, transitions from work to retirement between 2004 and 2006, and measures related to individual and contextual variables associated with these work-to-retirement transitions (see below). Data from each wave were merged, and the participants for this study were selected in two steps. First, we included the respondents who were receiving a pension income in 2004 and who continued to work at that time (N = 1189) and, second, we selected only those who also participated in W2 (N = 634).

Measures

T1 antecedents

We used the following variables measured at W1.

Health: respondents were asked to rank their health on a five-point scale ranging from 1 (excellent) to 5 (poor).

Optimism: we used the Life Orientation Test (LOT; Scheier and Carver, 1985), with five items and α = .77 (ex. ‘I pursue my goals with lots of energy’; ‘In uncertain times, I usually expect the best’).

Quality of life: the Control Autonomy Self Realization Pleasure-12 Scale (CASP-12) (Higgs et al., 2003) revealed a two-factor structure (Autonomy and Pleasure/Control) in Wave 1, both with acceptable reliability levels (α = .61 and α = .76, respectively). (E.g. ‘How often do you think your age prevents you from doing the things you would like to do?’ – autonomy, reversed scores – or ‘How often do you feel that your life has meaning?’ (pleasure/control factor). Responses were on a four-point scale, ranging from 1 (often) to 4 (never).

Life satisfaction has been measured with one item: ‘How satisfied are you with your life in general?’ Response scale ranged from 1 (very satisfied) to 4 (very dissatisfied).

Job satisfaction: we used one item both in W1 and in W2: ‘All things considered I am satisfied with my job. Would you say you strongly agree, agree, disagree or strongly disagree?’

T2-BE quality: was assessed according to the Effort Reward Imbalance model (Siegrist et al., 2004) using four items from the reward dimension, which consist in four reward measures (support, recognition, earnings, advancement possibilities) with acceptable reliability (α = .70) (e.g. ‘I receive the recognition I deserve for my work’, ‘Considering all my efforts and achievements, my salary is adequate’).

T2 consequences

We used the following variables measured at W2.

QoLT2: two factors (autonomy and pleasure/control) from the CASP-12 scale (Higgs et al., 2003) were used in Wave 2, both with acceptable reliability levels (α= .60 and α = .75, respectively).

Life satisfaction was measured with one item: ‘On a scale from 0 to 10, where 0 means completely dissatisfied and 10 means completely satisfied, how satisfied are you with your life?’

Analytic strategy

Structural equation modeling (SEM) was used to simultaneously examine the relationships between the antecedents, BE quality, and consequences. We specified T1 antecedents as exogenous predictor variables, BE quality as the mediator, and T2 consequences were specified as outcome variables. For the sake of clarity, neither the regression errors nor the corresponding observable variables are shown in Figures 1 and 2. To determine the fit of the proposed model, we report a number of absolute and relative fit indices. Adopting the approach suggested by Singh et al. (1994), we checked for the presence of a mediating effect by performing a competing model analysis such that two substantive models are estimated and evaluated for significant differences. The first model, referred as the direct effects model, examined the direct relationships between T1 antecedents and T2 consequences. The second model, the partial mediation model, examined the same relationships with BE quality acting as a mediator. The hypothesized mediating effect of BE quality is supported if, compared with the direct effects model, the partial mediation model shows: (a) higher percentage of explained variance in T2 consequences, (b) a significant relationship between T1 antecedents and BE quality, (c) substantially diminished or insignificant effects of T1 antecedents on T2 consequences, and (d) a significant relationship between BE quality and T2 consequences.

Results

The final sample of this study includes all those participants who received a pension from 2004 and who continued working in 2006. Mean age of participants ranged from 61.04 in the Netherlands to 61.8 in Germany, including in this range Austria. The rest of the countries’ mean age ranged from 62.2 for France to 62.9 for Spain, including Italy and Switzerland in this group. Finally, Greece, Belgium, and Denmark exhibited values around 63 years and Sweden exceeded 64.1 years of mean age. These participants worked a mean of 18.5 hours per week, showing a mean level of 10.2 years of tenure (SD = 9.2) at W1.

Years of education ranged from comprehensive school to multiple graduate degrees at the doctoral level and beyond. The mean of years of formal education was 11.09 (SD = 4.44).

Other relevant demographic characteristics are detailed in Table 1.

Table 1.

Demographic characteristics of participants (N = 634).

Characteristics	Percentage (missing values in brackets when available)
Gender
Males	38
Females	35.6 (26.3)
Type of pension received in 2004
Public old-age pension or a private occupational pension	49.8
Early retirement public or private pension	11
Disability insurance public or private pension	7.8
Public survivor pension from spouse or partner’s job or public war pensions	4.4
Type of pension received in 2006
Public old-age pension or a private occupational pension	49.8
Supplementary public pension	6.8
Early retirement public or private pension	7.3
Disability insurance public or private pension	8.5
Public survivor pension from spouse or partner’s job or public war pensions	4.4
BE in public/private sector
Public	18.3
Private	43.8 (37.9)
Main reason for retirement
Eligible for public or private pension	10.8
Offer of early retirement	1.3
Made redundant	0.8
Health	0.9
Same time as spouse	0.2
More time with family	0.6
Enjoy life	0.5
Type of job
Professional (managers, technicians, associate professionals)	50.1
Others (market sales workers, skilled agricultural or fishery workers, or craft and related trades workers)	31.2
Elementary occupations	2.2
Activity sector
Educational, health or social service	52
Agricultural, mining, manufacturing, and construction	21.4
Hotels, financial, transportation, and business activities, and social security system	26.5

Means and standard deviations of T1 antecedents, BE quality, and T2 consequences are presented in Table 2, which also depicts correlations among variables. We found that BE quality correlated significantly both with T1 antecedents and T2 consequences, except for health.

Table 2.

Descriptive statistics and correlation matrix.

Variables	M	SD	1	2	3	4	5	6	7	8	9	10	11
1. Age (2003)	62.75	6.86	–
2. Health T1	2.62	1.05	−.05	–
3. Optimism T1	2.18	.76	−.02	.22**	–
4. QoL T1 (autonomy)	2.09	.57	−.06	.22**	.41**	–
5. QoL T1 (pleasure/control)	1.49	.44	−.02	.29**	.38**	.44**	–
6. Job satisfaction T1	1.52	.67	−.18**	.09	.26**	.24**	.30**	–
7. Life satisfaction T1	1.58	.60	−.11**	.31**	.38**	.42**	.48**	.31**	–
8. BE quality T2	2.13	.67	−.25**	.08	.20**	.20**	.24**	.26**	.13*	–
9. QoL T2 (autonomy)	2.04	.59	−.06	.11**	.30**	.47**	.41**	.26**	.31**	.22**	–
10. QoL T2 (pleasure/control)	1.58	.49	−.02	.22**	.30**	.31**	.56**	.27**	.40**	.23**	.43**	–
11. Job satisfaction T2	1.47	.61	−.25**	.00	.24**	.10	.25**	.41**	.17**	.33**	.22**	.32**	–
12. Life satisfaction T2	3.02	1.47	−.07	.18**	.27**	.37**	.45**	.24**	.37**	.21**	.44**	.50**	.30**

Notes: N = 634. * p < .05; **p < .01.

Table 3 and Figure 2 show the results of the competing model analyses. Firstly, in terms of fit statistics, both models appear to be reasonable fits to the data (Direct effects model: χ² / (d.f.) = 79.7 (22), CMIN/d.f. = 3.62, NFI = .94, IFI = .95, CFI = .95, RMSEA = .06, ECVI = .30, MECVI = .30. Partial mediation model: χ² / (d.f.) = 63.3 (22), CMIN/d.f. = 2.88, NFI = .95, IFI = .97, CFI = .97, RMSEA = .05, ECVI = .32, MECVI = .31). although the χ² statistic for each model was significant, problems with this test statistic for large sample sizes (N > 300) are well documented. Other relevant fit indexes indicate good overall fit, as NFI, IFI, and CFI exceed .90 and residuals (i.e. RMSEA) tend to approach .05. Second, the partial mediation model explained more variances in T2 consequences than the direct effects model (.22 vs .15, for T2 job satisfaction; .20 vs .17, for T2 life satisfaction; .25 vs .20, for T2 QoL autonomy, and .32 vs .28, for T2 QoL pleasure/control). Third, positive relationships were found between T1 antecedents and BE quality, on the one hand, and BE quality and T2 consequences, on the other. And lastly, the partial mediation model significantly attenuated the effects of T1 antecedents on T2 consequences. Taken together, these four points suggested that the partial mediation model was a significant improvement over the direct effects model and supports the partial mediating role of BE quality.

Table 3.

Parameter estimates and determination coefficients for direct effects and partial mediation effects models.

Standardized regression coefficients	Direct model	Partial mediation model
Age → job satisfaction T2	−.15**	−.12*
Health T1 → life satisfaction T2	.09*	.10*
Health T1 → QoL (pleasure/control) T2	.07 n.s.	.09 n.s.
Optimism T1 → QoL (autonomy) T2	.11*	.09*
QoL (autonomy) T1 → life satisfaction T2	.24***	.21***
QoL (autonomy) T1 → QoL (autonomy) T2	.45***	.37***
QoL (pleasure/control) T1 → QoL (pleasure/control) T2	.36***	.35***
Job satisfaction T1 → job satisfaction T2	.33***	.29***
Life satisfaction T1 → life satisfaction T2	.29***	.20***
Life satisfaction T1 → QoL (pleasure/control) T2	.18***	.18***
Per cent variance explained
Job satisfaction T2	.04	.22
Life satisfaction T2	.13	.20
QoL (autonomy) T2	.02	.25
QoL (pleasure/control) T2	.11	.32

p < .05; **p < .01; ***p < .001.

Figure 2.

Standardized estimates for the partial mediation effects model.

The examination of the standardized estimates in the partial mediation model allows us to state that T1 antecedents have a significant impact on BE quality, except for health, optimism, and life satisfaction, providing partial support for H1 to H4. BE quality had a significant and positive effect on T2 outcomes, which ranged from .25 for job satisfaction to .16 for life satisfaction and QoL (pleasure/control), providing full support for H5 to H7. Finally, the partial unmediated effects of T1 antecedents on the T2 consequences reduced their insignificance in the partial mediation model, supporting H8. However, the unmediated effect of health on outcomes was positive, and the standardized coefficient increased compared to the direct effects model, contrary to our hypothesis. Notwithstanding this datum, the path from health T1 to QoL at T2 (pleasure/control) did not reach statistical significance.

Discussion

This work had two goals. First, to test a model of antecedents and consequences of BE using longitudinal data preceding from Waves 1 and 2 of SHARE. Second, to contribute to the literature by including specific job-related psychosocial variables to predict its impact on workers’ personal well-being. In general, we can state that both goals were achieved. Specifically, we contend that the availability of longitudinal data increases the feasibility of causal inferences. On the basis of these data, it was shown that personal characteristics (age, health, and optimism), QoL, life satisfaction, and job satisfaction assessed in 2004 predicted BE quality in 2006. It was also shown that BE quality predicted the consequences, such as life and job satisfaction and QoL, in 2006.

BE antecedents and consequences

The model tested reflects and confirms previous findings about BE (Beehr and Bennett, 2007; Davis, 2003; Feldman, 1994; Feldman and Kim, 2003; Gobeski and Beehr, 2009; Kim and Feldman, 2000). First, among personal characteristics, the most efficient predictors of BE quality are age, in the negative sense, and health, in the positive sense. Specifically, this result also confirms prior studies that indicate the higher probability of becoming involved in BE among younger people who enjoy better health (Wang et al., 2008; Weckerle and Shultz, 1999). Thus, we must consider that demographic characteristics act as proxies for other variables. Although using biographical variables in this manner is appropriate, it is important for researchers to be mindful of the different reasons these variables might be related to job seeking across different populations (Adams and Rau, 2004: 735). Despite this fact, recent research continues to show some recurrent patterns. For instance, among the major health conditions, participants who suffer lung disease are more likely to engage in BE in a different field than in career BE. At the same time, among the minor health conditions, participants who suffer from arthritis or diabetes are also more likely to choose BE in a different field (Shultz and Wang, 2007).

Second, we added specific job-related psychosocial variables. Among these antecedents, job satisfaction and QoL are the best predictors of BE in the positive sense. These results confirm other longitudinal analyses. For instance, using data from the HRS, Wang and his colleagues found a significant relationship between job satisfaction and engaging in a BE career (Wang et al., 2008). Detailed comparisons of empirical studies based on SHARE and the HRS are not possible because these studies pursued different objectives. However, our results suggest that extending the array of specific employment variables can further our understanding of the factors that motivate people to engage in BE. Such a position is in line with previous work focusing on the most attractive characteristics of organizations or on their personnel recruitment policies (Rau and Adams, 2005).

Third, the theoretical model has been extended with regard to previous studies to include the construct of QoL as a desirable outcome, in its facets of autonomy and control/pleasure. Although the nature of the transition to retirement, gradual or abrupt, had not showed its effects on happiness in retirement in some studies, we founded that the influence of BE on QoL is important, confirming previous findings from other extensive studies (Wang, 2007; Wang et al., 2008). These findings reveal that there are various patterns to withdraw from the workforce and not everyone feels satisfied with a brusque withdrawal at the age established by law.

Fourth, as expected, some direct relationships between antecedents at T1 and consequences at T2 are significant. Specifically, all the psychosocial variables assessed at T2 are significantly affected by the same variables that were assessed two years previously. In addition, age affects job satisfaction two years later, although its impact is lower than that of job satisfaction assessed at T1. The estimated standardized coefficient has a negative sign, but the way job satisfaction was assessed should be considered, because higher scores indicate lower satisfaction. Thus, the results are coherent with the first meta-analyses on the relations between age and global job satisfaction (Rhodes, 1983), although the debate continues. On the one hand, to consider retirement as a socially acceptable role that is economically feasible at old age can influence the global assessment of one’s own work. Thus, job satisfaction can also be affected by the attitudes toward full retirement. However, job satisfaction also reveals the impact of the older person’s status within the organization and of other factors that may lead older people to reappraise what they expect from their job.

Lastly, health has a direct impact on job satisfaction at T2, which remains statistically significant and even increases its value in the partial effects model. This finding is consistent with previous research. Empirical studies have constantly shown a strong link between workers’ health and their satisfaction and adjustment after retirement (Alcover et al., 2012; Pinquart and Schindler, 2007; Taylor and Shore, 1995). But the relationships between health and satisfaction would be complex, such as health simultaneously affects satisfaction both directly and indirectly. First, because people with better health are happier and can carry out more activities; but also because health allows access to more resources, such as social support networks, which, in turn, contribute to adaptation and satisfaction.

Theoretical and practical implications

Theoretically, first, BE before definite retirement favors semi-retired workers’ personal satisfaction and job satisfaction and their perception of adequately providing for their psychosocial needs in old age. But more important is that this influence of BE is not exclusively a function of the quantity of hours worked, or of the amount of salary. In contrast, the key factor is the perceived reward and the adequate balance between investment and retribution (Siegrist, 1996). This finding again reveals the stressful role of work relationships that lack balance and the importance of the adequacy of the rewards.

Second, extending the preceding models, we operationalized BE according to the recommendations of Kim and Feldman (2000), but these authors formerly stated that a more refined operationalization of BE than the simple consideration of the number of hours worked would be valuable. This assessment of the influence of BE may better account for the underlying mechanisms through which BE affects the results. Specifically, constructs such as job or career embeddedness (Feldman, 2007) could play a mediating role in the relationships between BE quality and long-term results. In the same vein, Raymo et al. (2010) found that several aspects of mid-life work experiences are associated with the likelihood of realizing one’s preferred work status at prime retirement ages.

Third, in our study we included all the consequences of BE in the same step, without considering the possibility that BE quality affects QoL, which in turn affects job satisfaction and life satisfaction. In this sense, it has been recommended that a dynamic perspective of BE would allow us to better understand that BE is a longitudinal workforce participation process (Wang and Shultz, 2010).

Practically, considering the complex nature of BE quality and its impact on personal well-being, we make some recommendations. First, we must consider Feldman’s (2007) suggestion of the importance of older workers’ moving out of undesirable employment. Thus, there is some possibility that older workers who are no longer satisfied with their jobs may be substituting moves into retirement with BE (Ng and Feldman, 2009). Related to this point, our results show a strong impact of T1 job satisfaction on BE quality. Therefore, low quality employment not only impacts on current outcomes, but also threatens older workers’ future possibilities of BE by reducing their positive attitudes toward work after retirement.

Second, variability in retirement wealth implies that many employees may need to engage in some form of BE to maintain adequate income. Therefore, financial planning is a critical step which would promote engagement in BE.

Lastly, our results highlight one of the current sociopolitical concerns: population ageing is one of the most urgent challenges currently facing developed countries (United Nations, 2011). Due to advances in medicine and care and the improvement of general living conditions, actual and future generations of seniors will be healthier and more competent and will have a higher life expectancy (Šimová, 2010). This trend calls for a later retirement age, although the balance between time spent working (and contributing to the social security system) and time spent in retirement is maintained approximately constant (Engelhardt, 2012). The active ageing concept captures these trends. The core idea is that societies should provide aged people with support, life-long learning, and opportunities to ensure their full participation in productive and social life based on their own choice and individual potential, thorough political actions and laws (Šimová, 2010). Older people should be able to fill their lives with activities beneficial to society as a whole, either continuing their work life appropriately – and BE is a key tool for this – or by engaging in other social and community-beneficial activities, such as volunteering, care activities, and socially productive activities, particularly those that offer opportunities of reciprocal exchanges (Siegrist and Wahrendorf, 2009). Moreover, as age-related morbidity declines, the strain on pension systems and related welfare programs has increased drastically (Börsch-Supan et al., 2009). In this regard, active ageing policies and practices attempt to prevent the negative social impact of a changing age structure of the population in developed countries by mitigating the burden of the social security and healthcare systems. As an anonymous reviewer suggested, some people might be working longer because they cannot afford to retire, or that their pension and partner’s circumstances might have been affected by the economic crisis. But, in sum, as older workers and countries as whole will face financial strains in retirement income in the years ahead, BE arrangements (wage-and-salary jobs or self-employment) may be a vital part of the pro-work solution (Giandrea et al., 2008), and to maintain the welfare state for future generations. In addition, BE aims to contribute to improving prospective QoL and well-being of aged people, as well as organizational effectiveness and productivity. Future research and empirical evidence of all these aspects would allow us to better know the form and timing of retirement processes, the labor market for older workers, the viability and quality of pension systems and the experience of retired life (Ekerdt, 2010), and the growing diversity and complexity of these life stages. And all of this will be very relevant both for researchers and policy makers.

Limitations of this work

This work has many methodological limitations. We used a longitudinal sample with participants from 11 European countries, obtained by rigorous procedures within the SHARE study, which allows us to analyze the dynamic nature of the phenomenon. This allowed access to information provided by many participants, and not limited only to workers from a few organizations or from a labor market or a restricted geographical setting. Size and heterogeneity of the sample averted the impact of variables of a specific organization or region on the results, and they compensated for the loss of information required by the need to reduce the questionnaire. However, although the participants in the SHARE study were selected by random sampling procedures, the results of the present work must be interpreted with caution because of the attrition problem. When, as in this case, sample selection is based on some variable of interest because of the specific goal of the study, the results could be biased. As a counterpart to these difficulties, it would not have been possible to obtain the sample characteristics without the resources and the research team in charge of SHARE. In the same sense, a comparative perspective across countries would be of interest, given the welfare regime variations and differences in pension entitlements and values across the countries studied and the possible consequences for decisions relating to continued participation in the labor market. Unfortunately, the sample size was not big enough to allow for national-level analyses.

Summing up, the present work makes an initial contribution to the theory and the empirical research on antecedents and consequences of BE. This was carried out with a European longitudinal sample and the work extends the model to include important job-related psychosocial variables to further our understanding of BE and its influence on people’s well-being.

Footnotes

Funding

This article was funded as part of the actions to aid the dissemination of the Research Promotion Plan of the Spanish University for Distance Education (UNED).

Author biographies

Gabriela Topa is a senior lecturer in social psychology at the Spanish University for Distance Education (UNED). Her research has been focused on retirement, bridge employment, and early retirement. She has developed both empirical and meta-analytical research. Her work has been published in Journal of Economic Psychology, Journal of Vocational Behavior, and Research on Aging, among others.

Carlos-María Alcover is a senior lecturer in social psychology at the Rey Juan Carlos University (Madrid, Spain). His research has been focused on the psychological contract and exchange relationships, bridge employment, early retirement and psychological well-being, and psychosocial risks in workers with disabilities. His work has been published in European Psychologist, Journal of Vocational Behavior, Economic and Industrial Democracy, Research on Aging, and Work and Stress, among others.

Juan Moriano is associate professor at the Department of Social and Organizational Psychology at the Spanish University for Distance Education (UNED). His research interests focus on vocational behavior and career development across the life span, including the changes and decisions that people make from work entry to retirement.

Marco Depolo is full professor of work and organizational psychology at the Department of Psychology at the University of Bologna (Italy). His main research interests are on well-being at work, work-related stress, and factors (both individual and organizational) related to retirement.

References

Adams

Rau

(2004) Job seeking among retirees seeking bridge employment. Personnel Psychology 57: 719–744.

Alcover

Crego

Guglielmi

. (2012) Comparison between the Spanish and Italian early work retirement models: A cluster analysis approach. Personnel Review 41: 380–403.

Alley

Crimmins

(2007) The demography of aging and work. In: Shultz

Adams

(eds) Aging and Work in the 21st Century. Mahwah, NJ: Lawrence Erlbaum, pp. 7–23.

Atchley

(1989) A continuity theory of ageing. The Gerontologist 29: 183–190.

Beehr

Bennett

(2007) Examining retirement from a multi-level perspective. In: Shultz

Adams

(eds) Aging and Work in the 21st Century. Mahwah, NJ: Lawrence Erlbaum, pp. 277–302.

Beehr

Bennett

(2008) Unemployment and retirement. In: Barling

Cooper

(eds) The Sage Handbook of Organizational Behavior Vol. 1: Micro Approaches. London: Sage, pp. 217–232.

Beehr

Bowling

(2002) Career issues facing older workers. In: Feldman

(ed.) Work Careers: A Developmental Perspective. San Francisco: Jossey-Bass, pp. 214–241.

Blanchet

Debrand

(2007) Aspiration à la retraite, santé et satisfaction au travail: une comparaison européenne [Retirement planning, health and job satisfaction: A European comparison]. IRDES. Working paper No. 1. Available at: www.share-project.org/ (accessed 25 May 2008).

Börsch-Supan

Jürges

(eds) (2005) The Survey of Health, Ageing, and Retirement in Europe. Methodology. Mannheim: Mannheim Research Institute for the Economics of Ageing.

10.

Börsch-Supan

Brugiavini

Jürges

. (eds) (2005) Health, Ageing and Retirement in Europe: First Results from the Survey of Health, Ageing and Retirement in Europe. Mannheim: Mannheim Research Institute for the Economics of Ageing.

11.

Börsch-Supan

Brugiavini

Croda

(2009) The role of institutions and health in European patterns of work and retirement. Journal of European Social Policy 19: 341–358.

12.

Cahill

Giandrea

Quinn

(2011) Reentering the labor force after retirement. Monthly Labor Review June: 34–42.

13.

Davis

(2003) Factors related to bridge employment participation among private sector early retirees. Journal of Vocational Behavior 63: 55–71.

14.

Debrand

Lengagne

(2007) Pénibilité au travil et santé des seniors en Europe [Working conditions and health of the elderly in Europe]. Économie et Statistique 403–404: 19–38.

15.

De Lucca

Lipps

(2005) Fieldwork and survey management in SHARE. In: Börsch-Supan

Jürges

(eds) The Survey of Health, Ageing, and Retirement in Europe: Methodology. Mannheim: Mannheim Research Institute for the Economics of Ageing.

16.

Dendinger

Adams

Jacobson

(2005) Reasons for working and their relationship to retirement attitudes, job satisfaction and occupational self-efficacy of bridge employees. International Journal of Aging and Human Development 61: 21–35.

17.

Ekerdt

(2010) Frontiers of research on work and retirement. Journal of Gerontology: Social Scienes 65B: 69–80.

18.

Engelhardt

(2012) Late careers in Europe: Effects of individual and institutional factors. European Sociological Review 28: 550–563.

19.

Feldman

(1994) The decision to retire early. Academy of Management Review 19: 285–311.

20.

Feldman

(2007) Career mobility and career stability among older workers. In: Shultz

Adams

(eds) Ageing and Work in the 21st Century. Mahwah, NJ: Lawrence Erlbaum, pp. 179–198.

21.

Feldman

Kim

(2003) Bridge employment during retirement: A field study of individual and organizational experiences with post-retirement employment. Human Resource Planning 23: 14–25.

22.

Giandrea

Cahill

Quinn

(2008) Self-employment Transitions Among Older American Workers with Career Jobs. Bureau of Labor Statistics Working Papers, No. 418. Washington, DC: U.S. Department of Labor.

23.

Gobeski

Beehr

(2009) How retirees work: Predictors of different types of bridge employment. Journal of Organizational Behavior 30: 401–425.

24.

Higgs

Hyde

Wiggins

. (2003) Researching quality of life in early old age: The importance of the sociological dimension. Social Policy and Administration 37: 239–252.

25.

Kim

Feldman

(2000) Working in retirement: The antecedents of bridge employment and its consequences for quality of life in retirement. Academy of Management Journal 43: 1191–1210.

26.

Kooij

De Lange

Jansen

. (2011) Age and work-related motives: Results of a meta-analysis. Journal of Organizational Behavior 32: 197–225.

27.

Marmot

Siegrist

Theorell

(2006) Health and the psychosocial environment at work. In: Marmot

Wilkinson

(eds) Social Determinants of Health. Oxford: Oxford University Press, pp. 97–130.

28.

Munnell

Sass

(2008) Working Longer: The Solution to the Retirement Income Challenge. Washington, DC: Brookings Institution Press.

29.

Feldman

(2009) Re-examining the relationship between age and voluntary turnover. Journal of Vocational Behavior 74: 283–294.

30.

Pengcharoen

Shultz

(2010) The influences on bridge employment decisions. International Journal of Manpower 31(3): 322–336.

31.

Pinquart

Schindler

(2007) Changes of life satisfaction in the transition to retirement: A latent-class approach. Psychology and Aging 22: 442–455.

32.

Quinn

(1999) Retirement Patterns and Bridge Jobs in the 1990s. EBRI Issue Brief 206 (February). Washington, DC: Employee Benefit Research Institute, pp. 1–23.

33.

Rau

Adams

(2005) Attracting retirees to apply: Desired organizational characteristics of bridge employment. Journal of Organizational Behavior 26: 649–660.

34.

Raymo

Warren

Sweeney

. (2010) Later-life employment preferences and outcomes: The role of midlife work experiences. Research on Aging 32: 419–466.

35.

Rhodes

(1983) Age-related differences in work attitudes and behavior: A review and conceptual analysis. Psychological Bulletin 93: 328–367.

36.

Schalk

(2010) Matching individual and organization needs to enable longer working lives. In: European Centre for the Development of Vocational Training (CEDEFOP), Working and Ageing: Emerging Theories and Empirical Perspectives. Luxembourg: Publications Office of the European Union, pp. 114–131.

37.

Scheier

Carver

(1985) Optimism, coping, and health: Assessment and implications of generalized outcome expectancies. Health Psychology 4: 219–247.

38.

Schmitt

Coyle

Rauschenberger

. (1979) Comparison of early retirees and nonretirees. Personnel Psychology 32: 327–340.

39.

Shultz

(2003) Bridge employment: Work after retirement. In: Adams

Beehr

(eds) Retirement: Reasons, Processes, and Results. New York: Springer, pp. 215–241.

40.

Shultz

Wang

(2007) The influence of specific physical health conditions on retirement decisions. International Journal of Ageing and Human Development 65: 149–161.

41.

Siegrist

(1996) Adverse health effects of high-effort/low-reward conditions. Journal of Occupational Health Psychology 1: 27–41.

42.

Siegrist

Wahrendorf

(2009) Participation in socially productive activities and quality of life in early old age: Findings from SHARE. Journal of European Social Policy 19: 317–326.

43.

Siegrist

Starke

Chandola

. (2004) The measurement of effort-reward imbalance at work: European comparisons. Social Science and Medicine 58: 1483–1499.

44.

Šimová

(2010) To work or not to work: Motivation for work after reaching retirement age. In: European Centre for the Development of Vocational Training (CEDEFOP), Working and Ageing: Emerging Theories and Empirical Perspectives. Luxembourg: Publications Office of the European Union, pp. 169–189.

45.

Singh

Goolsby

Rhoads

(1994) Behavioral and psychological consequences of boundary spanning burnout for customer service representatives. Journal of Marketing Research 31: 558–569.

46.

Taylor

Shore

(1995) Predictors of planned retirement age: An application of Beehr’s model. Psychology and Aging 10: 76–83.

47.

Topa

Depolo

Moriano

. (2009) Bridge employment and retirees’ personal well-being: A structural equation model with a European probabilistic sample. Psicothema 21: 280–287.

48.

Topa

Moriano

Depolo

. (2011) Retirement and wealth relationships: Meta-analysis and SEM. Research on Aging 33: 501–528.

49.

Ulrich

Brott

(2005) Older workers and bridge employment: Redefining retirement. Journal of Employment Counselling 42: 159–170.

50.

United Nations (2011) World Population Prospects: The 2010 Revision. Highlights and Advanced Tables. Working Paper No. ESA/P/WP.220. New York: UN, Department of Economic and Social Affairs, Population Division.

51.

Von Bonsdorff

Shultz

Leskinen

. (2009) The choice between retirement and bridge employment: A continuity theory and life course perspective. International Journal of Aging and Human Development 69: 79–100.

52.

Wang

(2007) Profiling retirees in the retirement transition and adjustment process: Examining the longitudinal change patterns of retirees’ psychological well-being. Journal of Applied Psychology 92: 455–474.

53.

Wang

Shultz

(2010) Employee retirement: A review and recommendations for future investigation. Journal of Management 36: 172–206.

54.

Wang

Zhan

Liu

. (2008) Antecedents of bridge employment: A longitudinal investigation. Journal of Applied Psychology 93: 818–830.

55.

Warr

Cooke

Wall

(1979) Scales for the measurement of some work attitudes and aspects of psychological well being. Journal of Occupational Psychology 52: 129–148.

56.

Weckerle

Shultz

(1999) Influences on the bridge employment decision among older USA workers. Journal of Occupational and Organizational Psychology 72: 317–329.

57.

Zaniboni

Sarchielli

Fraccaroli

(2010) How are psychosocial factors related to retirement intentions? International Journal of Manpower 31(3): 271–285.