ARTÍCULO

 

Early retirement: positive or negative for well being?

¿Es la jubilación anticipada positiva o negativa para el bienestar?

 

 

Kerstin Isaksson¹

Gunn Johansson²

¹Mälardalen University, School of Sustainable Development of Society and Technology, Västeras, Sweden.

²Department of Psychology, Stockholm University, Sweden.

 

 

ABSTRACT

Is the transition from work to early retirement generally a cause for distress? This question has been addressed in several research studies over many decades. It was also the focus of a longitudinal study of individuals of 55 years and older who retired as part of a downsizing program in a Swedish insurance company. The consequences in terms of health and well being during the first two post-retirement years were evaluated using two different approaches to data analysis. This paper brings together results from two different reports (Isaksson, 1997; Isaksson and Johansson, 2000) aiming to provide a concluding picture and to relate to recent research. Using longitudinal questionnaire data, the study compared early retirees and persons continuing to work over the years following downsizing.
The effects of voluntary/forced choice, employment status and gender on the subsequent adaptation of older individuals were evaluated. The results revealed no signs of a general retirement crisis. More importantly, voluntary (as opposed to a forced choice) choice of employment or retirement was directly and positively associated with satisfaction and psychological well being for both groups. Women showed lower values of work centrality, appeared to be more inclined to apply for retirement and were generally more satisfied with the outcome than men. Health problems were significantly lower 1.5 years after retirement but no similar effect was found among stayers.
Furthermore, an attempt was made to identify patterns of adjustment to early retirement by means of cluster analysis. Four stable patterns were described: continuous working, positive adaptation to retirement, one small group with signs of high distress and finally a group with improved health at T2.

RESUMEN

¿Es motivo de angustia la transición del trabajo a la jubilación anticipada? La investigación se ha planteado esta pregunta durante muchos decenios. En ella se centró también el estudio longitudinal con personas de 55 años o más que se jubilaban en el contexto de un programa de regulación de empleo de una empresa de seguros sueca. Se evaluaron las consecuencias tanto para la salud como para el bienestar durante los dos primeros años tras la jubilación mediante el uso de dos tipos diferentes de análisis de datos. Este artículo reúne los resultados de dos informes (Isaksson, 1997; Isaksson and Johansson, 2000) que pretendían obtener una visión concluyente, vinculados a la investigación reciente. Utilizando cuestionarios longitudinales, el estudio compara los datos de las personas jubiladas anticipadamente con las de aquellas que siguen trabajando durante los años posteriores a la regulación laboral.
Se evaluaron los efectos de la elección voluntaria/forzosa, la categoría laboral y el sexo sobre la adaptación de las personas de más edad. Los resultados no mostraban signos de una crisis general debida a la jubilación y, lo que es más importante, la elección voluntaria (frente a la elección forzosa) de seguir trabajando o de jubilarse guardaba una relación directa positiva con la satisfacción y el bienestar psicológico en ambos grupos. Las mujeres manifestaban valores más bajos en centralidad del trabajo, parecían más proclives a solicitar la jubilación y estaban por lo general más satisfechas del resultado que los hombres. Los problemas de salud eran significativamente menores al año y medio de la jubilación, efecto que no se apreció en los que continuaron trabajando.
Además, el intento de hallar patrones de ajuste a la jubilación anticipada mediante análisis de clusters, dio cuatro tipos de patrones estables: seguir trabajando, adaptación positiva a la jubilación, un pequeño grupo que tenía signos de angustia elevada y por último un grupo cuya salud había mejorado en el T2.

Key words

Early retirement, Adjustment over time, Gender.

Palabras clave

Jubilación anticipada, Adaptación con el tiempo, Género.

 

 

The transition from work to early retirement was the focus of a longitudinal study, which followed individuals who retired as part of a downsizing program in a Swedish insurance company. This was an unusual study for several reasons: First, the study managed to investigate employees who were offered the chance to apply for early retirement. These individuals could be followed over time, which gave longitudinal data, to be analysed in order to describe the change that took place over time in the adaptation of three subgroups differing in their preferences for retirement, that is, whether they wanted to retire or to continue working. Furthermore, the studied group consisted of almost equal proportions of men and women working on the same work tasks and with very similar occupations. This paper provides a summary of the results with the general aim of describing adaptation to early retirement (Isaksson, 1997; Isaksson & Johansson, 2000¹).

The study was performed in a Swedish insurance company employing about 4,000 individuals. As part of the downsizing process, all personnel of 55 years and older were invited to apply for early retirement, with a retention of 80% of their salary until the regular age for state pension (65 years). Following individual application, local branch managers decided whether to grant or deny retirement to their own subordinates. Since it was required that each case of retirement should effect a cost reduction, it could only be granted in cases where no replacement was necessary or when replacement could be obtained through in-house recruitment. On occasions, this strategy led to perceived injustice, either because of relatively subtle pushes towards early retirement or because people were denied retirement due to their specialized skills.

The three sub-groups—voluntary retirees, involuntary retirees and employees— who were denied retirement, were presumed to be fairly equal in terms of health status at the outset. The reason for this presumption was that employees with poor health had been encouraged to apply for a disability pension rather than for early retirement. We have no data about the health situation before the downsizing process but our first wave of questionnaire data collected during the process can at least provide some indication.

At this time, a very large majority, 87% of the whole group (aged 55+), regarded themselves as generally healthy. There were no significant differences in health complaints between the three subgroups. Among the voluntary retirees, the proportion reporting to be in good health was 90% compared to 81% for both involuntary retirees and those rejected. The same tendency was found among several health indicators but differences between groups were small and not significant. Although these results do not constitute proof, they do show that it is highly improbable that a negative selection to retirement based on health status had actually taken place.

The retirement offer was considered generous and attractive and by the end of the process, 55% of the entire staff above the age of 55 had been granted early retirement. They left the company during one year—20% during spring, 40% during the summer and the remaining 40% during autumn (the final group in December 1992).

Adaptation to early retirement

The decision to retire has become a more complex individual process as a result of societal changes in the labour force participation and the increasingly common practice of offering early retirement during downsizing to persons far below the age of the old age pension.There are an increasing number of groups of redundant people above the age of 55 who have to make the decision of whether or not they are willing to accept an offer of early retirement. For these relatively young retirees, the decision to retire is not necessarily a decision to stop working, but rather a decision to slow down and reduce the time and effort put into work, which could well involve a decision to seek parttime employment (Feldman, 1994). There is both theoretical and empirical evidence of continuity across the life span and across the retirement transition (Atchley, 1979; George and Maddox 1977). For a majority of retirees, previous lifestyle patterns as well as previous levels of selfesteem are maintained during retirement, and the transition to retirement appears to be fairly easy and is of no cause for distress. A recently presented study by Pinquart and Schindler (2007) used a latent class approach, analysing data in a large cohort over more than 15 years with a focus on the satisfaction before, during and after retirement. They identified three groups of people who had different experiences of retiring. For the largest group, satisfaction showed a temporarily small increase in retirement. A smaller group had a rather large increase but overall declining satisfaction with age. The third group demonstrated a decline in satisfaction for retirement but continued on a stable or increasing trajectory. A conclusion from this and other studies (see Kim and Moen, 2001) is that retirement tends to have a minimal impact on mental health and well being except among individuals with poor personal resources (health, financial concerns, few social contacts and marital quality).

A study by Palmore and co-workers (1984) is one of the few earlier investigations that specifically discuss problems associated with early retirement (that is, retirement before the regular age of the state pension) as compared with old-age retirement. The authors found early retirement to be associated with less satisfaction than late retirement and interpreted their findings as reflecting an over-representation of involuntary retirement and health problems among early retirees. Other studies (e.g. McGouldrick, 1989; McGouldrick & Cooper, 1994; Williamson, Rinehart & Blank, 1992) have shown generally higher levels of satisfaction among younger retirees. The latter studies emphasize the importance of a voluntary choice with regard to employment status in old age.

Feldman (1994) has pointed out the need for a better understanding of the links between early-retirement decisionsand later adjustment to retirement. Attention needs to be given to the decision to retire, which is influenced not only by individual factors (such as health and income) but also by organizational factors within a company. In a similar way, the decision not to retire but to continue working for a longer period of time can be perceived as more or less voluntary. Increased knowledge in this area is relevant both to companies planning earlyretirement schemes and to individuals faced with an offer to apply for early retirement.

A study by Herzog, House and Morgan (1991) also revealed the effects of an involuntary change of employment status. Individuals whose pattern of labour-force participation (or non-participation) at older ages (55+) was in concordance with their own preferences reported higher levels of both physical and psychological well being, than those whose participation was constrained by other factors. The finding was independent of both gender and occupation.

Research reported by Richardson and Kilty (1991) points to a relative lack of studies investigating short-term adaptation. In their study of the first post-retirement year, they found a variety of adaptation patterns—some stable and positive, others reflecting either a decline or an increase in satisfaction. Satisfaction decline was found to be related to income, age and gender. In particular, older women (60+) of low occupational status tended to be most vulnerable to adjustment problems.

Adjustment problems during retirement - An area of conflicting results

Conflicting results have been reported for adaptation to early retirement and the effects of factors such as social support, work involvement and status change, which seem to be critical for some but not for others. Results concerning gender differences also need further investigating (Hatch, 1994, Kim & Moen, 2001). There are several possible explanations for the conflicting results. One could be that the interplay between contextual factors and individual factors such as financial situation, education, family situation and social support is complex and difficult to control in large samples where men and women have different roles and positions in society.

Another part of the difficulty in resolving questions of contradictory results could be methodological, for example, by the use of cross-sectional studies, the comparison of different cohort groups (Nuttman-Shwartz, 2007) and lack of studies controlling for personal and situational variables in the analyses (Pinquart & Schindler, 2007). Relatively few studies have used a life course perspective or a holistic person approach as an alternative to the variable approach in the study of retirement adjustment over time. The purpose of the person approach would be to identify groups of individuals with similar patterns of adaptation, using profiles of several variables (Bergman, 1995; Magnusson, 1995), that is, aiming to describe different trajectories in the transition from work to retirement.

A few qualitative studies have made attempts to study persons instead of factors or variables. An interesting example was a study by Hornstein and Wapner (1985), which reported four patterns of adjustment after a qualitative analysis of interview data. The patterns were as follows: retirement as a transition to old age, as a new beginning, as a continuation of pre-retirement lift structures and as an imposed disruption. Age and general attitude to work in life differed among the groups.

Aim and research questions

The present study offered a unique opportunity for comparisons between ‘Retirees' and ‘Stayers' aged 55+ with similar employment and career backgrounds. The study allowed us to follow them for 18 months, when 55% made the transition to early retirement and the remaining group continued working after the downsizing of the organization. In addition, the study group comprised of roughly equal proportions of men and women. Moreover, it allowed the investigation of both organisational and personal factors outside the workplace that might influence the transition to early retirement.

The general aim of the present study was to investigate changes in well being and health among older employees with a special focus on the role of employment status and the voluntary choice to retire or to continue working. The second aim was to take a closer look at adapting to early retirement and trying to identify patterns of adaptation during the first few years after the transition.

For the first part of the research the following hypotheses were suggested.

Hypothesis 1: Voluntary choice will lead to better adjustment in terms of general satisfaction, health and well being than forced continuance of work or retirement both among retirees and stayers.

On the basis of earlier research, the transition to retirement was expected to be fairly easy and not a cause for distress. In a similar way, a voluntary decision not to apply for early retirement but to continue working was not expected to lead to any problems that might cause distress. Accordingly, no differences were expected to arise between retirees and stayers on ground of employment status per se. On the other hand, the possibility that the decision was not altogether voluntary but influenced by others was expected to have a major impact. Positive effects were expected for persons permitted with a voluntary choice with regard to employment status, and negative ones for those reporting a forced choice.

Hypothesis 2: The effects of voluntary/ forced choice of employment status will decrease after 1.5 years, and differences between persons in each choice category will decline.

No studies so far seem to have presented results concerning the durability of the impact of free/forced choice of employment status in old age. The most likely hypothesis is that the effects of this kind will decline as time passes.

Hypothesis 3: Stayers will generally report more distress and health complaints than retirees following the downsizing of the organization.

For adaptation in terms of distress and health complaints, a no-change situation was expected for retirees. In this case, the most important a priori reason for this was that older employees in poor health had been encouraged by the company to apply for a disability pension rather than for early retirement. Given this exclusion, our study group was relatively healthy compared with those of other studies. It was also expected that stayers would generally report more health complaints (on both occasions of measurement) because they had continued to work in a downsized organisation, presumably with a higher workload than previously.

Furthermore, differences in work values were expected on the first occasion of the measurement; in particular, retirees were expected to report lower values on work centrality than stayers. One important antecedent of the decision to apply for early retirement might be relatively low work motivation in old age, and a feeling that other values in life are equally or more important. This difference was expected to increase during the first 1.5 years following their departure from the company.

Hypothesis 4: Retirees will report lower scores on work centrality than stayers and the difference will increase over time. The hypothesis on gender differences was difficult to formulate and was based on several conditions. First, we found no obvious reason to expect gender differences among stayers. Among retirees, however, the fact that the average female salary was low—hence, the financial outcome of retirement was less favourable for women—might lead to poorer adjustment and lower well being among women (as indicated in earlier studies). On the other hand, the fact that work tends to be less central to the lives of women than to those of men might lead to smoother and less dramatic adjustment among women than among men. On ground of these contradictory expectations, a null hypothesis was chosen for gender comparisons.

Hypothesis 5: No gender differences were expected in terms of satisfaction, well being and health during the year following early retirement. The purpose of the second part was to describe patterns of adjustment to early retirement in terms of health and well being during the first two years after retirement. A person approach was used with cluster analysis combining several of the factors commonly found to be influential. Change over time in the adaptation and homogeneity of patterns during the first years of early retirement was studied. With a descriptive purpose, no hypotheses were suggested but the analyses aimed to answer the following questions:

• Is it possible to identify homogeneous patterns of adaptation to early retirement within the first two years after the transition?

• How stable is adjustment during the first two years?

 

Method

Participants

All employees, 55 years of age and older who had been offered to apply for early retirement, were invited to participate in the study. Of 633 men and women, 357 (55%) applied for retirement and were permitted to leave, whereas the remaining 276 (45%) remained in employment. Response rates are given in Table 1 below. The analysis of the first-wave data showed that the small group of nonrespondents at Time 2 did not differ from respondents at Time 1 in terms of age or gender. Further, the level of satisfaction with the outcome of the downsizing process at Time 1 was not significantly different between respondents and nonrespondents at Time 2.

The first wave of data collection, a questionnaire survey, took place in November 1992, immediately before the final group of retirees left the organisation. The second was conducted 1.5 years later for stayers and 1.5 years following retirement for retirees. The choice of the time interval was somewhat arbitrary and was made mainly for practical reasons. In addition, there were no obvious theoretical guidelines for a suitable time lag in the literature.

There was a significant age difference between these subgroups with both subgroups of voluntary and non-voluntary retirees being significantly older than the reject group (60 and 61 years compared to 58 years) (F = 12.9, 235, p < .001). There were no differences in how respondents evaluated the general importance of work in their lives. Group size, response rate, age, gender distribution and the perception of voluntary choice is shown in Table 1 below.

Questionnaire data was used both for cluster analyses and for analyses of variance. The variables chosen for the cluster analyses were some of those commonly assumed to have an impact on adaptation to retirement: age, reported health problems, satisfaction with retirement, psychological well being, work centrality and perceived social status.

Measures

In the analyses of variance, three variables were used as independent variables:

Employment status: A dichotomous variable retired/stayer was used as indicator of employment status. A total of 370 persons (58%: 226 retirees and 144 stayers), who delivered usable questionnaires on both occasions, were included in the analyses.

Voluntary/forced choice of employment status: In the questionnaire, participants chose among the different statements listed below describing whether or not their decision was the result of a voluntary choice.

On the basis of their responses, each employment-status group could be divided into two main subgroups: those reporting a voluntary choice of employment status (n = 257, 80%) and those reporting a forced choice (n = 64, 20%). The division was coded as a dummy variable (0, 1). After excluding some cases of missing data, an effective sample of 321 individuals was obtained for the analyses of variance.

A significant difference between employment-status groups was found in terms of proportion of voluntary/forced choices (χ²[21, n = 321] = 33.3, p < .001). A higher proportion of stayers belonged to the forced-choice group. Thus, a higher proportion of the retirees had made a voluntary choice of their employment status.

Gender Men made up a majority of both retirees (55%) and stayers (59%). There was a significant gender difference in employment status following downsizing with a somewhat higher proportion of women applying for and being permitted retirement (χ² [1, n = 192] = 3.79, p <.05). There was no gender difference with regard to the proportion of individuals being forced into retirement but a higher proportion of men than women among the stayers reported that their retirement applications had been rejected (χ² [1, n = 129] = 5.24, p < .05). Dependent variables for the analyses of variance were general satisfaction with downsizing outcome, psychological well being (GHQ), health complaints and work centrality—all measured on two occasions (at the end of the downsizing process and 1.5 years later).

Satisfaction with outcome of the downsizing process Reactions to downsizing were represented by a single item: ‘How satisfied are you personally at present with the outcome of the retirement program for you?' (range 1-5).

Psychological distress/well being The 12-item General Health Questionnaire (GHQ-12) was used to measure psychological distress/well being (Goldberg, 1979; Goldberg and Williams, 1988). Responses were provided on four-point Likert scales, entailing that the total scale ranged from 0 to 36. A high value on the GHQ represents a high level of distress. The reliability of the scale, as measured by Cronbach a, was .86 at Time 1 and .81 at Time 2.

Health complaints A summary index was constructed on the basis of seven items on a symptom checklist (e.g. cardiovascular, muscelo-skeletal and other psychosomatic symptoms), forming a scale originally developed and tested by Andersson (1986). The reliability (Cronbach a) of the scale was .66 on both occasions of the measurement.

Work centrality was defined in terms of the general belief about the importance of work to life (MOW, 1987, p.17). It was measured using an item from the Meaning of Working studies: ‘How important are the following areas at the present time?' Here, the points assigned to work, which could range in principle from 0 to 100, were used as measures of work centrality (MOW, 1987, p.82).

For cluster analyses, an additional indicator was chosen:

Perceived social status was one of the five latent functions of work defined by Jahoda (1982). The measure used here was taken from Henwood and Miles (1987), who used it in a study comparing employed individuals to unemployed individuals, retirees and housewives. It consisted of a single item for each function with responses provided on a seven-point scale. For perceived social status, the item was ‘Society in general respects people like me'.

Statistical analysis

The SPSS system was used for data analysis (Norusis, 1990). The differences between the employment status groups were tested by means of analyses of variance. Analysis of covariance (ANCOVA) was used to test whether employment status, voluntary choice or gender could explain variation in Time 2 measures of the dependent variables when Time 1 measures were entered as a covariate. This adjustment, which incorporates the pretest measure into the analysis, serves to increase the precision of the group effect (Cook and Campbell, 1979).

The SLEIPNER statistical package for pattern-oriented analyses developed by El- Khouri and Bergman (1995) was used for the descriptive cluster analyses. No missing values are allowed in the cluster analysis, and as attrition in some variables was relatively high, a method of imputation was used that identifies twin cases with similar values on all the variables involved for each case with missing values. Only one missing value was allowed. Cases for which no twin was identified or with more than one missing value were excluded from further analysis. In an effort to increase homogeneity of patterns, residue cases (‘outliers') were identified and excluded from the cluster analysis. Ward's method of cluster analysis was used with standardized variables. Finally, a relocation method recommended by Bergman (1995) was used to increase cluster homogeneity.

There are no definite rules of thumb for deciding the number of clusters in a data set (Aldenderfer & Blashfield, 1984). The rule applied here was recommended by Bergman (1996) and uses a limit of 67% of the explained error sum of squares. Bergman also discusses problems associated with a rather strong adherence to this principle in cases of reliability problems or where natural homogeneity should not be expected. Results from cluster analyses provided very heterogeneous results, however, resulting in as much as 14 clusters explaining 67% of variance on both occasions. Finally, a six cluster solution was chosen explaining 48% of error sum of squares with a homogeneity measure of .32 (point biserial) in order to obtain interpretable solutions.

To evaluate cluster stability over time, the EXACON program was used (El Khouri and Bergman, 1995). The program performs a cell-wise analysis of a contingency table (Time 1 * Time 2 clusters) on the basis of exact tests of significance (one-tailed hypergeometric probabilities). The results showed four stable patterns meaning that significant proportions of individuals belonging to cluster T1 were placed together at T2.

 

Results

Changes in adaptation between Time 1 and Time 2

Table 2 below shows mean values and changes in the four indicators of adaptation (satisfaction, distress, health complaints and work centrality) observed at T1 and T2 with an 18 month interval for both retirees and stayers.

The first impression given by the table is one of relative stability in the sense that similar mean values can be observed on both occasions of the measurement. The tendency is that retirees are generally more satisfied with the outcome of the downsizing process and report lower levels of distress and health complaints. In fact, the mean value of health complaints decreased significantly among retirees, indicating improved health in this group 1.5 years after retirement. Finally, as expected, the mean value of work centrality decreased for retirees but remained unchanged for stayers.

A second significant change was found. For stayers, the mean value of distress as measured by the GHQ-12 was significantly lower at Time 2, which indicates fewer symptoms of distress and a better sense of well being. By Time 2, the level for stayers had approached that of retirees. A further analysis, however, revealed that this outcome was an effect of the results of a small subgroup of 25 individuals who had left the company during the 1.5-year interval between Time 1 and Time 2 (for either retirement or other employment). This small group reported significantly higher well being at Time 2. When they were excluded from the stayers group, the change in well being was no longer significant.

The role of employment status, voluntary choice and gender

ANCOVA analyses aiming to test for the main and interaction effects of voluntary choice, employment status and gender were performed for Time 2 scores of the four dependent variables with Time 1 scores entered first as covariate. The results are shown in Table 3 below.

The results 1.5 years after retirement and 1.5 years after the first measurement occasion for stayers indicated both stability and change. The effects of the independent variables on general satisfactionat Time 2 largely resembled those at Time 1. There were significant main effects of employment status, voluntary choice and gender. Higher satisfaction scores were found for retirees than for stayers and higher scores for women than for men. An interaction effect of voluntary choice and gender was found at T1 (the original paper by Isaksson and Johansson (2000) provides more details of the analyses), indicating that among individuals who had to make a forced choice regarding their employment status, the satisfaction scores were lower for men than for women. This gender difference was no longer significant after 18 months and the interaction of voluntary choice with gender approached significance (p < .10). No significant interaction effects were found at T2 and these values are not shown in the table.

That the strongest effect on satisfaction after 18 months was found for employment status reflects that retirees have become increasingly satisfied, regardless of how the decision to retire was borne. With regard to level of distress as measured by the GHQ at Time 2, however, the possibility of voluntary choice still had a significant main effect, one that overshadowed any effects of employment status. As could be seen from the change in the mean values, adaptation had taken place, which is probably explained by the most distressed individuals having left the company on their own accord between our measurement occasions.

The pattern for health complaints had changed by Time 2, employment status becoming the only variable with a significant main effect. The mean value changes (see Table 2) indicated that the first 1.5 years of retirement had been beneficial for health, as reflected in a lower level of reported health complaints among this group. Finally, in the case of work centrality, employment status had a significant main effect after 1.5 years. In line with expectations, work had become less central to the life of retirees. The effect of gender, however, appeared to be stable over the first 1.5 years after retirement.

Patterns of adaptation after 1.5 years of early retirement

The second part of this paper presents results with a special focus on the adaptation to early retirement using cluster analyses to identify homogenous subgroups with a similar pattern of adaptation. A comparison of cluster solutions at T1 and T2 by using the EXACON program provided evidence of both stability and change. An indication of stability was that four of the six clusters at T2 consisted of significant overrepresentations of persons from four T1 clusters. This means that large proportions of persons who were similar to each other in terms of adaptation remained similar 1.5 years after early retirement and thus were grouped together in the same cluster at T2 as well. Tables 4 and 5 below show mean values for clusters at T1 and T2 for the variables used.

Plotted profiles were made on the basis of mean values transformed to z-scores in order to show the shape and level of the pattern for each cluster. The shape of the profiles from both occasions was compared to get an indication of stability or change of adaptation patterns over time. Inspection and comparing of profile shapes and level for clusters revealed three stable patterns. Besides similar shapes and level of profiles, a pattern considered stable also had a significant overrepresentation of persons belonging to the same cluster on both occasions. The four stable modes of adaptation together represented about 75% of the total group of early retirees.

The first stable adaptation pattern found indicated a general satisfaction with early retirement (Cluster B at T1 and C at T2). Persons in this cluster reported high satisfaction, low values for distress and health problems. The mean age was relatively high, work centrality low (although 34% were working) and the mean values for perceived status, high. This adaptation pattern represented 21% of the total group of early retirees.

Another stable pattern of adaptation to early retirement was to continue working (clusters F at both T1 and T2, 22% of those retired). The cluster consisted of the highest proportion of persons who had a new job (56%) and the highest mean values for work centrality. This mode of adaptation was positive, satisfaction was high and mean values for distress and health problems were low.

A stable but negative adjustment was also observed and described as highest distress and lowest satisfaction (cluster E at T1 and A at T2). The reluctance to retire appeared to be part of the explanation with 26% reporting that they were forced to retire. A relatively small group (12%) thus continued to show signs of poor adjustment after 1.5 years.

The fourth significant pattern indicated change and an indication of positive consequences of the transition from work to retirement. This group of retirees appeared to be relatively satisfied at T1 but reported high mean values for health problems. The profile at T2 was very similar on all variables with the exception of a markedly lower level of health problems (cluster D at T1 and E at T2). Adaptation thus appears to be improving perhaps as a result of a decrease in work-related stress symptoms. This group apparently benefited from a change to a less stressful life and represented 22% of the early retirees. The remaining two patterns (representing 25% of the group of early retirees) were clearly unstable over time and were also difficult to interpret.

 

Discussion

The first conclusion must be that there were no signs of a general crisis during the first two post-retirement years. The mean values of psychological well being were good and remained stable during the first 1.5 years. The results of the other health problems revealed a general improvement during the follow-up period among the early retirees, thus supporting recent results from McColdrick and Cooper (1994) and Midanik et al., (1995). A decrease in work-related stress symptoms was the most obvious explanation, especially since no similar improvement was found among the former colleagues of the same age group who continued working during the follow-up period.

The first hypothesis—better adaptation following downsizing in terms of general satisfaction—well being and health among individuals perceiving a voluntary choice rather than a forced choice of employment status, was supported. Retirees reported a more positive attitude and fewer symptoms of distress than those who remained at work. This was explained both by the fact that a higher proportion of retirees than stayers had chosen their course voluntarily, and by the experience of retirement, which appeared to be relatively favourable for most individuals (compared with continued working).

The high satisfaction scores for those who left the organization) compared with stayers might also be an effect of negative reactions to the downsizing process. This possibility makes it hard to assess the generalizability of the results. Nevertheless, it is clear that there are no signs of a general adaptation crisis following early voluntary retirement among the group of insurance employees studied.

The second hypothesis was of declining effects of voluntary/forced choice and better adjustment over time (in terms of higher satisfaction and decreasing distress symptoms). Satisfaction scores and well being scores were largely unchanged for both groups after 1.5 years. Signs of better adjustment were only found among a small group of persons who had left the company during the period between the measurements (by Time 2). The analyses of variance revealed that voluntary/ forced choice continued to have significant main effects on the second measurement occasion (1.5 years after downsizing), but that these effects were less strong.

The general conclusion, however, is that the effects of downsizing and the manner in which the decision to retire or to continue working is accepted affect the life of the individual for longer than expected. Efforts to restrict individual choice appeared to have negative effects both on those who were forced into retirement and on those for whom it was denied. Being able to personally influence the outcome of the downsizing process was found to be critical to the subsequent adaptation.

The third hypothesis concerning health complaints was only partly confirmed. The difference at Time 1 was related to free/forced choice rather than to employment status. After 1.5 years, however, differences between retirees and stayers were in line with the hypothesis. Decreasing values were found for retirees rather than stable ones as expected. Again, early retirement seemed to have positive effects.

There were no significant differences in work values between retirees and stayers at the time the decision to apply for retirement was determined, which clearly indicates that this factor alone did not determine the decision. Part of the hypothesis concerning work values was confirmed in that there was a decrease in values for retirees (compared with stayers) after 1.5 years.

Finally, the null hypothesis of similar adjustment to early retirement among men and women was not confirmed. Quite contrary to other results (for an example see Kim and Moen, 2001), women were found to report significantly higher satisfaction with the outcome of the downsizing process than men and there were no gender differences in terms of distress or health complaints. Further, male retirees appeared to be more sensitive to the rejection of their retirement application. One possible explanation is that the financial situation was rather also favourable for women in our study. The proportion of divorced or single women was higher among stayers than retirees. It may be assumed that divorced and single women tended not to apply for early retirement. Financial concerns were no more prevalent among female than among male retirees. Lower work centrality on the part of women was found but there was no interaction with employment status. It is possible that an explanation for the higher proportion of women among retirees lies in the interplay between factors such as family status, personal financial circumstances and work motivation.

In accordance with conclusions drawn by several researchers during the last decades (for example, Richardson & Kilty 1991; Williamson, Rinehardt & Blank, 1992; Kim & Moen, 2001), modes of adaptation in the first post-retirement year were found to differ between individuals. The circumstances under which the individual decides to retire, that is, the timing of retirement seems to be determinant not only of satisfaction with the immediate situation but also of psychological well being and health in the first year of retirement.

By and large, the study group may be regarded as representative of employees over 55 years of age at the company in question. For example, the degree of satisfaction with the final decision seemed to be equally distributed among responding and non-responding groups. The study suffered from a relatively high degree of missing data on some questionnaire items but on the final set of variables chosen for the analysis, there were no indications of anything but a random attrition. In light of the high proportion of voluntary retirees, the fact that they were in good health and the relatively favourable financial conditions offered by the company, a reasonable conclusion is that possible problems during the retirement process are underestimated by the results for this group.

On the whole, the study confirmed the profound significance of voluntary choice of labour-force participation for psychological well being in old age. An important conclusion for the management of downsizing, when aimed at older personnel, can be drawn. The company's strategy in this case was to provide the individual with the illusion of free choice by offering all senior personnel the opportunity to apply for early retirement. However, managers then rejected a number of applications with reference to unacceptable skill losses. Apparently, this lead to unintended negative effects (further discussed in Isaksson & Johansson, 2001).

Patterns of adjustment

The attempt to find homogeneous patterns of adjustment by means of a descriptive pattern-oriented approach yielded results that were clear in the sense that homogeneity was difficult to achieve. There were several indications of heterogeneity in the group: one was the relatively large residual groups consisting of cases that were similar to no other case.

One reason for heterogeneity could of course be the low reliability of the measures used. This was probably at least part of the problem since three of the measures used were single items. Another and probably more important explanation could be that the degree of homogeneity of adjustment during the first years after early retirement is, in fact, relatively low. Thus, individual variation of life course trajectories during the transition to early retirement seems to be substantial. Data here indicate that work values and health are factors that change during the transition from employment to early retirement. The low stability of clusters and the relative instability of adjustment patterns during the first years were a clear indication of this.

Three patterns of adaptation to early retirement were found that were stable over the first years. The first one was a generally positive adaptation to early retirement in terms of the transition to old age. Results further displayed a pattern with signs of poor adjustment negatively affected by a forced choice to retire early with very high levels of distress and low satisfaction at both T1 and T2.

It was also evident that early retirement did not necessarily mean the end of working. The third stable pattern of adjustment in fact involved continuing to work (cluster F), which was especially common among men and persons living in small towns and rural areas who still had a relatively high work centrality score. The more general pattern, however, showed that the significance of work in relation to other life spheres decreased considerably although working was far from uncommon in the other clusters.

A descriptive pattern analysis seems to be a promising path to follow for future research. A person approach is clearly needed in the area of retirement adaptation. The study here covered only the first two post-retirement years and a longer follow- up would be necessary for more definite conclusions of adaptive patterns. It would also be interesting to study the transition to old age retirement to see in what way it resembles the results of this study. There are probably some differences, for example, that a forced choice of retirement is not relevant, and a hypothesis based on the results of this study would be that the transition to an old age pension would be even smoother.

 

Referencias

Alenderfer, M., y Blashfield, R. (1984). Cluster analysis . Beverly Hills, CA:Sage.        [ Links ]

Atchley, R. (1989). A continuity theory of normal ageing. The Gerontologist, 29, 183-190.        [ Links ]

Bergman, L. R. (1996). A pattern-oriented approach to studying individual development: Snapshots and processes. In L.R. Bergman, R.B. Cairns, y K. Kagan (Eds.), The individual as a focus in developmental research. New York, NY: Sage.        [ Links ]

Cook, T.D., y Campbell, D. T. (1979). Quasi-Experimentation: Design and Analysis Issues for Field Settings. Boston:Houghton Mifflin.        [ Links ]

Davies, D., Matthews, G. y Wong, C. (1991). Ageing and work. In C. Cooper y I. Robertson, I. (Eds.) International Review of Industrial and Organizational Psychology, (pp. 149-211). New York: Wiley.        [ Links ]

El-Khouri, B. y Bergman, L. R., (1995). SLEIPNER A statistical package for pattern oriented analyses. Stockholm: Stockholm University, Department of Psychology. (Manual)        [ Links ]

Feldman, D. 1994). The decision to retire early: A review and conceptualization. Academy of Management Review, 19,285-311.        [ Links ]

George, L. y Maddox, G. (1977). Subjective adaptation to loss of the work role: a longitudinal study, Journal of Gerontology, 32, 456-462.        [ Links ]

George, L., Fillenbaum, L., y Palmore, E. (1984). Sex differences in the antecedents and consequences of retirement. Journal of Gerontology, 39, 364-371.        [ Links ]

Goldberg, D. (1979). Manual of the General Health Questionnaire. London: NFER Nelson.        [ Links ]

Goldberg, D., y Williams, P. (1988). A User's guide to the General Health Questionnaire . London: NFER Nelson.        [ Links ]

Hatch, L. (1992). Gender differences in orientation toward retirement from paid labor. Gender and Society, 6, 66-85.        [ Links ]

Henwood, F., y Miles, I. (1987). The experience of unemployment and the sexual division of labour. In D. Fryer y R. Payne (Eds.), Unemployed people (pp. 94-110). Oxford: Open University Press.        [ Links ]

Herzog. R., House, J. y Morgan, J.(1991). Relation of work and retirement to health and well-being in older age, Psychology and Aging, 6, 202-211.        [ Links ]

Hornstein, G., y Wapner, S. (1985). Modes of experiencing and adapting to retirement. International Journal of Ageing and Human Development, 22, 291-315.        [ Links ]

Isaksson, K. (1997) Adjustment to early retirement. I Kilbom, Å., Westerholm, P., Hallsten, L. y Furåker, B. Work after 45 (pp 177-185) Proceedings from scientific conference 22-25 September 1996. Arbete och hälsa 1997:29 vol 2.        [ Links ]

Isaksson, K. y Johansson, G. (2000). Adaptation to work vs early retirement after downsizing: Long-term effects and gender differences. Journal of Occupational and Organizational Psychology, 73, 241-256.        [ Links ]

Isaksson, K. y Johansson, G. (2001). Managing older employees after downsizing. Scandinavian Journal of Management. 19, 1-15.        [ Links ]

Jahoda, M. (1982). Employment and unemployment. Cambridge: Cambridge University Press.        [ Links ]

Kim, J.E. y Moen, P. (2001). Is retirement good or bad for subjective wellbeing? Current Directions in Psychological Science, 10, 83-86.        [ Links ]

Magnusson, D. (1995). Individual development: A holistic integrated model.In P. Moen, G.H. Elder Jr., y K. Lüscher (Eds.), Examining lives in context (pp. 19-60).        [ Links ]

Mc Coldrick, A. (1989). Stress, early retirement and health. In K.S. Markides y C. Cooper (Eds.), Ageing, stress and health. Tiptree, Essex: Wiley.        [ Links ]

Mc Coldrick, A., y Cooper, C. (1994). Health and ageing as factors in the retirement experience. European Work and Organizational Psychologist, 4, 1-20.        [ Links ]

Midanik, L., Soghikian, K., Ransom, L., y Tekawa, I. (1995). The effect of retirement on mental health and health behaviors: The Kaiser permanent retirement study. Journal of Gerontology, 50B, 559-561.        [ Links ]

Moen, P. (1996). A life course perspective on retirement, gender and well-being. Journal of Occupational Health Psychology, 1,131-144.        [ Links ]

MOW International Research Team (1987). The Meaning of Working. London: Academic Press.        [ Links ]

Norusis, M.J. (1990). SPSS Advanced Statistics User's Guide. SPSS Inc.        [ Links ]

Nuttman-Shwartz, O. (2007). Is there life without work? International Journal of Aging and Human Development, 84, 125-147.        [ Links ]

Palmore, E., Fillenbaum, G., y George, L. (1984). Consequences of retirement. Journal of Gerontology, 39, 109-116.        [ Links ]

Pinquart, M. y Schindler, I. (2007). Changes of life satisfaction in the transition to retirement: A latent-class approach. Psychology and Aging, 22, 442-455.        [ Links ]

Richardson, V., y Kilty, K. (1991). Adjustment to retirement: Continuity vs discontinuity. International Journal of Ageing and Human Development, 33, 151-169.        [ Links ]

Talaga, J., y Beehr, T.A. (1988). Retirement:A psychological perspective. In Cooper, C. y Robertson I. (Eds.). International Review of Industrial and Organizational Psychology, 6, (pp. 185-211). New York: Wiley.        [ Links ]

Williamson, R. Rinehardt, A. y Blank T. (1992) Early Retirement; Promises and Pitfalls. Plenum Books, New York.        [ Links ]