زمان بازنشستگی: شامل تاخیر تنزیل چشم انداز در مدل های بازنشستگی

This research examined the influence of delay and anticipated health and enjoyment on the amount of retirement savings sacrificed for early retirement. In addition to testing and supporting predictions that willingness to sacrifice retirement savings would be less with shorter delays to retirement, greater anticipated health, and greater anticipated enjoyment, an individual difference delay discounting measure derived from experimental studies was used to extend the Beehr, Glazer, Nielson, and Farmer’s (2000) model of retirement age. We predicted and found that oldest preferred retirement age related to personal characteristics, current work factors, future retirement factors, and individual differences in delay discounting, with greater delay discounting being associated with a younger preferred retirement age. Findings suggest that delay discounting is an important consideration in retirement planning and also highlight the significance that retirement expectations and being tired of work have in affecting one’s preferred retirement age.

The timing of older workers’ retirement is a matter of increasing concern for individuals and governments. For the individual, career-related retirement decisions involve complex evaluation of work, personal, and financial issues, with possibly greater consequences, at least in terms of health and finance, than other career-related decisions made earlier in life. Governments of many developed countries with aging populations realize that they face a growing welfare burden from retirees who lack sufficient funds for lives that are likely to last longer with medical advances (Bingham, 2003). Concern for the combined effect of an aging population together with an apparent increasing preference for early retirement (Feldman, 1994) stimulated considerable research during the 1990s (Adams, 1999; Adams & Beehr, 1998; Hanisch, 1994; Schultz, Morton, & Weckerle, 1998; Taylor & Shore, 1995). The current study added delay discounting to retirement models derived from this earlier research in an effort to improve the prediction of preferred retirement age. Predicting an employee’s exit from the workforce is important because the removal of a compulsory retirement age has introduced considerable uncertainty about staff profiles for human resource planning, and a likely welfare burden for governments, and has also required individuals to take a more active and strategic role in their own career planning (Hesketh & Considine, 1998). Ekerdt, Kosloski, and DeViney (2000) conceptualize retirement planning as an extended decision-making process. Individuals increasingly attend to retirement issues as the anticipated date approaches. As shown by a range of studies, choosing a retirement date involves personal, psychological, organizational, job, and environment factors (Adams, 1999; Beehr, 1986; Beehr et al., 2000; Feldman, 1994; Taylor & Shore, 1995). However, none of these studies has addressed the concept of time or delay discounting, a phenomenon known to affect a range of career-related decisions (Hesketh, Watson-Brown, & Whitely, 1998). The concept of time or delay discounting is well established (Kirby, 1997; Rachlin, Brown, & Cross, 2000). Much of this research examines how people choose between a smaller reward offered now and a larger reward to be received later. The evidence suggests that time (i.e., delay) can result in a subjective devaluation of the later reward causing people to prefer a smaller, early reward because they perceive the later reward as being worth less than its actual value and thus not worth the wait. In relation to retirement decisions, a key issue people face is the choice between early retirement with less money and delayed retirement with more money, but this decision is likely to be tempered by anticipated health and enjoyment considerations also potentially subject to the discounting phenomenon. We examine the extent to which financial, delay, health, and enjoyment factors affect retirement decision-making, either independently or interactively. These factors are examined within a delay discounting experimental paradigm. Subsequently, we investigate whether the inclusion of individual differences in delay discounting increases the predictive capacity of traditional retirement models such as that of Beehr (1986). In the context of retirement decisions, delayed retirement offers a larger monetary reward (i.e., more savings or superannuation money). Greater discounting of that reward would increase the tendency to retire early because a person would be less willing to wait for the larger delayed retirement savings, perceiving the delayed retirement payout as having much less value than it actually has. In reality, however, people must juggle multiple factors and not just finance when making a decision about when to retire. For example, although early retirement may offer less money for the duration of one’s retirement, this could be compensated to some extent by a better chance of good health in early retirement and more time to enjoy retirement. On the other hand, while delaying retirement can offer more retirement savings, it carries the risk that health may be poorer by that stage, reducing the ability to enjoy retirement activities such as travel. In this case, the potential income associated with delayed retirement might not only be devalued subjectively by the delay but also by the anticipation of poorer health or less enjoyment. We predict that people will be willing to sacrifice greater amounts of their potential retirement savings when the delay to retirement is longer, and the health or enjoyment advantages of retiring earlier are greater. The nature of discounting is such that devaluation of the delayed reward is much greater when the early reward can be collected immediately or before too much time has elapsed than when there is a long time to wait before the availability of the early reward ( Hesketh et al., 1998). Therefore, in terms of accepting the “reward” of an early retirement, we expect that people with less time before their preferred retirement age would discount more money than those who see their retirement as further away in time. However, in this study, this effect should be reduced because the sample is restricted to those aged 45–65 years. This study also places delay discounting in the broader context of the Beehr (1986) model of prediction of retirement age, which includes the personal variables of health and age, as well as work-related variables and retirement-related variables. Delay discounting, operationalized as an individual difference variable capturing the extent to which money is sacrificed as a result of delay (and the effect of health and enjoyment), is likely to influence the timing of planned retirement. Beehr (1986) distinguished the general category of Personal Variables, such as health and age, from Environmental Variables. Environmental variables include Job Variables such as job characteristics and attainment of occupational goals, and Nonjob Variables such as intent to pursue leisure activities. Later work adopting this model ( Adams & Beehr, 1998; Beehr et al., 2000; Taylor & Shore, 1995) clarifies the environmental variables as either those that relate to current work or those that relate to future retirement. 1.1. Personal variables The relationship between personal variables and the decision to retire has been studied extensively, with results consistently showing that health and age significantly predict retirement age. Typically, those with better health retire at an older age than those in poorer health (Hansson, DeKoekkoek, Neece, & Patterson, 1997; Kim & Feldman, 1998; Lund & Borg, 1999; Muller & Boaz, 1988). Obviously, ill-health makes continuing with work more difficult, and hence these results are unsurprising. However, some studies (e.g., Adams, 1999) fail to find a significant effect of health, and Feldman (1994) argues that results are confounded because measures of health do not distinguish between major and minor illnesses and the extent to which illness impedes one’s ability to work or, for that matter, to enjoy retirement. Furthermore, prior research has measured perceptions of, or satisfaction with, current health, without considering whether the prospect of future poor health might affect the decision to retire. Do people work longer to save more money to cover future health costs, or would they prefer to retire early to have at least some retirement time in good health, even if it means a reduction in retirement finance? In terms of age, previous research indicates that the older a person the later the planned retirement age (Adams, 1999; Taylor & Shore, 1995). Ekerdt, Rose, Bosse, and Costa (1976) suggested that a shift in horizon to a preferred later retirement age occurs in the early 50s age group. Nevertheless, Kim and Feldman (1998) found that age was positively correlated with actual acceptance of an early retirement incentive, a finding that supports our hypothesis that people closer to retirement will discount the value of a delayed payout to a greater extent. 1.2. Work-related variables Although early research (Hanisch & Hulin, 1991; Schmitt & McCune, 1981) indicated that job satisfaction predicts retirement age, the weight of recent evidence led Beehr et al. (2000) to conclude that it is not a significant predictor. Adams and Beehr (1998) showed that although job satisfaction is unrelated to retirement, it is associated with turnover intentions. They suggested that the differential prediction might occur because retirement withdrawal from work is a more life-changing decision than withdrawal from a particular organization and is therefore less affected by satisfaction with conditions of one’s current job. To date, much of the research has used broad, unidimensional measures of job satisfaction. However, one of the most widely used job satisfaction instruments, the Minnesota Satisfaction Questionnaire (Weiss, Dawis, England, & Lofquist, 1967), allows for measurement of both intrinsic and extrinsic satisfaction. Intrinsic satisfaction is a person’s response to variables such as ability utilization, autonomy, variety, and authority, whereas extrinsic satisfaction focuses on the pay, working conditions, and more physical aspects of one’s job. We investigate whether intrinsic and extrinsic satisfaction relate differently to preferred retirement age, while offering no hypothesis as the extant literature is inconclusive. Beehr (1986) suggested that attainment of occupational goals would predict retirement age but was unsure of the direction of the relationship. A test of this proposal (Adams, 1999) found that the more occupational goals are perceived to have been obtained, the earlier the age of retirement. Attainment of goals is therefore included in the present study along with an additional work-related variable, being tired of work. Beehr et al. (2000) found work fatigue to be the only work variable to significantly predict retirement age after controlling for health and wealth. Further support for its importance is given by Hansson et al. (1997), citing US census data, that showed retrospectively that being tired of work was reported as a more important factor than health in the retirement decision by already-retired people. We therefore predict that both being tired of work and having attained work goals will be associated with a younger preferred retirement age. 1.3. Retirement-related variables Four retirement-related variables are examined in this study: expected retirement wealth, planned activity level, expected enjoyment of retirement, and self-efficacy in retirement finance management. In past research, wealth has been operationalized as current financial status, retirement financial status, or a combination of both. A consistent finding is that those with the prospect of greater retirement wealth or those who are more satisfied with their likely retirement income will retire earlier (Adams, 1999; Maule, 1995). Current financial status is less consistently related to retirement age. For example, Adams (1999) and Adams and Beehr (1998) found that annual income was unrelated to retirement decisions, whereas Feldman (1994) cites research suggesting that individuals are more likely to retire early the greater their income, and Kim and Feldman (1998) found that the lower the income the more likely the acceptance of an early retirement incentive. In the present study, perception of expected retirement wealth is included within the retirement-related variables. Retirement affords the opportunity to spend time in a range of leisure and social activities. Beehr et al. (2000) hypothesized that those who expect to engage in such activity would be attracted to early retirement. Results were equivocal and suggested the need for further study. At a more general level, there has been consistent support for the predictive ability of a broad, positive attitude to retirement (Feldman, 1994; Hansson et al., 1997). Those who expect to enjoy retirement are more likely to retire earlier than those who expect to be bored in their retirement years. Self-efficacy is having a belief that one can perform certain behaviors. According to Bandura’s (1977) social learning theory, self-efficacy is not a generalized trait but a motivational construct that relates to a specific task. Feldman (1994) suggests that the reason why employees who receive pre-retirement programs retire earlier than those who do not is the programs that reduce uncertainty and raise self-efficacy about retirement; however, little research has directly examined self-efficacy in terms of retirement age. Indication that self-efficacy might be an important retirement predictor comes from Gowan (1998), who found that low general self-esteem was related to a lack of desire to retire early; and Fretz, Kluge, Ossana, Jones, and Merikangas (1989) who showed that higher self-efficacy was associated with lower pre-retirement anxiety. Taylor and Shore (1995) measured beliefs about the ability to adjust successfully to retirement. The results indicated that more positive self-ratings were related to earlier retirement. In this study, we examine self-efficacy for managing retirement finances because of the importance of finance in retirement and the increasing complexity of its management due to changing legislation and general uncertainty about financial returns, predicting that those with higher self-efficacy will prefer a younger retirement age. 1.4. Measures of retirement age The measurement of an appropriate criterion has not been a straightforward issue in research that examines predictors of retirement age. Although actual retirement age is considered a preferable measure to intended retirement age, research using the former ( Schultz et al., 1998) has collected retrospective predictor data. There is a need for longitudinal studies. Nevertheless, Adams (1999) argues for the fidelity of intentions as an outcome measure because there is sufficient empirical evidence to show that intentions to retire are related to actual retirement. Taylor and Shore (1995) found a high correlation between intended and preferred retirement age and suggested that respondents did not differentiate between the two. In addition, Feldman (1994) explains that the nature of retirement is changing from the traditional concept of a definite and complete end to one’s career to more gradual transitions between full- and part-time work and nonwork. Beehr et al. (2000), therefore, suggest that retirement exists on a continuum. To cater for this possibility, we measured oldest and youngest ages at which a person would prefer to retire permanently from the workforce. Although research over the last decade has identified a trend for people to retire early, some Western governments are advocating that people extend their working lives because recent analyses have highlighted both an imminent skilled labor shortage and the prospect for many of insufficient personal finance to cover long retirements ( Commonwealth of Australia, 2004). The oldest age at which a person would prefer to retire therefore becomes an important criterion. We make a number of hypotheses about retirement decision-making which will be tested in the current study by use of both a computer task that factorially combines delay, health, and enjoyment, and a survey that measures personal, work, and retirement factors and preferred retirement age. The amount of delayed retirement savings that participants were prepared to sacrifice was the dependent variable in the computer task while oldest preferred retirement age was the dependent variable in the survey instrument. Using the computer task: (1) People will discount the value of delayed retirement savings when making a decision to retire early. That is, the greater the delay the more money they will sacrifice to retire early. (2) People will discount the value of delayed retirement savings as the health advantages of early retirement increase and when there is a higher anticipation of enjoyment in early retirement. That is, the better the chances of good health in early retirement and the greater the anticipated enjoyment associated with early retirement compared with delayed retirement, the more money they will sacrifice to retire early. Using the survey: (3) In support of Beehr, 1986 and Beehr et al., 2000 model, personal, work, and retirement variables will predict preferred retirement age. It is expected that older preferred retirement age will be related to better current health, older age, lower attainment of goals, being less tired of work, and lower expected retirement wealth, activity, enjoyment, and self-efficacy for managing retirement finance. We make no prediction about extrinsic and intrinsic satisfaction due to the equivocal nature of past research. Using both the computer task and the survey: (4) Those who discount delayed rewards more (as measured by the computer task) are expected to have an earlier preferred retirement age (as measured by the survey instrument) than those who discount less. That is, delay discounting will explain variance in predicting retirement age. (5) People with less time before their preferred retirement age would discount more than those who see their retirement as further away in time. (6) The better people rate their current health and the more they expect to enjoy their retirement (as measured by the survey instrument), the more money they will sacrifice for the sake of better health in retirement and greater enjoyment, respectively (as measured by the computer task).

Delay discounting is a ubiquitous phenomenon with increasingly important policy ramifications in the light of changing workforce demographics. Sacrificing substantial additional retirement savings in pursuit of early retirement could jeopardize individuals’ financial independence, forcing their reliance on Government welfare underwritten by a dwindling pool of younger taxpayers. This study investigated delay discounting in the context of retirement decision-making. A novel, adaptive computer task assessed whether and how much participants were prepared to sacrifice a guaranteed retirement payout to retire early. While the task was hypothetical in nature, it did capture a more realistic situation where participants must evaluate multiple factors in making a retirement decision, rather than rate the importance of individual factors in isolation. The results supported our hypotheses that people would sacrifice greater amounts of retirement savings the longer they were delayed, and the better the chance of good health and enjoyment in early retirement compared to later retirement. However, it appeared that while the effect of health was equally important at both 2- and 5-year delays, enjoyment was more important when delay was longer. Furthermore, low levels of both health and enjoyment resulted in the smallest sacrifices, whereas it appeared that good health compensated for lack of enjoyment and vice versa. Considerable individual differences in discounting emerged. For a small proportion of individual participants, time spent in early retirement was worth sacrificing considerable sums of money but the bulk of participants were much more conservative. Whether or not participants’ responses were financially sound remains an economic issue rather than a psychological issue. If their responses were financially unsound, the social, economic, and psychological bases for the delay discounting of retirement savings should be addressed to help remediate some people’s decisions. We also proposed an extension to Beehr et al.’s (2000) model of retirement age predictors to include the extent that individuals discount with increasing delay together with personal, work, and retirement variables, hypothesizing that the more a person delay discounted, the earlier he or she would want to retire. Results from survey data combined with the computer task data supported this hypothesis, showing that discounting predicted the oldest age people are prepared to retire even after controlling for the effect of the personal variables of current age and health, and the work variables of job satisfaction, accomplishment of work goals, and being tired of work. However, differences in delay discounting did not explain variance in oldest preferred retirement age over factors related to one’s perceptions of retirement, which as a group, were the strongest predictors. Expectations that one would enjoy retirement and that one would be financially well-off in retirement were individually related to younger retirement age, supporting prior research (Adams, 1999; Hansson et al., 1997). Unexpectedly though, and contrary to the few previous studies that have examined self-efficacy, higher self-efficacy was associated with older preferred retirement ages. Unlike previous research, however, we did not measure self-efficacy for adjustment to retirement but self-efficacy to manage retirement finances. The items tapped confidence with understanding superannuation legislation and tax laws together with confidence to manage retirement savings investments. It could be that the more a person understands the complexities of this area, the more they accept the need to work longer to achieve the necessary financial security for retirement. Those with low understanding and low self-efficacy may not be aware of the risk they take in retiring early with less money. One of the strongest predictors in this study was the variable measuring how tired a person was with working, replicating the results of Beehr et al. (2000). Being tired of work or no longer having the energy to work, although a “simplistic” factor (Beehr et al., 2000, p. 209), is likely a direct measure of motivation. Interestingly, it is correlated with self-efficacy, which is thought to be related to motivation. Future research could investigate this as a mediating variable between retirement age and other predictors. A better understanding of the factors that result in a person being tired of work may clarify the process that leads to retirement. Retirement decision-making is clearly a complex process and we suggest that the addition of moderating and mediating variables into future models is necessary to advance understanding. The finding in this study that extrinsic job satisfaction was significantly correlated with preferred retirement age, whereas intrinsic satisfaction was unrelated, could explain the nonsignificant results in previous studies that have examined the relationship between retirement and satisfaction with one’s job. However, neither extrinsic nor intrinsic satisfaction was significant in the presence of other work or retirement variables. Interestingly, when it came to choosing amounts of money to retire early with an added health or enjoyment benefit for early retirement, people with higher levels of intrinsic satisfaction were less prepared to sacrifice money than those with lower levels of intrinsic satisfaction. Also supported was our hypothesis that people with less time before their preferred retirement age would discount more than those who see their retirement as further away in time. The discounting research identifies the existence of preference reversals ( Ainslie & Haslam, 1992). An individual may prefer a smaller reward to a larger reward simply because the smaller reward is available now or after a short delay, whereas the larger reward involves a long wait. However, if the delay until both the smaller earlier and the larger later reward were to increase substantially, preferences may reverse to favor the larger later reward over the smaller, earlier reward. Preference reversal would explain the current finding. That is, those close to retirement prefer the smaller early reward, but the further away from retirement, a person is, more likely to prefer a large later reward. By combining the survey data with the computer decision-making task, we showed that attitudes to future retirement and perceptions of current health status affect how one makes decisions related to retirement. Workers who were more optimistic in their expectations of enjoying future retirement made greater financial sacrifices to retire early when the early retirement option had higher levels of enjoyment than those who had lower expectations for enjoyment. Conversely, greater financial sacrifice for the sake of good health in early retirement was associated with lower levels of current health, suggesting that people with poor health are more influenced by prospective health status when making decisions than those currently in better health.