| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||
RESEARCH ARTICLE |
a U.S. Office of Personnel Management, Washington, DC
b Life-Span Development Research Center, Central Michigan University
Thomas A. Stetz, United States Office of Personnel Management, Personnel Research and Development Center, Room 6500, 1900 E Street, NW, Washington, DC 20415-9200 E-mail: tastetz{at}opm.gov.
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. The study relies on four waves (from 1986 to 1992) of data from the National Longitudinal Survey's Mature Women Cohort, and logistic regression was used to examine the extent of the proposed relationships.
Results. Retirement rate varied across industries and had a positive relationship with munificent environments, but retirement was not related to dynamism or unemployment rates.
Conclusions. The results suggest that environmental factors may influence retirement timing. Additional theoretical and empirical work is suggested to help sort out direct environmental effects and possible moderating third variables.
ALARGE amount of research has been conducted on the predictors of retirement. Reviews show that it has produced valuable insights into the underlying factors influencing the retirement decision. Reviews show age (Talaga and Beehr 1989
), health (Beehr 1986; Parnes 1988; Ruhm 1989), and personal finances (Beehr 1986; Parnes 1988; Robinson, Coberly, and Paul 1985; Ruhm 1989) are consistently the most important influences on the decision to retire. Other variables, such as education and occupational status (Talaga and Beehr 1989), have been found to have weak and inconsistent relationships with the decision to retire. Furthermore, sex and marital status (Parnes 1988) can have both direct and moderating effects on individuals' decisions to retire. Common to these findings is the emphasis placed on these personal variables or on individuals' job characteristics (e.g., Beehr and Nielson 1995). Little research has been conducted that examines broader organizational and environmental factors influencing the retirement decision.
The need for additional research at a broader level of measurement has long been noted. Based on a 1977 retirement research conference, Atchley 1979 identified seven areas in need of additional research. He stated that the effects of organizational characteristics on the decision to retire and on the timing of retirement are research questions that should be given high priority. Ten years later, Talaga and Beehr 1989 noted that researchers continue to be intrigued by the individual or personal causes of retirement, unfortunately, at the expense of the equally interesting potential effects of organizational and societal factors. Furthermore, still several years later, Feldman 1994 suggested the need for more research on organizational factors that influence individuals' early retirement decisions.
Older employees' career decisions are affected by present and anticipated future situations inside their organizations, and organizations are affected by their own external environments. Uncertainty in the organizations' environments is likely to affect both the organizations and their members, because it can be seen as a threat that makes the results of current actions unpredictable. At the individual level it has been theorized that a common element of an organization member's stress is uncertainty (e.g., Beehr and Bhagat 1985), and it has been shown that occupational stressors can lead to withdrawal from the organization (Gupta and Beehr 1979). At the organization level, uncertainty of the external environment is a major problem for organizations and those responsible for managing them. It makes planning more difficult, and this has implications for personnel staffing decisions and policies regarding pay, benefits, hiring, and downsizing.
Uncertainty of these, in turn, is expected to affect employees' career decisions, including the older employees' retirement decisions. This study assumes that (a) uncertainty varies across industries, (b) a munificent external environment reduces uncertainty, (c) dynamism or a changing environment increases uncertainty, and (d) unemployment increases uncertainty for older employees. Munificence, dynamism, and unemployment rate, therefore, are examined in relation to retirement status.
Retirement researchers have conducted a small amount of research examining the effect of environmental factors on the decision to retire. High unemployment rates create uncertainty by, for example, raising the specters of a recession or of lower increases in wages in the near future. This is uncertainty, but with a negative tone. Quinn 1977 found that living in an area of high unemployment increases the likelihood of retirement. Quinn, however, used male respondents to the Retirement History Study (RHS), and it remains to be seen if his findings hold true for women; knowledge about women's retirement decisions has lagged far behind knowledge about men's retirement (Gratton and Haug 1983; Quinn and Burkhauser 1990; Szinovacz 1983; Talaga and Beehr 1989). There is a need to examine women's retirement decisions, because they constitute a large and growing segment of the workforce.
In addition, it has been found that 60 year-old men who have lost their jobs are three times more likely to retire than are 60 year-old men who did not lose their jobs (Shapiro & Snadell, 1984, in Parnes 1988). Research looking at inflation and retirement has found that inflation does not affect retirement decisions (Parnes 1988). Changes in Social Security and other financial aspects of the environment have been found to have only very small effects on labor force participation and retirement (Burtless 1986; Burtless and Moffitt 1984; Fields and Mitchell 1984). Social Security's main role in the retirement decision is to cause a bunching of labor force withdrawal around the ages of 62 and 65 (Burtless 1986; Fields and Mitchell 1984).
A line of research closely related to retirement comes from the sociology literature and examines the effects of environmental factors, commonly industry growth and decline, on job mobility rates. Retirement can be thought of as job shift or transition, therefore, job mobility research is directly related to retirement decisions. Similar to retirement researchers, Haveman and Cohen 1994 commented that most research has focused on individual attributes of workers, such as education and experience, to explain job shifts. This is strikingly similar to the tendecy of retirement research to focus on individual difference predictors. Most studies assume that the environment, or opportunity structure, is stable. However, Haveman and Cohen 1994, using vacancy-chain and vacancy-competition models (White 1970; Sorensen 1975, Sorensen 1977), argued that organizational systems are unstable. Over time, organizations are born and organizations die, causing the large-scale creation and destruction of jobs and, thus, the opening and closing of job vacancies. They found strong support for their assertion. Organizational birth, dissolution, and merger had substantial and varied effects on internal and external mobility of California savings and loan managers. Thus, they concluded that socio-structural dynamics, which until recently have been ignored, have a strong and interesting effect on career processes.
Hachen 1992 argued that industry characteristics affect employer personnel strategies, which creates the opportunities and liabilities individuals encounter in the labor market. His results showed that industry characteristics are related to several types of job mobility. Labor intensity (defined as the total number of employees divided by total sales) and wage level affected several types of job movement. Hachen's two industry growth variables, growth in average establishment size and growth in number of establishments, had substantially fewer effects. Growth in the average establishment size was negatively related to involuntary exits and positively related to upward authority movement. Industry growth can lead older employees to predict a good future rather than uncertainty in their present occupations.
Finally, DiPrete 1993 found that industrial growth and decline affected several types of mobility (internal organizational movement, external movement within industry, external movement from industry, and unemployment) in upper white-collar, lower white-collar, and service and blue-collar workers. However, the effects were strongest for lower-level workers. Thus, sociological and retirement research has shown that environmental factors influence job mobility and the timing of retirement.
In organizational theory, the environment is virtually everything outside the organization (Mintzberg 1979). This includes the technology, the nature of its products, customers, competitors, its geographical setting, and the economic and political climate in which it operates. Munificence and dynamism are two commonly studied environmental characteristics. Munificence has been conceptualized as the extent to which the environment can provide sufficient resources to organizations and support their growth (Aldrich 1979; Dess and Beard 1984; Starbuck 1976). Dynamism, on the other hand, has been described as environmental instability and turbulence (Aldrich 1979; Dess and Beard 1984). Basically, dynamism is environmental change that is difficult to predict and thus increases uncertainty within the organization. Research has shown that environmental characteristics do predict organizations' behavior (i.e., their strategy, structure, and outcomes; Keats and Hitt 1988) and that industry characteristics influence job mobility (DiPrete 1993; Hachen 1992; Haveman and Cohen 1994). Therefore, it is believed that because workers are faced with different opportunity structures within industries, rates of retirement will vary with industry.
Hypothesis 1: The rate of retirement will differ across industries.
Feldman 1994 proposed that individuals who work for large firms in declining manufacturing industries are more likely to retire early. Consider, for example, an organization in a strong and growing industry. Often in such situations, industry unemployment is low and replacement workers are not readily available. As a result, organizations are likely to offer incentives to workers to continue working. However, in declining industries the opposite occurs. Instead of offering incentives to stay, organizations tend to offer workers retirement incentives as a way to reduce their workforce and payroll obligations. In addition, Hachen 1992 found that growth in the number of establishments in an industry was negatively related to worker departure rates. Thus, the following hypothesis is proposed:
Hypothesis 2: There will be a negative relationship between industry munificence and an individual's decision to retire.
A result of dynamism is uncertainty. In highly dynamic environments, workers do not have a clear and confident view of the future. Thus, they are unable to precisely plan for retirement and carry out those plans. As a result of the inability to plan, individuals may not have sufficient financial resources to retire. However, individuals in highly dynamic environments may choose to retire as soon as possible to avoid potential industry downturns and the financial hardships associated with them. In other words, some employees may want to get out while the getting is good! In such situations, retirement tends to lock in certain situations, such as future (retirement) income, and reduce uncertainty. After retirement has locked in stable income, the individual may then choose part-time work.
Hypothesis 3: There will be a positive relationship between industry dynamism and an individual's decision to retire.
Haveman and Cohen 1994 examined the relationship between industry environment and job mobility. They found a positive relationship between exits from the industry and (a) unemployment rate and (b) job moves within firms. Furthermore, Quinn 1977 found a positive relationship between unemployment rate and retirement. Quinn, however, used male respondents to the RHS, and it remains to be seen if his findings hold true for women. It has also been found that 60 year-old men who have lost their jobs are three times more likely to retire than are 60 year-old men who did not lose their jobs (Shapiro & Snadell, 1984, in Parnes 1988). Thus, the following hypothesis is proposed:
Hypothesis 4: There will be a positive relationship between local area unemployment rate and retirement.
|
Methods |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
The NLS began surveying 5,083 women, ages 30 to 44, in 1967. These women have been surveyed a total of 17 times, and the most recent year in which data was available was 1995. The Mature Women Cohort represented a national noninstitutionalized civilian population for its age and gender group; Blacks were oversampled to allow for computation of separate reliable statistics. The Center for Human Resource Research 1995 states that the retention rate for the Mature Women Cohort in 1992 was 58% (
). Rhoton 1984 showed that after 10 years the sample characteristics of the original cohorts (older men and mature women) were not distorted by attrition, and it is believed that sample representativeness still exists for the survey years used in this study.
The four panels of data were included in the analysis: 1986, 1987, 1989, and 1992. These were the most recent survey years in which the relevant data was available and respondents would begin to retire. Data were collected through telephone interviews in 1986, and in the remaining years data were collected through face-to-face interviews. Women who stated that in the week prior to the survey they were working, with a job but not working (i.e., on vacation), looking for work, or retired were eligible for this study.
Measures
Control variables.
Several individual characteristics served as control variables, because retirement has been shown to be affected by individual factors. These factors included Age, Race, Income, Marital Status, Education Level, Occupational Prestige, and the Presence or Absence of a Health-Related Work Limitation. Race was dummy coded into a Black category and "other" category, with White serving as the reference group in all logistic regression analyses. Income was measured as the total family income for the surveyed year. Marital Status was dummy coded into a widowed category, a divorced/separated category, and a never married category. Thus, married served as the reference or comparison group. Educational Level was measured as the highest grade completed, ranging from 1 year to 16 or more years of education. Prestige of the respondent's current occupation was measured with Duncan's index of occupational prestige (Duncan 1961), which assigns a two-digit prestige score based on the education and income distribution of the occupation. Finally, Health-Related Work Limitation was measured as the presence or absence of a self-reported health condition that limits the amount or kind of work that the respondent is able to perform.
Retirement status.
Measurement of retirement can be done in many ways, including the self-report that one is retired, current employment status (e.g., how many hours an older employee is working), and the receipt of retirement benefits. These different potential operational definitions of the major dependent variable can be problematic. On top of that, women's history of employment tends to be more irregular than men's, with more periods of interruption in employment (Gratton and Haug 1983), and this could make the transition from employment to retirement less clear. The present study's measure of retirement status is a self-report. Fortunately, some previous studies have measured retirement in multiple ways. Both Belgrave 1988 and Palmore, Burchett, Fillenbaum, George, and Wallman 1985 measured retirement status in three different ways and concluded that the results, including results for women, were similar regardless of which measure was used, and so they both reported results for only one measure (primarily a self-report for Belgrave and receipt of retirement benefits for Palmore et al.).
Each year the NLS asks "what were you doing most of LAST WEEK—working, keeping house, or something else?" On the basis of the respondent's answer, the NLS then codes the response into one of eight categories, one of those being Retired.
Unemployment rate.
The NLS contains information on the unemployment rate for the labor market of current residence. These measures are based on Current Population Survey's primary sampling units (PSU), which are geographical sampling areas developed by the U.S. Census Bureau and made up of one or more contiguous counties or Minor Civil Divisions (Center for Human Resource Research 1995). These measures were available for 1986, 1987, and 1989, three of the four years included in this study.
Munificence and dynamism.
The data to compute munificence and dynamism were obtained from Troy's Almanac of Business and Industrial Financial Ratios (Troy 1986–1995 eds.). Troy states that the source of the almanac's data is the U.S. Treasury, Internal Revenue Service. The approach used was similar to that used by Dess and Beard 1984. Using aggregate industry data at the national level, four measures of munificence and four measures of dynamism were computed.
Two measures, munificence–establishments and munificence–receipts, were calculated as the slope of the trend line of number of establishments and the slope of the trend line of total receipts divided by their respective mean value over a 10-year period (1982–1992). Two additional munificence variables (cost of operations and return on assets), based on financial ratios, were computed. Munificence–cost was measured as the cost of operations divided by net sales. This ratio shows for each dollar of sales how much is consumed by operating costs. A higher number indicates greater costs relative to sales. Return on assets was measured as the ratio of net income to total assets. This ratio provides an indication of how efficiently assets are being used to generate income. Next, these two ratios were each plotted over the 10-year period (1982–1992), and the slope of the trend line was taken as an indication of munificence. For the first two munificence measures, dividing the slope by the mean value essentially adjusts for the relative size of the industry. It was reasoned that because the slopes in the other two munificence measures were based on ratios, the relative size of the industry was already taken into consideration. Therefore, for the final two munificence measures, it was deemed inappropriate to divide the slope by the mean value. It should be noted that, unlike the other three munificence variables, a negative slope for cost of operations would be desirable because it indicates that cost of operations relative to sales has declined, thereby increasing the likelihood of increased profits.
Four measures of dynamism were computed, one corresponding measure for each munificence variable. Again, the approach was similar to that used by Dess and Beard 1984. For number of establishments and total receipts, dynamism was computed as the standard error of the trend line divided by the mean value for the 10-year period. Again, because cost of operations and return on assets were measured as ratios, it was deemed inappropriate to divide the standard error by the mean value. Therefore, dynamism for cost of operations and return on assets were measured simply as the standard error of the trend line for the 10-year period.
It was decided that manufacturing industries would be used to examine the relationship between munificence and dynamism and retirement. Manufacturing industries were selected for several reasons. First, accurate data on manufacturing industries are readily available from several sources. Second, Feldman 1994 proposition specifically identified manufacturing industries. Third, Hachen 1992 found that lower-level workers (i.e., blue-collar workers and lower-level white-collar workers) were the most affected by the environment.
Because 3-digit Standard Industrial Classification (SIC) codes from 1960 are the main industry codes used by the NLS, industry classifications had to be recoded as based on industry descriptions. This resulted in a few instances in which several industries were collapsed into a single category, and instances in which a single industry was expanded into multiple categories. Ambiguities arose and it was not possible to attach industry data to seven 1960 SIC codes. Therefore, a final count showed that 40 separate manufacturing industries were included. The munificence and dynamism measures were linked to each record based on current industry of employment.
Analyses
The analysis of longitudinal/event history data is complicated by two problems: censoring and time-varying variables. Censoring happens when the event of interest does not occur during the observation period. Time-varying variables occur when explanatory variables change over time. To solve these two problems, models were estimated using a method presented by Allison 1984. For each survey year that a respondent was in the labor force or had been retired during that observation period, a separate observational record was created. For example, a respondent who retired in 1986 would contribute one observational record, and a respondent who retired in 1989 would contribute three records—one for 1986, one for 1987, and one for 1989. Those respondents who were in the labor force and not retired in 1992 are censored cases, because the event (retirement) did not occur during the observation period (1986 to 1992), but they still contribute what is known about them, namely that they did not retire. Allison 1984 explains that this procedure solves the problems of censoring and time-varying explanatory variables. However, the statistical tests are based on the number of records, which is greater than the actual number of individuals in the analysis. This situation can influence the estimates of standard errors and associated statistical inferences.
The data set contained 5,407 records of data, representing 2,020 women, available to test Hypothesis 1. However, the analyses involving Hypotheses 2 and 3 only used women who were employed in a manufacturing industry. As a result, 756 records of data, representing 353 women, were used in the analyses for these tests. Finally, labor market unemployment rate was not available for the 1992 survey, which resulted in 4,126 records, representing 2,002 women, available to test Hypothesis 4. An analysis of variance (ANOVA) was used to test Hypothesis 1 and logistic regression was used to test the remaining three hypotheses. For the logistic regression analyses we entered predictors in two groups. First the control variables were entered and then the remaining environmental variables. We then examined the chi-square change between models to determine the strength of the relationship between retirement and the environment.
|
Results |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
), average years of education was 11.97 (
), and the average yearly income was $30,953.70 (
). In addition, 18% of the sample had reported the presence of a health limitation and 12% of the sample was retired. Marital status showed that 58% of the sample was currently married, 17% widowed, 20% divorced or separated, and 5% were never married. The racial makeup was 73% White, 26% Black, and 1% "other."
|
|
found strong factor analytic support for the uniqueness of these constructs. Examination of the bivariate correlations involving retirement status showed that age, health limit, income, and unemployment rate significantly correlate with retirement. Thus, older respondents and respondents with a health limit are more likely to report being retired, and respondents with higher incomes and respondents who live in areas with higher unemployment rates are less likely to report being retired. Of the eight munificence and dynamism measures, only one measure, dynamism–cost of operations, correlated significantly with retirement status. It exhibited a weak negative correlation indicating that as dynamism of cost of operations increases, retirement decreases.
Hypothesis 1
Major industrial groups were used to test Hypothesis 1. The NLS used a modified coding scheme of the 1970 Census that identified 14 major industrial groups. The NLS, however, collapsed durable and nondurable goods into a single manufacturing industry and wholesale and retail trade into a single industry. The result was 12 major industrial groups: (a) agriculture, forestry, and fisheries, (b) mining, (c) construction, (d) manufacturing, (e) transportation, communication, and other public utilities, (f) wholesale and retail trade, (g) finance, insurance, and real estate, (h) business and repair services, (i) personal services, (j) entertainment and recreation services, (k) professional and related services, (l) public administration. Results of an ANOVA showed support for Hypothesis 1 and a statistically significant difference across industries for the percentage of respondents retired, 
. The 
2, however, showed that industry accounted for trivial 0.2% of the variance in retirement status. Future research might examine this issue again to determine if this is typical.
Hypothesis 2
To test Hypothesis 2, only records of respondents employed in a manufacturing industry were used. Because of missing data, the initial analysis resulted in 389 usable records out of a possible 756. Inspection of the data revealed that missing data on one control variable, Income, caused most of the records to be excluded from the analysis. When Income was removed from the equation there were 622 usable records. Although research shows that income can have effects on retirement status (Beehr 1986; Parnes 1988; Robinson, Coberly, and Paul 1985; Ruhm 1989), we feel that some of what we were trying to measure is captured in the Duncan index, which is a 2-digit measure of Occupational Prestige based on the education and income distribution of the occupation. Table 3 shows the results of the logistic regression analysis. The significant improvement of the chi-square between Model 1, containing the control variables, and 
, indicates that the addition of the munificence variables significantly improved the fit of the model. Further examination of the data shown in Table 3 showed us that two of the four munificence variables, Number of Establishments and ROA, had significant logistic regression coefficients. However, the coefficients are positive, the opposite of what was predicted. The logistic regression coefficient gives the change in the log odds of being retired for an increase of one unit in the independent variable while adjusting, statistically, for all the other variables in the equation (Hosmer and Lemeshow 1989). With the large coefficient for munificence–establishments it may be pertinent to mention that the range for this measure is .18 (.11 to -.07). In addition, the reader is reminded that the number of observations can represent a single woman multiple times. Thus, Hypothesis 2 was not supported, but environmental munificence did predict retirement.
|
. Thus, as a group, the dynamism variables did not significantly improve the fit of the model. It was noted earlier that in Table 1 dynamism–cost of operations had a significant bivariate correlation with retirement, although this provided some support, Hypothesis 3 was not supported by the logistic regression.
|
stated that his R2 is interpreted as the proportion of explained variation and, thus, appears to have the same interpretation as the R2 for multiple linear regression. When the four munificence and four dynamism variables were added to the equation, the R2 increased to .46. In comparison, total R2 for the prior munificence analysis was .44 and the final R2 in the earlier dynamism analysis was .43. Thus, these results suggest that munificence and dynamism each contribute uniquely to the prediction of retirement status. Because of multicollinearity among some of the predictors, it would be misleading to present and interpret their logistic regression coefficients.
Hypothesis 4
Inspection of the bivariate correlations (Table 1 ) revealed that local area unemployment rate had a weak negative correlation with retirement status (
, p < .05). However, the logistic regression analysis testing Hypothesis 4 indicated that after controlling for the other variables, unemployment rate was not significantly related to retirement. The data in Table 5 shows that the model improvement from Model 1, containing only the control variables, and Model 2, which adds the local area unemployment rate, failed to reach significance, 
. Thus, the results failed to support Hypothesis 4.
|
|
Discussion |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
After we controlled for a variety of personal factors, local area unemployment rate was not related to retirement status. This is in contrast to findings by Quinn 1977 that after a variety of personal and financial characteristics were controlled for, labor force participation was 3.3% higher for men who lived in low unemployment cities (below 3.5% unemployment rate). The difference could be a result of Quinn's use of male respondents to the RHS, in contrast to this study's use of the Mature Women Cohort of the NLS. Thus, gender may moderate the relationship between the decision to retire and unemployment rate. In one study, Talaga and Beehr 1995 found that there are gender differences in predicting the retirement decision. In addition, there could be a cohort effect moderating the relationship between retirement and unemployment rate. Retirement is a cultural phenomenon, and, as a result, cohort effects are likely. The women examined in this study are reaching retirement age approximately 20 years after the men in the RHS. Sherman 1985 showed that reasons for retirement dramatically changed between 1968 and 1982. Thus, attitudes toward retirement may differ across cohorts, and the predictors of retirement may vary with cohorts.
Environmental dynamism was not related to retirement status, but environmental munificence was positively related to retirement status. It was believed that individuals in highly munificent environments would continue to work, however, the results indicated the opposite. A straightforward explanation is that when times are good individuals retire while they can, and when times are bad individuals cannot afford to retire. When considered in tandem with the results regarding unemployment, dynamism, and the explanation offered to explain the results of Hypothesis 1, this seems to go against environmental uncertainty as the basis of retirement. Instead, environmental munificence seems to be an environmental level reflection of the individual level hypothesis that wealth results in retirement (e.g., Quinn, Burkhauser, and Myers 1990).
After testing our munificence and dynamism hypotheses, we conducted an additional analysis that included the industry measures in a single regression equation. The results suggested that munificence and dynamism each contributed uniquely to the prediction of retirement status. However, with a more complete data set, it would be particularly interesting to include the unemployment variable in the equation to determine if unemployment rate contributes uniquely to retirement status, while the industry measures are controlled for. Furthermore, future research on munificence and dynamism also seems warranted as general conceptual constructs, because their intercorrleations and the factor analysis suggests they may not clearly be unidimensional constructs and their elements may have systematic relationships with each other. If they can be considered separate variables, future research could consider interactions between the variables. For example, does a more dynamic environment have a larger effect on retirement decisions when the environment is less munificent.
There are several limitations of the present study that could have affected the findings, and future research would benefit from addressing each. First, the sample consisted of women only. Traditional gender roles suggest that women and men have different employment histories, and gender differences have been found in predicting retirement decisions (Talaga and Beehr 1995). Thus, it would be beneficial to further examine gender differences in the relationships between the environment, especially unemployment rate, and retirement status. This suggests that mixed-gender samples and moderator analyses would be useful. Second, broad industry classifications (3-digit SIC codes) were used to compute environmental munificence and dynamism. The use of 4-digit codes would more accurately measure the environment. However, there is still substantial variation within 4-digit SIC codes; aggregate measures of industry characteristics may not accurately describe all companies within a given classification. Perhaps measuring company rather than industry growth and instability may be more desirable. Third, because of missing data problems, Income was not used as a control variable, but the Duncan index of occupational prestige largely atoned for this omission. Another potentially important financial variable, however, is Wealth or Assets, for which the study had no measure. Fourth, only manufacturing industries were used in the analyses examining munificence and dynamism. Obviously, it would be desirable to determine if these findings apply to other industries. Fifth, objective measures of environmental munificence and dynamism were used. Weick 1979 argued that there is no such thing as an objective environment. Instead, he believes that all environments are enacted and essentially perceptual in nature. Thus, investigating the relationship between an individual's perception of his or her environment and decision to retire would be interesting.
Received for publication June 14, 1999. Accepted for publication December 23, 1999.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
