Chapter 16 Patterns of work-to-retirement transition among Canadian public-sector employees1
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
With special assistance from Harpreet Randhawa
Systematic inter-sector variations in patterns of transitions to retirement
Presence of flexibility in the work-to-retirement transition
Major variations in patterns of transitions to retirement among key sub-groups of public sector employees
In 2004, the Institute of Public Administration of Canada conducted a survey of very senior public servants: federal and provincial Deputy Ministers, and municipal Chief Administrative Officers. They were asked to identify the top ten issues facing public sector managers. Number one, by a wide margin, was called "retirement, recruitment, retention and succession planning (cited 75 times)" (Marson and Ross 2005). This finding means that many public sector managers are focusing upon the consequences of the major wave of retirements that Baby Boomers will unleash during the next 10 years.
Long ago analysts used to think that the Baby Boomers' retirement wave would begin about the time when their youngest members reached age 65, about 2011. But the wave will become substantial well before that time; because the average age of retirement has fallen below 62 in recent years, as Chart 16.1 shows.
To help those leaders prepare for the said retirement wave, the existing body of basic knowledge about patterns in work-to-retirement transitions should be substantially improved. Part of that improvement should focus upon the public sector and this for two reasons. The public sector is perhaps the venue of the most advanced planning concerning gradual retirement in Canada. Yet, what is available as information about retirement patterns in the public sector has been limited largely to brief commentary about the trends shown in the following Chart 16.1.
What are the persistent and important differences in patterns of transition from work to retirement between public-sector employees and those in the private sector? To what extent can these differences be explained by variation between these two sectors in aspects of employee attributes that greatly influence patterns of work-to-retirement transitions? What major variations in patterns of work-to-retirement transitions exist among key occupational or "employer" subgroups within the public sector? These are the key research questions that have guided the work in this chapter. The answers to these questions will contribute to the foundation of basic knowledge that human resource policy makers and managers of all sectors will find useful as they begin to confront the baby boomers' coming retirement wave.
Answers to the questions will also help to advance fundamental social -science-based knowledge that is needed in teaching about and understanding about a major aspect of life courses in Canadian society. Definitive answers require much more research than that done for this book, and that research needs to be based on data that provide more adequate samples for key population subgroups than the ones now available. This chapter is, therefore, a small contribution toward the needed answers.
To make this contribution, we have drawn upon the data for Panel Two of the Survey of Labour and Income Dynamics (SLID). The target population for SLID is all persons living in Canada, excluding people in the Yukon or Northwest Territories, residents of institutions, persons living on reserves, and full-time members of the Canadian Armed Forces living in barracks. The exclusions constitute about 3% of the population.
The members of Panel Two of SLID were repeatedly interviewed; about 12 times, from 1996 to 2001.The data for the first two years have been used to determine which panel members had begun their transitions to retirement during 1996-1997. Focusing on these members, we then mapped transitional patterns from 1998 to 2001. (Additional information about SLID is given in Appendix A.) See also Statistics Canada 2004.
To benefit from reading this chapter, the reader will need to absorb several new concepts. This section presents meanings for these concepts using plain language as much as feasible (the more technical definitions have been placed in Appendices A and B).
Central to what follows are the terms "public sector employee" and "private sector employee". These are widely used concepts and it seems sufficient to note here that in Canada "public sector employee" generally refers to someone employed in a government agency, or in a government-financed hospital or school. All levels of government are included.
"Private sector employee" generally means someone who is employed by a non-government organization, and excludes the self-employed. In Canada such an organization is most likely to be a profit-making entity.
In classifying a person to the category named "public sector employee" or that of "private sector employee", we have required that she be employed in the same sector in each of the years 1996 and 1997. Moreover, we excluded those who had any self-employment in those years.
To compare patterns of transition to retirement between public sector employees and those of the private sector, one must define the phrase "transition to retirement", and devise a scale to identify who began their transitions to retirement with a specified time period. The scale, named TRANSCOR, is presented in Appendix A.
In this chapter, "transition to retirement" refers to a process that often involves multiple movements among designated positions toward the state of being retired. ("Retirement" here means the state of being retired, which is often marked by a long-term departure from the labor market, accompanied by the receipt of some kind of retirement-related income.) The specific sequence of positions that comprises one person's transition to retirement is called her "trajectory of transition to retirement". (See Appendix A for more details.)
Trajectories have properties that we can define and measure. And we compare patterns of transition to retirement between public and private sector employees in terms of these properties of trajectories of transition to retirement.
This chapter deals with three properties of trajectories of transition to retirement. These properties are speed of closure, presence of flexibility in the work-to-retirement transition and exposure to events that threaten to reduce standard of living in retirement. The following paragraphs will offer brief remarks about each of these properties, to allow the reader to understand the research findings that follow without having to spend a lot of time studying the more detailed information given in Appendix A.
A person (identified as being in transition during 1996 to 1997 is said to have closed her trajectory when (a) she has left the labor market and has been in receipt of some form of retirement-related income for at least six consecutive months, and (b) following those six months she did not return to the labor market up to the end of 2001. The sooner the person begins this period of uninterrupted departure from the labor force (while receiving some form of retirement-related income) the faster is her speed of closure.
Closure can begin in any quarter of the 15 quarters from the first of 1998 to the third of 2001. A sixteenth category holds those whose trajectories were unclosed as of the end of December 2001 (for details see Appendix A). Although this implies 16 levels of speed of closure, based on the quarter as the unit of time, they are grouped into four broad speeds, when comparing the patterns of public sector employees with those of their private sector counterparts.
In this chapter, flexibility in the transition to retirement is measured by means of counting certain movements that form part of a person's trajectory of transition to retirement. The movements in question are those that indirectly point to certain voluntary actions - those that reflect the use of available options or choices in how one's retirement process unfolds.
Whereas the index of flexibility deals with movements that suggest voluntary changes undertaken by the SLID respondent, the index of vulnerability focuses on job loss and involuntary job change. In this text "vulnerability" means risk of loss or setback. Here we are referring to risk of setback to whatever plans or arrangements the person has made concerning standard of living in retirement.
This section describes the results of our research aimed at finding persistent and important differences in patterns of transition from work to retirement between public-sector employees and those in the private sector.
The plan of the exposition that follows is to first highlight the overall pattern of inter-sector difference between the public and private sectors. Then follows a review of how much this pattern varies among key segments of the target population, in the mode of bivariate analysis. Finally, the text addresses the issue of whether the patterns shown in bivariate analysis are maintained in multivariate analysis where many relevant variables are held constant statistically.
Speed of closure of trajectories
Bivariate Analysis. On average, private sector employees, aged 45 to 69 in 1996 and beginning their work-to-retirement transitions during 1996-1997, closed their trajectories more rapidly than their public sector counterparts (Chart 16.2). Just over 25% of the former group had closed their trajectories by the third quarter of 1998, in contrast to nearly 20% of public sector employees. At the next category of speed of closure (closing between the fourth and ninth quarters), the trajectories of private sector employees again closed at a markedly higher rate than those of public sector employees.
Distributions according to speed of closure of trajectories for public-sector and private-sector employees, by sex, cohort aged 45 to 69 in 1996, Canada, 1998 to 2001
Thus, in the last quarter of 2001, the public sector employees were more likely to have unclosed trajectories. (That is, they had not left the labor market for six consecutive months ending in December 2001.) Close to 70% of public sector employees had unclosed trajectories, which was more than 10 percentage points greater than the figure for private sector employees. These figures apply to those who were in the labor market in the first quarter of 1996, and who were judged to have begun their transitions to retirement during 1996 to 1997.2
The overall pattern, then, is that, among those that had begun their transitions to retirement during 1996-1997, the private sector employees were more likely than their public-sector counterparts to have closed their trajectories within the first two of the four years that are covered by the trajectories. Thus at the end of 2001 the public sector employees who began their transitions during 1996-1997 were more likely to be still in the labor market.
However, among age groups this overall pattern is shown only for those aged 60 or more in 1996. Below that age, employees in the two sectors had similar percentages with unclosed trajectories at the end of 2001 (see Chart 16.3).
Percentages with unclosed trajectories, for public-sector and private sector employees, by age in 1996, Canada, 1998 to 2001
Because the population aged 45 to 54 is normally a high proportion of those aged 45 to 69, the pattern of variation by age just cited may be surprising. However, it is explained by the fact that the data refer only to those that had begun their transitions to retirement during 1996-1997. Within this sub-population the percentage of those aged 45 to 54 is much lower than the corresponding percentage in the whole population aged 45 to 69.
Since the overall pattern cited above is being heavily weighted by those aged 60 or more in 1996, a key question arises. Have we estimated the pattern for the 60-plus-year-olds reliably? For example, is the sample too small for great confidence in this result?
A standard test of statistical significance is the usual way of addressing this question; but taking this route places undue emphasis on the precision of the estimate of one single percentage. This is not appropriate when our focus is in fact on a pattern of variation formed by a series of numbers, without particular attention to their actual values. Given this focus, the statistical significance issue becomes the following: how easily would this pattern arise by chance?
This question can be answered by a computation that uses bootstrap principles, where patterns, rather than single numbers, are examined; but there are also rough and ready tests that are useful. First, the analyst can sub-divide the sample into various meaningful sub-groups, and recompute the pattern within each one. Second, the analyst can step back to larger samples by relaxing some of the restrictions used in arriving at the original finding, and once again recompute the pattern. If the same pattern tends to recur across all those recomputations, it is probably a pattern that would not arise easily by chance. The next few paragraphs provide some of the results of the various recomputations done. (A further recomputation will be done within the context of multivariate analysis presented below.)
Both Charts 16.2 and 16.3 are for persons we judge to have started their transitions to retirement during 1996 to 1997. To make that judgment we devised a special scale, named "TRANSCOR". We adopted a threshold value of 3.0 for TRANSCOR. SLID panel members were judged to have started their transitions to retirement if their ratings on TRANSCOR were at or above this threshold. (For details see Appendix A.)
An initial step in the said recomputations is to lower the threshold. We lowered the TRANSCOR threshold to 2.1, and found no change in the pattern. There were changes in individual numbers but none in the pattern of variation.
The second step was to ignore TRANSCOR altogether. Taking all people aged 50 or more in 1996, we recomputed the pattern. Once again, there was confirmation of the basic pattern. Among those aged 60 or more, rapid closure of trajectories was more prominent among private sector employees.
Next we took the sample of those we judged to have started their transitions between 1996 and 1997, and broke it down into meaningful sub-groups. The said computations were repeated within each sub-group.
Subgroups defined in terms of gender, education and income show a strong tendency toward repetition of the overall pattern cited above. This pattern described above is repeated in each of three separate levels of education (less than high school graduation, high school graduation with no university degree, and university degree), as well as in the first and fourth quartiles of household income in 1996.3Chart 16.2 shows the patterns for men and women separately.
Chart 16.2 shows that the pattern is much more noticeable for men than for women. At the end of 2001 the percentage with unclosed trajectories is substantially higher for public sector male employees than for their private sector counterparts. This divergence is much smaller among women employees (although the direction of the intersector difference remains the same for males).
In summary, as regards patterns in speed of closure, among those that started their work-to-retirement transitions between 1996 and 1997, there is a divide between those aged less than 60 in 1996, and those aged 60 or more. Among the former, public-sector employees were more likely, than their private sector counterparts, to have closed their trajectories before the end of 2001. However, this pattern is reversed among those aged 60 or more in 1996. In this group, speed of closure was slower for the public sector employees - they tended to stay longer in the labor market.
Multivariate Analysis. Does the just highlighted intersector difference in pattern of speed of closure of trajectory arise because of factors that explain both the probability of being employed in one of the two sectors (public versus private) and the speed of closure? If the answer is "yes" it would mean that the association of sector of employment with speed of closure is the result of their separate associations with a third variable. If this third variable is an aspect of the composition of the employee populations, then the speed-of-closure variations just cited cannot be attributed to differences between the work environments of public and private employees.
To address this issue, we have devised a multivariate model that generates predicted probabilities for the different speeds of closure, based on combinations of values of several predictor variables. Among these variables is sector of employment, as well as several others that should be taken into account in attempting to explain speed of closure of the trajectory of work-to-retirement transition. The selection of these variables is based on our theoretical considerations about the processes involved in determining the gradualness of retirement processes, or on findings from other research projects.
Appendix C presents the model and the variables, and discusses the overall performance of the model as well as the contributions of important variables that are not part of the story line of this chapter. A few key variables should be noted here, however. Among them are several measures of change; all of them having as their reference date the year before closure of when the trajectory began. They include marital status change, whether there was an increase in responsibility for providing family care, health status decline, a proxy for increase in wealth, and whether another economic family member began receiving retirement-related income. There is also a measure of irregularity in the person's work history.
Table 16.1 shows that the odds favoring a particular speed of closure are very similar between the two sectors. However, there is a slight tendency for public-sector employees to close trajectories more rapidly than those of the private sector, when all other predictors are held constant simultaneously.
Adjusted patterns of association of public and private sector employment with speed of closure of trajectory of work-to-retirement transition during 1998 to 2001, for three nested models, Canada
The first two positive values in column C indicate that, relative to the reference category of class of worker (category number five), being an employee of either sector tends to increase the probability of having an unclosed trajectory. However, the tendency is slightly greater for private sector employees (see the odds ratios of 5.5 and 5.2 in column C). 4
Moreover, although the results in the first two sub-models (see columns A and B) are subject to great variability due to sampling (see the values of the Wald chi-squares and footnote 4 to the table), they too show a slight advantage for private sector employees as regards probabilities for the slower speeds of closure.
However, if we held age constant in the 60 to 69 range, we would see a different pattern than that shown in Table 16.1. Table 16.2 holds age constant in that range, and, in contrast to Table 16.1, it supports the observation that speed of closure of trajectory of work-to-retirement transition was slower for public sector employees than for their private sector counterparts. In contrast to the very close odds ratios found for the two sectors in Table 16.1, the corresponding ones in Table 16.2 are 5.5 for the public sector and 2.9 for private sector employees. The odds in question are those favoring an unclosed trajectory relative to a trajectory that closes rapidly.
Adjusted patterns of association of public and private sector employment with speed of closure of trajecory of work-to-retirement transition during 1998 to 2001, for three nested models, Canada
There is a caution to be sounded here. The Wald chi-squares in Table 16.2 tell us that these estimates are very fragile, relative to sampling variability - the culprit is probably small sub-sample size. Yet, the very high odds ratio for the self-employed aged 60 or more, which is expected (see the preceding chapter for details), suggests that this pattern is reasonable, even though it is statistically unstable.
In summary, both the bivariate associations presented earlier and the multivariate analysis just reported concur as regards systematic differences between the two sectors of employees concerning speed of closure of the trajectory of transition from work to the state of being retired.
Holding constant other key variables such as gender, marital status, cultural group, education, occupation and household income does not have a substantial impact on the pattern of association of sector of employment with speed of closure of trajectory of transition from work to retirement. (Patterns of association involving these variables are presented in Appendix C.)
With regard to trajectories, flexibility in the transition to retirement is indicated by changes of position that indirectly point to voluntary actions that reflect the use of available options or choices in how one's retirement process unfolds. (See Appendix A for elaboration of this idea.)
Public sector employees are more likely, than their private sector counterparts, to have either Medium or High levels of the index of flexibility in the transition process. However, the advantage for public-sector employees is concentrated at moderate values of the index; because at the high values both sectors' trajectories tend to have similar percentages (see Chart 16.4). Slightly more than 20% of public sector employees had trajectories with Medium or High levels on the scale of flexibility of the transition to retirement, and this was about six percentage points higher than the figure for their private sector counterparts. However, close to 5% of each group is found at the High level on the scale.
Index of flexibility in the work-to-retirement transition, for employees in the public and private sectors, cohort aged 45 to 69 in 1996, Canada, 1998 to 2001
The overall pattern5 of intersector variation just described tends to recur among the broad age groups being studied (unlike the case of speed of closure) and within various levels of gender, education, income and occupation. The pattern is confirmed both when the TRANSCOR threshold is lowered to 2.1; but is not supported consistently when TRANSCOR is disregarded and the sample comprises all persons aged 55 or more in 1996.
Among those with a university degree, the "lead" of the public sector, as regards the index of flexibility of work-to-retirement transitions, is especially notable. At this level of education, the difference between the two sectors in terms of the percentage of trajectories with Medium or High values on the scale of flexibility is distinctly greater than is the case for the whole sample. And this observation is made even in the population of persons aged 50 or more in 1996.
To what extent does this pattern recur after we hold constant several variables that may simultaneously explain both (a) sector of employment and (b) level of the index of flexibility? To answer this question, we have devised a multivariate model that generates predicted probabilities for the different levels of the index of flexibility, based on combinations of values of several predictor variables. Among these variables is sector of employment, as well as several others that should be held constant in attempting to explain flexibility of the work-to-retirement transition.
The variables held constant include sex, age, an index for irregularity of work history, marital status in 1996, whether there was another economic family member who began receiving retirement-related income in the year before closure of the trajectory started, health status in 1996, education level in 1996 and occupation group in 1996. When all of these were held constant, the parameter estimates and odds ratios were so similar between the two sectors that they cannot be said to be substantively different (see Table 16.3).
Adjusted patterns of association of public and private sector employment with flexibility of the trajectory of work-to-retirement transition during 1998 to 2001, for two nested models, Canada
The model also fits the data poorly in the sense that it does little better than the null hypothesis model, which asserts that all the predictor variables have parameter estimates valued at zero. The error of prediction of the latter model is reduced by only 10% (the tau-a measure of goodness of fit is 0.13, n = 525).
In short, we do not have large enough sub-samples from each sector to conclude that the multivariate analysis conveys any message about whether the intersectoral differences are substantial after taking into account potentially confounding variables. The model does suggest, however, that these differences are not large, as did the bivariate analysis. It is important to add that the most useful predictor variables may be absent from the model, due to their absence from the SLID file.
Exposure to events that threaten to reduce standard of living in retirement
The index of flexibility deals with movements that suggest voluntary changes undertaken by the SLID respondent while the index of vulnerability focuses on job loss and involuntary job change.
The index of exposure to events that increase risk of reduction in standard of living shows the two sectors with very similar levels of concentration at the Low level; both being in the vicinity of 85% (Chart 16.5). The public sector employees have a slightly higher concentration at the Medium or High levels (taken together). However, this slight divergence disappears when we resort to the much larger sample of all persons aged 50 or more in 1996.
Index of exposure to events that increase risk of reduced standard of living in retirement, for employees in the public and private sectors, cohort aged 45 to 69 in 1996, Canada, 1998 to 2001
Both samples (that with TRANSCOR of 3.0 or more, and that comprising all persons aged 50 or more in 1996) show the following pattern: public sector employees whose trajectories are rated at Medium or higher levels on the index were heavily concentrated at the Medium level. At the High level of the scale, the percentage of private sector employees was greater than that for public sector employees (substantially above 5% for the former, versus well under 5% for the latter, and that in both samples). Thus, public sector employees were less likely to be found with trajectories that had High scores on the vulnerability scale.
The different levels of education show broadly similar variation between public sector employees and their private sector counterparts. Chart 16.6 shows a systematic pattern. As one shifts from those with less than high school education toward those with university degrees, the overall level on the vulnerability scale falls; and the gap between public sector employees and their counterparts from the private sector has the same direction as that reported above.
Percentage of persons at Medium or High levels of the index of exposure to events that increase risk of reduced standard of living in retirement, for employees in the public and private sectors, cohort aged 45 to 69 in 1996, by sex and education, Canada, 1998 to 2001
Such persistence across subgroups of the population would be difficult to dislodge in multivariate analysis of the association of employment sector with the index of vulnerability, unless there is a strong variable that simultaneously affects both. No such variable has been found in our analysis; only one parameter estimate, among the levels of the class-of-worker variable, is statistically significant at the15% level, and none at the 5% level (all Wald chi-squares are below the value of six). This problem is not resolved by using the sample of all persons aged 50 or more; because in this sample we have selected only those that were employed throughout the last quarter of 1997. Thus, once again, small sub-sample sizes are preventing the multivariate analysis from conveying useful information to validate the patterns shown from the bivariate analysis above.
Major variations in patterns of transitions to retirement among key sub-groups of public sector employees
In beginning this chapter, we advocated that human resources managers in all sectors will benefit from a focus of analysis upon the public sector, where the governmental labor-market environments have much in common, and where there is a relatively advanced state of deliberations concerning retirement-related policies and programs among a small number of employers. The benefits of this focus are enhanced by considering variations among key "employer" and occupational groups within the public sector.
The phrase "employer group" is a shorthand reference to a subdivision of the public sector that is based partly upon the different levels of government (federal, provincial, municipal) as well as on major blocks of services such as those pertaining to education and health. In attempting to define meaningful "employer groups", we are limited by very small sub-samples in SLID for this sector and we make use of properties of the industrial classification system as it is used by SLID (the North-American Industrial Classification System - NAICS).
The closest approach to usable (but often inadequate) sub-sample sizes is achieved with the following three categories:
1 federal government administration (the phrase used in the listing of NAICS classes)
2 provincial and municipal administration
3 schools (all levels) and hospitals.
It is possible that some elements of the public sector are missed in these three groups. Also, the third category may include some private schools. In any event, those are the three "employer groups" to be used below, and they will be called "federal government", "other governments", and "schools and hospitals".
An occupational breakdown of the sample has three classes that deserve attention in this study: managers, Professionals Type A (generally, professionals excluding those in the arts, sports, food preparation and personal care), and a collection of clerical and technical workers. There is a residual class of "other occupations".
The remainder of this section presents variations among the stated categories of public sector employees with regard to patterns formed by their trajectories. The properties of trajectories that will be examined are speed of closure and flexibility. The sub-samples are too small to permit presentations concerning either the vulnerability index (where the sub-sample is limited to those who had a job in the last quarter of 1997), or the propensity to return to the labor market after departure (which deals with those that had left the labor market between 1996 and 1997).
Among the "employer groups", the federal employees show the slowest speed of closure of trajectories (or the most gradual transitions to retirement - see Chart 16.7). Their level on the flexibility index is second to that of employees in schools and hospitals (Chart 16.8). The margins of difference among the three employer groups are not large, however.
Distributions according to speed of closure of trajectories for employees in three public-sector employer groups, cohort aged 55 to 69 in 1996, Canada, 1998 to 2001
Percentage of public-sector employees at Medium or High levels of the index of flexibility in the work-to-retirement transition, cohort aged 55 to 69 in 1996, for three employer groups, Canada, 1998 to 2001
Another notable pattern is that for school and hospital employees. Their speed of closure is fastest. Also, as noted above, they have the highest score on the scale of flexibility of their trajectories.
Among the occupational groups, that called "Professionals Type A" (defined above) has slower than average speed of closure (the greater than average prevalence of gradual retirement), and higher than average levels of flexibility. However, the inter-occupational variations are very modest.
In sum, the federal government, among the "employer groups", and professionals, among occupation groups, seem to exhibit the greater than average tendencies towards gradual and more flexible transitions to retirement.
Whether these patterns can be explained away by reference to the attributes of employees is the question to which we now turn via multivariate analysis. To the extent that these attributes are insufficient to account for the variations, there would be an indirect indication that some study of the groups' institutional environments may be needed, in the effort to understand their differences in patterns of work-to-retirement transition.6
Given the small sample size involved in studying the public sector employees only, the multivariate analysis is limited to the association that is most striking in the data shown above - that between "employer group" and speed of closure.
For this association, the overall pattern described above is confirmed in the multivariate analysis; but none of the parameter estimates is reliable (the Wald chi-squares all indicate levels of significance below 15%).
In summary, the bivariate and multivariate analyses concur that the federal government, among the "employer groups" of the public service, seems to exhibit the greatest tendency towards gradual transitions to retirement. The bivariate associations also indicate that flexibility in the transitions is greater for federal government employees than for other government employees; but less than that of employees of schools and hospitals.
Information that public-sector employees retire at an earlier age, on average, than those in the private sector has been in the public domain for some time, based on questions asked in the Labour Force Survey (Chart 16.1). At first glance, it seems to run counter to our finding that private sector employees were quicker to close their trajectories of work-to-retirement transitions.
However, there is a route toward reconciling these findings. Consider the way that the LFS-based estimate is produced. Respondents who have left the labor market within the last 12 months are asked why they did so, and among the possible answers is "retired". The subset who said "retired" included a substantial proportion who would soon return to the labor market. In contrast to the LFS approach, our finding relies upon identifying a group who were in the labor market in the first of 24 quarters during which they were repeatedly interviewed, and whom we identified as having started their transitions during the first eight quarters. Only those who subsequently left the labor market and did not come back (up to the end of the 24 quarters) are said to have closed their trajectories. Hence the two approaches use substantially different measures that offer little theoretical basis for the expectation that their estimates be similar.
Yet both data sources show a similar pattern of intersectoral differences when the SLID sample is restricted to persons below the age of 60 at the starting date. The subset of our panel that was designated as beginning their transitions during 1996-1997 had a relatively large proportion of people aged 60 or more, compared to the general population of older Canadians. And it is this set of 60-plus-year-olds who have generated our finding that public sector employees are more likely to delay retirement.
This result becomes more interesting when we recall that within the public sector, the federal government seems to be the venue most likely to show more gradual and flexible retirements. One may suggest the hypothesis that it is primarily in the federal government where those aged 60 or more are more likely than the average (for that age group) to have gradual retirements. This hypothesis would suggest that the employees of provincial and municipal governments, and especially of hospitals and schools, account for the widely publicized tendency for public service employees to retire earlier.
This idea brings up a notable question for consideration and analysis in the future. What is it in the public sector environment, and perhaps in the federal government especially, that may be inducing the oldest group of those in transition to retirement (ages 60 to 69 at the start of the transition) to delay retirement to a greater degree than their private sector counterparts?
We find that flexibility in the process of making the transition from work to retirement is more likely to exist among public sector employees than among their private sector counterparts. However, the margin of difference is very slight, although substantive significance is indicated by the recurrence of the pattern among the broad age groups studied and within various levels of gender, education, income and occupation.
The two sectors seem to be on a par as regards their percentages at the Low level of the index of exposure to events that increase the vulnerability of incomes in retirement. Among those above the Low rating on this scale, the public sector has a higher profile at the Moderate level while the private sector has a higher profile at the High level of the scale. Once again, however, the difference is small.
Why aren't the differences between the two sectors, public and private, more dramatic? As regards the indexes of flexibility and vulnerability in the transition to retirement, to answer this question we need to start with an underlying theory about the determinants of flexibility or vulnerability in the transition to retirement. We have not had time to try and produce such theoretical work.
As regards speed of closure, our theoretical work offers some grounds for discussion of the question just posed. The essential ideas are as follows. At the individual level we postulate that the speed of closure is the outcome of three processes:
(A) making choices to reach goals under certain constraints (see, for example, Parker and Rougier 2004),
(B) negotiating in response to behaviour changes made by significant others in your social network (see Rasmusen 1995, Lin 2003),
(C) coping with major intrusive life events and their consequences (see Ma and Zhang 2004; Clark et al. 2004).
These processes apply to individuals with probabilities that vary from one person to the next.
The second two forces, B and C, are quite important, evidence suggests, in any effort to understand retirement behavior. Force B is, in fact, a sort of rising star among determinants of retirement behavior. It is causing a depreciation of the value of the large body of theory which assumes that lone persons are timing their retirements in order to optimize personal lifetime consumption utility subject only to a budget constraint (this is covered in force A).
Forces B and C are largely beyond the influences of employer policies, union policies, or the corporate working environment. And force A is not entirely determined by these macro-level factors. Consequently, a substantial proportion of the determinants of speed of closure are not affected by the working environment or by corporate policies. This means that corporate policies and work environments should be expected to have limited influence on speed of closure of the trajectory of transition to retirement..
The following questions arise in the context of related policy deliberations. "How far can policies succeed within the margin of possible influence that is available? And is this much success worth the cost of achieving it?"
New research is needed to validate the findings just discussed, especially in terms of using more comprehensive measures of flexibility and vulnerability in the work-to-retirement transition process. Given that validation, explanatory theory and hypotheses should be developed and tested.
Finally, there is an implication for policy-relevant statistical analyses. Statisticians and other researchers should look beyond the global figures (which indicate public sector employees retiring earlier), and break down the data to show trends for different age groups and for major segments within the public sector.
Accenture. 2005. Creating Public Sector Value in a Rapidly Aging World.
Clark, P.M. et al. 2004. "A model to estimate the lifetime health outcomes of patients with Type 2 diabetes: The United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDSno. 68)." Diabetologia. 47, 10:1747 to 1759.
Department of the Premier and Cabinet Government of Western Australia. 2003, March. "Phased retirement in the western Australian public sector." Ageing Workforce. Discussion Paper Series. (amended March 2004).
Kieran, Patrick. 2001. "Early retirement trends." Perspectives on Labour and Income. Online edition. 2, 9, September. Statistics Canada Catalogue no. 75-001-XIE.
Lin, Raymund J. 2003. Bilateral Multi-Issue Negotiation.
Ma, Xin and Yanhong Zhang. 2004. A National Assessment of Effects of School Experiences on Health Outcomes and Behaviours of Children: Technical Report.
Marson, Brian and Peter Ross. 2005. "Targeting managers: Demographics, citizen expectations, fiscal pressures keep the pressure on Deputy Ministers, CAOs." Public Sector Management Magazine. 16, 1.
Parker, S.C. and J. Rougier. 2004. The Retirement Behaviour of the Self-Employed in Britain. Working Paper in Economics and Finance no. 04/08. Durham, UK. University of Durham.
Policy Research Initiative (PRI). 2004. Population Ageing and Life-Course Flexibility: The Pivotal Role of Later Retirement. Draft Discussion Paper. Ottawa.
Rasmusen, Eric. 1995. A Model of Negotiation, Not Bargaining.
Statistics Canada. 2004. Survey of Labor and Income Dynamics. Ottawa.
Statistics Canada. 2002. Guide to the Labour Force Survey. Catalogue no. 71-543-GIE. Ottawa.
The Graham Lowe Group. 2003. Phased-In Retirement Options Needed for Skill Shortage Challenge.
- The authors thank the peer reviewers for their contributions to improving earlier drafts of this chapter. The comments of Bob Baldwin, Kevin Cahill, Hervé Gauthier and Bruno Rainville were especially helpful. Assistance from Harpreet Kaur Randhawa is gratefully acknowledged. All opinions and errors herein are our sole responsibility.
- In the much larger sub-samples obtained when all persons aged 50 or more in 1996 are considered, this pattern is reversed. The percentage with unclosed trajectories is just below 75% for private-sector employees, and this is welll above the figure of above 65% for public-sector employees. The reason for this reversal is given in the text that follows.
- To save space and in some cases when the sub-samples are too small to permit publication of specific numbers, data patterns for sub-groups are cited without display of supporting tables or charts in this book. The reader may request unpublished tables or charts where inadequate sample size is not an issue to release data.
- Readers who need help with interpreting the parameters and odds ratios shown in this table should review the short text desinged to offer this help in Appendix C.
- References to an "overall pattern" in this text do not include specific numbers cited; but rather they point to the relative magnitudes of numbers, directions of differences or change, etc. The focus is on a pattern of variation formed by a series of numbers, rather than to the numbers' specififc values.
- Unfortunately, the exercise that follows cannot provide a definitive answer on this issue because it is not demonstrable that all pertinent employee attributes have been taken into account in the model.
- Date modified: