Bruce R. Bacon

New measures developed to assess the burden which future workers will have to bear in providing for retired persons indicate that this burden will increase by about 50 per cent in the first three decades of the 21st century.

This paper looks at the dynamics of work and retirement in Australia and its interaction with the ageing of the population. In particular, the paper develops measures for the number of retired and provides estimates of retirement rates as the basis for generating projections of dependency rates into the twenty-first century. These measures and estimates are fundamental to an understanding of the public policy dilemmas of financing the dependent-aged population as Australia’s population ages.

INTRODUCTION

Over the last decade there have been major changes in the working patterns for both males and females in Australia, including:

Notwithstanding the obvious differences between male and female labour-force attachment, a general proposition which underlies this paper, is that the work/retirement behaviour of males and females is slowly converging. The relative stability of all of these underlying trends will determine how well we are able predict work/retirement outcomes.

WORKING LIFE

Total participation rates in Australia have been rising over the last two decades. They reflect an increased female participation which is partially offset by a falling male participation.

These trends have their origins in a number of fundamental supply and demand factors occurring in our society.¹

On the labour demand side there have been:

• attitudinal shifts by employers with regard to employing women,
• growth of industries which favour female employment, and
• increased use of part-time (and casual) employment (which fits the lifestyle requirements of many women).

• a narrowing of the relative pay gap between males and females,
• increased access to child care along with smaller families,
• delay in marriage,
• delay in childbirth, and
• changes in marriage rates including increasing numbers of never-married and divorced females.

These factors appear to be producing a convergence of male and female labour-market behaviour. In particular, unmarried women are behaving more and more like men and the behaviour of married women is converging, albeit more slowly, towards that of unmarried women. This general proposition underlies much of the projection work of the Retirement Income Modelling (RIM) Task Force.

Figure 1. Labour Force Participation by Age Group

Source: ABS 6203.0 and Young 1994

The convergence in labour-force behaviour can be easily seen in Figure 1 which shows the age-specific participation rates of females over the last fifty years.² Over this period female labour-force participation has been shaped by child-bearing/raising responsibilities which currently account for over fifty per cent of the spells out of the work force.³ However, when one discounts for child bearing/raising, women (particularly younger women) are behaving more and more like their their male counterparts as regards their participation in the labour market.⁴

Aggregate data can, however, mask the differences between males and females. The aggregate trends in labour force participation disguise the changes that have occurred between full-time and part-time employment. In particular, male part-time employment has been increasing, admittedly from a low base, and is only partly offsetting the significant falls made in full-time participation. Women, on the other hand, have increased their participation rates for both full-time and part-time work.

In summary, there appear to be four major mechanisms underlying the labour force participation patterns:

• gender shifting, with more female employment at the expense of male employment,
• part-time growth, with the share in part-time and casual employment increasing,
• female re-entry, with more females re-entering the work force after child-bearing/raising, and
• early retirement, where a larger proportion of people are retiring before the pension age.

RETIREMENT

Retirement can be a complicated process. People may not just move directly from work to full retirement. For example, those in full-time work may take part-time work, become unemployed or possibly become a discouraged job seeker before taking the decision to leave the work force permanently.

We define full retirement as the point when a person leaves the work force and never re-enters it.⁵ We do not, however, have data which directly capture this concept, but we do have data on people who are not in the labour force and who do not intend to enter (re-enter) the labour force. These data provide an operational definition of full retirement. For many applications the concept of partial retirement may be more appropriate than full retirement. There is an increasing tendency for people to reduce the number of hours worked as they approach retirement. This is observed in the data as a shift from full-time to part-time work.⁶ It may therefore be important to distinguish between retirement from full-time employment and retirement from part-time employment.

Figures 2 and 3 provide a snapshot of the magnitude of the phenomenon for persons late in working life. The number of males working part-time who have retired from full-time work, although a relatively small proportion, increases from around the age of 53 and stops around 73 years of age. For females, however, a significant proportion are working part-time after full-time employment by the time they reach 45 years of age. This proportion then decreases in importance through to about 67 years of age in line with, but not as fast as, retirement from full-time work. We can expect significant changes in these profiles as the younger cohorts, with possibly different work patterns and life expectations, move through.

There have been a number of studies in the US researching the dynamics of partial retirement. A recent analysis by Ruhm⁷ found that:
 

The ‘job-stopping’ process of older workers often includes some combination of post career ‘bridge’ employment, partial retirement, and reverse retirement. Fewer than two-fifths of household heads retire directly from career jobs, over half partially retire at some point in their working lives, and a quarter re-enter the labor force after initially retiring. In addition, post career employment is frequently located outside the industry and occupation of the career job, and there are important differences in post career labour force experiences by gender, permanent income, and career-job pension status.

Figure 2. Labour Force Status of Males - Nov 1994

Source: unpublished data ABS Retirement and Retirement Intentions Survey (1994)
 
 

Figure 3. Labour Force Status of Females - Nov 1994

Source: unpublished data ABS Retirement and Retirement Intentions Survey (1994)

For Australia in 1994, eight per cent of males over 45 years of age who had retired from full-time work, were in part-time employment. This number is double for females at 17 per cent. Although these static statistics are not directly comparable to those developed in Ruhm’s ‘dynamic’ US analysis, they do indicate the lower limits of the extent of partial retirement in Australia.

LATE RETIREMENT

Figures 2 and 3 also show the number of people who remain in employment after the pension age of 65 for males and 60 for females.⁸ The participation rate for all males over 65 years of age is around ten per cent and it is around six per cent for all females over 60 years of age. Further, the ABS Income Distribution Survey 1990 (Table 1) shows that self employment becomes significant during the pension ages and suggests that people working for themselves in their own business retire later.⁹

RIM projections show a significant increase in full-time and part-time employment at older ages as the baby boomers move through their middle years. Even with flat projections of participation rates for those 65 and over, the number of wage and salary earners at this age will double over the next 30 years, although self employment will remain the dominant form of late employment.

Table 1. Labour Force Status Composition - Five Year Age Groups - Nov 1990
 

		MALES						FEMALES
		60-64	65-69		70-74		75+		60-64		65-69		70-74		75+
Wage & Salary	FT	55.0%	21.2%		5.4%		0.0%		30.4%		5.2%		8.0%		5.5%
	PT	10.0%	17.5%		24.9%		3.4%		32.5%		30.3%		23.1%		0.0%
Own Business		35.0%	61.3%		69.7%		96.6%		37.1%		64.5%		68.9%		94.5%
Participation Rate		48%	17%		8%		3%		17%		6%		2%		1%

Source: unpublished data ABS Income Distribution Survey(1990)

HOW MANY PEOPLE ARE FULLY RETIRED?

Fully retired persons include all those not in the labour force who do not intend to (or did not know if they intended to) enter the labour force and who were not in an educational institution. Persons who are permanently unable to work were also included. Estimates based on these statistics will, however, be overestimates since some fully retired persons will in fact re-enter the labour force. On the other hand, the number of retired may be underestimated because the estimates miss people who are effectively retired but misclassify themselves (particularly for persons who are unemployed and not in the survey). Finally, we believe that many wives (or husbands), who may be effectively retired, may not take retirement action until their spouse retires. We do not have any direct information on this phenomenon.

Using the ABS Persons Not in the Labour Force (Sept 1994) survey, we estimate that some 3.2 million people (1.1 million males and 2.1 million females) were fully retired in 1994, out of a population of 14 million aged 15 years old and above. These estimates imply that some 17 per cent of males are fully retired with about a third under 65 years of age. For women, however, 30 per cent were fully retired with about a third of those retired under 60.

RETIREMENT RATE PROFILES AND BEHAVIOURAL PEAKS

We have used the age of retirement data from the Retirement and Retirement Intentions Survey to calculate age-specific retirement rates from full-time work (Figure 4). These are average rates for all persons

Figure 4. Age of Retirement from Full-time Work - Persons Over 45 - Nov 1994

retirement age

in the survey. They do not reflect the retirement behaviour of a specific age cohort. Abstracting from the recall bunching where respondents round their age of retirement to the nearest five years, the most significant features of this chart are the behavioural peaks at 55, 60 and 65 years of age. Males have a large peak at 65 years of age, a slightly smaller one at 60 and even smaller one at 55. Females have their large peak at 60 years of age with smaller peaks at 55 and 65. These peaks are possibly generated by two mechanisms. The first is the direct link to the availability of social security (pension) and to superannuation lump sums. The second is the persistence of cultural patterns based on the historic mandatory age for retirement.

Figure 4 also highlights other differences between males and females. Male retirement from full-time work does not occur to any degree until they pass 40. The male retirement rate progressively increases until pension age of 65. Females show significant retirement rates from full-time work during their child-bearing years. Then they appear to have a relatively constant, but significantly higher, retirement rate from 30 years of age until they reach 60. Given that these data are average data for persons who are over 45 years of age, and given the recent trends in female participation rates, we would expect the rate of retirement for females retiring between 20 and 30 will have fallen for those now under 45. This trend is illustrated in Figure 5.

Figure 5. Retirement Rates from Full-time work for different cohorts - 45-59 and 60+

Source: unpublished data ABS Retirement and Retirement Intentions Survey (1994)
 

RETIREMENT DYNAMICS

Unfortunately Australian authorities do not currently collect the data required to study the dynamics of full retirement. The question on the respondent’s intention to enter the labour force was not included prior to the 1994 Persons Not in the Labour Force Survey and it was dropped from the 1995 survey. The Retirement and Retirement Intentions Survey does, however, provide some information on the dynamics of retirement from full-time work.

Full-time work and retirement patterns of males and females, whilst converging, are still quite different. This is particularly true when we look at the older cohort aged 45 and over. Table 2 shows that, while only some seven per cent of males aged 45 in 1994 and over had retired from full-time work before they reach 45, nearly 60 per cent of females had done so.
 

Table 2: Percentage of persons aged 45 and over who had retired from full-time work before they were 45
	1983	1986	1989	1992	1994
Males	3.6	5.2	5.7	5.2	7.2
Females	61.9	60.2	60.7	59.9	56.5
Source: ABS, Retirement and Retirement Intentions Survey 1994, unpublished

Figure 4, as already discussed, displays the average retirement rate from full-time work for persons over 45. These data can be decomposed into finer cohorts. Figure 5, for example, graphs smoothed data for two broad cohorts: persons aged 45 to 59 and persons aged 60 years and over. This chart dramatically illustrates the changes that have occurred between older and younger cohorts (as indicated by the arrows). We expect that these trends will continue for persons under 45 years of age.

EARLY RETIREMENT

In this paper early retirement is defined as retirement from career employment before a person reaches the pension age. Because of lack of data, we use data on retirement from full-time work to analyse early retirement. The analysis assumes that retirement behaviour from full-time work is a useful indicator of early retirement. For males this is clearly valid as over ninety per cent of men have a full-time career before they retire. For females the case is not as clear cut. Figure 3 shows, however, that the female decline in part-time employment is similar to the female decline in full-time employment. If we assume that that proportion of part-time workers who might be considered to be in career employment behave in a similar manner to those in full-time careers, the results for women will not be significantly distorted.

Rising early retirement rates will therefore be characterised by falling retirement rates at pension age and rising retirement rates before retirement age. Estimates of retirement rates from full-time employment using data from the ABS Retirement and Retirement Intentions Survey show that retirement rates at pension age have halved since 1974 (Figure 6)¹⁰ and have quadrupled since 1960 for both men and women between the ages of 45 and 59 (Figure 7).¹¹ All in all the charts show that early retirement is a real phenomenon for both males and females. The results suggest, however, that the increase in early retirement has slowed and that it might even have stabilised.

Figure 6. Retirement Rate at Pension Age

Source: unpublished data ABS Retirement and Retirement Intentions Survey (1994)
 

Figure 7. Average Age-specific Retirement Rates from Full-time Work
(percentage of single year age-specific population)

Source: ABS, Retirement snd Retirement Intentions Surveys, 1986, 1989, 1992 (Cat. No. 6238.0) and 1994 (Cat. No. 6238.040.001)

PARTICIPATION RATES ARE NOT INDICATORS OF EARLY RETIREMENT

Most analyses of early retirement, both in Australia and overseas, equate the fall in age-specific workforce participation rates with a shift towards early retirement. This methodology is highly suspect and, in some circumstances, totally misleading. First, being classified as not in the labour force does not necessarily mean retired. Our estimates indicate that the retired only make up some 60 per cent of those classified as Not in the Labour Force. Second, but more importantly, the movements in participation rates and retirement rates can be in the opposite directions. This is particularly true for females. In general, age-specific participation rates for females are rising at the same time as age-specific retirement rates are also rising.

This phenomenon is easily explained when one considers what is happening to the four mechanisms driving labour force participation rates, as stated earlier. For women, the retirement dynamics are swamped by gender shifting and the changes in re-entry to the work force after child bearing/raising.

Using simple assumptions, RIM estimates that, since 1978, around 25 per cent of the fall in participation rate for males aged 45-54 was due to early retirement. Similarly, around 75 per cent of the drop in participation rates for males aged 55-59 was due to early retirement for males. Further, we estimate that the increase in participation rates for females aged 45-54 would have been five per cent higher had it not been for early retirement and that it would have been 25 per cent higher for females aged 55-59.

One lesson to be learned from this analysis is that participation rates should not be used as indicators of early retirement.

RETIREMENT MODELLING AND PROJECTIONS

Charting persons not in the labour force using single-year age data reveals patterns which are not otherwise seen with group data (Figure 7).¹²

By combining the smoothed representation of the non participation rate (NPR) (Figure 8) with the smoothed retired ratio (Figure 9) we can model the retirement rate and project future retirement profiles.¹³

The resulting retirement peaks are separated from the underlying base (Figure 10), and each component is modified to capture underlying trends and policy changes The peaks are shifted and/or flattened to reflect the change in the shift away from and/or the change in pension age, whilst the base is broadened to capture early and/or late retirement trends.

Figure 8. Smoothed Nonparticipation Rate (NPR)

Figure 9. Retired as a Proportion of Not in the Labour Force

Figure 10. Estimated Retirement Rates - Base and Peaks

The retirement model reproduces plausible behavioural peaks at 55, 60 and 65 years of age for both men and women. The peak around 65 years of age for women may reflect the hypothesis that some females delay retirement until that of their spouse.

The existence of these behavioural peaks is supported by data from the Australian Taxation Office on Eligible Termination Payments (ETPs) (Figure 11). Although ETPs can be obtained for reasons other than retirement,¹⁴ the peaks associated with retirement behaviour are clearly identified.

Further details of the RIM retirement projection model, RETMOD, along with a detailed bibliography on labour-force dynamics, retirement and ageing can be found in Bacon 1996.¹⁵

Figure 11. Distribution of Tax Filers with Eligible Termination Payments - 1992/93

Source: Unpublished data from the ATO - 1992/93

DEPENDENCY RATIOS

Dependency ratios are used to indicate the burden that economically active persons carry by supporting dependent persons. Figure 12 shows the usual demographic measures of age, child and total dependency ratios for the population. Here, for example, age dependency is simply defined as the ratio of the population aged 65 and over to that aged18-64 years.

Figure 12. Projected Demographic Dependency Ratios

RIM estimates that the aged dependency ratio will rise from 14 per cent in 1971 to 36 per cent by 2031. Concurrently, child dependency is predicted to fall from 59 per cent in 1971 to 36 per cent in 2031. In terms of the total dependency measure, it is argued that the burden on the working-aged population is not expected to be much worse in the 2030s than it was leading up to the 1970s.¹⁶ It is clear, however, that this outcome occurs through the offsetting effects of child and age dependency. Depending on the intended use, there are a number of reasons why the child and aged dependency ratios should not be added together. First, the relative burdens placed on society by the young vis-à-vis the aged are not the same. The aged place a significantly greater burden. Second, whilst all children can be considered dependent, this is not true for all aged persons.

Further, demographic measures of dependency assume that the dependency burden is carried by all those aged 18 to 65 years. These measures do not recognise that many in the working-age group are also dependent, nor do they capture other changes that are occurring in the labour force for persons in this age group.

Two alternate classes of dependency ratio are the labour force and the retired dependency ratios.¹⁷ The labour force dependency ratio is simply the ratio of persons not working to those who are. It is designed to overcome the limitations of demographic dependency ratios. However, this ratio goes too far the other way. It includes many people in the not-working category who are not dependent. Further, cycles in the nonparticipation rate and in unemployment, show up in both the denominator and the numerator producing pronounced swings in the measure. Figure 13 shows how these cycles are amplified. Figure 13 also plots a labour-force dependency ratio adjusted for the growth in part-time employment. This ratio calculates the amount of full-time equivalent work using the average hours worked for full and part-time employment. Notice how the ratio of those not working to the number of full-time equivalent workers increases in the 21st century with the presumed increase in part-time employment.

The retired dependency ratio is the ratio of those retired to those working (Figure 14). The number of retired, the number of people not in the labour force who do not intend to enter/ re-enter the labour force, is projected into the next century using RETMOD.¹⁸ If retirement is considered to be the point in a person’s lifecycle where many shift to greater reliance on public support, then the retirement dependency measure might be an appropriate indicator. This indicator can be used to assess the burden that future workers will face in supporting the growing number of retired people. Also plotted for comparison is the retired dependency ratio for persons who retire before they reach 65 years of age. This ratio falls slightly into the next century reflecting the fact that female re-entry to the labour force far outweighs the increases arising from early retirement.

RIM estimates that the retired dependency ratio, adjusted for part-time work, is currently around 48 per cent and is projected to rise to over 70 per cent in just thirty years time. It is this concept which underlies the costing analyses that the RIM Task Force undertakes using its RIMGROUP model.¹⁹

Figure 13. Labour Force Dependency Ratios     Figure 14. Retired Dependency Ratios

CONCLUSION

The paper argues that, even though women will continue to carry most of the child-bearing/raising responsibilities, the labour-force behaviour of males and females is converging. Most of the this convergence comes from changes in the working/retirement decisions of females which have been evident since at least the 1940s. The paper identifies four mechanisms driving this convergence, of which early retirement was found to be a significant phenomenon for both males and females.

The paper develops a more appropriate dependency measure to analyse the ageing problem than alternative age dependency measures. This measure, the ratio of those retired to those working, when applied to projections of Australia’s population, indicates the burden which future workers will face in supporting the growing number of retired persons. It is likely to increase by about 50 per cent into the next century.

Note

This is a revised version of a paper given to the Eighth National Conference of the Australian Population Association, Adelaide, December 1996. The views expressed in this paper are those of the author and do not necessarily reflect the views of the RIM Task Force, the Government or any of the Sponsoring Departments of the RIM Task Force. The RIM Task Force is jointly sponsored by the Commonwealth Departments of the Treasury, Finance and Social Security. Author’s email: brb@treasury.gov.au

References

Back to People and Place Home Page