Determinants of time allocation combinations among non-employed older persons: evidence from Australian time use diaries.

Brandon, Peter D. ; Temple, Jeromey B.

Introduction

Two emerging demographic trends have captured the attention of many industrialized countries. Firstly, the proportion of those aged 65 years of age and over is projected to increase dramatically. In the United Kingdom, the United States, and Australia, for instance, the proportion of those aged 65 years and older is expected to increase between 2000 and 2030 by 7.8 percent, 7.4 percent, and 8.7 percent, respectively (Kinsella and Velkoff, 2001:126). Secondly, in these same three countries there has been a trend towards earlier retirement; that has lead to increases in the number of older persons outside the formal labor market (Lumsdaine and Wise, 1990; McDonald and Kippen, 2000; Gendell, 1998). Although this second trend is now expected to plateau, there are still more early retirees than in the past (Kinsella & Velkoff, 2001: 96).

The changing age distribution and early retirement in these countries has stimulated many debates. One important debate has centered upon the provision and funding of services required in an aging economy. For instance, in Australia, the federal government in 2002 spent $5.5 billion on aging and aged care. Of that amount, $4.3 billion was spent on residential care subsidies (Commonwealth Department of Health and Ageing, 2002: 9). Similarly, in the United States in 1998, the federal government spent $87,826 million on nursing home care and $152,629 million on Medicaid for those 65 years and older (U.S. Census Bureau, 2002). The debate about the provision of human and social services in these countries is partly driven by the related assumptions that older people are no longer productive once detached from the formal labor market and routinely utilize government services.

Older people are traditionally perceived as dependent upon government and private support services and no longer spend their time contributing to the economic and social well-being of themselves, their households and society (U.N. Volunteers, 2002). However, as research has shown, many older people spend time doing productive activities (Herzog and Morgan 1992; Gallagher, 1994) and the majority continue active lives without severe disabilities (Kinsella and Velkoff, 2001). Despite this growing body of research that contradicts this conventional view, a more detailed understanding is required about how older people allocate their time to activities and the factors associated with that allocation (Mutchler, Burr, and Caro, 2003). Moreover, much of the existing data on time allocation of older people is out-of-date and requires revision (Knapp and Muller, 2000:25).

Using recent time use data from Australia and a non-linear time allocation model, we show older persons who have left the formal labor market: (1) remain active, and that (2) many allocate their time to human service activities that benefit others and their communities. These findings contradict the stereotype that older individuals are sedentary, no longer contribute, and dependent, and embody an important aspect of productive aging (Mutchler, Burr, and Caro, 2003; Knapp and Muller, 2000).

Our findings provide new knowledge of the factors associated with older persons allocating their time to human services rather than allocating their time exclusively to household activities, recreation, and socializing. By identifying the factors associated with time allocation of older persons, a deeper understanding is gained of the importance of demographic and economic factors leading older persons to provide services that could potentially substitute for government funded ones) This study, therefore, broadens the literature on the time allocation of older people as well as pointing out that the true cost of providing human services is underestimated due to the unvalued services provided by older Australian's time allocation to human service activities.

Background and Theoretical Perspectives

This study builds upon previous research that has investigated the allocation of time among older persons (Gauthier and Smeeding, 2001; Gershuny, 2003). The past research has provided several important insights. Significant among them is that marital status, levels of human capital, age, gender, income, and ethnicity have been found associated with older persons' allocation of time. Glass (1995) found that being married protects against declines in productive activity in old age as the former are at half the risk of productivity declines compared with the latter. Chambre (1993) found that older people were more likely to volunteer if married rather than not. Pezzin and Scone (1999) found that those who have been divorced have looser parent-child ties than those who have never divorced. Besides marital status, Herzog et al (1996) found that education level influences the frequency of participation in productive activities. The importance of education attainment to productive activities is particularly relevant given the higher educational attainment of future aged cohorts. Chambre (1993) argues that higher levels of education provide a larger potential pool of skilled volunteer workers in the future. Lastly, among the important demographic and socio-economic factors, Danigelis and Mcintosh (1993) pay special attention to showing the effects of gender and ethnicity. Their study shows that the allocation of time to paid work, unpaid work at home, and unpaid work outside the home depend on the person's gender and ethnicity. Klumb and Bahes (1999) add to Danigelis and McIntosh's (1993) work by demonstrating that age and gender interact with other variables reflecting events over the life course.

Further, our study is motivated by the only Australian study on the time use of older Australians. De Vaus, Gray, and Stanton's (2003) study sought to measure the value of unpaid services provided by older Australians. They find that Australians aged 55 years and over contribute over $74 billion dollars per year through unpaid work. They conclude that "While there are many reasons why some older people undertake considerable unpaid work while others do little, one challenge for research is to learn why there are such wide variations" (p 21). Our study pursues this challenge and provides new knowledge of the factors that explain the variation that De Vaus, Gray, and Stanton (2003) observe.

Although these studies and others now discussed have been pivotal to understanding the allocation of time among older persons, few have explicitly sort to explore the ramifications of the time allocation to the provision of public services. Interestingly, many of the studies have noted that older individuals time allocation must matter to the provision of public services. For instance, in Pezzin and Sconce's (1999) study, they suggest that given the higher rates of divorcees in future cohorts of the aged it is likely that greater demands will be placed on publically funded programs. Also, in a key piece of research by Herzog et al (1996), and Klumb and Bakes (1999), each set of researchers suggested that what older person do with their time may have implications for the level of government outlays. Similarly, Joshi, Denton et al (1997:24), whose major focus was identifying, measuring, and quantifying the value of unpaid services provided by older people, still note the potential link between older persons time allocation and the economy. In the same paper, these researchers also noted that if governments continue to cut funding to social services, the role of the volunteer workforce will become more significant (Joshi, Denton et al, 1997:24).

The major theories guiding the studies that we cited on time allocation among older persons include Becker's theory of time allocation (Becker, 1965) and the more general gerontological theories of "continuity" (Ravanera and Fernando, 2001) and "activity and disengagement" (Lemon, Bengston and Peterson, 1972). The important insight that Becker's theory offered was the key roles that incomes and prices play, i.e., economic factors, in explaining the allocation of time. Many contemporary studies of time allocation have confirmed his argument that the income available to a household is a constraint on time allocation (Gronau, 2003 ; Biddle and Hamermesh, 1990). Gauthier and Smeeding (2000), who focused explicitly on samples of older persons, built upon Becker's insights by arguing that constraints beyond incomes and prices, like, health status and education influence time allocation among older persons, as well. Apart from the economic perspective on time allocation, gerontological research has considered the change in time allocation upon retirement.

One gerontological viewpoint posits that time allocation among retirees is influenced by how the retirees spent their time during their working life (Atchley, 1972). For example, Chambre (1993) showed that people who volunteered prior to retirement are more likely to volunteer in retirement compared with those who had not volunteered during their working life. Though this viewpoint on the continuity of time allocation is debated (Burr and Caro, 2000), the perspective has motivated research that seeks to examine whether individuals transfer patterns of time allocation from their working life into retirement.

Our study is influence by these conceptual frameworks, especially those offered by Gauthier and Smeeding (2000), Mutchler, Burr, and Caro (2003), Danigelis and Mcintosh(1993), and Klumb and Baltes(1999). These studies are relevant to our approach because they underscore the significant role that socioeconomic and demographic factors play in the allocation of time among older persons.

Notwithstanding the contributions made by these studies, more knowledge is required about the effects of specific socioeconomic and demographic factors on time allocation among older persons. For instance, we theorize that the allocation of time to activities is associated with the type of living arrangement of older persons. We hypothesize that older persons who are single will allocate time to many activities that contribute to the well-being of the economy and community. More specifically, time allocation to activities should vary by whether an older person is a widow, divorcee, or has never been married.

Contrasting the latter type of arrangement, older persons who are partnered will have a different time allocation compared with singles, e.g., widows. The former hypothesis is consistent with research that shows differential financial and support networks (Davis and Denton, 2001; Pezzin and Schone, 1999; Choi, 1995) available to widows, divorcees, and those never married. This hypothesis is motivated by time use studies showing couples are more likely to volunteer (Chambre, 1993) and less likely to suffer productive declines in later life compared with single persons (Glass, 1995).

Our other chief hypothesis is that sources of income--not simply the level of aggregate income--will determine the allocation of time to activities. Two main income sources for older Australians are income from private investments and the aged pension--a public income transfer (Kelly, Harding, and Percival 2002). We hypothesize that older persons who receive income from investments, (a source of wealth), will allocate their time differently to other older persons who do not receive income from investments. We further conjecture that pension recipients will allocate their time differently to all other older Australians because the aged pension is age- and asset- tested and thereby acts as a crude indicator of the lack of accumulated private wealth over the lifecourse. These hypotheses about income sources are based upon the well-known association between wealth and time allocation (Becker, 1965; Gronau and Hamermesh, 2003).

Lastly, we theorize that, at any point in time, constellations of demographic and economic factors, as well as preferences, affect how older persons group their activities into time budget combinations. We argue that a time budget combination is a set of activities to which an individual allocates time to meet their preferences. For instance, we expect to observe some older persons who cluster a set of activities to recreation and household production only, so that their preferences are met. By contrast, we also expect to observe older individuals who do not to socialize, but nonetheless cluster a set of activities to satisfy their needs to spend time recreating, maintaining a household, and contributing to their communities. Furthermore, we expect to identify those with a time budget combination that encompasses an even fuller range of activities that not only satisfy their own needs to spend time on recreating, socializing, and maintaining household production, but also want to spend time caring for others and doing volunteer work. By pursuing this theoretical framework on the allocation of time to particular time budget activity combinations, we can exploit a non-linear modeling strategy that is described in the methods section.

In summary, by more closely examining the effects of investment income, pension receipt and particular living arrangements on time budget activity combinations, this study distinguishes itself from its predecessors, expands the literature on time allocation among older persons, and highlights the implications of that allocation across activities for government-funded human services.

Methodology

Data Description

Data for this study are from the 1997 Australian Time Use Survey (TUS). The TUS is a cross-sectional study conducted on a multi-stage area sample of private dwellings in Australia. The survey excludes persons living in non-private dwellings, such as, nursing homes and retirement villages (ABS 2000). The purpose of the TUS is to provide data on the allocation of time by individuals aged 15 years and over to work and non-work activities (ABS 2000). The public use file, known as the "Confidentialised Unit Record File", i.e., the CURF, includes information at the person, household, and activity levels. (2) Out of 8,618 respondents, 7,260 individuals from 4,059 households, (i.e., a response rate of 84.2 percent), responded to the survey. Combined, these individuals contributed information on 406,133 activity time episodes.

The TUS was conducted over four, 13-day periods during 1997 in order to account for the effect of seasonal variation on time use activities. Since different activities can occur according to whether the day is a weekday or weekend, all days of the week were surveyed in equal proportions. Data for the survey was collected by both interview and self completed diary. The information on persons includes information on their socio-demographic characteristics. Each individual is provided with a 'time diary' to record seven components of a 'time episode', which includes: (1) what the activity is; (2) who it is being done for; (3) when it begins; (4) when it ends; (5) whether anything else is being done at the same time, i.e., secondary time; (6) where it takes place; and (7) who else is present. To be recorded, the duration of each time episode must exceed five minutes. For fitting our statistical model to these data, we focus on primary time only. By using only primary time, we concentrate on the main activity that individuals allocate time too; and, beyond the substantive usefulness of a primary time focus, we also avoid the difficulty of not knowing whether total time results from secondary and primary time activities interacting additively or multiplicatively.

The time diary approach has an advantage over alternate methods that rely on stylised, usually retrospective questions. The retrospective methods have been shown to yield inconsistent estimates of time use (Gauthier and Smeeding, 2000:4). Although alternative time categories are possible, the categories of time use that this study exploits are the nine classifications determined by the ABS (ABS, 2000), namely: (1) personal care (2) domestic (3) purchasing goods and services (4) education (5) employment (6) socializing (7) recreation (8) child care and (9) volunteering and care. (See Table 1.)

Merging the person and activity level data in the TUS enables us to examine the allocation of time to a broad range of activities by older Australians. In particular, we examine the time allocation of 1,350 Australians aged 55 and over, i.e., 18.6 percent of the full TUS sample, who are not in the laborforce. These 1,350 older individuals account for 73.56 percent of older Australians in the TUS aged 55 years and over, N = 1,835. The TUS estimate that about 74 percent of older Australians were not in the labor force in 1997 is remarkably similar to the percent reported in the 1998/1999 Household Expenditure Survey, which was about 75 percent.

The 1,350 older individuals not in the labor force we focus upon contributed 78,306 activity time episodes to personal care, domestic work, purchasing goods and services, socializing, recreation, child care, volunteering and care, which are seven of the nine classifications created by the ABS. Of the latter TUS time episodes, we conjecture that if older individuals chose not to allocate their time to child care, volunteer work, or adult care, then some form of public (or private) (3) sector substitute may be necessary. Thus, older individuals' decisions to allocate time to these types of service activities potentially reduce government outlays. We further conjecture that if older individuals restrict their time allocations to personal care, domestic purchasing, socializing, and recreation then government would have to increase expenditures to fill the unmet demand for certain caring work and services. Thus, by distinguishing among activities, we gain a better understanding of the factors associated with the allocation of time among older Australians and a deeper appreciation of how older individuals substitute for public sector services.

Statistical Model and Variables

Given our research objective, which is to better understand time allocation of older Australians and its importance for government-provided services, we specified a multinomial logistic regression model that calculated the probabilities of older Australians undertaking various combinations of activities based upon the seven ABS classifications noted above. Heretofore, empirical strategies have used either cross-tabular descriptions of time-use data or regression models, mostly logistic and ordinary least squares regression (Gauthier and Smeeding, 2001; Gallagher, 1994; Garfein and Herzog, 1995; Danigelis and Mcintosh, 1993).

This study builds upon such modelling strategies by using a multinomial logistic regression model to examine the probabilities of older Australians allocating time to mutually exclusive time budget combinations. Using the seven ABS classifications identified above and based upon our theoretical perspective explicated earlier, we collapse the seven classifications into four time budget activity categories and label the dependent variable "Time Budget Activity Combination." Although other dependent variables with fewer or more time budget activity categories are possible, the time budget activity categories used are based upon the extensive literature on socializing (Lund, Caserata et al, 1990; Hoonard and Kestin, 1994) and productive activity (Garfein and Herzog, 1995; Herzog and Morgan, 1992), two essential components of later life.

Specifically, if older persons allocate their time to household production and recreation activities only, then the first category, "household orientation", is coded 1. If they allocate their time to household production, recreation, and socializing activities only, then the second category, "household and social-life orientation", is coded 2. If they allocate their time to household production, recreation, child care, adult care, and volunteering activities only, then the third category, "household and service orientation", is coded 3. Lastly, if they allocate their time across all activities, i.e., household production, recreation, socializing, child care, adult care, and volunteering, then the forth category, "multiple activity orientation", is coded 4. Category one, "household orientation", is the comparison group in the multinomial logistic regressions.

We assume that a multinomial logistic model generates our observations and that the probability of observing an older person in one activity category relative to another category is associated with variations in the independent variables of theoretical interest. The model as described by Greene (1997) for the multinomial logit is:

Pr([Y.sub.i] = k) = exp([p.summation over (j=0)] [x.sub.ij] [[beta].sub.jm])/[r.summation over (m=1)] exp([p.summation over (j=0)][x.sub.ij] [[beta].sub.jm)

Living arrangements, income level, source of income, education level, and disability status are the chief independent variables that are hypothesized to influence time allocation across these four categories. Other independent variables controlling for demographic characteristics such as age, gender, geographic location, speaking English, and transportation are also included in the model.

Findings and Discussion

Table 1 shows the time allocation by activity grouping and time for older Australians who are no longer in the labor force during the period of data collection. Once out of the labor force, 27 percent of older Australians, (N = 370), allocated all their primary time (4) exclusively to household production and personal recreation. Another 31 percent, (N = 422), allocated their primary time to the latter activities and socializing only. Of the remaining 42 percent, twelve percent, (N = 160), allocated primary time to volunteering and child care rather than socializing activities and 30 percent (N = 398) spread their primary time across socializing, household production, recreation, child care, and volunteering, i.e., all the activities to which their time can be allocated.

Across the four groups for primary time, and for total time too, (see Table 1), older Australians who have left the labor force spend approximately equal time doing personal care, purchasing, and domestic cleaning. Interestingly, the activities from which time is withdrawn as older people engage in a greater range of activities are personal care and recreational activities, not purchasing goods and services or domestic work. Specifically, Table 1 shows that older Australians who allocate their primary time across all activity groups spend slightly less time on personal care, but much less time on recreation compared with other older Australians. The pattern is repeated for total time allocation, except for the comparison of time spent recreating for those who engage in household production only relative to those who add socializing activity. (Total time is primary and secondary time combined.)

Comparisons of time allocation in Table 1 offer important insights into time allocation to combinations of activity categories among older Australians not in the labor force. However, they are misleading without introducing controls for socioeconomic and demographic differences among the groups. To resolve the concern, a multivariate model was employed: the dependent variable that contained the four activity groups shown in Table 1 was regressed on age, education level, disability status, income, and living arrangements, being the chief variables of theoretical interest. Other control variables, shown in Table 2, were included.

Table 2 presents descriptive statistics for the independent variables used in the multivariate analyses by each activity grouping. The age distribution of the four groups of older Australians involved in different activity combinations differ considerably, with older Australians who are household oriented being much older than the other groups of Australians. Those who have a multiple activity orientation are much younger compared with those who are household oriented. In turn, the household oriented are most likely compared to the others not to have completed high school, have the highest rates of disability, and receive government income support. The multiple-activity group, by contrast, is the most highly educated. Controlling for wealth, car ownership is an important indicator of mobility and fewer in the group of household oriented only own cars. On the other hand, those in the multiple-activity group are the most likely to have at least one vehicle, though these data suggest they are also those most likely to live in regional or rural areas where cars are more necessary. The table also indicates variation across activity grouping by gender. Whereas there is an equal split by gender for those household oriented only, much higher proportions of females compared with males engage in addition activities. The pattern suggests that males are more likely to have a household orientation only rather than combine household work with child care, volunteering, and socializing. This is consistent with continuity theory that hypothesizes individuals continue allocating their non-labor market time to activities that resembled their time allocation prior to leaving the labor force (Ravanera and Fernando, 2001). Lastly, those that are household oriented are less likely to be widowed and more likely to have a partner. Time allocation differences by marital status are clearer once controls are introduced as shown in Table 3.

Table 3 shows results from fitting a multinomial logistic regression model that estimates the likelihood of engaging in particular activity groupings that extend beyond having a household activity orientation only. Our major interest is to understand the effects of living arrangements. The estimated coefficients for the variables measuring the different living arrangements suggest that living arrangements are associated with the likelihood of having other than a household only activity orientation. For instance, if an older person is separated or divorced, that person is about 70 percent more likely, exp{.532} = 1.70, than a person with a partner to have a household and social life orientation. Similarly, a separated or divorced older person is 90 percent more likely, exp{.643} = 1.90, than a partnered person to focus upon their household and community. On the other hand, separation or divorce appears not to influence the likelihood of engaging in all activities, (i.e. an all purpose orientation), relative to engaging in only a household orientation. Widowhood also has a large effect on how older persons allocate their time. A widow is twice as likely, exp{.708} = 2.03, to allocate time to a household and social life orientation as a partnered person. Likewise, these persons are nearly 90 percent more likely, exp{.638} = 1.89, than partnered persons to have a household and community orientation. Moreover, the effect of widowhood continues and its effect is strongest for those that have an all-purpose orientation. Widowed older persons are 120 percent more likely, exp{.808} = 2.24, than coupled persons to engage in all activities relative to engaging in only household oriented activities. By contrast, estimates for having never married suggest that the allocation of time to activity grouping does not differ from the time allocation of those partnered.

Our other chief interest is estimating the effects of particular sources of income on older persons' allocation of time to activity groupings. Estimated coefficients for receiving government income support suggest that older persons who receive government income transfers are two thirds less likely, exp{-1.103} = .33, than those who do not receive government income transfers to have a household and social orientation relative to a household activity orientation only. Similarly, an older person who receives government income support is approximately 40 percent less likely to have a household and community orientation, exp{-.946} = .39, than a person who does not receive such support. The sign of the coefficient for the variable stays negative across activity groups, though it weakens and found insignificant for time allocation for an all-purpose orientation relative to a household orientation only.

In addition to estimating the effects of receiving government income support, we estimate the effect of receiving private business investment income--our only measure of wealth. (5) Estimated coefficients for receiving business income show that older persons who receive such income are about 60 percent more likely, exp{.453} = 1.57, than those who do not receive it to have a household and social life orientation relative to a household orientation only. By contrast, receiving business income is not associated with spending more time on caring activities and volunteering than spending time on household oriented only. Yet, income from a business increases the likelihood of membership in the all-purpose group relative to membership to the household oriented only group. These findings for receipt of business income may reflect differences across individuals within activity groupings. Lastly, the effect of regular weekly household income, net of business and government income, on the allocation of time to activity groupings is estimated. The estimated effect for the coefficient suggests that regular weekly household income does not affect the allocation of time to activity groupings. This result may occur because of the lack of variability in the income measure, or alternatively its insignificance is due to having already controlled for two major sources of income in old age once out of the labor force, namely, income from investments, retirement accounts, or government pensions (Kelly, Harding, and Percival, 2003:8).

Besides these key findings, the multivariate model estimates the effects of other key demographic factors that are associated with the allocation of time to activity groupings among older Australians who are not in the labor force. Those in the older age groupings, i.e., those between 65 and 74 and those over 75 years of age, for example, are found less likely compared with those who are younger than 65 years to focus on activities beyond household activities. As expected, those over 75 years of age are even least likely than those 64 to 74 to spend time on activities beyond household ones. These results for age are consistent with other studies that have found an age grading effect in time allocation (Mutchler, Burr and Caro, 2000:9). The level of education attained among older Australian not in the labor force also has a significant effect on how they spend their time. Older Australians who have earned a college or university degrees compared with those who had not completed secondary school are more likely to spend time on service activities. For example, the most highly educated older persons are almost 150 percent more likely, exp{.894} = 2.44, than their least-educated peers to provide volunteer and caring services to the community. Moreover, these same individuals are nearly 7 times as likely, exp{1.914} = 6.78, compared with the least educated to combine socializing with these community service activities. These results for educational attainment underscore the importance of investment in education to the types of activities to which older Australians allocate time.

The literature has suggested that gender matters (Dangelis and McIntosh, 1993; Etther, 1995; Walker, Pratt and Eddy, 1995). Consistent with that literature, this study finds that there are differences in the allocation of time to activities by gender. Older women compared with older men are more likely to engage in activities beyond those with a household orientation. They are especially more likely than men to offer time to community service activities. Older women are over one and a half times more likely than older men, exp{.449} = 1.57, to offer time to community service activities. Also, they are more likely than men to combine many activities that benefit themselves, socializing, and others, by combining all activities, exp{.411} = 1.51. Lastly, the literature has made clear than disability in older age matters to time allocation. We expected and found that disability is associated with the allocation of time to activities among older Australian. Those with a severe or moderate disability are over a third less likely, exp{-.463} = .63, to spend time on community activities compared with those who are either not disabled or have only a minor disability. Likewise, the former are nearly half as likely than the latter, exp{-.588} = .56, not to try to do all activities combined. This result is significant because it shows the capacity to combine across many types of activities is constrained by the degree of a person's disability.

Findings discussed above are robust with or without controls for geographic region, access to a vehicle, and English as a first language. The latter controls are important, but discussion of their effects are excluded since this study concentrates on estimating the effects of living arrangements and income sources on older persons' time allocation. Furthermore, Hausman's test for the "independence of irrelevant alternatives" (Hausman and McFadden, 1984) confirmed that the classifications of activity groupings for the dependent variable are statistically independent, i.e., the hypothesis that the activity groupings are substitutes for one another is strongly rejected ([] = -7.69, d.f. = 38, p = 1; [] = -0.356, d.f. = 38, p = 1; [] = -28.87, d.f. = 38, p = 1).

Conclusions

A major conclusion of this study is that the typecasts that older Australians out of the labour force spend their time in passive, leisure pastimes only or are prevented from having an active life due to incapacitation are both strongly rejected. Few older Australians in this representative sample were found to have severely disabling conditions, (16 percent), and most were engaged in a variety of activities beyond household-oriented activities. Significantly, among those older Australians who had expanded their range of activities beyond those oriented solely on the household, about 57 percent allocated time to activities that not only benefited themselves, but also benefited others, or communities. Hence, over half of older Australians not in the labor force provided services, such as, child care, adult care, or volunteer work, which are activities contradicting negative stereotypes of disengaged, older people.

Evidently, according to our findings, most older Australians who are not in the labor force remain active, stay engaged, and contribute to the overall well-being of the Australian community. This depiction is an optimistic one, especially since research shows that leading an active life in older age promotes health and well-being (Wang, Karp, Winbald, and Fratigloni, 2002; Glass, Leon, Marottolo, and Berkman, 1999). We speculate further that remaining active encourages mental and physical well-being that could indirectly reduce demand for government health and human services. Overall, these data provide governments with important information about the potential positive aspects of older persons' lifestyles. In the Australian case, the federal government has sought after much needed evidence to propel its strategy of recognizing the roles and contributions of older persons to the economy and community (Andrews, 2002). This study and one other (DeVaus, Gray and Stanton, 2003), furnish the government with the evidence it needs.

Besides dismissing unwarranted stereotypes, this study concludes that many older Australians' time allocation, particularly to caring activities, occurs in the shadow of government services. The caring and voluntary services provided by older Australians conceivably subsidize, to some extent, the public expenditures on health and other community services. Policy makers need to remain aware of the factors that influence the supply of caring and voluntary activities, if for no other reason than to recognize that an element of hidden demand is meet by the service provision of older Australians. Importantly, the changing demographic profile of the aged may very well alter the supply of this valuable economic resource as indicated by the importance of demographic variables in our analysis.

Although our study provides no direct evidence, the findings strongly suggest that investments earlier in the lifecourse increase the likelihood of an active lifestyle in older age. The study sample shows that those older individuals who invested more years in education earlier in their lives were more likely to engage in activities that assist themselves and their communities. Beyond educational investments, those who had accumulated sufficient financial assets as they aged were also more likely to engage in activities that mirrored government-provided human and social services.

Beyond these significant findings, the living arrangements of older Australians are found to determine their allocation of time to activities. Specifically, single older people, i.e., they are either separated or divorced, are more likely to engage in activities beyond the household compared with married or de facto couples. The one exception to this generalization is that the divorced and separated were no more likely compared with couples to allocate their time to activities with an all-purpose orientation. Possibly, reciprocity in social exchange in later life offers an explanation for this finding as married older people rely on spouses and children for social support in later life (Hogan, 1995), whereas single older Australians' may need to rely on other sources of social support, such as, friends and community organizations.

Importantly, our study further suggests that widows are far more likely than any other group to combine all activities, i.e., have an all-purpose orientation. The perception is that widowhood is an ongoing negative phenomena, but our findings suggest that widows maintain active lives in their communities. Moreover, our finding lends support for Wilcox, Aragaki, Mouton etals (2003) notion that widowhood frees older people from the task of constant caregiving, thereby allowing them time to engage in other social activities. Lastly, never married persons are no more or less likely to engage in activities beyond having a household orientation only compared with couples. This finding perhaps suggests that the lives of the never married are more similar to those of married couples than previously thought, or it could be an artifact of possessing a small sample of never married persons. Certainly, the research is mixed on the lifestyles of older persons who never married and raises an issue for further research.

Undoubtedly, our paper raises several important new research questions. Among them, is the central question of discovering how activity patterns in old age are associated with earlier accumulations of social, economic, and human capital? For future studies to adequately address this core question of the effects of earlier life-course investments on activity patterns in old age, measures on the timing of earlier life course investments are required. Thus, this line of inquiry requires future time use studies to measure current as well as earlier investments in social and economic capital. Practically, this data necessity means that future cross-sections of ABS time use surveys should be designed to incorporate retrospective questions about earlier life course investments. Alternatively, new longitudinal time use data would enable analyses to examine whether time use in old age is a departure from earlier time use patterns or a continuation of them as proposed by Atchley (1972).

Overall, this study has broadened the literature on time allocation among older individuals not in the labor force. Using representative data on the time use of older Australians and an innovative empirical strategy that shows how older persons combine activities, this study dismisses several unwarranted depictions of older persons in Australia, highlights the potential relevance of work outside the formal labor market to government provided services, and points towards further research on the value of unpaid caring services performed by older persons that remained unrecognized and undervalued by society.

References

ABS (2000) Time Use Survey Users Guide, 1997 Cat no. 4150.0, Australian Bureau of Statistics, Canberra.

ABS (2000a) 1998/99 Household Expenditure Survey Confidentialised Unit Record File

Andrews, The Hon. Kevin. (2002) National Strategy for an Ageing Australia. Canberra, Australian Government Printing Office.

Atchley, R (1972) The Social Forces in Later Life Wadsworth Publishing co, California.

Becker, G (1965) 'A Theory of the Allocation of Time' The Economic Journal 75(299)

Biddle, J and Hamermesh, D (1990) 'Sleep and the Allocation of Time' The Journal of Political Economy 85(6) p1099-1124

Chambre, S (1993) 'Volunteerism by Elders : Past Trends and Future Prospects' The Gerontologist 33(2)

Choi, Namkee (1995) 'Long Term Elderly Widows and Divorcees : Similarities and Differences' Journal of Women and Aging 7(3)

Commonwealth Department of Health and Ageing (2002) Aged Care in Australia Commonwealth Department of Health and Ageing, Canberra.

Danigelis, N & Mcintosh, B (1993) 'Resources and the Productive Activity of Elders : Race and Gender as Contexts' Journal of Gerontology 48(4)

Davis, S and Denton, M (2001) The Economic Well-Being of Older Women Who Become Divorced or Separated in Later Life SEDAP research Paper No. 66

Department of Family and Community Services (2001) Annual Report 2000-01 Commonwealth Department of Family and Community Services, Canberra.

De Vaus, D, Gray, M and Stanton, D (2003) 'Measuring the Value of Unpaid Household, Caring and Voluntary Work of Older Australians' Australian Institute of Family Studies Research Paper No. 34.

Ettner, S (1995) 'The Impact of Parent Care" on Female Labor Supply Decisions' Demography 32(1)

Freeman, R (1997) 'Working for Nothing : The Supply of Volunteer Labor' Journal of Labor Economics 15(1)

Gallagher, S (1994) 'Doing Their Share : Comparing Patterns of Help Given by Older and Younger Adults' Journal of Marriage and the Family 56(3)

Garfein, A & Herzog, A (1995) 'Robust Aging among the Young-Old, Old-Old and Oldest-Old' Journal of Gerontology 50(Bn2)

Gauthier, A and Smeeding, T (2001) 'Historical Trends in the Patterns of Time Use of Older Adults' Paper commissioned by the OECD. URL: http://www.oecd.org/dataoecd/21/5/2430978.pdf Date Accessed: 16/12/03

Gauthier, A & Smeeding, T (2000) 'Time Use at Older Ages: Cross National Differences" Paper commissioned by the OECD. URL: http://www.oecd.org/dataoecd/20/24/2431647.pdf Date Accessed: 16/12/03

Gendell, M (1998) 'Trends in Retirement Age in Four Countries, 1965-95' Monthly Labour Review August, 1998.

Gershuny, J (2003) 'Time, Through the Lifecourse, in the Family' ISER Working Papers, Number 2003-3.

Glass, Seeman, Herzog, Kahn and Berkman (1995) 'Change in Productive Activity in Late Adulthood: MacArthur Studies of Successful Aging' Journal of Gerontology 50(2)

Glass, Leon, Marottoli and Berkman (1999) 'Population Based Study of Social and Productive Activities as Predictors of Survival Among Elderly Americans' British Medical Journal Aug 21, 1999.

Greene, W (1997) Econometric Analysis Prentice Hall, Upper Saddle Ricer, N.J.

Gronau, R and Hamermesh, D (2003) 'Time vs. Goods : The Value of Measuring Household Production Technologies' National Bureau of Economic Research Working Paper Number 9650

Hausman, J & McFadden, D (1984) 'Specification Tests for the Multinomial Logit Model' Econometrica 52(5)

Herzog, Franks, Markus & Holmberg (1996) 'Productive Activities and Agency in Older Age' in Baltes, M & Montado, L (1996) Produktives Leben im Alter Campus Verlag GmbH, Frankfurt/Main.

Herzog, A & Morgan, J (1992) 'Age and Gender Differences in the Value of Productive Activities' Research on Aging 14(2)

Hogan D & Eggebeen, D (1995) 'Sources of Emergency Help and Routine Assistance in Old Age' Social Forces 73(3)

Hoonaard, V and Kestin, D (1994) 'Paradise Lost : Widowhood in a Florida Retirement Community' Journal of Aging Studies Summer 1994.

Hulbert, J and Acock, A (1990) 'The Effects of Marital Status on the Form and Composition of Social Networks' Social Science Quarterly 71(1)

Joshi, A, Denton, M, Gafni, A, Lian, J, Robb, R Rosenthal, C & Wilssison, D (1997) 'Evaluating Unpaid Time Contributions by Seniors : A Conceptual Framework' IESOP Research Paper No.26.

Kelly, S, Harding, A & Percival, R (2002) 'Live Longer and Prosper? Projecting the Likely Superannuation of Baby Boomers in 2020' Paper presented at the 2002 Australian Conference of Economists, Business Symposium, 4th October, 2002.

Kinsella, K & Velkoff, V (2001) An Ageing World: 2001 U.S. Census Bureau, Series P95/01-1, Washington, DC.

Klumb, P & Baltes, M (1999) 'Time Use of Old and Very Old Berliners: Productive and Consumptive Activities as Functions of Resources' Journal of Gerontology: Social Sciences 54b(5)

Knapp, K & Muller, C (2000) Productive Lives : Paid and Unpaid Activities of Older Americans International Longevity Center-USA, New York.

Lemon, B, Bengston, V & Peterson, A (1972) 'An Exploration of the Activity Theory of Aging : Activity Types and Life Satisfaction Among In-movers to a Retirement Community' Journal of Gerontology 27.

Lumsdaine, R & Wise, D (1990) 'Aging and Labour Force Participation : A Review of Trends and Explanations' NBER Working Paper Series.

Lund, D, Caserata, M, Van Pelt, J and Gass, K (1990) 'Stability of Social Networks after Later-Life Spousal Bereavement' Death Studies 14(1)

McDonald, P & Kippen, R (2000) 'The Implications of Below Replacement Fertility for Labour Supply and International Migration 2000-2050' Paper presented to 2000 Annual Meeting of the Population Association of America, Los Angeles, California, March 23-25.

Mutchler, J, Burr, J & Caro, F (2003) 'From Paid Worker to Volunteer: Leaving the Paid Workforce and Volunteering in Later Life' Social Forces 81(4)

Mutchler, J, Burr, J & Caro, F (2000) 'Structure and Intensity of Productive Activity Over the Life Course'. Paper presented at the American Sociological Association.

Pezzin, L & Schone, B (1999) 'Parental Marital Disruption and Intergenerational Transfers : An Analysis of Lone Elderly Parents and Their Children' Journal of Applied Gerontology 12.

Ravanera, Z and Fernando, R (2001) 'Integration at Late Life: Inclusion, Participation and Belonging Among the Elderly' Paper presented at the meeting of the Federation of Canadian Demographers, Ottawa, Ontario, December 14-15 2001.

UN Volunteers (2002) 'Voluntary Action by Older Persons' Paper presented at the Second World Assembly on Ageing, Madrid, Spain, 8-12 April 2002.

U.S Census Bureau (2002) Statistical Abstracts of the United States U.S. Census Bureau. URL: http://www.census.gov/prod/www/statistical-abstract-02.html Date accessed: 11/17/03.

Walker, A, Pratt, C & Eddy, L (1995) 'Informal Caregiving to Aging Family Members : A Critical Review' Family Relations 44(4)

Wang, H; Karp, A; Winbald, B & Fratigloni, L (2002) 'Late-Life Engagement in Social and Leisure Activities is Associated with a Decreased Risk of Dimentia: A Longitudinal Study from the Kungsholmen Project' American Journal of Epidemiology 155(12)

Wilcox, S, Aragaki, A, Mouton, C, Evenson, K, Wassertheil-Smoller, S and Loevinger, B (2003) 'The Effects of Widowhood on Physical and Mental Health, Health Behaviours, and Health Outcomes : The Women's Health Initiative' Health Psychology 22(5)

Endnotes:

(1.) Presently, the Australian government spends millions of dollars funding services for young and old Australians. For example, in 2000-2001, the Australian Commonwealth government spent $1,356,380 thousand on child care support services (Department of Family and Community Services, 2000). However, this figure is an underestimate because the informal services provided by older Australians remain unaccounted for and unvalued.

(2.) The CURF, by definition, collapses many interval- and ratio- level variables into categorical level variables to prevent identification of survey respondents. This aggregation procedure that is conducted by the ABS before public release of these data means that researchers' empirical approaches are constrained by the amount of detailed information contained in the publicly available data. For example, variables, such as, income, have been highly aggregated, thereby masking potential variation.

(3.) Although we recognize that private sector responses occur, our primary focus is to understand how time allocation among older Australians not in the labor force relates to government provision of services, such as, government nursing homes or child care centres that are often heavily subsidized.

(4.) As noted, the CURF collected information on primary and secondary time use activities. This study concentrates on primary time use activities, although for completeness, Table 1 provides information on primary and total time allocation. Secondary time is total time minus primary time, with primary time defined as time spent on the main activity for that time episode. If any other activity occurred simultaneously, then this was recorded as secondary time (ABS, 1998).

(5.) The CURF has limited information on wealth, thereby restricting this study's ability to identify sources of wealth. An illustrative example is home ownership. The time use data fails to provide an indictor of home tenure, that is, whether respondents' own or rent the property in which they live. Possessing an indicator of income from business investments, at least enabled the study to measure one aspect of an older person's wealth.

Please direct correspondence to Peter D. Brandon, [email protected]. The authors wish to acknowledge that the Customised Unit Record File used in this research was made available through the Australian Bureau of Statistics and Australian Vice Chancellors Committee Agreement. The opinions expressed in this paper do not necessarily reflect those of the ABS or AVCC. The authors thank an anonymous referee, Dr. Don Rowland for comments on an earlier draft, Robert M. Hauser at the University of Wisconsin, as well as workshop participants at the Center for Demography of Health and Aging at the University of Wisconsin and seminar participants at the Department of Sociology, Brown University.

Table 1: Time Allocation (1) of Older Australian's by Time Budget
Activity Grouping and Time, in minutes (Standard Errors in
parentheses)

 Time Budget Activity Grouping

 Household and
 Social Life
 Household
 Orientation Orientation

Time Sent in: Mean (3) % Time Mean % Time

Primary Time (4)

Personal Care 1417.39 51.4 1438.33 50.4
 (16.43) (11.37)
Domestic 385.72 13.8 392.18 13.7
 (14.39) (10.59)
Purchasing 110.08 3.9 10141 3.5
 (6.98) (5.29)
Education 0.00 0 0.00 0
 (0) (0)
Employment 0.00 0 3.02 0
 (0) (1.22)
Socialising 0.00 0 155.23 *** 5.5
 (0) (6.92)
Recreation 84049 30.5 766.05 *** 26.9
 (17.43) (14.21)
Child Care 0.00 0 0.00 0
 (0) (0)
Volunteering 0.00 0 0.00 0
And Care (0) (0)

 100% 100%

Total Time

Personal Care 1434.55 25.9 1457.27 25.5
 (16.97) (11.27)
Domestic 394.34 7.0 40.78 7.0
 (14.67) (10.95)
Purchasing 111.12 2.0 102.23 1.8
 (7.06) (5.35)
Education 0.00 0.0 0.00 0.0
 (0) (0)
Employment 0.00 0.0 0.00 0.0
 (0)
Socialising 0.00 0.0 156.69 *** 2.8
 (0) (6.98)
Recreation 3587.66 64.6 3595.69 62.9
 (31.05) (17.12)
Child Care 0.00 0.0 2.37 0.0
 (0) (0.84)
Volunteering 0.00 0.0 0.00 0.0
And Care (0) (0)

N = 370 (27%) 422 (31%)

 Time Budget Activity Grouping

 Household and
 Community Multiple-Activity

 Orientation Orientation (2)

Time Sent in: Mean % Time Mean % Time

Primary Time (4)

Personal Care 1362.6 ** 48.9 1345.94 *** 47.0
 (19.92) (9.04)
Domestic 424.24 15.0 430.55 ** 15.0
 (19.43) (10.92)
Purchasing 97.24 3.4 109.13
 (7.91) (4.98)
Education 0.00 0 0.00
 (0) (0)
Employment 0.00 0 0.00
 (0) (0)
Socialising 0.00 0 145.09 *** 5.1
 (0) (7.35)
Recreation 697.21 *** 24.7 657.34 *** 22.9
 (21.33) (12.83)
Child Care 45.89 *** 1.8 22.63 *** 0.8
 (7.94) (2.84)
Volunteering 172.01 *** 6.1 153.83 *** 5.4
And Care (14.94) (8.42)

 100% 100%

Total Time

Personal Care 1378.46 ** 24.5 1369.17 *** 23.9
 (19.79) (9.42)
Domestic 440.65 7.8 446.89 *** 7.8
 (20.32) (11.45)
Purchasing 97.65 1.7 110.65 1.9
 (7.95) (5.13)
Education 0.00 0.0 0.00 0.0
 (0) (0)
Employment 0.00 0.0 0.00 0.0
 (0) (0)
Socialising 0.00 0.0 150.29 *** 2.6
 (0) (8.03)
Recreation 3422.22 *** 60.9 3427.53 *** 59.7
 (40.31) (17.89)
Child Care 107.98 *** 2.0 76.59 *** 1.3
 (21.89) (11.15)
Volunteering 172.14 *** 3.1 156.14 *** 2.7
And Care (14.95) (8.55)

N = 160 (12%) 398 (30%)

SOURCE: 'ABS (1997) Time Use Survey, Confidentialised Record File',
N=1,350. Notes: (1) results for secondary time are available from
the authors upon requests (2) Other Activities include child care,
voluntary work and adult care, and the combinations; (3) cells
were the mean confidence interval includes 0 have been recoded to
0; (4) Statistical tests for mean difference between "Household
Orientation" and the three other activity groupings. Other mean
differences among groups available from authors upon request;
*** p < .01; ** p < 0.05; * p < .10.

Table 2: Characteristics of Older Australians by Time Budget
Activity Grouping

 Time Budget Activity
 Grouping

 Household
 and Social
 Household Life
 Orientation Orientation

 Mean Mean

Age group (55-64) 0.23 0.32
Age group (65-74) 0.44 0.39
Age group (75+) 0.33 0.29
Less than high school 0.41 0.30
 (0.03) (0.02)
High school graduate 0.31 0.36
 (0.02) (0.02)
Vocational 0.26 0.26
 training/education (0.02) (0.02)
College/University 0.03 0.09
 (0.01) (0.01)
English first language 0.22 0.17
 spoken (0.02) (0.02)
With Disability 0.22 0.18
 (0.02) (0.02)
Gov't pension or allowance 0.58 0.54
 (0.03) (0.02)
Household Income X Gov't 1.11 1.11
 pension or allowance (0.06) (0.06)
Household Income 1.97 1.97
 (0.06) (0.06)
Vehicle 0.76 0.84
 (0.02) (0.02)
City 0.59 0.62
 (0.03) (0.02)
Regional 0.25 0.30
 (0.02) (0.02)
Rural 0.16 0.08
 (0.02) (0.01)
Female 0.50 0.60
 (0.03) (0.02)
Coupled/ Defacto 0.69 0.63
 (0.02) (0.02)
Seperated/Divorce 0.07 0.08
 (.01) (.01)
Widowed 0.19 0.25
 (0.02) (.02)
Never Married 0.06 0.04
 (0.01) (.01)

N = 370 (27%) 422 (31%)

 Time Budget Activity Grouping

 Household
 and Multiple
 Community Activity
 Orientation Orientation Total

 Mean Mean Mean

Age group (55-64) 0.39 0.44 0.34
Age group (65-74) 0.46 0.39 0.41
Age group (75+) 0.15 0.18 0.25
Less than high school 0.37 0.26 0.33
 (0.04) (0.02) (0.01)
High school graduate 0.38 0.32 0.33
 (0.04) (0.02) (0.01)
Vocational 0.19 0.27 0.25
 training/education (0.03) (0.02) (0.01)
College/University 0.06 0.15 0.09
 (0.02) (0.02) (0.01)
English first language 0.24 0.17 0.19
 spoken (0.03) (0.02) (0.01)
With Disability 0.13 0.11 0.16
 (0.03) (0.02) (0.01)
Gov't pension or allowance 0.55 0.45 0.52
 (0.04) (0.03) (0.01)
Household Income X Gov't 1.16 0.91 1.05
 pension or allowance (0.09) (0.06) (0.03)
Household Income 1.86 2.13 2.01
 (0.08) (0.09) (0.04)
Vehicle 0.85 0.90 0.84
 (0.03) (0.02) (0.01)
City 0.63 0.55 0.59
 (0.04) (0.03) (0.01)
Regional 0.27 0.31 0.29
 (0.04) (0.02) (0.01)
Rural 0.10 0.14 0.12
 (0.02) (0.02) (0.01)
Female 0.65 0.60 0.58
 (0.04) (0.03) (0.01)
Coupled/ Defacto 0.65 0.67 0.66
 (0.04) (0.02) (0.08)
Seperated/Divorce 0.09 0.08 0.08
 (.02) (.01) (0.01)
Widowed 0.21 0.21 0.22
 (.03) (.02) (0.01)
Never Married 0.05 0.04 0.05
 (0.02) (.01) (0.01)

N = 160 (12%) 398 (30%) 1.356 (100%)

SOURCE: 1997 Australian Time Use Survey; Notes: 'Household
Orientation' is the base category in the multinomial logit
model; Primary time only; * p < .10; ** p < .05; *** p < .001.

Table 3: Multinomial Logistic Regression Model of Time
Allocation Among Older Australians

 Time Budget Activity Groupings (1)

 Household & Household & Multiple-
 Social Life Community Activity
 Orientation Orientation Orientation

Independent Variables
Age 65-74 -.503 ** -.504 ** -.794 ***
Age 75+ -.552 ** -1.330 *** -1.297 ***
Separated/Divorced .532 * .643 * .491
Widowed .708 *** .638 ** .808 ***
Never Married -.097 .339 .053
High School Graduate .357 ** .160 .210
Vocational Training .332 * -.204 .361 *
College/University 1.378 *** .894 * 1.914 ***
English First Language Spoken -.456 ** -.128 -.312
Severe/Moderate Disability -.118 -.463 * -.588 **
Government Pension/Allowance -1.103 ** -.946 * -.354
Household Income X Government
Pension/Allowance .562 ** .567 ** .127
Household Income -.112 -.163 .005
Vehicle .855 *** .942 ** 1.128 ***
Business Income--Dividends .453 * .267 .828 ***
Regional .079 -.035 .286
Rural -.851 *** -.674 ** -.164
Female .317 * .449 ** .411 **
Constant -.485 -1.20 -.915

-2 * Log likelihood = -1,684.53

N = 1,350

SOURCE: 1997 Australian Time Use Survey; Notes: 'Household Orientation'
is the base category in the multinomial logit model; Primary time only;
* p < .10; ** p < .05; *** p < .001.