Introduction

Wellbeing is an important part of a fulfilling life, and while indicators of wellbeing for the overall population are increasingly available (see the ABS Measures of Australia’s Progress and the Australian National Development Index [ANDI], as examples of national indicators), there are currently few indicators of wellbeing for older people (for examples, see Lui et al., 2011; Miranti and Yu, 2015). These indicators are important as they can provide a basis for research on what characteristics are associated with higher wellbeing for older people. If calculated at a community level, they can also provide important information on what community factors impact on wellbeing for older people. Similar indicators for children have been used to identify the community level factors that affect educational outcomes (see Goldfeld et al., 2015).

Older people form an important part of any community. They may, for example, be involved in community organisations, assist with childcare for their family, provide care and support to other family members and act as mentors to younger generations. Everyone in a community contributes to the community in different ways, and older people help to provide diversity and life experience in any community.

However, just as there are public and private financial costs specific to each life stage, there are particular costs associated with later life. These can stem from increased health and support costs, as well as reduced tax revenue flowing to governments as this age group retires. Population projections from the ABS show that the proportion of people aged 65 and over is expected to increase from 14 per cent on 30 June 2012 to 22 per cent in 2061 (ABS 2013).

An increasing number of older people also means potentially higher levels of some other forms of government assistance, including age pension payments and aged care. While older people may mean higher welfare payments and health costs, they also contribute to significant benefits in the community. It has been estimated that the uncosted contribution of older Australians as carers of people with a disability and carers of grandchildren is $22 billion per annum and their contribution through volunteering is valued at $16.3 billion per annum. These are substantial figures that are equivalent to nearly one sixth of total Commonwealth expenditure in 2013/14 (National Seniors Australia 2015).

There is a significant saving to the Government of people ageing in place (Bridge et al. 2008). The average annual cost to the Government of care for older Australians who receive assistance through home care packages was $15,525, whereas this cost was $66,512 for those in residential aged care. A review of the costs of ageing in place in the US found that assisted living technology to enable ageing in place may reduce costs, though the studies finding this were limited by little and poor-quality data. Gains or losses in quality of life were not measured, so potential impacts on health and wellbeing may have been missed (Graybill et al. 2014). However, there are disadvantages of ageing in place. One of these is the potential social isolation it can create as elderly residents become less mobile. This can be reduced through community transport

The increasing proportion of older people, the increasing costs associated with this group and their potential economic contribution, mean that it is important to have an understanding of the wellbeing of this group, and more importantly, where areas of high and low wellbeing are, to assist policy development and service provision to this group.

Thus, in this article, we develop an index of wellbeing using various data sources. The aims of this work are to identify areas where older people have low wellbeing; and to show how the domains and indicators can be used to identify why older people in an area have low wellbeing. The areas of low overall wellbeing for older people may require more services, and the domains and indicators can then help to identify what types of services are required to improve the wellbeing of older people in these areas.

This article is structured as follows: section two discusses the literature review, section three covers the data and methodology. Section four outlines results; and section five has some discussion and analysis on domains and indicators. Section six deals with limitations, followed by future work. Section seven provides a conclusion.

Literature Review

There is a growing body of work which discusses increasing economic and social inequalities among older people in Australia (Kendig 2000; Olsberg and Winters 2005; and Faulkner 2007). Research suggests that some groups of older people are among the most disadvantaged Australians. Miranti and Yu (2015) have measured the extent of social exclusion among older people in Australia using the Household Income and Labour Dynamics in Australia (HILDA) data at the individual level and examined why social exclusion persists in Australia.

Other research has found that while many older people are outright homeowners, others, unfortunately, are still renters or pay mortgages. While some older people achieve financial independence through substantial superannuation balances, many others are wholly dependent on government pensions. Older people (including those who rely on the age pension as their main source of income) are among those at greatest risk of poverty (Saunders et al. 2008; Tanton, et al. 2009 and Miranti et al. 2011). Miranti et al. (2011) found that almost half of lone persons aged 65 and over were in poverty (47.4 per cent). Further, while the poverty rate for a couple with a reference person aged under 35 was 2.1 per cent, Miranti et al. (2011) found that the poverty rate for couples with a reference person aged 65 and over was nine times higher at almost 19 per cent.

For an older person, where you live represents an important dimension of ageing. In this context, the physical environment provides community infrastructure such as closeness to amenities like supermarkets and doctors. Clifton’s (2009) discussion of the built environment relates to older people’s well-being and includes the possible impact of a range of factors including a sense of safety, opportunities for physical activity and the maintenance of mobility and contact with the natural environment. In an Australian context, Kendig (2000) has noted the importance of an ’age-friendly environment’ where older people prefer to live in locations close to amenities, usually located in the middle and outer suburbs where people first settled as young families in the 1960s and 1970s, and which have now experienced generational change. Conahan et al. (2004, pp. 1-3) argued that the attachment of older people to their location depends on three factors which are (i) physical environment, (ii) social environment and (iii) support services. These factors are significant as older people are far less mobile in comparison to younger people due to difficulty with physical mobility. This is the case regardless of where they live (at home or in an aged care home).

Previous research on well-being and disadvantage has increasingly emphasized the need for spatial and environmental indicators to add to an understanding of locational disadvantage and the extent to which disadvantage varies across small areas (for examples, see Harding et al., 2009; Miranti et al., 2021; Tanton et al., 2010). However, we have been able to find very little work which has analysed older people’s wellbeing at the small area level. Among these limited studies, Gong et al. (2012, 2014) estimated indicators of advantage and disadvantage among older Australians at a small area level, and indexes of wellbeing for older Australians using 2011 Census data were produced in 2016 (Tanton et al., 2016).

Thus, to fill in the gap, the importance of the interplay between place and wellbeing is an important part of this article. Abbott and Sapford (2005) have researched the importance of place and the geographical polarisation of neighbourhoods in relation to older people. They found that some older people lived in areas with concentrated deprivation and thus were especially affected by some forms of disadvantage including basic access to services as well as health and personal safety (fear of crime).

Differences in wellbeing between different areas may determine the quality and level of care and services to which older people have access. The Australian Institute of Health and Welfare (AIHW) notes that ‘ageing in regional areas is affected most by the proportion and age structure of people entering or leaving an area, rather than influences such as fertility and mortality which underlie population ageing in Australia as a whole’ (AIHW 2007, p.7). This concentration may then have implications for service delivery. This examination of differences in wellbeing between different small areas is conducted through the development of an index of wellbeing for older Australian.

Data and Methodology

This section describes the domains and indicators chosen for the index, the data used for the indicators, and the index method.

The Domains

The method we have used to develop the index of wellbeing for older Australians uses a domains approach. The choice of domains has been informed by previous studies on key indicator frameworks that focused on the wellbeing of older people (for example, Brotherhood of St Laurence 2009; Australian Institute of Health and Welfare (AIHW) 2007; UK Government 2006; United States Federal Interagency Forum on Aging and Related Statistics 2008). We also used the wellbeing framework published in a literature review by Miranti et al. (2010). The index covers five domains which are participation, education, resources, wealth and housing and functional ability.

The participation domain is about how well older people can participate in society. The main indicator is employment. Davey and Davies (2006) emphasise the benefits of working in later life in New Zealand and the United Kingdom. For older people, working in later life can boost confidence, self-esteem and health. Other indicators in this domain reflect how older people may participate in society, and include volunteering, access to motor vehicles and internet, and whether the person is caring for others.

The education domain is about the level of education that the individual has. Education has been considered as an investment in human capital and has been linked to higher incomes (Becker 1962). We have followed the approach taken by the Brotherhood of St Laurence (2009) and included an education domain which covers completed year 10, completed year 12, and post school qualifications indicators.

The functional ability domain is about the person’s physical capabilities. As Peel et al. (2004) argued, physical capability is the ability to function independently without reliance on care. In the index, it includes indicators of service use and provision including the proportion of older people in the area who need assistance with core activities; and administrative data on older people using Commonwealth Home Support Program services.

The resources domain covers what resources the person has. It provides a framework for thinking about the positive aspects of ageing and includes evidence of older people’s strengths and contributions (Miranti et al., 2010). One of the most important indicators under resources is income, which for the elderly can include superannuation payouts and the pension. Indicators in this domain include income, financial stress, and whether the person is paying housing costs on a low income.

Another key resource of older people in Australia is wealth with home ownership the most common asset among older people (Olsberg and Winters 2005). Legge and O’Loughlin (2000) assert that home ownership is important not only to provide security/protection, but also for intergenerational transfers of the use of the parental home by adult children. The wealth and housing domain is about what wealth the person holds, as well as their housing situation. The indicators include the proportion of older people in the area in housing stress, in public housing, or homeless.

The final list of indicators used, the domains, and the source of the data, are shown in Table 1.

Table 1 List of indicators, domain and source, 2016 IWOA
Full size table

Data Sources

Older people were defined as people aged 65 and older. This choice is supported by Australian and international literature (OECD, 2015). Lack of data means that we cannot consider the oldest old in our index, but we acknowledge that they may have a different pattern of wellbeing due to greater health issues. The oldest old are still included as people aged 65 and older.

All the indicators were collected or derived for areas called Statistical Area 2 (SA2) by the ABS. These broadly match suburbs in capital cities but tend to be larger areas in regional and remote Australia.

The complexity of the Index, and the framework identified above, means that there are a number of sources of data for the indicators. The main source of data was the 2016 Census. This data was extracted using the ABS Tablebuilder package. The data were mainly for a person’s place of usual residence, so that we are looking at wellbeing for the area where the person normally lives, rather than where they were on Census night. Table 1 shows that 12 out of total 21 indicators were from the 2016 Census.

The next source of data was a spatial microsimulation model. Spatial microsimulation is a statistical technique that calculates estimates for small areas from survey and small area Census data. As indicated in Table 1, there were a third of total indicators (such as poverty rates and housing stress) estimated using this technique as the data at the small area level was not available.

The method uses a generalised regression reweighting algorithm (Tanton et al., 2011) that reweights a unit record file (in this case the ABS Household Expenditure survey 2015-16) to 2016 Census benchmarks in the same way a national survey is reweighted to national totals. The SAS program GREGWT is used to conduct the spatial microsimulation, and the R SURVEY package has a similar command. The Census benchmarks for each SA2 are provided to the GREGWT program, as well as the survey file with the original weights scaled down to the population of the SA2. These start weights are then adjusted to fit the benchmarks, and the program iterates until estimates with an acceptable error level are derived.

The method provides a reweighted survey for every SA2, which can then be used to derive estimates from the survey for each SA2 for indicators in the survey that are correlated with the benchmarks. The results by SA2 are then aggregated to State/Territory level and validated against the aggregated HES survey data, a standard way of validating these models (Edwards and Tanton 2013). The results from this validation showed that the estimates we had derived were reasonable, and these graphs are available on request from the authors.

The model automatically eliminates any SA2 that fails to achieve the threshold of acceptable error (the Total Absolute Error from the model is greater than the population of the SA2), in which case the model would not provide an estimate. The generalised regression model is also iterative, and if convergence isn’t achieved after 30 iterations, then the number of benchmarks is reduced until convergence is achieved. If this doesn’t happen at 7 benchmarks, then the area is removed from the analysis.

A technical description of the model can be found in Tanton et al. (2011). The model relies on a number of benchmarks, and these benchmarks determine what indicators can be accurately estimated from the model. The benchmarks used are shown in Table 2.

Table 2 Benchmarks for the spatial microsimulation model
Full size table

After estimates for each indicator were derived from the spatial microsimulation model, the results were then validated against other data where available. This can be difficult (the reason for doing the modelling is because the small area data are not available), but the results showed estimates that were reasonable. The graphs of this validation are available from the authors.

The final set of data were administrative data from the Australian Institute for Health and Welfare on users of the Commonwealth Home Support Program services and Home Care Packages Program services - low level and high level.

Because incomes are shared across families, all the income data were aggregated to an income unit level before calculating the poverty rate (the proportion of older people living in poverty). The Australian Bureau of Statistics defines an income unit as “A unit based on the degree of income sharing between couples and their dependent children.” (ABS 2019). So, for all indicators, the person was the base level – so for example, for housing stress, it is the proportion of older people living in households experiencing housing stress, rather than the proportion of households in housing stress. This was because all indicators were for people aged 65 and over, so they all needed to be measured at the person level.

To reduce variability in the final Index, any areas with less than 30 people aged 65 and above were excluded. The reason for this was that areas with low populations provide unreliable results when calculating a proportion – so with 20 people in the area, an additional 1 person with the characteristic being measured by the indicator adds 5% to the indicator value. This is reduced to 3.3% when there are 30 people in the area, reducing the variability of the indicator.

Index Methodology

The method used for calculating the Index was principal components analysis for each domain, and then adding the domains together using a log transformation following the standard formula described in Noble et al. (2004) for indexes of deprivation in South Africa and Bradshaw et al. (2009). The log transformation standardises all the domain indexes into a normal distribution, allowing them to be added.

If there are missing values (i.e., where data are not available) for at least one domain in an area, the whole area is removed from the analysis as the log transformation cannot be calculated for these areas. This is similar to the method used for child and youth social exclusion indexes calculated in Australia (Miranti et al. 2015; Abello et al. 2016). Principal components analysis is the same method used for calculating the Socio-Economic Indexes for Areas (SEIFA) indexes by the Australian Bureau of Statistics (ABS).

The first step in the method is to run a correlation matrix for all the indicators in the domain. Indicators that are too highly correlated are dropped from the domain. Indicators that are not correlated with other indicators will also be dropped in the next step as they will have a low loading against the overall Index.

The next step is to run an initial principal components analysis and look at the loadings of each indicator against the first component. Indicators with a loading less than 0.3 are removed as they do not contribute much to the final index. This is the same cut-off as used by the ABS for their SEIFA index. This is an iterative process, so the indicator with the lowest weight is removed and the principal components analysis is re-run until all indicators have weights above 0.3.

The next step is to look at the proportion of the correlation explained by the index. The first component should explain most of the correlation, with the following components explaining less. If the second component still explains a lot of the correlation, then this can be used as another component in the final index. Only the Functional Ability domain had a reasonable loading on the second component, but interpreting this component was difficult due indicators loading onto the first two components, so only the first component in each of the domains was used for the index.

The final step was to ensure that the direction of the domain indexes was the same. For our index, a lower value meant a higher proportion of older people in the area with low wellbeing, and a higher value meant a higher proportion of older people in the area experiencing high wellbeing. So, our Index includes indicators of high wellbeing (like volunteering and the employment rate) as well as indicators of low wellbeing (like the unemployment rate and the poverty rate).

These steps were conducted for all the domains, and then the domains were transformed using a log transformation, as described above. The final Index was then calculated by averaging the five domain indexes after the log transformation.

As explained earlier, the final index will consist of five domains: participation, education, resources, wealth and housing and functional ability. The advantage of the domains approach is that it allows drilling down from the overall index. If an area is showing as low wellbeing, the domains can be used to identify whether there is one domain driving the low wellbeing in an area; and the indicator can be mapped to identify the indicators that are contributing to low wellbeing in the domain. So, for example is it to do with incomes in the area, or participation or some other factor. Thus, the use of the index, domains and their indicators then provide a powerful tool for additional analysis, and this will be demonstrated later.

Results

The chosen final indicators and weights for each domain are presented in Table 3. The results show that in the participation domain, the proportion of older people who are volunteers has the highest contribution followed closely by the employment rate of older people.

Table 3 List of Indicators, Domains and Weights
Full size table

On the other hand, older people who cannot speak English well or not at all is the main contributor to lower wellbeing, which is in line with the finding from Miranti and Yu (2015) that older migrants from non-English speaking countries tend to experience persistence in social exclusion. No access to a car to drive and then no access to internet from the dwelling follow as the next contributor to lower wellbeing.

In the education domain, the post school qualification is the main indicator that increases wellbeing. In terms of resources, the poverty rate for older people and living in income units that rely on the age pension as the main source of income are the major factors that reduce wellbeing. In functional ability the proportion of older people who need assistance with core activities is the main factor that reduces wellbeing, reinforcing the finding from Miranti and Yu (2015). In the wealth and housing domain, receiving rent assistance and being in housing stress are the major factors reducing wellbeing and these factors have a greater impact on wellbeing than living in public housing. Older Australians who rent in private markets have been largely identified as a group who are highly likely to experience housing stress (Tanton and Phillips 2013).

The results reflect both vulnerability and capability concepts of wellbeing among older people and these may be associated with different cohorts of the older population. The need for assistance reflects the vulnerability of the oldest old cohort, who are more prone to sickness and frailty than the younger cohort of older people. Volunteering reflects the capability of the younger group of older people who are experiencing the transition from formal participation in the labour market to volunteering as their way to participate actively in society. Volunteering is a way for older people to make a contribution by participating socially and engaging in community life, and it has been argued that this not only improves morale, self-esteem and creates a larger social network, but also increases life satisfaction (Burr et al. 2005) and wellbeing (Morrow-Howell et al. 2003).

The spatial results for the final Index are shown in Figure 1. In this map, we show population weighted quintiles of wellbeing for people aged 65 and over. A population weighted quintile splits the population into 5 equal groups, with the same number of people in each quintile rather than the same number of areas. Higher values are where a higher proportion of older people experience high wellbeing, and lower values are where a higher proportion of older people experience low wellbeing.

Fig. 1
figure 1

Map of older person wellbeing index, 2016. Source: Authors’ summary

Full size image

For some areas in remote Australia, estimates for the index could not be derived. This was either because there were too few people aged 65 and over in these areas; that some of the data were not available for these areas; or that there were technical problems with the small area modelling procedure (see Tanton et al., 2011, for a technical description of the problems with convergence in the model used). These areas are shown speckled on the map.

It can be seen that some of the highest levels of wellbeing for older people are in regional Australia, particular in Victoria and Western Australia, as well as in the outskirts of cities. Most of the higher levels of well-being in these areas are due to higher functional ability and participation. The participation domain in these areas was driven by employment and volunteer activities among older people. It may not be surprising that the proportion of volunteers among older people are higher in the outskirts of and outside the cities. Nevertheless, the employment level in the outskirts of the cities is also contributing to the result. Despite this, employment for older people seem to be higher in rural areas. The higher participation of older people outside the cities may suggest that it is important to provide older people opportunities to be active and contribute to the community. As suggested by Davey and Davies (2006), there seems to be correlation between this participation with functional ability. Nevertheless, the lack of aged care facilities may also contribute to this estimate.

Large cities show more diversity in terms of wellbeing. Many of the areas of the lowest and highest wellbeing are in the cities. Many areas of low wellbeing in the cities are areas of gentrification in the inner city, and this is seen most clearly in Sydney and Melbourne. Areas of low wellbeing are generally seen outside the central suburbs, like Sydney’s Western suburbs, and Melbourne and Perth’s East. The education and resources domains play an important role in the level of wellbeing in this part of the cities, while the participation rate also contributes in the inner-city areas. This highlights the importance of education for older age groups. Mermin et al. (2007) have found that higher educational attainment is one of the determinants of older people working longer. However, it is not necessarily true that gaining an education at an older age will contribute to this since Crystal and Shea (1990) have argued that this may just reflect social stratification, and sources of retirement income.

A generalisation could be made that the further the area is from the city centre, the lower the wellbeing for older people. This is not always the case (e.g., North of Sydney central is all high wellbeing), so there tends to be corridors of low wellbeing through Western Sydney; north Melbourne; and East Perth.

These could be examples of the areas described by Randolph and Freestone (2012) as middle-ring suburbs and which Forster (2004) identified as places ‘strongly affected by ageing in place’. The gentrification in these outer suburbs is far less than the inner-city areas leaving more older people in the area while the local government (councils) may have difficulty maintaining services to older people in this usually high-density area.

In terms of remote areas, there is a range of values, with areas in northern NSW having low values, and areas in central Victoria having high values. Remote communities in North Queensland have the lowest levels of wellbeing.

The maps show where older people are experiencing high and low levels of wellbeing. After low wellbeing areas are identified, the next step is to decide what services are required by older people in areas with low wellbeing. The next section shows how the domain method used to calculate the index can contribute to this identification of the services required by showing what is contributing to low wellbeing in an area. Further quantitative and qualitative work, including cost/benefit analysis of the service being provided, then needs to be done before providing the service.

Discussion

The intergenerational reports from the Commonwealth Treasury have highlighted the increasing costs of providing services and assistance to an ageing population in Australia, including health costs, income support costs, costs of community care and aged care. Both Commonwealth and State governments are increasingly focussing on the impact of the ageing of the population and changes in services needed to prepare and respond to it.

This means that services need to be provided to older people in the most effective and efficient way possible and in the right place. It is essential to identify areas of low wellbeing for older people, and the factors contributing to low wellbeing in these areas as this will allow governments at all levels to address any failures in the provision of services to older people in areas of low wellbeing and to more finely target service provision such as health, housing assistance, community care services or low income support services.

This section shows how the method used for creating the final index of wellbeing for older Australians can be used to assist in the identification of why an area is experiencing low wellbeing. The analysis uses the overall index to identify areas of low wellbeing; and then uses the domains to identify whether an area has low wellbeing on all domains, or whether the area has high wellbeing on one domain and low on the others. This helps to target particular services to particular areas.

Figure 2 shows the index quintiles for Gunnedah, a regional town in NSW, along with the surrounding area. Because we have focussed on one small area, the maps only show two colours: one colour for the index quintile in Gunnedah, and another colour for the index quintile in the area around Gunnedah. Figure 2 shows the town is in the bottom quintile of the index (low wellbeing) while the surrounding region is in the second highest quintile. Figure 2 also shows the indexes for each of the domains and shows that both areas are in the bottom quintile of education and second bottom quintile in resources, the domain which is mainly driven by income. Figure 2 shows that the main driver of low wellbeing in Gunnedah town is housing, which is estimated to be better in the surrounding area. Other differences are that participation is average in Gunnedah town (Q3) while it is in the highest quintile in the surrounding area. The functional ability domain is also higher in the surrounding region as it is in the second highest quintile while in Gunnedah it was in the second lowest quintile. This shows that the index isn’t simply tracking income.

Fig. 2
figure 2

IWOA index and sub-index values for Gunnedah. Source: Authors’ summary

Full size image

Investigating the housing domain further using the indicators that contribute to this domain can help to show what is driving low wellbeing in this domain. Figure 3 shows the indicators that contribute to the housing domain. These need to be read in conjunction with Table 3 which shows the important drivers of wellbeing in each domain. The main driver of low wellbeing in the housing domain is the percent of people in housing stress, and Table 3 shows that a higher proportion in housing stress is associated with lower overall wellbeing in an area (a negative weight). Receiving rent assistance is also a driver, and Table 3 shows a higher proportion of those receiving rent assistance is associated with lower wellbeing in an area (a negative weight). There is a low proportion who are still paying a mortgage, and this would be associated with higher wellbeing as this indicator has a negative weight in Table 3.

Fig. 3
figure 3

Indicators in the housing domain for Gunnedah. Source: Authors’ summary

Full size image

From this analysis, we can say that low wellbeing in the area is driven partly by housing, and in particular relatively high housing costs compared to incomes (as indicated by the housing stress indicator). This would be partly offset by a high proportion of people receiving rent assistance, but not fully offset as the rate of housing stress is still high. This appears to be renters rather than mortgagees, as the percent still paying mortgages is low. This means that policies around lowering rents for older people in Gunnedah would be an effective policy to increase wellbeing for older people in the area.

These indicators also allow government and non-government providers of services to older people to target their services at the spatial level more precisely and appropriately – for example, areas with low levels on the participation domain may be areas where improvements to public and community transport will be most useful, allowing older people to continue to stay active members of their community although they have no access to a motor vehicle.

To make access to this type of analysis easier, online maps have been provided to allow any service provider to conduct a similar analysis for their community. These online maps use the ARCGIS online mapping capability, and are available at:

https://canberra.maps.arcgis.com/apps/webappviewer/index.html?id=071c7b8e4ff54eff811b944e66de97a5

All indicators that were modelled or were open data were provided in the online maps. Some indicators were derived from special data requests and were not available through the online maps.

Obviously, the issues surrounding where services should be provided to older people is much more complex than looking at some indexes and indicators, but this type of analysis using the online maps can provide important input into a discussion on where to provide services, along with community consultation, cost/benefit analysis, and other considerations.

Limitations and Future Work

Limitations of the Index

Despite its benefits, we acknowledge there are some limitations of this index. The development of the index was limited by what data were available for small areas. The index required data for older people that covers small areas. Most surveys in Australia will not have the coverage required to produce estimates for a small area for a specific age group. This meant that most of the data was sourced from the ABS Census of Population and Housing, administrative data, or modelled small area data. For some domains, there were some missing data due to the modelling process which could not provide robust and reliable estimates for all areas. This meant that the index could not be calculated for some remote areas, and some parts of both regional and metropolitan Australia.

Further, the functional ability domain was initially a health domain, but no small area estimates of self-assessed health were available across Australia. This means the health domain has been called functional ability, recognising that the domain contains indicators of functional ability, not health.

Due to the unavailability of data, we also could not create an index for sub-groups of the population like the oldest old, indigenous persons or cultural and linguistically diverse (CALD) populations.

Finally, like all summary indexes, it is intended for general identification of wellbeing in an area, and other indicators should be used for particular case studies. For example, the index could be used to identify areas of low wellbeing for potential increased service provision to older Australians, but then demographic change and economic decline over time would need to be studied to identify whether it is worth investing in the area. The index can act to broadly identify areas of low wellbeing to then focus on further using specific indicators and qualitative analysis in the community.

Further Work

This work has highlighted several areas of potential further study. They include comparing the index of older person wellbeing to the ABS Socio-Economic Index for Areas (SEIFA), an index of general disadvantage in an area, and deriving modelled small area estimates of self-assessed health.

An interesting question that came out of this research was which areas have low wellbeing for older people but high wellbeing overall (using, for example, the SEIFA index). We expect these areas might be in cities where gentrification is occurring, or advantaged areas with a concentration of low wellbeing older people in aged care homes. If the situation is the former, then in terms of service provision, identifying these areas is important because services for low wellbeing people would normally not be provided in an advantaged area.

There is also significant work required on indicators for the health domain. Modelled estimates of self-assessed health were derived for this work, but validation showed they were unusable. Further work would look at using different imputation techniques (for example, imputation methods as used in Namazi-Rad et al. 2017) to attempt to derive reasonable estimates of self-assessed health.

Conclusions

The Index of wellbeing of older Australian contributes to the limited studies which examine economic and social inequalities among older people. There is intrinsic social value in better understanding the lived experience of older Australians; to know more about the quality of later life of our older citizens. In addition, Australia, like many developed countries, has an ageing population, and the financial and social implications of this will be both a challenge and a valuable opportunity for future generations, as identified in recent intergenerational reports from the Australian Government.

Older retirees add to the diversity of the future population and will continue to make an important contribution to society. The Index results show that one of these contributions, volunteering, in turn contributes to wellbeing among this group. The intergenerational reports have begun to understand and acknowledge the essential economic contributions made (now and increasingly into the future) by older people, both as consumers and in a range of employment, volunteer, and caring roles.

However, higher health care, aged care and income support costs associated with an ageing society are projected. Social changes also mean older people may have more difficulty accessing government services as departments promote online provision of services, and fewer family support networks survive in regional areas as younger generations move to cities for work.

The implications of these changing financial and social trends will affect older people experiencing low wellbeing greater than they will affect older people who experience high wellbeing. Those with high wellbeing have financial assets and possibly strong health to fall back on. Those who experience high wellbeing will also be able to use preventative health services and will be able to travel more easily for medical procedures.

From a policy perspective, the index of wellbeing for older Australians and the associated domain indexes and indicators can be used to identify where older people experiencing low wellbeing live. In this article, we have demonstrated a method of drilling down from an index to a domain index to an indicator to identify why an area has low wellbeing and help identify what services might assist low wellbeing older people in the area. This will help to target specific services to those most in need, meaning more efficient service provision in the future. For example, areas with low levels on the participation domain may be areas where improvements to public and community transport will be most useful, allowing older people to continue to stay active members of their community when they have no access to a motor vehicle. Housing in terms of rental issues contributing to low wellbeing indicates that functional assistance needs may become an increasingly important concern for older Australians. This suggests that social care assistance provided to older people needs to be a growing priority for governments.

To conclude, our results show how the index of wellbeing of older Australians is an important tool for current and future government and non-government service provision. It is older people experiencing low wellbeing who suffer the most from lack of services now and into the future, as they do not have the health, transport or money to access government services. This risks a very real demarcation of high and low wellbeing, as older people experiencing low wellbeing may not be able to access services provided in other areas due to low mobility or access to transport. The online map allows both government and non-government service providers to target services to particular locations that need these services the most, enhancing the lives of older people experiencing low wellbeing.

While our findings are based on a case in Australia, this work can be replicated using indexes calculated in the same way in other countries, so the method is applicable internationally.