Budget 2009-10 - Pension Review Report
4. Indexation
Overview and findings
This chapter considers the adjustment of the rate of the pension and other parameters of the pension system over time. The Review considers that effective indexation arrangements are essential to maintaining the capacity of income support payments to provide a basic acceptable standard of living, and therefore to provide pensioners with ongoing financial security.
This chapter outlines the current indexation arrangements; discusses the merits of different approaches to price adjustment, such as the Consumer Price Index and the Analytical Living Cost Indexes; and considers issues around indexation and sustainability. It also outlines the current benchmarking arrangements and discusses the relative merits of different benchmarking measures.
In this consideration, the Review notes that while there are strong arguments for determinations on changes in the rate of pensions to be taken as deliberative decisions by government, automatic indexation provides greater security for pensioners.
In addition to the benchmarking and indexation of the rate of the pension, this chapter considers how other income support parameters should be adjusted. The chapter notes the potential of ‘fiscal drag’ to introduce unintended distortions into structures, and concludes that indexation arrangements that create distortions between different components of the payment system should be avoided or, if they cannot, they should be subject to regular review.
The Review has concluded that there is a need for reform of the current indexation arrangements and that better measures are required of the price changes experienced by pensioners and of the community benchmark.
The Review has therefore developed specific findings in relation to both the indexation and benchmarking mechanisms that are used to set pension rates.
The Review’s first finding concerns the general approach to indexation and the relative roles of the community living standards benchmark and the index of price changes.
Finding 8: The Review finds that automatic indexation of pensions and a two-part approach of benchmarking and indexation should continue. Benchmarking pensions relative to community standards should be the primary indexation factor, with indexation for changes in prices acting as a safety net over periods where price change would otherwise reduce the real value of the pension. (Section 4.4.2)
This broadly conforms with the current structure of indexation, although the relative roles of the two components are differently structured to place an emphasis on the pension being benchmarked to a community standard.
However, the Review’s analysis of the different available measures of price changes has led it to conclude that reforms are needed to the price indexation mechanism for pensions to ensure that it is fully responsive to changes in pensioners’ purchasing power due to price changes.
Finding 9: The Review finds that pension indexation for price change would be better undertaken through an index that more specifically reflected cost of living changes for pensioners and other income support recipient households. (Section 4.4.3)
The Review’s analysis also indicates that reforms are needed to the current approach of using Male Total Average Weekly Earnings (MTAWE) to benchmark pensions to community living standards because of its lack of transparent relevance to the experience of the wider community.
Finding 10: The Review finds that no single measure to benchmark the pension to community living standards is without limitations. However, the Review considers that a measure of the net income of an employee on median full-time earnings may be a more appropriate measure than the existing Male Total Average Weekly Earnings benchmark. (Section 4.4.4)
Indexation is, however, a highly technical area of policy, and the Review has therefore found in relation to the timing of the implementation of reform.
Finding 11: The Review finds that, while reform to indexation and benchmarking is important to the financial security of pensioners, implementation may need to be phased in to account for policy developments that may arise out of the Australia’s Future Tax System Review and to allow for the development of appropriate mechanisms for benchmarking and indexation. (Section 4.4.4)
4.1 Terms of reference
The question of how rates of payment should be adjusted over time is the second part of the Review’s consideration of the first term of reference concerning ‘appropriate levels of income support’.
This adjustment process is central to the question of how to maintain ‘appropriate levels of income support’. Effective indexation methodologies are a critical component of reporting on ‘measures that might be adopted to strengthen the financial security of carers and seniors’. Maintaining the purchasing power of the pension and ensuring that pensioners can share in improving living standards are fundamental to the concept of financial security.
4.1.1 Current indexation and benchmarking mechanisms
A complex set of arrangements currently applies to the indexation of components of the pension system. While most payments and some allowances are indexed twice annually, others are only adjusted once each year and a few are not indexed at all. Indexing of various components occurs on four different dates each year: 1 January, 20 March, 1 July and 20 September, although no payment is indexed more than twice yearly.
Legislative provisions and benchmarking to community standards
The Social Security Act 1991 provides for the rate of the pension to be reviewed twice annually, in March and September. Technically, this is undertaken by a two-stage process:
- The base rate of the pension (known in the legislation as the ‘Maximum Basic Rate’) is indexed by calculating an indexation factor (the ratio of the CPI index number of the most recent reference quarter over the index in the base quarter), applying this factor to the current rate of payment, and rounding the resultant rate (called the provisional indexed amount) by the rounding base.
- For the single rate, this result is then compared with 25 per cent of the annualised MTAWE for the specified quarter. If this is higher, a ‘top-up amount’ is added to the pension rate so that it equals 25 per cent of MTAWE, and the resultant rate is rounded up by the rounding base.
- For the partnered rate, where a top-up amount has been added to the single rate, 83 per cent of this top-up amount is added to the partnered rate, and the resultant rate is rounded up in the same way as for the single rate.
Indexing other components of the pension system
Most other components of pension payments are subject to CPI indexation, which occurs once or twice each year depending on the component. The indexation periods of the Pension GST Supplement, Rent Assistance and Utilities Allowance are on the same schedule as the base rate; Carer Allowance, Pharmaceutical Allowance and Mobility Allowance are adjusted each January, and Telephone Allowance each September.
Some supplementary payments, including the Pensioner Education Supplement, are not subject to any indexation or formal review. Indeed, the rate of the Pensioner Education Supplement, which was introduced in 1998, has been adjusted only once, in July 2000, when it was increased by 4 per cent, from $60.00 to $62.40 a fortnight to take account of the impact of the introduction of the GST and other taxation changes.
Most means testing and related parameters for the payment of pensions are indexed to changes in the CPI. As with the indexation of rates, a series of different approaches is applied to the timing and frequency of indexation of other parameters. There are, however, some parameters that are not indexed: the ‘income test free’ areas for allowances and student payments, the child amounts of the pension income test, the additional child amounts of the Youth Allowance parental income test, and the Student Income Bank and Working Credit maximum accrual limits and fortnightly maximum accrual amounts.
There are also some direct links between the pension rate and other components of social assistance. These include the benchmarking of certain Family Tax Benefit rates to the pension, and the consequence of changes in the rate of the pension for the income-based concessional rents charged by public housing authorities and the basic daily fee charged by aged-care facilities, which is capped at 85 per cent of the annual single basic Age Pension, which is currently $32.95 a day.
4.2 What the consultations told us
Although some specific views on how the pension should be adjusted over time were put to the Review in consultations, most of the contributions took the form of concerns by individual pensioners that the cost of living for them had risen more rapidly than the pension rate.
4.2.1 Consultations—experiences of living cost changes
Submissions to the Review indicated a strong belief by pensioners that the cost of living has exceeded increases in the pension and that their living standards had fallen. This view was firmly put in the public forums and was raised in 28 per cent of individual submissions and 46 per cent of submissions from organisations. It was more frequently reported by single or widowed pensioners and by those on Carer Payment and Disability Support Pension.
Many pensioners provided detailed information on household spending and how the prices they have paid for various items have changed over time. Areas where particular concerns were voiced, both because of the apparent rate of price change and importance of the item to pensioners’ budgets, were the cost of food, especially fresh fruit and vegetables and meat, the price of fuel, and the cost of utilities, particularly electricity and water.
The issue of price change was also highlighted in many of the submissions from organisations. A clear message was that these organisations considered, from their experience, that there is a significant group of pensioners who are totally reliant on the pension who are finding it extremely difficult to manage from day to day. There was a strong view expressed that increases to the Age Pension had not kept pace with the cost of living and that this was often placing pensioners in precarious situations. It was also claimed that the purchasing power of the pension has diminished to the point where it is completely inadequate.
A number of the submissions also made specific comments about the appropriateness of different mechanisms for adjusting pensions, these are considered below. Some submissions raised specific concerns about the CPI. Chief among these was a concern that, because it is designed to indicate general price inflation, the CPI does not recognise distributional impacts of changes in costs of goods and services and differing consumption patterns across various household types.
Less comment was made on the use of MTAWE as the basis of pension benchmarking, although one submission from a national organisation proposed that Full-Time Adult Ordinary Time Average Weekly Earnings be used as a new benchmark. It was argued that MTAWE was significantly distorted by the inclusion of part-time earnings and the effect of a growing proportion of part-time workers. The submission also suggested that it would be appropriate to adopt a measure that recognises the role of women in the workforce.
4.2.2 Consultations—proposals for alternative indexation mechanisms
Most of the submissions addressed the question of indexation by expressing their view that current indexation arrangements have failed to maintain living standards in the light of changing prices. A number made specific proposals as to how pensions should be adjusted in the future. These included:
- ongoing indexation through benchmarking (at a higher level) to MTAWE
- regular updating of a budget standards benchmark, either using an existing measure such as the Westpac–ASFA or a new measure (as discussed in Chapter 3)
- linking to the minimum wage
- redevelopment of an extended version of the ABS Analytical Living Cost Index for age pensioners.
Another proposal was for more frequent pension adjustments. Several submissions suggested that quarterly changes were more appropriate, given rapidly changing price levels.
In addition to voicing criticisms of the current indexation of pensions, in a number of submissions the claims made clearly indicated that the way in which pensions are currently indexed is not well understood. For example, many suggested that the pension had only been indexed for price increases, while others suggested that although it had been adjusted for wages, no account had been taken of changes in prices.
4.3 What the evidence shows
This section initially describes the current structure and outcomes of pension indexation in Australia and internationally. It then considers the following three major areas of analysis of indexation conducted by the Review:
- the nature of the CPI and the marked discrepancy in the views received in submissions and consultations on changes in the cost of living, and price change as measured by the CPI
- the nature of an earnings benchmark and the interrelationship between earnings and pension indexation
- the modelling the Review has undertaken of the sustainability of indexation arrangements.
4.3.1 International approaches
There is considerable diversity, and flux, in the way in which different countries index income support. A recent OECD study (OECD 2007) described the process in many countries as follows:
Pension uprating policy is a classic example of ad hoc policy-making. Even if most countries now have a formal link to prices, indexation is still often suspended as an emergency measure to relieve strong financial pressures on the pension system.
The same study reported that many countries had changed their approach to maintaining the level at which pensions are set, noting that ‘there is a clear underlying trend towards a reduced pension promise for today’s workers compared with past generations’. More specifically, the study reported that many countries had moved away from indexation based on earnings to price-based indexation, a strategy characterised as one that ‘preserves the purchasing power of pensions, but means that pensioners do not share in the general growth in living standards’. This is an approach that is also advocated in a recent OECD working paper (Whitehouse 2009).
These findings, although indicative of the problems many other OECD countries are experiencing in financing their retirement income systems, are generally not directly applicable to Australia.
A central reason for this is that in many countries indexation of the contributory pensions that form the main support for the aged has two components. The first is ‘valorisation’, which is how the individual’s pension is initially set relative to their own earnings or a measure such as average earnings. The second is the indexation of this pension over time.
Valorisation is not a part of the Australian pension system. In Australia, indexation sets the original value of the pension relative to a community benchmark and maintains that relative value over time. That is, while in Australia simply using price indexing would mean that the overall level of pensions relative to other members of the community would decline, in other countries this would not be the case because of the additional valorisation element.
4.3.2 The Australian Consumer Price Index
The CPI produced by the ABS is specifically designed to provide a general measure of price inflation for the household sector as a whole. This section considers issues associated with the use of the CPI as a measure of living cost change for those in receipt of pensions. These questions relate to the underlying concept of the index and its applicability for understanding the experience of price changes by pensioners.
Calculating the Consumer Price Index
The CPI comprises two components: a weighted pattern of consumption of goods and services—referred to as the ‘basket’—and price change data, which is then applied to the basket to derive a quarterly CPI.
- The weighting pattern for the basket of goods and services is largely based on the detailed information on household spending derived from the ABS Household Income and Expenditure Survey. Adjustments are made for the underreporting of some expenditure, especially alcohol and tobacco, and to housing, banking and insurance, to reflect the conceptual approach of the CPI. The basket contains 90 expenditure classes grouped into 33 subgroups and 11 major groups.
- Price change data are collected by the ABS from a wide range of retail and other sources, with over 100,000 separate price quotations collected each quarter. These price data, along with some ABS modelling of some components and the offsetting role of subsidies, are then aggregated into an estimate of price change within each of the 90 expenditure classes. Over the decade to June 2008, while the prices of some of these expenditure classes have increased very rapidly, for others they have fallen. At one extreme is the cost of automotive fuel, which has increased by 118.8 per cent. At the other, the price index for audio-visual equipment and computers has fallen by 81.3 per cent. The CPI effectively balances these different movements, based on the pattern of consumption of different items, to estimate the overall impact of prices across the community.
Analytical Living Cost Indexes versus the Consumer Price Index
In addition to the CPI, the ABS produces ‘Analytical Living Cost Indexes’ (ALCIs). These are described by the ABS as having been designed to answer the question: ‘By how much would after tax money incomes need to change to allow households to purchase the same quantity of consumer goods and services that they purchased in the base period?’
Such indices are produced for four household types: employee households, age pensioner households, other government transfer recipient households, and self-funded retiree households.
There are two major differences between the CPI and the ALCIs. The first difference is that while weights for both are based on the ABS Household Income and Expenditure Survey, as the overall survey sample size is not large enough to provide reliable estimates for the specific ALCI households at the capital city level, the expenditure weights are estimated at the national level only. Expenditure weights for the CPI (all private households) are estimated separately for each capital city. The second difference is that, in contrast to the ‘acquisitions’ approach of the CPI, the ALCIs use an ‘outlays’ approach. This affects the treatment of purchases of housing and financial and insurance services in the indexes.
The consumption patterns of the different types of households for which ALCIs are constructed vary considerably from those of Australian households overall, as illustrated in Chart 15. For example, age pensioner households are estimated to spend 21.1 per cent of their household budgets on food, compared with 15.4 per cent for households in the CPI. As well as these differences at the aggregate level, the more detailed composition of spending within the groups varies. For example, with the ‘food’ group, the age pensioner households spend a much higher proportion of their budget on tea, coffee and food drinks; meat, including beef, lamb and pork; and eggs. They spend much less on soft drinks and fast food.
Chart 15 Relative weighting of expenditure items in the age pensioner ALCI and the CPI

Notes: Expenditure weights calculated excluding financial services.
Source: Derived from ABS 2005, 2007a.
Although the ALCIs have only been available since June 1998, a longer-term estimate of the series for age pensioner households is available by linking an earlier experimental index for age pensioner households with the age pensioner ALCI. The results of this estimate are shown in Chart 16. Over the almost 28 years for which the two indexes are available, the CPI estimates that prices have risen by 244.4 per cent, while the index for age pensioner households indicates that the prices these households face have grown by 251.8 per cent. In annualised terms, these increases represent growth rates of 4.56 per cent and 4.64 per cent respectively. While noting that the two series are designed to measure different economic phenomena (that is, price inflation in the case of the CPI, and out-of-pocket living expenses in the case of the ALCIs), detailed analysis indicates that:
- On average, the rate of changes in the living costs of households mainly reliant on the Age Pension has been marginally higher than the rate of inflation recorded by the CPI. The cumulative impact of the difference over the almost 28-year period is, however, only 3 per cent
- At certain times, the rates of changes in the out-of-pocket living costs experienced by age pensioner households have moved faster or slower than the CPI. Over most of the 1980s, living costs for age pensioner households as measured by the age pensioner ALCI rose more slowly than for other households. This pattern was reversed in two stages between 1991 and 1994, and again in 1997 and 1998. There was a further, but somewhat smaller, relative increase in the age pensioner ALCI in 2006.
Chart 16 Comparison of age pensioner ALCI and CPI from 1980 to 2008

Sources: Derived from ABS 2000, 2008a,2009a
However, in comparing these two series from the perspective of pension indexation, particular attention needs to be given to the shorter-term movements in the two series. The reason for this is that when the CPI is used as an indexation factor for pensions, it essentially operates as a safety net to maintain the real value of the pension at times when wages growth is slow, or where there are particular spikes in prices that may impact on living standards.
Hence, while the two series may show similar patterns over time, they may behave quite differently in this short-term role. The degree to which this occurs is shown in Chart 17, which presents, for each six-month indexation period, the difference between the measurement of price change between the age pensioner ALCI and the CPI. The chart plots the percentage point difference in the change. A positive figure indicates that the ALCI showed higher price movements, and a negative that the CPI was higher. The chart indicates that:
- Over any six-month period, there can be considerable variation in the rate of price change recorded by the CPI and the age pensioner ALCI. The magnitude of these differences exceeds 1 per cent on a number of occasions.
- While there is no consistent pattern, on several occasions these discrepancies have been consistently in one direction over a number of indexation periods. This, in the absence of stronger earnings growth, would have a cumulative effect in terms of the implications for those on pensions.
Chart 17 Percentage point difference in estimated six-monthly cost increases as measured by age pensioner ALCI and CPI, 1981—2008

Note: Positive indicates high ALCI growth
Sources: Derived from ABS 2000, 2008a,2009a
The Consumer Price Index and changes in living costs
The CPI is designed as a measure of overall price inflation for the household sector. There are some differences when measuring changes in the out-of-pocket expenditure needed to gain access to consumer goods and services.
- While consumers tend to vary their pattern of spending depending on price movements (for example, purchasing more rice if potato prices increase), the CPI uses a fixed basket of goods that assumes no such behaviour. This is known as ‘substitution bias’ and leads to the CPI overestimating the cost of living.20
- The CPI as a measure of price inflation over households does not reflect the actual way price changes may impact on any particular household. As a consequence, price change can be experienced by individual households at rates quite differently to the rate predicted by the CPI.
- As a measure of overall price movements for the household sector, the CPI is weighted by total goods and services consumption across all households. This means that it is more heavily weighted by the spending patterns of higher income households. Reflecting this weighting pattern, the CPI is what is termed a ‘plutocratic index’.21 A contrasting conceptual approach is to weight the CPI on a household basis so that it represents the average experience of ‘equal’ households. This produces a ‘democratic index’. Analysis by the Review,22 using a restricted basket of goods, estimated that over the decade to June 2008, while the annualised rate of price change for a ‘CPI’ calculated on a plutocratic basis was 2.7 per cent over the decade price change measured by a democratic ‘CPI’ was 3.1 per cent. This suggests that the average experience of price change by households was higher than the average price change overall. Moreover, when the distribution of price change at the individual household level was considered, one-half of all households experienced price changes outside of the range of 2.5 per cent to 3.5 per cent.23
- As the objective of the CPI is to measure pure price change over time, identical goods and services should be priced from one period to the next (called ‘pricing to constant quality’). This can lead to some apparent discrepancies in the way price changes are perceived and the way in which they are measured.
- Over the past 20 years, the CPI index for the purchase of motor vehicles has increased by just 7.3 per cent. This in part reflects the extent to which cars today are better equipped. From a consumer perspective, where such quality upgrades are often taken as granted or involve changes that consumers may not necessarily have chosen to purchase, the estimate is seemingly at odds with the prices of any particular model of new car, which may have risen by some 25 to 50 per cent over the same period.
- The largest decline in prices at the expenditure class level of the CPI has been in the ‘audio-visual and computer equipment’ category, where prices have declined by 81.3 per cent. In some cases, this reflects declining nominal prices, especially as items such as DVD players have changed from being expensive luxury products to commonplace consumer items. In other cases it represents the marked improvement in the quality of products. For example, although the prices of computers may not have changed all that much in nominal terms, their power, storage capacity and other specifications have increased enormously. The pricing of computers to constant quality in the CPI reflects this improved computing capacity, rather than the computer as an item per se. Such treatment, while appropriate for the assessment of pure price change, is different to measuring the impact of these changes on the cost of living experienced by consumers, where quality changes may add little to the utility of the product for many consumers.
- A third area in which the pricing to constant quality approach to the CPI may differ from consumer experience concerns the cost of renting. While the CPI rental component increased by 22.9 per cent between December 1994 and December 2005, the average rent paid by private tenants, as surveyed in the ABS Survey of Income and Housing over the same period, increased by 41.2 per cent (ABS 2005–06). Again, much of this difference can be related to improvements in the quality of the rental stock.
- Similarly, the conceptual treatment of some other items in the CPI does not reflect common conceptions of consumer price change. For example, the land value of housing is excluded, as this is seen as investment rather than consumption. There are similar exclusions of the price of established housing, as this is seen as a transaction between households with no net impact across the household sector. Banking services reflect the cost of fees and the margin between deposit and lending rates, and are hence insensitive to changes in interest rates, and insurance is similarly the difference between premiums and claims.
While some of these issues do not apply in the case of the ALCIs (for example, a different approach to home purchase and financial services is taken in the ALCIs), and more homogenous populations with a lesser plutocratic bias are used, issues such as those concerned with quality adjustment remain in the ALCIs.
In making these comments the Review emphasises that these are not issues with the way in which the ABS operationalises the CPI. The Review recognises that the ABS approach is consistent with international best practice and for the purposes of measurement of generalised price inflation across the household sector. Rather, the Review questions whether a measure of generalised price inflation is the best way of indexing pensions.
It is also emphasised that many of these distortions are not necessarily consistently in one direction or another.
This is particularly true of the distinction between democratic and plutocratic estimates of price change. International data show that the direction of difference varies between countries and between periods—largely as a result of the specific factors that are driving price change at that particular point in time.
With regard to the other biases, substitution bias always tends to lead to an overestimate of price change, while the discrepancy between the constant quality approach and the actual capacity to purchase products is likely to be in the other direction.
The cost of living in contemporary society
In addition to the cost of purchasing a set basket of goods and services, considering the ‘cost of living’ raises a broader question about the capacity of people to participate in society. This is particularly important with regard to changing living standards and changing social and technological structures in society.
For example, a decade ago, a computer with internet connections was useful for most households but not necessary. This is increasingly changing, in part because many government and commercial transactions that were previously undertaken through face-to-face contact, or even call centres, are being streamed into electronic formats. Similar transformations have occurred in many other fields including, for example, community organisations moving from paper-based to electronic newsletters. With this type of change it can be considered that simply ‘maintaining’ a living standard now increasingly requires individuals to purchase a computer and internet connection.
This type of shift is not easily measured through conventional approaches to the cost of living, which simply treat households acquiring these new capabilities as improving their living standard. There are similar issues in addressing changes in charging regimes by government and other bodies and in the provision of services. For example, an increased supply of services may have a considerable impact on the wellbeing of householders, especially those with a disability or other high service need, while a contraction of services will have the opposite impact. Neither of these will be identified in a price-based measure of living standards.
These types of changes are one rationale for benchmarking pensions to community living standards. However, they are discussed here to emphasise that price changes and changes in living costs as experienced by a specific group may not coincide with price changes as measured by mechanisms like the CPI.
Clearly the adoption of an ALCI to index the pension rather than the CPI may result in a lesser or greater price indexation factor at any given point in time. Either way, the effect is likely to be temporary due to the operation of the wages benchmark, which is the dominant factor in determining pension increases. This is consistent with price adjustment operating as a safety net to protect purchasing power at a time of rising costs. The essential question is what the better measure of the cost change is for pensioners over these periods.
Perceptions of price changes
Another insight into the relationship between price changes and perceptions of changes in living standards has been given in recent analysis by Bradbury (2009a). In ‘Regular inflation in Australia’, he suggests that the way in which price change has unfolded has resulted in consumers seeing the price of more regularly purchased products increase much more quickly than more discretionary and less regularly purchased items. He estimated that between June 2005 and March 2008, the price of ‘regular goods’ increased by 4.3 per cent compared with a rise of just 0.5 per cent in the price of ‘irregular goods’.
It is likely that this effect may also be compounded by normal patterns of risk aversion which mean that people are more likely to be aware of losses—such as those associated with rising prices—than the gains they make, for example when prices drop. In addition, the longer time periods between purchases of irregular goods may mean that people are less likely to recall the previous price they paid, and the duration may mean that there is less of a tendency to realise the real change in the price of the product.
Analysis such as this highlights the way in which divergences between perceptions and actual trends can occur. While clearly this type of experience is not a basis for a policy change, it points to the importance of public confidence in pension setting arrangements.
4.3.3 Benchmarks—measures of change in living standards
The goal of benchmarking is to maintain the relativity of the pension to a measure of living standards or income of another group. Currently the pension is benchmarked to MTAWE.
Impact on pension rates
Since the introduction of legislative provisions for benchmarking which became operative in September 1997, while the CPI alone would have seen the pension increase by 37.1 per cent to take account of rising prices, the base pension has increased by 56.1 per cent. That is, the single rate of the pension has risen by $1,978.60 a year (including the Pension GST Supplement)— $38 a week—more than if it had been subject to CPI indexation alone. Over the last 23 pension indexation points, pension increases have been driven by wages growth as measured by the MTAWE benchmark on 15 occasions and by CPI increases on eight occasions.
Over the longer term, the rate of the pension has not only reflected the impact of indexation, but also the impact of policy changes, such as the inclusion of compensation for the introduction of the GST, the introduction of additional allowances such as the Utilities Allowance, and ad hoc initiatives such as the payment of bonuses. The real value of the pension over time is shown in Chart 18.
Chart 18 Real value of Age Pension, 1970–2009

Notes: Calculations are based on Age Pension at 1 January each year, and December quarter CPI, including Seniors Bonus when paid.
A difficult question faced by the Review in considering the adequacy of the indexation mechanisms for pensions is the relationship between these ad hoc adjustments and the effectiveness of indexation. That is, to what extent were some of these additional payments and changes a response to the recognition that the indexation of pensions was inappropriate to maintain the relative wellbeing of pensioners?
There is no clear answer to this question. However, one interpretation would be that, to the extent that a number of these payments have become embedded in the structure of the pension and into the expectations of pensioners, the historical indexation approach has not been wholly adequate.
Taken together, these various changes in the rate of the pension have not only resulted in an increase in its real value, but also (as noted in Chapter 3) an improvement, in most cases, in its value relative to the take-home earnings of working Australians.
Approaches to benchmarking
The base rate of the pension is benchmarked to 25 per cent of MTAWE, however, the value of the total pension package is higher than the benchmark. In September 2008, when the pension rate was last indexed against the May 2008 MTAWE of $1,069.00, the combined pension and allowances for a single age pensioner was 27.7 per cent of MTAWE rising to 28.6 per cent when the weekly equivalent value of the Seniors Bonus is included.
The Review notes, though, that MTAWE is only one of the measures of community living standards available in Australia.
A range of different reference series could be considered as being appropriate for benchmarking. These include household incomes and various earnings measures. Within these approaches, the use of earnings benchmarks introduces an additional level of conceptualisation, namely that of payments being set as a ‘replacement rate’ for what people may have been able to earn if they were able to participate in the labour market or had not retired. Obviously, such an approach is a broad conceptualisation relating to some average outcome not at the individual level.
Household income
One way to achieve a direct link to the living standards of the community would be to use a measure of household income. Two different types of measure are available for this. The first is ‘household disposable income’ from household surveys. This reports the incomes received by households from earnings, investments and transfers, less income tax. The second is National Accounts–based measures.
Survey-based equivalised household disposable income
Between 1994–95 and 2005–06, median equivalised household disposable (after-tax) income grew by 75.6 per cent, in real terms a 31.9 per cent increase. This is considerably higher than the growth in the rate of the pension. The growth in household income reflects higher earnings, a slight decline in taxation and considerable shifts in the structure of households. During this period, the average number of employed people per household increased from 1.22 to 1.26 and the average number of people in households declined from 1.96 to 1.88 adults and from 0.73 to 0.63 dependant children (MIAESR 2008a). In essence, much of the apparent growth in this measure of income is a consequence of these households working more and having fewer people to support.
Although there is some intuitive appeal to benchmarking pensions to a measure of household income as being representative of the community as a whole, the nature of the various drivers of household income, as seen above, casts doubt on this option. In particular is the extent to which changes in household income have been affected by changes in the composition of households and the nature of labour force participation. This type of change makes the estimates of income sensitive to a wide range of social changes, such as decisions by children as to when to leave the parental home. It is also hard to see a rationale for such changes to impact on the rate of the pension.
National Accounts measures of income
National Accounts–based measures of household income take a broader approach to the concept of household income. In addition to aggregating the total value of wages and salaries of employees, it includes the incomes of unincorporated enterprises, including an estimate of imputed rent for owner-occupied housing. Offset against these are the interest payments on dwellings and consumer debt, insurance premiums, fines and transfers to non-residents. As such, it includes many items that are not immediately available to individuals as consumption such as superannuation savings, and, as with the CPI, various items are netted out across the household sector as a whole. While this is a very comprehensive concept, it is not clear whether it provides a useful reference point for the adjustment of cash incomes of pensioners, given the way in which it is constructed, and the degree to which it does not focus on the day-to-day capacity of households to consume.
The approach has two other limitations. The first is that as a per capita measure it does not take account of the way in which household income needs vary on the basis of household composition and age. For example, a decline in fertility with fewer children being born in one year would show up as an increase in per capita income. Secondly, as with other National Accounts measures, the estimates published by the ABS are subject to considerable revision over time.
Male Total Average Weekly Earnings
Over the 20 years to August 2008, earnings (seasonally adjusted) as measured by MTAWE have increased by 122.6 per cent. The real value of MTAWE, taking into account price change as measured by the CPI, has increased by 20.6 per cent. Over the past 10 years, the nominal value has grown by 50.8 per cent and the real value has increased by 9.9 per cent. As indicated by its title, ‘Male Total Average Weekly Earnings’, it is an estimate of the average total (ordinary time and overtime) earnings of all male employees, including full-time and part-time workers, and both adults and juniors. As an average of currently paid wages, it is affected by changes in the composition of the workforce, hours worked and changes in the structure of employment, including occupation. The series also does not take account of the value of amounts salary sacrificed by employees, although this will be available in a new experimental series in 2009.
While use of MTAWE as a benchmark has delivered significant real increases in the value of the pension, it also has some limitations:
- MTAWE has grown relatively slowly compared with total adult male full-time earnings (which includes overtime) and male adult full-time ordinary time earnings. Over a longer term, the value of earnings as measured by MTAWE has fallen from being 92.8 per cent of full-time male total earnings in 1972 to 84.0 per cent in August 2008.
In large part, this is a consequence of the increase in the proportion of men working part-time, including the extent to which a considerable proportion of young people combine some part-time employment with their studies. The proportion of all employed men who are working part-time increased, between January 1989 and January 2009, from 7.4 per cent to 15.1 per cent. - The use of male earnings alone in the measure is anachronistic in that it ignores the participation of Australian women in the workforce.
- While the concept of ‘average weekly earnings’ is often conflated with the idea of the earnings of an average employee, this is not the case because most employed people earn less than average earnings. Using data from the 2005–06 Survey of Income and Housing, it is estimated that 56 per cent of full-time male employees and 75 per cent of full-time female employees earn less than the average of full-time earnings estimated from the series.
Earnings growth also varies across the income distribution. Data from the ABS Survey of Employee Earnings and Hours shows that between 1996 and 2006, average full-time adult earnings grew on an annualised basis by 1.7 per cent. Looking at the rate of growth at points in the income distribution, earnings grew at the median (that is, the ‘middle’ employee) by 1.4 per cent, by 1.1 per cent for a lower-paid worker (25th percentile) and 0.9 per cent for a low-paid worker (10th percentile). - The measure is affected by changes in the workforce including enhanced skills and does not reflect what is happening to the earnings of individual employees. In comparison, data from the Wage Price Index, which is adjusted to eliminate the impact of such changes, shows lower growth. Over the decade from June 1998, average real full-time ordinary time earnings increased by 14.2 per cent but the real wage rate for a ‘constant quality position’ only increased by 5.0 per cent.
The comparative movements in these series over the past decade are illustrated in Chart 19.
Chart 19 Comparative trends in earnings, 1997—2008

Notes: Index numbers, June 1998=100, real values derived using CPI. MFTAOTE = Male Full-Time Adult Ordinary Time Average Weekly Earnings; MTAWE = Male Total Average Weekly Earnings; PFTAOTE = Persons Full-Time Adult Ordinary Time Average Weekly Earnings; EEH = employee earnings and hours.
Sources: Derived from ABS 2007b, 2008b, 2008c
4.3.4 Indexation and sustainability
Decisions on the indexation of pensions have a range of consequences. In the case of automatic changes, this occurs without any consideration of the state of the budget or current economic conditions. It also impacts on the pattern of income distribution across the community.
- Increasing pensions at a rate less than that of overall growth in community incomes can result in increased income dispersion and rising inequality, as the gap increases between the incomes of pensioners and those of others in the community who gain their income from employment increasing.
- Conversely, increasing pensions faster than the net income growth, say, of working households, will close the gap between these households. This in turn may raise questions about the rewards for work as the incomes available to those on income support grow relative to earnings, and the relative net gains from employment decrease. As noted in Chapters 2 and 7, pensioners receiving a part rate of pension can already have a higher disposable income than low-income taxpayers who are funding pension spending through their taxes.
- The extent to which indexation arrangements seek to maintain a link with increasing living standards, as opposed to simply maintaining purchasing power, also directly affects budget sustainability. More generous indexation arrangements have a higher fiscal cost over time because their effect on payment rates compounds with the aging of the population. Less generous indexation arrangements have less of an effect.
- A further issue, highlighted by the current disruption to global markets, is that guaranteed increases to pensions, which are a means of improving the financial security of pensioners, are in contrast to the level of security of those who rely on private investments or earnings. A consequence of this is that indexing of pensions during periods of economic downturn may see the real value of pensions maintained while earnings and returns on investment for others in the community decline.
Relationship between the pension, wages and taxes
In addition to its impact on the rate of pension payment, the most immediate impact of pension indexation is on the government’s budget.
The payment of pensions and other transfers relies on taxation revenue, especially income tax. A consequence of this is a relationship between the tax rate and the rate of payment. More specifically, if as a result of demographic or other change, the size of the population eligible for income support increases relative to the size of the population with employment (who are taxed to provide this income support), then maintenance of the rate of pension payment at some relativity to earnings can only be achieved by increasing the rate of tax required to fund these benefits. To the extent this is funded by taxes paid by those with employment, this in turn results in a relative increase in the net income of those on transfers compared to the net income of those with employment.
Chart 20 illustrates a simple hypothetical model that shows this effect. It comprises two population groups, employed people and pensioners, and assumes that the only government expenditure is on pensions and the only tax is income tax on employees. What is modelled is the effective tax rate on employed people to pay for a pension that is set at a ratio of gross earnings and the relationship between the net incomes of these two groups.
The model shows that as the pensioner share of the population changes, the tax rate needs to increase to pay for the larger number of pensions, but that the value of the pension increases relative to the take-home pay of the working population who are having these taxes deducted from their earnings. In the illustrated scenario, a pension set at 25 per cent of gross earnings would increase in value from 26.1 per cent of the disposable income of an employed person to 33.3 per cent as the ratio of pensioners to workers increases from 1:6 to 1:1.
Obviously, the relationship is much more complex in contemporary Australia than illustrated in the model. The convergence between the pension and net wages may be less obvious if governments choose other approaches to finance Age Pension pressures, including the use of indirect tax increases, spending cuts, and policies that increase aggregate participation such as immigration. There may be some offsetting savings from the shift in the share of the population, such as lower demand for education, as well as the impact of the Superannuation Guarantee in reducing pension reliance. However, the broad relationship remains, and in the long run it may prove that a pension benchmarked directly to a net rather than gross wage may be more sustainable as it is likely to be a better measure of community living standards.
Chart 20 Model of relationship between earnings, value of pension and population structure

Source: Pension Review modelling.
As discussed in Chapter 2, the issue has direct relevance to the indexation of pensions in Australia. The Intergenerational Report 2007 indicates that the ratio of people of working age to people over the age of 65 years will decrease from 5.0 in 2007 to 2.4 in 2047 (Treasury 2007). Again, in a simple two-sector model taking into account current and projected participation and pension take-up rates, this would involve an increase in the tax required to fund a 25 per cent gross income pension from 5.5 per cent to 10.4 per cent; and on the basis of this being the only tax paid, the relative value of the pension would increase from 26.4 per cent to 27.9 per cent of the disposable income of employed people. However, this effect would, in addition to the factors discussed above, be in part offset by the increasing proportion of the aged who will have superannuation and other savings.
While the linking of pensions to earnings is often argued as being a mechanism to share the benefits of productivity gain across the community, the indexation of pensions is not the only claim on this gain:
- Employed people who have generated this increase through their efforts including investing time in education see themselves as being entitled to much of this gain. This is particularly so when the higher productivity results from their own efforts and investments.
- The return on capital to companies and investors is important, not just to ensure adequate flows of capital for the future, but increasingly to the funding of retirement incomes of those with superannuation.
- As indicated in the Intergenerational Reports, these productivity gains are also critical to fund increasing levels of health care and other services needed across the community, in part because of the ageing population, as well as the costs of health services as a result of technological and pharmacological development.
A key question is how these competing claims can be balanced, taking into account issues of both equity and sustainability, while ensuring sufficient incentives remain to achieve future productivity growth.
4.4 Reform directions
The indexing of pensions is critical to maintaining the standard of living of pensioners. This section outlines a series of principles that the Review considers should guide future policy decisions on indexation. Building on these, and the technical analysis provided in this chapter, the Review has developed findings in relation to price adjustment and benchmarking to community living standards.
Before discussing the principles, the Review considers that it is important to address two specific proposals for indexation approaches that were made in the consultations.
4.4.1 Consideration of specific indexation proposals
There was widespread support in the consultations for two alternative approaches to indexation. These were to link the pension to the Federal Minimum Wage and to utilise an existing or new budget standard. Neither of these is supported.
Linking the pension to the Federal Minimum Wage has a strong intuitive appeal in the context of considering the pension from the perspective of a replacement rate for what a person may have earned. It has, however, the potential to significantly distort the processes and considerations of the setting of the Federal Minimum Wage. In particular, the 3.3 million pensioners who would see their pensions affected by the determination of the Federal Minimum Wage would well outweigh the 1.3 million Australians who rely directly on the Australian Fair Pay Commission’s decision. Further, it is noted that several of the criteria used by the commission for determining the Federal Minimum Wage as a ‘safety net for the low paid’ and the impact of decisions on ‘employment and competitiveness across the economy’ and the capacity of the ‘unemployed and low paid to obtain and remain in employment’ may not be considered appropriate for pension setting (Australian Fair Pay Commission 2008).
The use of a budget standard as the basis for indexation has four significant problems. The first concerns the question of the representativeness of the standard for the diversity of living circumstances of Australian pensioners. It would in essence result in indexation being based on a single normative set of purchases. As well as being unrepresentative, it would be considerably more volatile than a broader-based measure. Secondly, regular repricing of the budget would involve either extensive data collection in order to price the many hundreds of individual products, in different locations, or the use of broader indexes such as the CPI class-level indexes, which are not product specific. Thirdly, as a fixed basket of goods, the measure would totally ignore product substitution, as well as be insensitive to new products. Fourthly, is the question of how the budget would be readjusted over time for changes in living standards.
The Review notes that these types of measure can have a role, along with other indicators, in social policy analysis, but relative cost effectiveness would rule them out.
4.4.2 Indexation principles
The Review proposes that the following principles be used to guide future policy development for pension indexation.
Automatic indexation
Changing pension rates can be undertaken automatically through the inclusion of indexation and benchmarking criteria in legislation, or through regular review, typically in the budget process. Such reviews can be guided by particular benchmarking objectives such as the MTAWE benchmark, or be more ad hoc.
Each of these approaches has merit. Automatic legislative indexation provides for a greater degree of certainty and security. At the same time, because legislative provisions can be changed, they are not a guarantee of a particular outcome over time.
A downside of legislated automatic change is that it can result in policy being set on ‘autopilot’ without serious consideration of questions of adequacy and affordability. In such circumstances there is a risk that little attention will be paid to actual outcomes with regard to these issues, as it is assumed that the indexation process will ensure appropriate outcomes.
Where changes are not done automatically, governments must consider the question of the appropriate rate of payment on a regular basis. In some areas of government policy—such as the determination of taxation rates and thresholds—indexation is not done automatically. This approach provides a greater degree of flexibility to governments to take decisions on when and to what degree adjustments should be made to program settings, thereby allowing decisions to take into account the range of economic circumstances at the time, including the affordability of changes in rates.
There are strong arguments for decisions on indexation of pensions to be taken as deliberative decisions by government. However, taking account of the current outcomes for pensioners, trends in living standards across the community and the ability to fund any changes, the Review considers that automatic indexation is a better approach. The major advantage of this approach is that it provides greater financial security for pensioners, the focus of the terms of reference for the Review.
This neither removes the need for indexation to be regularly reviewed, nor precludes the capacity of government to intervene if circumstances require this. Automatic mechanisms should also not be considered as sufficient to ensure that payments and payment structures are meeting their goals or responding to changes in the broader economic and social environment. Consequently, a need remains for regular reviews of the rates and structure of pensions.
An alternative view to this is that automatic indexation limits government flexibility, particularly in the light of changing economic circumstances. In cases where other groups in the community may be facing constraints on their living standards, this approach would tend to fully shelter pensioners, and where governments may face cost pressures, it tends to ‘lock in’ expenditure.
While the Review recognised the force of these arguments, it does not consider that they provide sufficient grounds to overrule the broader principle. In particular, where these constraints are significant and questions of national priorities demand it, governments would remain free to make decisions through suspending or otherwise modifying these policies and any related legislation.
Continuation of the two-part approach
Given the goals of reflecting improvements in the quality of life of the society and in protecting pensions from falls in purchasing power, the Review considers a two-part approach should continue to be used—that is, adjustment of the pension both against a community benchmark and to account for price change.
This dual approach has advantages in that each component acts as a stabiliser to the weaknesses of the other. Benchmarking can provide an effective link to changes in living standards, something a price index cannot do, while the use of a price index protects pensioners in the case of any surge in costs that might otherwise act to reduce their living standards. The dual mechanism also reduces the risk associated with the compounding of any small discrepancies in the price index over time.
Under this arrangement, the benchmarking of the pension would be the principal mechanism, and the price mechanism would be limited to safety net adjustments to maintain the purchasing power of the pension at times when it would otherwise be eroded.
Finding 8: The Review finds that automatic indexation of pensions and a two-part approach of benchmarking and indexation should continue. Benchmarking pensions relative to community standards should be the primary indexation factor, with indexation for changes in prices acting as a safety net over periods where price change would otherwise reduce the real value of the pension.
Relativities of components in the pension system
Because different approaches to indexation have been taken for different elements of the pension system, along with various ad hoc changes, the relativities between components have changed over time.
| September 1997 | September 2000 | September 2008 | |
|---|---|---|---|
| Newstart Payment Rate | % relative to base payment single pension | ||
| Over 21 no children | 92.4 | 89.0 | 79.9 |
| With children or over 60 years | 100.0 | 96.2 | 86.5 |
| Rent Assistance | % relative to base payment single pension | ||
| Single person | 21.5 | 21.9 | 19.6 |
| Minimum rent | 20.6 | 19.0 | 17.4 |
| Income test | % relative to base payment single pension | ||
| Free area for single pensioner | 28.8 | 26.9 | 24.9 |
| Cut-out single pensioner | 231.9 | 280.4 | 281.5 |
| Free area single pensioner with child | 35.7 | 33.1 | 28.9 |
| Free area single allowee | 17.3 | 15.7 | 11.0 |
| Assets test | Multiple of fortnightly rate of pension | ||
| Threshold (single home owner) | 362 | 338 | 306 |
| Cut-out (single home owner) | 700 | 676 | 979 |
Source: FaHCSIA modelling.
Note: The relativities are with the single rate of pension not including the value of Pharmaceutical Allowance, Utilities Allowance or Telephone Allowance.
Table 3 shows how these factors have impacted on a number of the relativities between components of the income support system and introduced imbalances in the structure of payments and other program parameters. For example, the Rent Assistance threshold was 20.6 per cent of the single pension rate in September 1997. It is now 17.4 per cent.
In addition, because the threshold at which means testing cuts in grow less quickly than wealth and living standards, the use of the CPI for the means test thresholds tends to increase the proportion of the pensioner population receiving a part rate of pension. In fact, some of these parameter changes represent a form of ‘fiscal drag’ similar to the effect in the taxation system, where the lack of automatic indexation means that average tax rates will increase unless decisions are made to change the tax thresholds and rates. Of course, in the pension system in most cases this issue is generated by the two different components of indexation, rather than the lack of indexation. Therefore, fiscal drag will be less aggressive in the pension system than in the taxation system.
There are three main arguments for continuing with the current approach:
- The lack of automatic indexation of taxation rates provides a greater degree of flexibility to governments to take decisions and therefore allows decisions to take into account the range of economic circumstances of the time. The same is true for the indexation of many pension components to the CPI, though the magnitude of the effect is much less.
- The approach of not indexing all payments and parameters by a community benchmark such as wages has been used as a means of reducing the effect of policies or program parameters that are likely to become less relevant over time.
- While CPI indexation of thresholds has led to changes in pension relativities, CPI indexation is a way of gradually increasing the aggressiveness of the means tests for individuals who have real increases in their level of income and wealth.
The Review considers that indexation approaches that result in distortions between components of programs should be avoided. This is also the case with the use of fiscal drag, other than as part of an explicit policy adjustment process, such as being a part of a formal grandfathering mechanism.
One approach to avoiding the creation of distortion is to index all program parameters, including rates and thresholds, on a consistent basis. In pure policy terms this is considered the best solution as it maintains the deliberative intent of all policy settings, while permitting changes by formal policy decisions. In some cases this approach may be seen as generating an inappropriately high level of ‘lock in’ to parameters that may have been set with more than one policy objective in mind. The fiscal cost of indexing all program parameters on a consistent basis (such as wages growth) is difficult to determine as it would operate differently on different components, and would need to be judged against what might otherwise be done through ad hoc and other adjustments. Locking in means test parameters would tend to increase costs, while moving payment thresholds up (for example, for Rent Assistance), will increase targeting and potentially reduce costs.
An alternative to a unified approach to indexation would be to maintain different indexation approaches but establish a formal process to review relativities on a regular basis. Because the creation of distortions tends to be a gradual process, this would enable any emerging imbalances to be examined before they have a deleterious effect on the program structure. This is the approach the Review has adopted.
The Review considers that its findings with regard to specific settings of the pension system, including the current treatment of supplementary payments, changes to Rent Assistance, the treatment of earned income and other changes to the means test, provide the potential to fix most of the distortions currently within the pension system. These issues are considered in Chapters 5 and 7.
4.4.3 Price adjustment
The Review considers that price adjustment is a ‘safety net’ and anticipates that the wages benchmark, in whatever form, will generally drive pension rates and will almost certainly be the dominant indexation factor over the medium and long term.
However, the Review notes that the times when price adjustments are the dominant indexation factor are those when pensioners are at most risk of experiencing a fall in their living standards. Hence, a price measure that is responsive to their costs in the short term is important. As seen in the Review’s analysis, while the CPI moves similarly to the age pensioner ALCI in the long term, it can differ markedly over shorter periods.
Therefore, the Review considers that a specific price index, along the lines of the ALCI, should be developed for income support recipients.
- The index should be based on the expenditures of households that are mainly reliant on government transfer payments. Restricting the measure to these lower-income households limits the distortionary impact of the expenditure pattern of higher-income households from the measurement of price change.
- The measure should treat insurance and consumer debt on an acquisitions basis and exclude rent and home purchase. The ABS should also undertake and publish further analysis of the issues associated with the purchase of items that are significantly affected by quality adjustments.
- Given that this index will specifically consider the impact of price change on low-income households, it should be the only price index used for the adjustment of pensions.
However, even a pensioner-specific index will not eliminate the inherent limitations of any index measuring the actual price impacts on all households within a specific group.
As the Review’s preferred approach involves significant changes in the Australian approach to price indices involving the construction of a second purposive price index, and raises several conceptual issues, there may be merit in considering these issues in a way that can draw on both expert input and the views of the community, as well as the interests of key users of current price data, to provide for a more thorough and public analysis of the issues.
Finding 9: The Review finds that pension indexation for price change would be better undertaken through an index that more specifically reflected cost of living changes for pensioners and other income support recipient households.
Indexation of rent assistance
Rent Assistance plays a very specific role in the income support system since it is conditionally linked to a particular expenditure incurred by a minority of pensioners. As discussed in Sections 3.4.6 and 4.3.2, the evidence indicates that the pattern of changes in the cost of private rental housing is not well reflected in broader measures of price change. This suggests that the basis for using such measures for the indexation of Rent Assistance is not strong. However, to the extent the payment is associated with actual expenditure on rent, there similarly is no clear reason to link the rate of assistance to a more general community benchmark.
For these reasons the use of a separate index that more appropriately reflects changes in the costs of private rent would have merit. Such an index could, for example, be based on the actual rents paid by income support recipients to obtain a good measure of the actual changes experienced in the particular segment of the market in which this group operates.
However, it was also noted in Section 4.4.2, and is further discussed in Chapter 5, that the structure of this payment has been significantly distorted over time as a result of differential indexation that apply across the range of social security parameters. In line with the Review’s general approach to fiscal drag, which involves reviewing and adjusting review relativities on a regular, deliberative basis, the Review has developed findings on the targeting of Rent Assistance. The Review’s findings on Rent Assistance are discussed in Chapter 5.
4.4.4 Benchmarking
Benchmarking requires the linking of the pension to a measure that reflects changes in the wellbeing of the community. This has traditionally been done through the benchmarking of the pension to MTAWE.
The Review considers that the key objectives of the benchmark should be to reflect:
- a relativity to a consistent work effort of an employed person, which is not affected by changes in the proportion of full-time and part-time employees
- the earnings of the ‘average’ worker rather than the average earnings of all workers (which is disproportionately affected by income change among high-income earners), as well as the earnings of both men and women
- the disposable after-tax income of an employed person. This is a more appropriate reference point, especially from the perspective of maintaining a relative replacement rate between the disposable incomes of those on pensions and the disposable incomes of those in employment. It also links pensions to the experience of the wider community and the impact of changes in income taxation over time.
On this basis, the continuation of benchmarking to MTAWE is inappropriate in the long term. It is proposed that the measure used for this purpose should be the net income of a single person earning a wage paid at the rate of the median earnings of a full-time employee.
Finding 10: The Review finds that no single measure to benchmark the pension to community living standards is without limitations. However, the Review considers that a measure of the net income of an employee on median full-time earnings may be a more appropriate measure than the existing Male Total Average Weekly Earnings benchmark.
Implementation
In coming to these findings, the Review recognises that there are two factors that may limit immediate implementation. The first is that several of the reforms would involve some work to ensure that appropriate and timely measures of price change and earnings are available. The second concerns the degree to which benchmarking may be affected by other reforms to the tax-transfer system.
As the goal of benchmarking is to create a link between community living standards and the rate of the pension, the issue of benchmarking is closely associated with a range of other policies that impact on wellbeing. Hence, the final determination of the level of the benchmarking of the pension may be more appropriately delayed until the Australia’s Future Tax System Review Panel has made its report. This reflects the extent to which:
- considerations and recommendations of the Australia’s Future Tax System Review Panel may have a direct impact on the level and distribution of resources across the community and between sectors
- more specific consequences of policy reforms, such as any changes to the personal income tax system, may have direct impacts on a benchmark that reflects net earnings.
The Review considers that such a delay in the implementation of any reforms to indexation does not pose any major concerns around the adequacy of pensions in the short term, as long as existing indexation approaches were maintained, as reform in this area is primarily concerned with long-term adequacy and sustainability.
Finding 11: The Review finds that, while reform to indexation and benchmarking is important to the financial security of pensioners, implementation may need to be phased in to account for policy developments that may arise out of the Australia’s Future Tax System Review and to allow for the development of appropriate mechanisms for benchmarking and indexation.
Implications of revised benchmark
The proposed approach of using a net median full-time employee benchmark builds two linkages into the rate of the pension:
- It would pass on to pensioners the full rate of any productivity gains that are reflected in the earnings of employees, less the degree to which some of this is received by government through the tax system.
- It thus links the pension system into any changes in the income tax system and the income tax experience of the median worker.
Both of these features have positives and negatives:
- The linking of the pension rate to productivity growth as reflected in full-time earnings would ensure that the pension remains closely linked to changes in living standards. While indexing to a benchmark that is likely to grow less strongly can be argued for, over the longer term it can only result in a widening gap between the wellbeing of those on pensions and others in the community, and therefore pressures for ad hoc increases. As discussed earlier, although a number of overseas countries have moved away from full wage indexation, this is usually associated with a dual valorisation–indexation approach that still maintains an earnings link for the initial setting of the pension rate.
- Linking the pension to the tax system is considered by the Review to be important for long-term sustainability in the context of an ageing population. It means that if tax increases are required to fund higher support for an older population, then pensions will increase more slowly than they would if they were indexed to gross earnings. That is, if current workers have growth in their take-home pay reduced by a higher tax burden, this is reflected in a slower pension growth. This is important for sustainability, is more transparent, and sets the pension against the actual standard of living experienced by the community. An implication of this approach is that decisions on taxation structures and levels would need to consider the impact on pensions. In some cases the nature, or purpose, of the taxation change may appropriately require an adjustment to the benchmark.
In considering benchmarking, the Review was primarily concerned with developing an approach that is conceptually robust and fair, rather than with the specific outcome it would generate relative to other measures. Nevertheless, the question of how this approach would compare with ongoing MTAWE indexation is a legitimate one. In large part the answer depends on three factors: the proportion of men working part-time, changes in the earnings distribution, and future income tax rates.
- If the proportion of men employed part-time increased relative to full-time male employment, the benchmark based on full-time earnings would increase more rapidly than MTAWE because part-time employment brings down the average earnings in that series.
- If earnings inequality increased, in particular if driven by a continuation of the more rapid growth in high incomes seen over the past decade, then the median measure, all other things being equal, would grow less quickly than MTAWE since it is driven by changed earnings in the middle and not the extremes.
- If the income tax burden on ‘middle earners’ increased, then the net median measure, again all other things being equal, would not increase as quickly as a gross measure such as MTAWE.
Unfortunately, only a relatively short time series of net median earnings of all full-time employees is available.24 (A longer time series is available for the subset of ‘non-managerial workers’; this was used in Chart 4.) Over the shorter period between 2002 and 2008, as illustrated in Chart 21, the net median measure and MTAWE tracked quite closely until the end of the period, when the effect of the tax cuts introduced in July 2008 kicked in.
Chart 21 Relative growth in net median full-time earnings and MTAWE, 2002–2008

Notes: Index numbers, July 2002=100. Net median calculated as at July each year using interpolated values of earnings in years for which data are not available from the ABS Survey of Employee Earnings and Hours, and projected using FTAOTE for 2007 and 2008. MTAWE original for August quarter.
Source: Review modelling using ABS 2007b, 2008b.
The Review recognises that these historical changes are not necessarily indicative of any future change.
A further matter of implementation arises because of the more limited availability of data on median earnings. Unlike MTAWE, which is available quarterly, median earnings data are currently available only every two years, but could be available annually. Unless the survey of earnings were undertaken on a rolling basis—with a ‘moving average’ earnings estimate—this would suggest that six-monthly benchmarking would not be possible. To address this, a reasonable approach may be to benchmark annually and to utilise a cost adjustment every six months to maintain the purchasing power of the pension. In a relatively low inflation environment this would not appear to have any material impact on adequacy. It would still place the security of the purchasing power of the pension on a much more robust basis than earnings, where most employees have, at best, an annual adjustment.
- The Boskin Commission reported an estimate of this bias at 0.15 per cent a year for the United States CPI (Advisory Commission to study the Consumer Price Index 1996). The ABS is unable to quantify the possible impact on the Australian CPI, but employs a number of strategies to minimise item substitution. Analysis by the Review estimated it at around 0.22 per cent a year.
- This terminology reflects the weighting approach adopted by the difference indexes. The democratic index weights each household’s expenditure equally, while the plutocratic index weights it by the level of expenditure of the household. The use of a plutocratic approach is necessary in a price index that seeks to identify price changes across the household sector as a whole, as this would not be identified as the sum of democratically weighted households.
- The Review’s analysis of price changes was undertaken in order to investigate the extent to which the CPI was affected by various methodological approaches, and the results are merely illustrative. Because the analysis did not use the full basket of goods, and there are limitations imposed on the distributional analysis because household expenditure patterns are relatively short-term snapshots, these estimates should not be considered to be alternative estimates of actual price changes over the period.
- The analysis may overestimate the actual extent of dispersion of price change experiences due to the limited time period for which expenditure data are available at the individual household level and hence its sensitivity to abnormal as opposed to average expenditures.
- As the data are only available up to 2006, later values have been estimated using trends in full-time earnings from Average Weekly Earnings.
[ top ]
