A recent paper by the Australian Institute of Superannuation Trustees, using Melbourne Institute’s Household, Income and Labour Dynamics in Australia (HILDA) survey, concluded that retirees’ spending increases as they age. The implication is clear: if retirees spend more as they age, they need to save more before retirement to fund that higher spending.
But there’s a problem. The HILDA survey misses a lot of the spending story. It captures only half of total household expenditure captured by the Household Expenditure Survey (HES) conducted by the Australian Bureau of Statistics (ABS).
Overall, HILDA implies that households across almost all ages spend an average of $20,000 a year. That’s implausibly low, given median household incomes are $74,776. In contrast, the HES suggests the average household spends about $40,000 a year, possibly still on the low side compared to incomes, but much more accurate than HILDA.
The problems with the HILDA survey arise because it ignores some expenditure, and collects data in a less reliable way.
HILDA omits several important spending categories (See Table below). Recreation, for example, is the third largest category of household spending and a critical source of expenditure in retirement when people have more leisure time. Yet recreation is largely absent from HILDA spending data.
The producers of the HILDA survey at the Melbourne Institute acknowledge some of these issues. They estimate that HILDA’s questionnaire asks households questions about roughly 80 per cent of household spending.
Yet the HILDA survey also appears to collect less reliable data on the categories that it does cover. For a number of spending categories (such as alcohol consumption), the HILDA survey asks every bill payer to guess their entire household’s expenditure over a recent survey period. A Melbourne Institute paper explains in great detail how there are big variations in what different family members think. When faced with conflicting accounts, HILDA simply averages the guesses.
In fact, retirees’ spending falls sharply as they age
The HES provides much more reliable data on how Australians spend their money. Unlike HILDA, which is designed to track households’ incomes and work patterns, the HES is specifically designed just to measure household spending. People aged 15 and over are asked to keep a diary of every item they buy during a two-week survey period. The expenditure of all respondents in the household is then added up to find total expenditure.
Grattan Institute’s 2018 Money in Retirement report uses the HES to analyse retirees’ spending patterns. But the HES, unlike HILDA, is not a longitudinal dataset: it doesn’t track the same households over time. So how can we track retirees’ spending patterns as they age using the HES?
We use successive waves of the HES to track how the spending of cohorts of retirees changes as they age. Since each wave of the HES is drawn from a representative sample of the population of that age, we can infer changes in spending patterns for households of given ages over time.
Our analysis shows that Australians’ tend to spend less after they retire. Even the wealthy eat out less, drink less alcohol, and replace clothing and furniture less often. Spending tends to slow at around the age of 70, and decreases rapidly after 80.
This fall in overall spending is mainly a result of lower spending on transport, recreation, food and furnishings. Retirees who own their home tend to have paid off the mortgage, and retirees no longer need to spend money on children or on work-related expenses.
Pensioners also spend less because they get discounts on council rates, car registration, electricity and gas bills, public transport fares, and pharmaceuticals. Public transport concessions apply to all retirees – not just pensioners. Retirees’ spending also tends to be lower because they have more time, and so cook at home more and eat out less.
They spend more on health care as they age, but Medicare largely shields them from the full costs. The modestly higher out-of-pocket costs they do pay are mainly due to rising premiums for private health insurance.
Not only do most retirees not draw down their savings throughout retirement, many add to them. One recent study found that the typical pensioner still had 90 per cent of what he or she retired on after eight years. Retirees say they actually feel more comfortable financially than the Australians younger than them who are still working. And retirees are less likely to suffer financial stress such as not being able to pay a bill on time. Even retirees who rent are less financially stressed than people who are working and renting.
International studies make similar findings. Reports using the British Family Expenditure Survey and The American Income Dynamics and the Consumer Expenditure Survey both found spending decreases into retirement. Another prominent US study found that real spending falls by around 1 per cent each year in retirement.
What this means for retirement incomes policy
Our analysis using the HES shows that retirees’ spending needs fall as they age, but not because they can’t afford to spend more, or in a way that leaves them unsatisfied. The fall in spending appears to reflect declining recreation and other discretionary spending as retirees age and their health declines. Meanwhile governments foot most of the bill for retirees’ rising health and aged care costs.
This means that calculations about the adequacy of retirement savings ought to be based on whether they are enough to maintain buying power (at best). Yet much of the existing Australian research assumes that retirees need to save enough to enable their incomes to keep climbing throughout their retirement, in line with general wage growth. Implicitly, these studies assume that a retiree needs to spend 22 per cent more at age 90 than at age 70, after accounting for inflation.
In contrast, we assume that retiree spending needs to keep up with inflation, and we find that almost all workers of all income levels will retire on incomes at least 70 per cent of their pre-retirement earnings – the so-called replacement benchmark used by the Organisation for Economic Cooperation and Development and the Mercer Global Pension Index.
Ultimately our work shows that retirement incomes are likely to be adequate for most Australians. But it also shows that the data source matters a lot to the answer. Grattan Institute regularly uses HILDA for analysing income, wealth and employment dynamics over time. But when it comes to analysing household expenditure, the HILDA survey isn’t really designed for the job.