Why relying on averages when writing a financial plan is a mistake
When it comes to planning for our financial future we almost always rely on “averages”, which is a flawed approach, writes Andrew Barnett.
Nobel laureate Milton Friedman, the man described by The Economist as one of the most influential economists of the 20th century, is credited with saying “never try to cross a river because it is, on average, only four feet deep”.
The average may be four feet but the depth of the river in the middle is likely to be well over your head. The lesson here is that averages won’t help you to ford rivers, and there are plenty of other circumstances where they won’t be of much use.
Take the weather. The average temperature across the year in Melbourne is 15 degrees. You’re not going to dress for this every day.
You’ll think first about the season and then about the forecast for the day (although this is generally more helpful in Sydney than in Melbourne) before making your decision.
When it comes to planning for our financial future, however, we blissfully rely on averages. Averages appear to tame what are inherently and ultimately unpredictable paths.
They simplify what would otherwise be a bewildering array of possible future outcomes.
This is manifest in almost all financial planning models, which display the wealth of an individual rising to the point of retirement, and then falling over retirement, as a single path.
It seems most models conveniently have us running out of assets and life at the same time.
George Box, an English statistician, wrote that “all models are wrong, but some are useful”. Our virtually exclusive reliance on averages makes our models just wrong. Half of our plans will fail.
There are four assumptions explored below that euthanise our planning: life expectancy, market returns, correlations, and insurance costs. Let’s take life expectancy first.
Life expectancy
The life expectancy of a 65-year-old male is 85 (assuming some mortality improvement).
In many financial plans, we use this to determine the level of assets we need at retirement to provide for, say, a comfortable lifestyle for the next 20 years.
But using this as the basis of planning is more than likely to be wrong.
The proportion of males who will actually shuffle off this mortal coil at age 85 is 5 per cent. The other 95 per cent live somewhat shorter or longer.
There isn’t even a large cluster around the age of 85 itself. If it was a useful planning benchmark, then the majority of individuals would live to this age, plus or minus a couple years. Instead what we find is that 20 per cent of males don’t live to 75, and 20 per cent of males live past 92.
Planning based on an average life expectancy is, on average, going to be wrong.
Market returns
The second pervasive average we use is market returns. To a large extent this is based on mean reversion or the observation that, over the long run, the average market return is fairly stable.
Periods of underperformance will be followed by periods of outperformance.
This is borne out by the data: rolling returns for the All Ordinaries index over one-year periods from 1984 vary between -44 per cent and 59 per cent, but annualised rolling returns over a 20-year period vary between a narrower range of 4 per cent and 9 per cent.
Most modelling makes the seemingly logical step of assuming that, because superannuation is a long-term investment, we should assume long-term average returns.
There are many problems in using a long-term market average. First, it doesn’t account for timing risk to individuals.
We have far shorter timeframes; what the market does during 20 years is important, but so is whether there is a significant market correction in the years preceding or following our retirement.
Related to this, and also unaccounted for, is sequence risk where a series of negative returns precedes a series of positive returns. Both these risks have a profound effect on an individual’s wealth.
There are two alternatives to managing risk: avoid it or insure it.
Many of the investment decisions since the global financial crisis have been about avoiding risk. Cash and term deposits account for 30 per cent to 40 per cent of new platform flows.
Industry dialogue is also emerging on target date funds (TDFs, also known as lifecycle funds) that de-risk as an individual approaches retirement by becoming more defensively invested.
On average, these should indeed have lower volatility.
But avoiding risk also brings with it lower long-term returns than equities. In this, they swap market risk for longevity risk.
There is a greater chance of outliving your assets if you earn a lower return.
The correlation assumption
The reliance on diversification of TDFs to manage risk reveals the third assumption of an average: the risk that correlations between asset classes are constant through time.
On average this is true, but during times of turbulence – the very periods when you’re relying on diversification to stabilise returns – correlations between asset classes tend to increase dramatically. Nothing goes up in a falling market except for correlations.
This is the reason why in 2008 in the US, even the most conservative TDFs lost between 9 per cent and 41 per cent.
The Department of Labor and the SEC held Joint Congressional Hearings to understand the nature of this failure of risk mitigation.
Somewhat less of the industry dialogue has been about insuring risk. Effectively, the individual transfers the risk from themselves to a financial institution.
The two primary risks are longevity risk, which can be insured using lifetime annuities or variable annuities; and market risk, which can be insured using variable annuities and options.
There is a cost, of course, to any insurance. For a one-year put option, this might be 10 per cent of the notional amount insured.
For a variable annuity, this might be 1 per cent per annum.
The fourth flaw
The fourth flaw of averages is the common mistake of assuming that the cost of insurance should be netted out against the average expected return.
Taking the variable annuity example, if the average market return is 7 per cent, then we’re tempted to estimate the return net of the insurance premium is 6 per cent.
This is wrong. The insurance only costs this in instances where there is no claim.
In instances where there is a claim, there is no cost of the insurance but instead a large benefit that may be multiple times what the premium was.
Another way to think about this is that the average ‘cost’ of the premium is the value of all premiums collected, less all claims paid which might be less than half what the headline premium is.
To improve planning and avoid the fallacy of averages, there are several techniques we can use.
First, we should use scenarios that allow for variability in the assumptions, especially market and mortality, to understand the range of potential outcomes.
Second, if you want to reduce the risk of plan failure, then use a base assumption that is conservative, such as longer longevity and poor market outcomes.
Third, if you want to avoid the risk of plan failure, consider using insurance and differentiate between product fees and premiums.
Finally, understand the key levers to pull under different scenarios, such as contribution rates, withdrawal rates, retirement age, and the value of insurance, and how these affect the outcomes under different scenarios.
Andrew Barnett is the general manager of MLC Retirement Solutions.
Recommended for you
The FSCP has announced its latest verdict, suspending an adviser’s registration for failing to comply with his obligations when providing advice to three clients.
Having sold Madison to Infocus earlier this year, Clime has now set up a new financial advice licensee with eight advisers.
With licensees such as Insignia looking to AI for advice efficiencies, they are being urged to write clear AI policies as soon as possible to prevent a “Wild West” of providers being used by their practices.
Iress has revealed the number of clients per adviser that top advice firms serve, as well as how many client meetings they conduct each week.
