Friday, August 08, 2008

Superannuation - the assumptions

Regarding Super-man, below, commentator Tej writes:

"I guess this is what to expect when you get librarians to do financial calculations. If you assume a normal profile of earnings, 40 years in the work force, 15% tax on contributions, 12% tax on earnings which is typical of super funds, 1% fees, 6.5% real returns before fees (typical of the last 100 years), then a person whose only savings are the 9% super will have 35% of their peak income available to draw.

This is based on the fact that if you draw more than 3% you have a big chance of exhausting your money. I would be interested to see the librarians' calculations - I suspect they are assuming 10% plus investment returns which is unrealistic after taxes, fees and inflation.

So, what are the Parliamentary Librarian's assumptions?

The document is:

Richard Denniss, The Crisis of Cash or Crisis of Confidence - the Cost of Ageing in Australia, Australian Journal of Political Economy, July 16, 2007

Here's the table: (click to enlarge)

And here are the assumptions:

"An individual making 9 per cent contributions to superannuation for 40 years, living for 18 years post retirement and receiving a real rate of return of 3.5 per cent. The retirement incomes for low income earners include income from both superannuation and the relevant part pension to which they are entitled. The reasonable benefit limit has been ignored."


WT said...

While making no judgement on Tej, what he says sounds a lot like the convoluted crap that fund managers sprout in an attempt to look like they're worth the outrageous amounts they're sucking out of the funds through fees.

economan said...

On the assumptions of the Parliamentary librarian paper, I would say the assumed 3.5% real rate of return is very conservative. However, the 18 years after retirement is probably too low, even if you're assuming people wait to retire until they are 65. Life expectancies have grown consistently over the last 50 years, despite repeated predictions that the improvement inevitably had to stop or at least slow.

I think Tej's assumptions are pretty good, but Tej doesn't give details on age at retirement / life expectancy.

However, I don't know what Tej is trying to say. The 35% and 3% (is that a typo?) are not clear. If it is saying after 40 years of super at 9% you only get 35% replacement income, then that's just wrong. Unless s/he is assuming a VERY long life expectancy.

(And by the way, let's not pretend that the report was written by 'just a librarian'. The parliamentary library puts out good papers on a variety of topics!)

Jon said...

Peter, I'll declare my self interest up front, as I work for a super fund. That doesn't mean I disagree with everything you say (I enjoyed listening to you on the Adelaide ABC the other morning).

One point I'd like to make is that the problem with the SG is that it concentrates on inputs, not outputs. 9% may be perfectly adequate, depending on what returns the funds make. The super industry (at least the fund I work for does) do their best to get the highest return for the cheapest price, but, yes, there is a market out there, and yes, it goes down as much as it goes up.

Because of this concentration on inputs, any report about what retirement income is to be produced by a given contribution is based on assumptions that are very unlikely to hold true for a given individual. Eg the JAPE report assumes a 3.5% real return. The NATSEM report that you list assumes 4.5%. Over the 5 years to 30 June 2007 many funds have averaged a real return of 8% plus, with many get -8% or less over the last year. Because of this uncertainty, and the risk averse nature of most people, is it better to overshoot on the contribution side?

As I'm sure you're aware, superannuation existed a long time before the SG, but usually only for government employees (like, say, the ABC :) ) or large corporations. The old funds were generally defined benefit, meaning the employer picked up the tab and the employee got a benefit as a multiple of salary. This had the benefit of concentrating on outputs, though it required something that doesn't happen anymore, ie staying 40 years with one employer. The ideal superannuation would be a portable defined benefit fund, say like a government pension that wasn't $26 pw above the poverty line. David Knox, an actuary, proposed sometime ago a 3 tier system involving a universal pension (ie no means testing), plus employer contributions plus voluntary individual contributions. I thought it was a good idea, but the government wasn't so keen.

One problem with the current pension is that when you and I get to old age, who's going to pay for it? Yes there is a cost in tax concessions to super now, but at least we are paying for our own retirement. If we leave a tax burden for our children, will there be enough of them (along with net immigration) to fund it appropriately?

I agree that the tax concessions on super are heavily skewed to the higher paid. My preferred response to that would be to remove all input tax on super (ie contribution and investment tax) but tax all withdrawals as income (including death benefits to non-dependents). While this would increase current tax concessions, it would encourage super to be used as income, rather than for lump sum withdrawals.

Anyway, give ASFA heaps, because I think they could do a better job at hearing both sides of the argument, and keep up the good work.

Anonymous said...

Tej (T i m J o s l i n g) here...

I have just been through an extensive exercise of working out how much money I need to retire. That is my interest in the topic and the reason why I have an option that is actually based on data and analysis.

I was very surprised to find that the amount needed is a lot higher than I had expected. If you want less than 1% chance of running out of money, you cannot safely draw more than 3% per year of your savings.

I have no links to the funds management industry, other than using the services of a fee-for-service financial planner. I agree with the opinions of many that the industry charges fees that are out of proportion to the value added. They also have a vested interest in maximising the money in super.

My figures on life expectancy are based on the figures provided by the ABS. Figures on stock market returns are based on historical data.

In my calculations I assumed retirement at 57, but I also ran the monte carlo analysis based on age 60 and it made very little difference. If you look at actual retirement ages, this is realistic.

My assumption of 40 years in the workforce is quite high, compared to the population at large. You need to consider time out of the workforce due to education, childcare and illness. As people age, they find it harder to get work, and often illness intrudes.

The parliamentary library study assumes you and your spouse die 18 years after retirement. In other words, they are drawing down capital and after 18 years you run out of money. There is a high risk of living longer than 18 years. So the study does not contradict my claims.

I also note the study makes the ridiculous assumption that income is constant during the working life. For most people, income rises strongly to middle age. This means the super contributions are skewed to later in life, with little time for compounded investment returns to work for you.

The full study does not seem to be available anywhere so there may be other assumptions. One common false asumption is that market returns are constant, This is not true and the variability of returns is a major risk for retirees. Ask anyone who retired before the recent crash how they feel.

I note no-one has asked for my spreadsheet. Anyone interested in some data and analysis, rather than opinion, is welcome to a copy - email tej at

My figure of 3% is the percentage of your wealth that you can safely draw in retirement. You *may* live a long time, so you need to plan for that possibility. The stock market *may* do badly so you need to plan for that possibility. 3% ensures you will not run out of money (1% chance). If you live a long time and the markets do badly, are you happy to run out of money?

35% is the percentage of final income that you can safely draw, based on the assumptions I made.

Anonymous said...

tej here again...

I have reverse engineered the librarians' calculation to get the same result. That is, their calculation with all the money gone after 18 years.

Points to note

1. They assumed it's OK for all the money to be gone after 18 years. Bad luck if either you or your spouse live longer than that.

2. They assumed no variability of returns. This is a major flaw. For a detailed discussion of this issue, see "The Intelligent Asset Allocator" by William Bernstein. During the draw-down phase, volatility of earnings greatly reduces the amount you can safely withdraw, due to a phenomenon called "volatility drag".

3. They assumed that income is uniform during your working life. For most people this is quite unrealistic. Earnings grow as skills grow and people are promoted. What this means is that the super contributions are skewed to later times, when there is less time for the nest egg to grow over time. This has a major effect.

4. They also assumed that people would access the social security system. The "earnings" they quote actually include social security payments! This is OK, but not if you are trying to make an argument that 9% superannuation is enough. If you have to put your hands into the taxpayers' pockets to live after saving 9%, then 9% is not enough.

The distribution over working life was put in by hand - it may not be exact but as Keynes said, "It is better to be roughly right than exactly wrong".

Note that if there are two of you, you need to cater for the longer of the two life expectancies. In most couples, the
woman is younger, and women live longer than men - so typically the woman dies 10 years later. Also, it is important to remember that expectancy of age at death at 65 is a *lot* longer than life expectancy at birth.

I used past stock market returns (US/AU) and threw random years from that into the pot to get out simulated future returns.

There have been some really bad returns in the past. For example, the All Ords in Australia did not exceed the January 1970 level again until about 2002 AD in real (inflation adjusted) terms. The US market has had similar and highly correlated periods of bad returns (eg 1929-1946). Also, the US and Australia, which I used in the simulations, have been in the top 5 top stock markets of the past 100 years. Results look a lot
worse if you use the world index (return more like 5% even excluding Russia and China) or countries like Belgium (2.5%) instead.

All are welcome to the s/sheet if interested. See the email address in my other posting.

T i m J o s l i n g

Jon said...

Tej, a lot of commentators, including Peter and yourself, have a go at the "funds management industry" for trying to maximise funds under management. And, yes, a lot of the industry does that, and a lot of the fees are excessive (but I work for an industry super fund, so I would say that, wouldn't I).

However, as with a lot of things in life, you can always do it yourself, and a large number of people (500,000 I think at last count) choose to run their own fund. And good on them if that's what they want to do.

As for how much you need to retire, you hit the nail on the head as far as the issues of risk go. That is, the risk of poor returns, of living too long, of needing some spare cash, etc. There is lots of lovely assumptions one can make in calculating the 'right' amount, and none of those assumptions will apply to you, me or anyone else.

PS Nothing personal, but I see enough spreadsheets during the week. I also suspect that it will confirm that I agree with you.


Anonymous said...

"However, as with a lot of things in life, you can always do it yourself, and a large number of people (500,000 I think at last count) choose to run their own fund. And good on them if that's what they want to do."

Tej again...

I have a DIY super fund. This does help but there is a lot of red tape and there are many limitations and extra costs involved. For example you have to (in practical terms) have an accountant AND an auditor. If you are running a pension you probably have to get an actuary to sign off as well. Etc etc.

The retail funds management industry is actively trying to make it as hard and expensive as possible for people to run their own funds. Have a look at their submissions to government enquiries.

You hit the nail on the head Jon. The simulation by the librarians ignored the two major sources of risk - market risk, and longevity risk. If we knew exactly how long we will live and if the stock market provided constant returns life would be a lot simpler. But we don't.

I actually tried to find a way to deal with longevity risk. This is a classic insurance problem. It is quite predictable how long the average person will live, so insurance companies should be able to insure you against longevity risk. If you could buy a policy that gave you say $30k every year you live after 82 years old, that would solve the problem. The cost need not be high.

Unfortunately this is not possible. Government regulations do not allow an annuity that starts after retirement ago. So no solution is possible. I spoke to two insurance companies and two financial planners and it seems you just have to bear the longevity risk yourself.

You can buy a lifetime annuity. The problem with that is that the returns are pathetic because they are based on bond yields less hefty commissions. The net return on the annuity is a couple of percent below term deposit interest rates, or to put it another way, close to zero after inflation. So you will probably be a lot worse off.

Thanks once again to our friends in Canberra for messing things up - again.

Post a Comment