A while ago it dawned on me that risk affects a pensioner in a fundamentally different way to a wealth accumulator. There is much more to it than the traditional stuff about income being non-replaceable for pensioners and hence needing to be extra cautious, in fact volatility tends to do opposite things depending on whether a portfolio is being accumulated or distributed.

Peter Bernstein makes the point that, "the determining question in structuring a portfolio is the consequence of loss; this is far more important than the chance of loss." [Bernstein, P. L. Management of Individual Portfolios. The Financial Analysts Handbook; Levine, S. Ed.; Dow Jones Irwin Inc: Homewood Il, 1975.]

Risk is a function of a portfolio's assets and its liabilities, in particular the cash flow between the two over time.

A more workable definition of risk would be a calculation of the probability of not having sufficient cash with which to buy something important. Obviously the income needs of a wealth accumulator and a self-funded retiree are drastically different and require a vastly different approach to risk management.

Ability to purchase should not be confused with the need to purchase - the latter is a portfolio's raison d'être. The consequences of loss are dramatically different for a wealth builder and a retiree, and different forms of loss affect them differently.

Volatility is clearly less important to the wealth accumulator because commonly used dollar cost averaging tactics lead the investor to purchase more units/shares at low prices than at cyclical tops. The end result for an accumulator is that volatility is a relatively benign force and in some cases may actually improve the lot of the investor by providing regular buying opportunities that will be automatically capitalised on with a dollar cost averaging strategy.

On the other hand, the exact opposite situation applies for a pensioner. Arguably, volatility is significant from more than just a psychological point of view because drawing a pension is a dollar cost averaging strategy in reverse. When a pension is drawn from a volatile portfolio the net effect is that the most units/shares are sold at cyclical lows, rather than highs.

Volatility aversion is not new for pension planning, after all the traditional portfolio for a pensioner is indeed full of bonds and mortgages and cash and other relatively stable investments. Volatility reduction techniques really work only to an extent and carry no guarantee, the typical strategy of finding a portfolio with a low covariance is a hazardous exercise given the lack of stable enduring fixed correlations between securities classes. I think that what is needed is instead a method of portfolio construction that sidesteps volatility rather than tries to reduce it, though by all means a diversified portfolio makes sense anyway.

If volatility can't be reliably or completely reduced, you can to an extent make it irrelevant. If you hold a bond to maturity and the bond is a relatively good credit risk, receiving the face value of the bond back is virtually assured. Along the way interest rate changes will play havoc with the face value of the bond, but if held to maturity none of this will really matter much. There could be an opportunity cost by locking in low interest rates on the bond before yields rise, but this should be seen in the context of any hedging decision, you always trade security for potential returns and a hedger learns to accept the occasional missed windfall in return for a guaranteed return. Naturally the investor should be careful to maintain sufficient liquid reserves in the form of cash investments to cover short term contingencies to avoid having to sell any asset at a capital loss.

Being a long term investor I am also aware that while in the short term stocks and other growth asset classes are very volatile, over a much longer term the chance of loss is reduced. On the other hand, the longer you hold a fixed interest security for the greater your opportunity cost of being out of better performing growth assets and the more severely inflation is likely to devalue your portfolio.

If there could be some way to allocate funds to take advantage of the assured return of high quality fixed interest investments for the short term and yet preserve a portion of the portfolio for long term liabilities then an ideal compromise could be found that combined the best aspects of defensive and growth investment. Could there be a way to allocate funds so that pension recipients could have bonds for their short term needs and shares for their longer term needs?

Traditionally financial advisers like to pigeonhole clients into "risk profiles", and usually these advisers will thus allocate a fixed asset allocation strategy depending on this risk profile. While this is obviously an applaudable first step in figuring out how to make a portfolio suitable to an investor, it still remains evident that this is a less than ideal solution because of the different ways that volatility affects a pensioner and a growth investor. The typical wisdom is of course to say that pension investors are low risk people and hence only make use of the lower risk profiles, but I consider this inadequate on the grounds that the term of the pension and the amount of capital to invest would lead to great variation in the needs of the individual investor.

So I went about devising a way to calculate asset allocations for different pensioners, taking into account psychological risk tolerance, the number of years the pension is to be provided, inflation and other factors.

I question for a start if volatility is a relevant parameter to fix a portfolio by. In my time as a financial planner I have not yet met a client that was able to express their needs in the context of the standard deviation of the returns achieved. I immediately settled on a different measure of risk that has a real meaning for an investor:

A high risk investment strategy for a pensioner is one that is unable to provide a guaranteed income stream for any length of time, and will be susceptible to the shorter term fluctuations of the market. A low risk strategy is one that is able to provide a guaranteed income stream for many years to come, even if the stock market were to fall substantially in the very near future.

You might wonder who I ripped that idea off. No one actually, it is my idea and although the statement is obvious when someone makes it, almost trite, I can't say that in my readings I have ever heard of risk expressed in exactly those terms. So how do I go about making risk profiles out of such a statement?

I decided that it would be foolhardy for a pensioner reliant on income from a portfolio to have 100% exposure to growth assets, unless of course the portfolio were so large compared to daily needs that an income stream from the dividends would be sufficient to match living requirements without needing to dig into capital, an outcome I find undesirable given the reverse-dollar cost averaging effect mentioned earlier. Therefore even though it might be argued that a high growth investor could invest only in growth assets as an accumulator, at least some defensive assets are needed for a pension portfolio.

So I decided to define risk profiles in terms not of volatility, but in terms of number of years of guaranteed income before you need to start selling capital growth investments:

For example, when constructing a pension portfolio for a "Balanced" investor, you could buy bonds with staggered maturity dates and face values that correspond with annual payments, a secure mortgage trust, cash management trust or other low risk investment that can provide a guaranteed income stream for 8.5 years. This is easily programmable as a spreadsheet, and explicit assumptions about returns, taxes and inflation are simple to build in to the model. For the section of the portfolio that provides for needs following on from the 8.5^th year a completely different asset allocation can be adopted, either a standard allocation used conventionally for balanced investors, or, because it is long term money, a portfolio made entirely of growth assets.

Even a high growth investor can rest assured that if the stock market were entirely wiped out tomorrow he or she would at least have another six and a half years of secure income, assuming that the same calamity did not also wipe out investment grade bonds, cash management trusts and mortgage trusts. A more conservative investor knows that this security extends out over a decade.

This method allows bonds and cash to be explicitly budgeted for spending in the near future. There is no reason to have bonds as a part of the portfolio earmarked for spending in 25 years time and there are no shares in the portion of the portfolio that will be spent in the next few years. In this way a diversified portfolio can be made that has long term growth investments for long term consumption and short term defensive investments used for short term needs.

As sensible as the idea sounds, this is not how portfolios are presently made up: the typical conventional portfolio has a mix of assets that is permanent, a single risk profile is determined and then a fixed asset allocation used, meaning that a portion of long term money is always invested in defensive assets and a certain amount of the growth portfolio is to be sold in the short term - not a logical way to do things but nevertheless universally accepted.

My choice of years was purely arbitrary, I could have chosen for example to make the capital guaranteed portion of the portfolio for a high growth investor last only 2 years or the conservative investor to have 20 years of security, but the idea of defining risk in terms of a "term certain" guaranteed income stream as opposed to capital fluctuations is to my knowledge a new idea.

Annually the portfolio will need rebalancing. Bonds will mature on schedule, asset allocations will change as markets move their separate ways, cash will be spent and dividends, interest and rent received. The growth assets will need to be rebalanced back to their relevant benchmarks and new bonds bought to provide for future years.

If an extreme short term market fall occurs, growth asset sales may be temporarily suspended. For example given the buffer period made available by the capital guaranteed investments it is not necessary to sell any growth assets to fund a new purchase of bonds. At the discretion of the investment manager money can be left in the growth assets during hard times (which are really the best buying opportunities) and growth assets sold during boom times.

In this way there is a certain amount of "fat" in the portfolio, which provides an automatic reserve against short term adversity. This is a good example of using the time honoured principle of saving funds during the proverbial seven years of plenty, to tide you over for the seven years of drought.

Value averaging, as opposed to dollar cost averaging, may therefore be successfully used in the rebalancing of this portfolio. This will provide a near optimal use of assets that explicitly (as opposed to incidentally) takes into account such real world factors as market cycles, inflation, pension drawings and the suitability of different investments for short or long term investment strategies. This type of robust risk modeling is virtually absent in most discussions of portfolio theory.

The interesting thing about this method is that no two portfolios are really exactly the same unless they are only intended to be short term investments. A "High Growth" portfolio is actually identical to a "Conservative" portfolio if the term of the pension is less than 6.5 years, but a high growth portfolio planned to last for 10 years is very different to a high growth portfolio calculated to last for 20 years.

In addition, as you change assumptions about returns and inflation the portfolios are adjusted to reflect this. It is my contention that a fixed asset allocation strategy schedule is unsuited to a pension portfolio, instead asset allocation needs to be calculated each and every time.

I have constructed a spreadsheet that performs these calculations. You can change the names of the funds, asset allocation of these funds and assumed return, inflation, starting capital and number of years to run in the yellow fields of the asset allocation worksheet. The blue field shows the rate at which drawings can be theoretically made, based on this data.

The way I have built the worksheet I am assuming that the first fund you choose is some sort of highly liquid investment like a cash management trust, used for the next year or two's income. The second fund could be direct bonds with staggered maturity dates corresponding to when you need the money or another low risk investment. The third fund onwards are all the other funds or securities you might choose from, specifically either direct shares, property trusts, share funds or various other growth assets.

If you estimate (conservatively) the long term growth rate for each of these investments the spreadsheet will take this into account when allocating capital over the term of the pension. You will notice that if you select a portfolio made up of high return growth funds, the spreadsheet will allow a higher initial drawing for the higher growth risk profiles. This reflects the assumed higher growth possible with the growth funds and hence the lower asset allocation given to later years (if you assume higher growth you can spend a bit more now because you don't have to allocate as much to future needs as you do when you assume low growth).

Tip: one way to guarantee that the portfolio is being constructed in a way that ensures the long term viability of the income stream would be to make a very high estimate of inflation. This will automatically budget more money to later years and thus a shortfall becomes less likely.

Click here to have a look at the spreadsheet.