Taking the luck out of investment

Ellie Bennett asks how stochastic modelling can help both individual savers and contract-based schemes

Rolling some dice or spinning a roulette wheel would not be everyone's preferred method of planning for retirement. Such activities are usually associated with frivolous games of chance, played out in the casinos of Monte Carlo and Las Vegas – a world away from the serious business of asset allocation.

Surprisingly, however, the same principles of probability apply to a strand of statistics called stochastic modelling, which is increasingly being used to help map pension investment strategies.

Clearly, it is impossible to accurately predict how stock markets, bonds, gilts and property will be performing twenty years down the line.

Stochastic forecasting tries to account for this by running thousands of simulations, which are based on a sound economic model, in order to produce probability distributions that represent the likelihood of a variety of different events happening.

Advocates of stochastics often claim that it is better than its alternative, deterministic modelling, which is still used for FSA compliant business quotes. This apparent poor relation selects a single rate of interest to predict what investments might yield in the future, typically at five, seven and nine per cent. It produces the most likely average return but takes absolutely no account of how money is invested, which can, of course, be
highly dangerous.

"The chances of the UK equity market delivering nothing or completely negative returns are practically non-existent, but even that is more likely than the market delivering seven per cent each year for the next 40 years," explains Steve Rumbles, head of defined contributions at BlackRock.

Rob Fisher, head of sales and marketing at FundsNetwork, the investment supermarket owned by Fidelity, likens the restrictions of deterministic modelling to an internet route planner. The application predicts how long the journey will take, but does not take crucial factors such as time, traffic and weather conditions into consideration. During rush hour the trip might take two hours, whereas in the middle of the night it could take half that time.

Similarly, deterministic modelling does not take account of a whole raft of situations that could affect a person's investments along the way.

Stochastic modelling, on the other hand, provides a far more structured way to construct a client's portfolio. One of the key benefits is that investors get an objective view of what the future might hold. This includes factors such as inflation which cannot be controlled by the individual. It helps choose the best asset allocation for a customer's own risk tolerance and gives them the best and worst scenarios based on their current situation.

The investor needs to know, for example, that equities are likely to produce a much higher reward, but that the spread will be wider than with bonds, which are safer but less profitable.

"It is endeavouring to show investors in a clear way what they are likely to get from their investments so they don't get surprises and you don't get people ending up at retirement and suddenly discovering that their savings are completely inadequate to provide a decent pension," comments Bruce Moss, principal at Towers Perrin Tillinghast.

He describes stochastic modelling as a way of "empowering investors to look after their own affairs," by giving pension savers early warning signs that they might not be doing enough to provide a decent income in retirement.

History
The concept of stochastic modelling has actually been around for years, usually known more simply as statistical sampling.

The process uses Monte Carlo simulation, named in reference to the legendary casino in Monaco, to generate random numbers upon which a stochastic formula is applied to show, in the case of pensions, what impact that has on different asset classes.

Prior to the emergence of the first electronic computers in 1945, the history of stochastics is a little sketchy, although in 1930 Enrico Fermi famously used a random sampling method to calculate the properties of the newly discovered neutron. The use of stochastics moved up a gear after the emergence of widespread computerisation, although for many financial advisers it has only been during the last few years that it has entered the mainstream.

Towers Perrin Tillinghast, for instance, has been applying stochastic modelling to client's portfolios since 1993. This was in response to the general UK-wide shift from defined benefit to defined contribution pension plans. Contract-based pension scheme members do not have the luxury of being able to rely on a board of trustees to make the underlying investment decisions for them.

"The onus is now being put more on the individual. We really need to give them the tools to understand from a risk perspective how much is inherent in their strategy," says Robert Childs, senior business analyst at DST International PAS.

The gradual slide over to money purchase pensions has also altered the role of the financial adviser. Ten years ago they were mainly distributors of financial products, says Fisher, but their roles have now morphed into advising clients more broadly on how to achieve their financial goals in life.

However, not all stochastic tools are the same and Chris Harrington, financial planning consultant at Cowgill Holloway, warns that different products can produce variable asset allocations for the same individual.

It is therefore essential that all advisers receive an appropriate level of education. "There's a tendency for them to be seen as black boxes and you need to understand how the model is working; what sort of assumptions the model is based on; and look at the outputs to see that they make sense."

Not just for advisers
Stochastic techniques are not exclusively for advisers either. There are some products now emerging, such as BlackRock's Target Plan, which enables pension scheme members to enter their details into an online engine that calculates the probability of them achieving their goals, independent of an adviser.

If the results come as a shock, the diversification of assets, retirement age and contribution levels can all be altered. Such initiatives are opening the door for stochastics to benefit
a wide range of people, not just the high-net-worth or mass affluent groups who might well have a SIPP or similar product.

However, it is important not to forget that a stochastic model is just that: a model and one underpinned by complex mathematics so the value of accurate advice cannot be over-stressed. Furthermore, the stochastic simulations are themselves very complicated and would be far too difficult to understand if presented in a raw form.

"Part of the trick is to try and simplify the output, but by doing this you're actually taking away some of the essence of the underlying model. So there is a balance to be sought with this kind of tool," says Chris Read, chairman of Dunstan Thomas.

Another drawback pointed out by Harrington, is that the stochastic engine needs to be tinkered with constantly to ensure that it remains a true representation of the present economic situation. It would be extremely unwise to perform a stochastic forecast just once and then leave it; you must re-balance the asset allocation on a regular basis.

Products are now evolving to remind advisers and clients to do just that, such as Standard Life's wrap platform, which automatically flags up when asset allocation has strayed more then two per cent away from where it should be.

Some providers try to glean the best of both worlds by combining stochastic and deterministic projections within their products. Childs explains: "It's horses for courses. Deterministic gives you a reasonable measure of the various charges in different products and what stochastics gives you is a much better understanding of your likely return based on the actual asset classes that you invest in."

Until now, most of the stochastic modelling products have been built for financial advisers and by definition their main clientele has tended to be the mass affluent consumer upwards.

However, the application of stochastics is indiscriminate of wealth. As long as there is a probability to work out, it can be used for a pension pot of £200,000 or that of £200. And making more tools available via the internet would help reach a wider range of people.

There is a growing belief that pension savers should receive more guidance on how to sensibly diversify their investments and Harrington suggests that the FSA is already calling for an investment process that all advisers should follow. Stochastic forecasting offers the more disciplined approach that people are looking for and it may only be a matter of time before it becomes a compulsory part of pension planning.

- Pensions Age July 2007