The world of an actuary is a strange one. We place monetary values on things that may happen in the future and then help companies and clients prepare for these uncertain events based on our professional, but subjective, outlook.
Financial planning professionals of course attempt to do the same thing for individual clients, to construct a portfolio that will grow and shrink over time and capture the client’s financial journey through life to meet their financial goals and obligations.
Traditional deterministic models for lifetime cashflow planning are, quite rightly, falling out of favour, with the financial planning world embracing uncertainty and choosing stochastic modelling. Investments don’t grow steadily with a single rate of return and so neither should the models. Stochastic modelling is a particularly powerful tool to ensure that the clients’ financial position during periods of volatility, like March 2020, can be tolerated and recovered from.
Uncertain events are almost always modelled stochastically. However, it’s important to remember that one individual path from a stochastic model is almost certainly not going to reflect the actual path a client’s finances will take. Instead, it is more an illustration of the type of volatility they should expect. It is the median result, out of many thousands of investment paths, that can guide us to a financial result that has an equal chance of being overstated and understated at a specific point in time, and is the one that should be used to judge if adjustments to the lifeplan are required.
Not all stochastic models are created equal
With the adoption of stochastic modelling for cashflow projections and the ever-increasing computing power available to us all, a wide range of software solutions are available, but not all stochastic models are created equal. Without standards from the FCA to fall back on, how then, do you decide which stochastic model is right for you?
The traditional actuarial view would be to construct a “perfect” model that projects funds, contributions, withdrawals, charges and taxation in line with the real world. However, the real world is more uncertain than just investment returns. Indeed, the entire construct of taxation, the asset allocation of indices or funds, or the characteristics of a named fund may be entirely different 30 years into the future.
Attempting to address any of these fundamental changes to the financial planning world in a stochastic model is the start of a journey into insanity, and leaves the model exposed to an insurmountable number of subjective assumptions that must be made to achieve any semblance of a reasonable output.
Attempting to model how the underlying asset allocations of a fund you are currently invested in may change over time, for example, is near impossible. In fact, there is no guarantee that this fund will even exist in 20 years, let alone if it’s performing well enough to be invested in. If I could predict which funds will perform well 20 years from now, I would own my own island from which I could wave at Chris Jones.
So it seems reasonable that we should model the characteristics of a fund suitable to the client’s risk profile, rather than model the fund itself, as this is far more likely to be relevant so many years into the future.
It is inevitable, and correct, that approximations will need to be made in the modelling. It is therefore key to pick a model that has reasonable approximations that ensures the output adds value to the client and adviser interaction.
A modern-day actuary must weigh up the benefits of accuracy (or the potential pitfall of spurious accuracy) against the value added from introducing complexity into any model. It is here that I agree with the US Navy and the KISS principle, which states that most systems work best if they are kept simple rather than made complicated; therefore, simplicity should be a key goal in design, and unnecessary complexity should be avoided.
As an actuary, I am bound by the Actuaries Code and required to adhere to Technical Actuarial Standards (TAS) which addresses modelling specifically. TAS 100 requires models to be fit for the purpose for which they are used, be subject to sufficient controls and testing, and be adequately documented.
Any model that has been signed off by an actuary, such as our MonteCarlo forecaster here at Dynamic Planner, should give you confidence that the model you are using is a realistic projection of future events and can be relied upon to help build a financial plan for your client. There is also thorough documentation available to help you understand how and why the model is working as it is.
A stochastic model can answer a lot of questions for you, but as with all modelling, the answer you receive is only as valuable as the model in question. Keep it simple.
Steph Willcox is head of actuarial implementation at Dynamic Planner