A common concern among DB plan sponsors these days is the sustainability of their pension plan. This question typically translates into whether or not the markets will give us what we need to be able to pay for the pension promise.
The market returns are all important, considering the following:
- contributions tend to become a smaller piece of the puzzle over time. As a plan grows and matures, the impact of the investment returns, positive or negative, tends to increasingly outweigh the impact of the contributions; and
- most of the fund returns will be dictated by what the markets gives us. Of course, we can assume away any funding problem by anticipating a large amount of value-added through investment manager outperformance. But believing manager brilliance will overcome the plan’s funding challenges is similar to believing there will be no problem when you jump off a cliff because you have assumed that you can fly.
While a lot of sophisticated modelling can be brought to bear on the question of the affordability of the pension promise, there are a number of general questions to which we can provide general answers, even before the modelling begins.
Can we be guaranteed that we won’t have to pay significantly higher contributions?
No. The volatility of the capital markets is quite large, so the range of potential outcomes from any model of future investment returns will also be very large. Any realistic stochastic model is going to produce a certain number of disaster scenarios. When we combine a large range of possible outcomes with a highly leveraged financial situation (i.e. large pension assets relative to the plan sponsor’s available cash flows), we can often end up with an uncomfortably high probability of ruinous future contribution requirements.
Is there any chance that the plan is affordable?
Yes. For the same reason that we cannot guarantee affordability we can also not say with certainty that the plan won’t be affordable in future. In fact, some extreme scenarios will likely result in run-away surplus.
Will a liability-driven investment approach help to ensure affordability?
Probably not. This will, of course, depend on the definition of affordability. Theoretically, if the plan sponsor could afford to lower the going concern liability discount rate to the long bond yield, and fund the plan on this basis, then it might be possible to lock in this level of affordability, ignoring any inflation-linked elements like a final-earnings plan design. But for any plan that is still a going concern and needs to earn a higher return than 2.5% to be “affordable,” moving closer to solvency or accounting liabilities is not really “de-risking,” because it increases the risk of not achieving a high enough return.
What will the model tell us?
What might happen. When we set about to predict future investment returns, the only thing we know for certain is that our prediction will be wrong. The odds of completely accurately predicting future investment performance are infinitesimally small. We may have the most sophisticated stochastic model ever created, but the outcomes can only be as accurate as the assumptions, so greater mathematical sophistication will only take a vaguely wrong answer and make it precisely wrong.
So the important information coming out of a model will be the range of potential outcomes, not any specific expected outcome.
Beyond the frustratingly large range of possible future results, a good asset-liability model can provide insight into the various factors that will affect the plan’s future funding levels, the relative importance of these factors, and how they interact with each other. This information can provide a good basis for managing the plan. For this reason alone, the time and resources required to construct a good model might be worth the cost.
But for the model to be really useful, it should not be treated like some arcane ritual to be performed every five or 10 years and then forgotten. Going forward, it is important to evaluate actual results as they unfold relative to the predictions made by the model. In this way, the model can provide insights into these results, and the results can provide insights into the efficacy of the model.
The work of determining an appropriate asset mix policy should not end with the output from the model, because any answer given will be far from definitive, and dangerous if followed blindly.