In pension scheme funding, the discount rate assumption is key. Our latest research suggests that the vast majority of schemes set the discount rate used in their funding valuations with reference to the yield on gilts. This seems to hold true, irrespective of the investment strategy the a scheme adopts - the return on all non-bond asset classes are effectively assumed to generate a fixed level of return above the yield on gilts. This is not surprising, as the Pensions Regulator benchmarks the discount rate that schemes adopt against the yield on gilts. However, the “gilts plus” method is not the only approach to setting the expected future return on non-bond assets.
In our report entitled “The discount rate quandary: The impact of different approaches to discounting pension scheme liabilities”, we consider two alternative methods for setting the expected future return on UK equities. The return on other non-bond assets can be derived using similar principles. These expected return assumptions can then be fed into the calculation of the discount rate, taking appropriate margins for prudence.
The two alternative methods we considered produced similar results to the “gilts plus” method up until 2011. However, after the real yield on gilts turned negative, the gilts plus method started to give a much lower expected return on equities than the other approaches, as can be seen from the graph below:
In our report, we consider possible explanations for this divergence between the output from the “gilts plus” model and the other models. We also highlight just how significant accepting this divergence and continuing to use the “gilts plus” method has been for pension schemes and consider the potential implications of adopting one of the alternative methods.
We have also developed a simple model which allows pension scheme managers and their sponsors to very quickly see just how much higher the discount rate might have been at the date of their scheme’s last valuation had they used the “inflation plus” or “intrinsic value” models rather than the “gilts plus” approach.