A new approach, which I believe helps provide a much more intuitive framework for decision-making and that focuses on a more long-term measure than the predominantly short-term measures currently employed, should be considered. The approach recommends carrying out asset-liability modeling on thousands of different scenarios for the entire life of the plan (i.e., until the last payment is made). It then looks at how successful the plan has been at paying all pensions in full and on time across all simulations.
The approach uses a measure of success for a scenario that we call "proportion of benefits met," or PBM. As a simplified example, suppose the trustees have promised payments of £100 a year for the next 20 years. The plan's assets meet these promises for the first 10 years but then the sponsor defaults, the scheme is terminated, and only 50% of the remaining benefits are met on buyout (the likely scenario in the result of a sponsor default). In this instance the PBM value would be 75%.
Trustees can explore a range of asset allocations and choose the one with the most attractive distribution of PBM. A benefit of this approach is that it recognizes pension plans are long-term investors and ensures the benefit of being able to hold an asset over the long term is considered. This is not the case using shorter term measures of risk and return. For example, with corporate bonds it will focus on the risk of default rather than shorter-term mark-to-market impacts.