There has been some discussion in the FT over the last week regarding the issue of longevity bonds. David Blake proposed that the Government could raise capital by issuing longevity bonds whilst Paul Dare proposed that these bonds could be sold direct to individuals.
Let’s look at the case for selling longevity bonds to individuals first. I’m not convinced. There’s already a healthy competitive market for pension annuities, a market where competition is leading to increasingly accurate pricing. Why should the Government want to join this market? It’d be more helpful if it were to issue a greater proportion of long-dated Gilts, since the shortage of securities at this duration results in bond – and annuity – prices being artificially high. Should the Government instead be offering non-pension annuities? If it did, it would find the take-up disappointing. Whilst around 450,000 pension annuities were sold in 2008 (to individuals with little choice but to buy them), only 800 non-pension annuities were sold in the same period. When given a choice, people do not want to buy annuities.
There is a demand for longevity protection from the institutions that hold longevity risk, in particular pension schemes and insurance companies, but are Government longevity bonds the answer? Such bonds would need to be based on the mortality experience of the general population, limiting the extent to which they could reduce risk for any individual institutions. (There is some more research on issues with longevity indices here.) Longevity bonds are also capital-intensive – to hedge the risk you need to buy the bond. More importantly, though, does the Government really want to take on more longevity risk? It already has responsibility for over £1,000 billion of State pension liabilities and the longevity risk that they contain.
Longevity markets are still in their infancy and they need help to develop, for the sake of institutions rather than individuals. However, the source of that help should not be the Government – leave it to the market.