ALL MODELS ARE WRONG …

I was reading Amy Kessler’s excellent article in Professional Pensions this morning ‘Understanding the limitations of the CMI longevity model’.

Amy is right to remind us that the model has limitations. All models do; they are necessarily and by definition, simplifications of reality. (For the time being anyway until we can run simulations of reality within the simulation of reality that we all ‘exist’ if you subscribe to the hypothesis popularised by Elon Musk recently…). I digress.

So let’s just remind ourselves for now that all models are wrong and that some models are useful.

Faced with this certainty, the best that we can do is to familiarise ourselves with a range of models – and their limitations – so that we can understand the outcomes that apply to us under different simplifications of the world.

That is how we like to work with our clients. Discrete blocks of functionality that you can drop into a system and observe the impact. This can be done narrowly (changing parameters) or more broadly (changing models). Build a picture.

The final argument in the article is interesting. That longevity is an unrewarded risk… and so you should hedge it… Hmm, possibly, but Amy is the head of longevity risk transfer at Prudential Financial and very well meaning I am sure but we probably shouldn’t take that on blind faith.

Surely that depends on the price.

‘Unrewarded’ means, I would say, that you can remove the risk without increasing the expected cost. Now this statement only holds true if the market price of removing a risk matches [your] best estimate cost. Which is likely to require a well-functioning market of willing buyers and sellers.

It will be great for pension schemes if we are there. Are we? Our advice would be to review the impact of longevity solutions on the expected cost of funding your pension scheme under a range of models to build up a picture of the likely expected cost or saving. But then we do provide a range of longevity models so we would say that wouldn’t we…