[**UPDATE:** Thanks, David, for your reply in the linked comment. This is just to note that, in my original linked blog post to which Sam also links, I was also careful to characterise your 1 in 20 claim as regarding the failure to receive one’s *promised* pension. I wrote: “This fear that ‘self-sufficiency’ leaves members with a 5% chance of failing to receive their promised pensions is unfounded.” I agree that you didn’t claim that there was a 1 on 20 chance of receiving no pension whatsoever. Rather you were claiming that there was a 1 in 20 chance on not receiving a pension up to the promised level.]

Regarding Sam’s fourth point in his reply below, which bears on David Miles’s scepticism over whether USS’s prudent adjustment from 50% to 67% is too small, what Guy Coughlan says at min 33:30 to the end of the ‘A technical overview of the 2017 valuation’ video — especially the convergence of their modelling on very different Brownian motion modelling — should provide some assurance that their adjustment is credible:

In this first example (33:30), we run the ORTEC model forward over a 30-year period using 2,000 different scenarios for different market environments. What you see in 30 years’ time is that you’ve got a very skewed distribution and that the average is much higher than the median.

If we look at a similar simulation (34:00) over a 30-year period but we use a completely different stochastic model for example, Brownian Motion, what we end up with is a distribution which is shaped very similar to that which we’ve used in the ORTEC model. It is skewed in the same way and it has a very similar mean and median from the ORTEC model.

So the details from the actual simulation model, when appropriately diversified, and when used with linear financial instruments (i.e. without options) gives you a very similar outcome.

Above quotation taken from USS’s pdf transcript. The accompanying graphs indicate the 67th percentile from the modelling as well as the median and the average.