Simon Wren-Lewis points out a paradox - that on the one hand we have good evidence that older people tend to be more risk averse than younger ones, but on the other hand they supported the risky prospect that is Brexit. How can this be?

It's not - or at least not wholly - because older people, being retired, are insulated from some of the costs of Brexit. People tend to vote sociotropically (pdf) - for the common good as they perceive it rather than their own narrow interest - and so older folk should have considered the interests of their children and grandchildren.

What's more, there's a good reason why older people should be more risk averse. They have learned - often at that most effective pedagogic establishment, the school of hard knocks - that big ambitions often fail. They should be more aware than most of our cognitive limits, not least because many older folk are so limited (pdf). They should be small-c conservatives, and hence Remainers.

So why weren't they? Part of the answer, I suspect, lies in a paper (pdf) by Michael Woodford and colleagues. They point out that attitudes to risk are shaped by how we code (pdf) prospective wins and losses.

Take, for example, the question: how much would you pay for a 50-50 chance of winning £1000? The expected value of the bet is £500, but most of us would pay less because we're risk averse.

Just how much we'd pay, though, depends on how we think about the win and the loss - how we code them. We ask: what could I do with £1000? How much would the loss of (say) £400 hurt? Differences in these codings generate differences in how much we'd pay - that is, differences in risk aversion.

This seems trivial. But it explains a lot. It explains why more intelligent people are more willing (pdf) to take on good bets. They can more clearly translate pay-offs into prospective mental well-being, whereas for others the coding is noisier and so they avoid such bets. It also helps explain why older folk are more risk averse: knowing their cognitive limits (and perhaps knowing that money doesn't buy happiness) they avoid bets that others would take.

It also explains one element of prospect theory; the tendency to seek risks when you've lost money. We see the chance of £1000 and think "yay, I can get even" and think of the loss as "I'm in so much trouble already a little extra won't make any difference."

And it also explains why our attitudes to risk vary across domains: why we hold equities and buy insurance; why some avoid financial risk but play risky sports; and so on. It's because the codiings differ from domain to domain.

Which brings me to Brexit. Brexit was not presented simply as a choice between a risky pay-off (Leave) and a safe one (Remain). Brexiteers also offered what they claimed to be intrinsic goods such as national self-determination. For various reasons - being brought up on 1950s war films, discomfort with immigration, whatever - these goods appealed more to oldsters than youngsters. And this appeal offset the tendency for oldsters to regard Brexit as risky.

In a sense, this is consistent with the coding view of risk aversion: oldsters saw Brexit as the offer of valuable intrinsic goods. They therefore coded it as a safe option. Youngsters, being less attracted to those goods, saw it as riskier.

I offer this only as a theory. But I think Simon's point is sound: there is a paradox here that needs some sort of explanation, and this is the best I have.