Skip to main content

Insurance Companies Were Bad At Predicting How Dumb Their Customers Would Be

  • Author:
  • Updated:

Retail clients are not typically paragons of rationality or possessors of Black-Scholes calculators so a good way to make money is to bamboozle them with mispriced derivatives. The classic way banks do this is with structured notes, where you combine a bond worth $75 and an S&P option or whatever worth $15 and sell the combination for $100 because who has time to check your math, really. In the early years of this century life insurance companies came up with a clever variation on this idea. The variation was:

  • Combine a bond worth $75 and an S&P put option worth $15.1
  • Sell the combination for like $80.
  • Hope everyone forgets about it.

This was an amazing plan. You can see why it sold well? You can also see why it did not really work, for the insurers? It totally totally did not work, for the insurers, and yesterday Moody's issued a report about it saying basically "a lot of life insurers are kind of fucked because of this," which, sure, but what were they expecting?

Here's what they were expecting:

That's from the Moody's report. Basically the insurers knew that, the worse the market performed, the more in-the-money their sold put would be, so the more valuable the contract would be, so the more likely policyholders would be to keep the contract alive by continuing to pay premiums. They also knew that some customers would forget, or not be able to come up with the money one month, and so would let the contract lapse. They just thought that more customers would forget than actually did. (Similarly: they knew that if the market went up, the contracts would be worth less, and so more policyholders would cancel their contracts. They just thought that more customers would forget to cancel than actually did.)

The insurers didn't screw up the derivatives: they had Black-Scholes calculators and hedged their market exposure just like anyone else would have. As Moody's says, "Though equity market declines are generally seen as the biggest risk in VA contracts, most insurers effectively hedge that risk via derivatives." They just bought didn't buy enough derivatives, because they screwed up "the less-easily hedged and more unpredictable policyholder behavior, and particularly lapses," predicting that more policyholders would make uneconomic decisions than actually did.

There's a part of me that wants to say this was a clever product in conception if not in ultimate result? The thing is that no bank would sell an unhedged structured note, or vanilla call option for that matter, meaning that the typical retail client can only get downside protection on the market by overpaying for it. But insurance companies' comparative advantage is that they can take an actuarial approach to risk: if you can reliably predict that 50% of customers will let their valuable puts lapse, then you only need to charge 50% as much for a put. So you can beat fully hedged banks on pricing and take market share away from them. As insurers did.

The problem comes when you can't reliably predict what you're trying to predict. That occurred.2 Perhaps strangely, the insurers erred on the side of lower prices. Badness ensued.

The last time we talked about variable annuities was in the context of the glorious Joseph Caramadre, who found some dying people to buy these contracts so he could wildly underpay for the puts and clip a lot of free money.3 What I said then was that retail financial services are built around the expectation that customers will make uneconomic decisions. Those decisions can be as simply uneconomic as overpaying you for a structured note, or they can be things like incurring overdraft fees. If those uneconomic decisions are stable and predictable, and if you can predict them, then you can turn them into a pretty nice revenue stream. If you can't, then you end up making some pretty uneconomic decisions yourself.

Misjudged Annuity Guarantees May Cost Life Insurers Billions [WSJ]
Unpredictable policyholder behavior challenges US life insurers' variable annuity business [Moody's]

1.Ooh I feel bad about this one, that's not right at all. Reeeeeally you're selling, like, a combination of the S&P and a put on the S&P. (Or, like, bonds and a put on those bonds, or whatever.) I'm just trying to abstract away from that to draw out the comparison to structured notes. In any case the thing that matters is the put - in the form of a guaranteed minimum withdrawal, etc. - which is the thing that lapses or does not lapse depending on what the policyholder does.

2.Moody's again:

Virtually all insurance products have the risk that actual experience will deviate from pricing. For example, life insurers have pricing risk with respect to mortality on term insurance, morbidity on long term disability and longevity on payout annuities. However, for mortality, morbidity and longevity risk, actuaries have vast experience data to draw upon and a benefit payout not generally under the control of the policyholder. Thus, these types of risks are more predictable and can be priced within a reasonable range of expectations. What differentiates variable annuities with guaranteed living benefits is that ultimate profitability—unknown until many years after pricing—is very sensitive to policyholder behavior.

Policyholder behavior can vary widely, especially under different circumstances, and policyholders may or may not behave as efficiently as pricing actuaries or investment professionals would assume. Predicting the degree of efficiency in policyholder behavior and incorporating it into a product’s pricing can therefore be extremely difficult. The variable annuity guarantees in question are still relatively new, at least in life insurance industry terms, and insurers lack sufficient historical policyholder experience to guide pricing with great certainty.

3.Some Googling reveals that he's now in jail for it. Oh well.