On Saturday I said, without spelling things out, that all insurance companies are leveraged financial institutions. This resulted in pushback from Cap Vandal:
Property casualty companies don’t tend to use leverage, unlike some life companies. They fund assets using their own capital and policyholder funds. Nothing close to the type of leverage used by banks or even some of the life insurers. Berkshire with huge amounts of excess capital to support their insurance operations and billion in cash and cash equivalents is remarkably unleveraged.
The question here is really how you think of leverage. If you think of it as borrowing money, then yes Berkshire — like many insurers — “is remarkably unleveraged”. But I think of leverage as something rather bigger than that. For instance, if you borrow money and buy stocks, hoping that the rate of increase in the stocks will be greater than your cost of funds, that’s classic leverage. If you replicate that trade using call options and no borrowed money, have you really made the leverage disappear?
When I say that insurers are leveraged, what I’m thinking about is their equity-to-liabilities ratio. At a bank, that ratio is normally about 10%, and as we’ve seen, it can erode to zero with alarming speed. At an insurer, if you consider that the liabilities comprise the face value of the policies in effect, then the equity-to-liabilities ratio can be much lower, below 3%. Which makes insurers, on their face, much more leveraged than banks.
That doesn’t mean they’re riskier than banks. A bank which marks its liabilities to market can easily see them fall in value by more than 10%, thereby wiping out its equity. An insurer’s market-linked investments, by contrast, tend to be in the numerator, not the denominator, of the equity-to-liabilities ratio. So in that sense market declines can’t wipe an insurer out in the same way that they can a bank.
Insurers and banks both rely on diversification to shore up their future: if a bank’s loans all soured at once, it would go bust overnight, and if an insurer’s policies all became due simultaneously, the same thing would happen. So banks count on different borrowers defaulting at different times, and insurers count on a lack of disasters. A big hurricane hitting Miami or New York would wipe out many property insurers; a serious bird flu pandemic could do the same thing for life insurers. Berkshire has been quite good at minimizing its event risk — there aren’t many disasters which cause a huge surge of car-insurance claims, for instance — but the fact remains that its contingent liabilities are many multiples of its claims-paying abilities, which is why I consider it leveraged.
Being leveraged is not the same as being risky, and Berkshire’s triple-A credit rating might well be justifiable, if there really is no conceivable way that a lot of its insurance policies will ever be asked to pay out at once. I’m just not sure that we have any way of knowing that: in the past, when people have tried to quantify the benefits of diversification, disaster has resulted.
So it’s worth looking quite closely at triple-A ratings, to see exactly how they’re arrived at. One reader asked me this weekend how asset-backed securities get triple-A ratings, and whether it was always something to do with Gaussian copulas: the answer is no. There are basically three ways that an asset-backed security can get a triple-A rating: diversification, overcollateralization, and wraps.
Ultimately, all three methods, if you follow the string far enough, rely to some degree or other on diversification and a lack of correlation: even the US government wouldn’t take long to default if everybody stopped paying taxes at once. But some sources of reassurance are more reassuring than others: if a company has been consistently profitable for years, is likely to be consistently profitable in the future, and has more cash on hand than it has in debt, for instance, then the chances of that company defaulting on its bonds are slim.
At Berkshire Hathaway, I’m unclear on exactly what its real and contingent liabilities comprise, and I’d be happier if it didn’t have a triple-A rating which it felt a strong need to protect. When you invest money or buy insurance, there’s always some risk that you won’t get the payout you’re expecting. And if anybody tells you that there’s no risk at all, it’s time to get a little suspicious.
MBIA’s New Structure
Extra Credit, Wednesday Edition
Bond Insurer Break-Up