2003: Why are bond insurers rated AAA like Berkshire despite their 140-to-1 leverage?

AUDIENCE MEMBER: Bill Ackman from New York City, and my question is as follows:

Insurance companies — could you comment on insurance companies taking on credit risk through the sale of credit derivatives, the adequacy of the accounting for these derivatives?

And finally, could you explain why the financial guarantee insurers, who are the primary sellers of these derivatives, have the same triple-A rating Berkshire has, despite their more than 140-to-one leverage, and the correlated nature of the risks that they take on?

WARREN BUFFETT: Well, I think you should go to work for Standard and Poor’s or Moody’s.

The question about credit insurance or credit guarantees of one sort or another, you know, that’s become very popular.

And it’s become — actually, popular with, sort of, the standard insurance, property-casualty insurance, companies in recent years.

And I would say that, in many cases, the people participating in that business don’t really know what they’re doing.

It’s so easy in the insurance business — it’s the curse of the insurance business — it’s also one of the benefits of it — is that people hand you a lot of money for writing out a little piece of paper.

And what you put on that piece of paper is enormously important. But the money that’s coming in that seems so easy can tempt you into doing very, very foolish things.

We had a situation here in Omaha 15 or 20 years ago in the mid-’80s where Mutual of Omaha — largest health and accident association in the world, at least at one point — and they decided to go into the reinsurance — property-casualty reinsurance business.

And in a very, very, very short time they wrote not very many contracts, and it resulted in wiping out half of the net worth of everything that had built up over many, many decades.

If you are willing to do dumb things in insurance, the world will find you.

I mean, you do not — (Laughter)

You can be in a rowboat in the middle of the Atlantic and just whisper out, “I’m willing to write this,” and then name a dumb price, and you will have brokers swimming to you, you know — (laughter) — with their fins showing, incidentally. (Laughter)

It is brutal. I mean, if you are willing to do dumb things, there are people out there, and it’s understandable. But they will find you, and you will get the cash up front.

You will see a lot of cash and you won’t see any losses, and you’ll keep doing it because you won’t see any losses for a little while. So you’ll keep taking on more and more of this, you know, and then the roof will fall in.

And I mentioned in the annual report how GEICO had taken in, you know, 70-odd thousand dollars — $70,000 — of premiums in the early ’80s for a few policies, and they thought they were just picking cherries at the time, and they reinsured a lot of it. And so far we’ve lost $93 million.

Now, the most we could make was 70-odd thousand, and I don’t know what the most we can lose is. But I know that 93 million has gotten my attention. (Laughter)

When you’re playing in a game like that, you can’t afford to make a mistake. I mean, it’s — the mistake — a single mistake or a few mistakes that are correlated, as you’ve mentioned — because these things do correlate — a few mistakes will overcome a lifetime of savings.

I mean, it is — you will make a few cents on the dollar when you’re right, and you will lose incredible sums when you’re wrong.

And in credit insurance, when you go around — a lot of people went around guaranteeing credits based on ratings.

And they said, well, we’ll guarantee a whole bunch of single-A ratings, or we’ll create these structured arrangements that involve A-rated credits.

And they would use a lot of studies that would show that X percent of A-rated credits defaulted per year, and you go back to the ’30s, and all these back-tested arrangements.

But the problem with that is that what the questioner mentioned is correlation. And when things go bad, all kinds of things correlate that no one ever dreamed correlated.

And what you had, of course, in the debt field was you had a whole bunch of, say, telecoms or energy companies, that were all rated similarly. But they were correlated in a huge way and you weren’t getting a diversification. You were getting a concentration that would — you didn’t realize.

And there’s nothing more deadly than unrecognized concentrations of risk, but it happens all the time.

So I would say we see a B-double-A credit enhanced to a triple-A credit by somebody guaranteeing it, and they may guarantee it for 10 or 15 basis points. And yet the spread in market yield might be 100 basis points.

That does not strike us as smart.

And I would say this about the triple-A rating. They have a triple-A rating for claims paying, but they don’t carry, I don’t think, general triple-A.

There’s only, I think, eight or nine triple-As left in the United States. Berkshire Hathaway’s one of them. But I believe there’s only one other insurance company, which is AIG, and then there’s a half a dozen other companies.

So those companies are not in the same class, credit-wise, as Berkshire, nor are they recognized as being in the same class.

But I would say you could get into a lot of trouble at 140-to-one — at some point — insuring credits.

Charlie?

CHARLIE MUNGER: Yeah, he also asked about the quality of accounting. And in my view, at least, the quality of accounting in America for derivative transactions is still terrible. And it’s terrible in that it’s too optimistic.

And one of the places where it’s most terrible is when you talk about guaranteeing future credits way, way out — years ahead.

That sort of thing just lends itself to people getting very optimistic in their assumptions and in their audited figures.

And people pay attention to the audited figures, not the underlying reality. So therefore, if the accounting is lousy, the business decisions are lousy. And I think that’s going on mightily as we sit here.

WARREN BUFFETT: Yeah, there are dozens of insurance organizations or trading organizations in the country that have written credit guarantee contracts in derivative form in the last few years, in fact, on a huge scale.

I will guarantee you, virtually, that every single one of those contracts that was written, in the first week, whoever wrote it, you know, recognized some sort of an income account or an income entry, and that somebody got paid a little bit of money for writing each one of them.

And you know that many of those are going to go bad, and maybe, as a category, that it’s going to be a terrible category. But nobody ever wrote a contract and recorded a loss at the time they wrote it. I mean, they just don’t do it.

And I will tell you that there are a lot of those contracts that if somebody wrote them for me, 10 seconds later I would’ve paid somebody to take them off my hands, so that I would’ve regarded them as having a built-in loss. Nobody ever records a built-in loss on a derivative contract.

In fact, I find it extraordinary that you have two derivative dealers, and dealer A and dealer B write a ticket, and dealer A records a profit and dealer B records a profit, you know, particularly if it’s a 20-year contract, you know? I mean, that is the kind of world I’d love to live in, but I haven’t found it yet.