Start with my disclaimer: I don’t know for sure. Buffett says that he didn’t know about the details, and certainly didn’t approve of the deal. From the Dow Jones Newswires:

Buffett said in a 2005 statement that he “was not briefed on how the transactions were to be structured or on any improper use or purpose of the transactions.”

Buffett’s attorney, Ronald Olson, said in a recent statement that Buffett “denies that he passed judgment in any way on the challenged AIG/Gen Re transaction in November 2000 or at any other time.”

Personally, I find this amazing for a few reasons. 1) In any dealings with AIG, a smart insurance executive would want to know what was going on. AIG has had a history of getting the better end of the deal in working with reinsurers. Buffett is not dumb, and there had been a decent amount of rivalry between the two companies over the years. 2) Buffett was not “hands off” on the insurance side of the house when it came to large insurance contracts. From his 2001 Shareholder Letter (page 8 ):

I have known the details of almost every policy that Ajit has written since he came with us in 1986, and
never on even a single occasion have I seen him break any of our three underwriting rules. His extraordinary
discipline, of course, does not eliminate losses; it does, however, prevent foolish losses. And that?s the key: Just as
is the case in investing, insurers produce outstanding long-term results primarily by avoiding dumb decisions, rather
than by making brilliant ones.

Now, maybe Buffett was overstating the case of how much he knew about what Ajit did. It is clear that he spent more time with Ajit than the managers at Gen Re, but I find it difficult to believe he didn’t review a major contract of a client who was also a major competitor known to be tough reinsurance negotiator.

3) He understands finite insurance very well. From this article of mine at RealMoney about the 2004 Shareholder letter, my last point:

12) Finally, what was not there: a discussion of Berkshire’s activities in the retroactive (or retrocessional or finite or financial) reinsurance business. This is notable for two reasons: first, in 2003, he split out the retroactive reinsurance in order to give a clearer presentation of the insurance groups operating results. This year the data is only presented in summary form. Second, Buffett made a big positive out of the retroactive reinsurance results, going so far as to explain the business in both the 2000 (page 8 ) and 2002 (page 9) shareholder letters.

Now, to varying degrees, Buffett made effort over the prior four years to explain the profitability of Berky’s retroactive reinsurance business, because it skewed the loss ratios of Berky upward. In the 2004 Shareholder letter, it was too much of a hot potato to give similar coverage to, even eliminating the entries that would have allowed one to see the accounting effect. In 2000 and 2002, he gave mini-tutorials on the business. In 2000 (page 8 ):

There are two factors affecting our cost of float that are very rare at other insurers but that now loom large at Berkshire. First, a few insurers that are currently experiencing large losses have offloaded a significant portion of these on us in a manner that penalizes our current earnings but gives us float we can use for many years to come. After the loss that we incur in the first year of the policy, there are no further costs attached to this business.

When these policies are properly priced, we welcome the pain-today, gain-tomorrow effects they have. In 1999, $400 million of our underwriting loss (about 27.8% of the total) came from business of this kind and in 2000 the figure was $482 million (34.4% of our loss). We have no way of predicting how much similar business we will write in the future, but what we do get will typically be in large chunks. Because these transactions can materially distort our figures, we will tell you about them as they occur.

Other reinsurers have little taste for this insurance. They simply can?t stomach what huge underwriting losses do to their reported results, even though these losses are produced by policies whose overall economics are certain to be favorable. You should be careful, therefore, in comparing our underwriting results with those of other insurers.

An even more significant item in our numbers ? which, again, you won?t find much of elsewhere ? arises from transactions in which we assume past losses of a company that wants to put its troubles behind it. To illustrate, the XYZ insurance company might have last year bought a policy obligating us to pay the first $1 billion of losses and loss adjustment expenses from events that happened in, say, 1995 and earlier years. These contracts can be very large, though we always require a cap on our exposure. We entered into a number of such transactions in 2000 and expect to close several more in 2001.

Under GAAP accounting, this ?retroactive? insurance neither benefits nor penalizes our current earnings. Instead, we set up an asset called ?deferred charges applicable to assumed reinsurance,? in an amount reflecting the difference between the premium we receive and the (higher) losses we expect to pay (for which reserves are immediately established). We then amortize this asset by making annual charges to earnings that create equivalent underwriting losses. You will find the amount of the loss that we incur from these transactions in both our quarterly and annual management discussion. By their nature, these losses will continue for many years, often stretching into decades. As an offset, though, we have the use of float — lots of it.

Clearly, float carrying an annual cost of this kind is not as desirable as float we generate from policies that are expected to produce an underwriting profit (of which we have plenty). Nevertheless, this retroactive insurance should be decent business for us.

The net of all this is that a) I expect our cost of float to be very attractive in the future but b) rarely to return to a ?no-cost? mode because of the annual charge that retroactive reinsurance will lay on us. Also ? obviously — the ultimate benefits that we derive from float will depend not only on its cost but, fully as important, how effectively we deploy it.

Our retroactive business is almost single-handedly the work of Ajit Jain, whose praises I sing annually. It is impossible to overstate how valuable Ajit is to Berkshire. Don?t worry about my health; worry about his. Last year, Ajit brought home a $2.4 billion reinsurance premium, perhaps the largest in history, from a policy that retroactively covers a major U.K. company. Subsequently, he wrote a large policy protecting the Texas Rangers from the possibility that Alex Rodriguez will become permanently disabled. As sports fans know, ?A-Rod? was signed for $252 million, a record, and we think that our policy probably also set a record for disability insurance. We cover many other sports figures as well.

And 2002:

Ajit Jain?s reinsurance division was the major reason our float cost us so little last year. If we ever put a photo in a Berkshire annual report, it will be of Ajit. In color!

Ajit?s operation has amassed $13.4 billion of float, more than all but a handful of insurers have ever built up. He accomplished this from a standing start in 1986, and even now has a workforce numbering only 20. And, most important, he has produced underwriting profits.

His profits are particularly remarkable if you factor in some accounting arcana that I am about to lay on you. So prepare to eat your spinach (or, alternatively, if debits and credits aren?t your thing, skip the next two paragraphs).

Ajit?s 2002 underwriting profit of $534 million came after his operation recognized a charge of $428 million attributable to ?retroactive? insurance he has written over the years. In this line of business, we assume from another insurer the obligation to pay up to a specified amount for losses they have already incurred ? often for events that took place decades earlier ? but that are yet to be paid (for example, because a worker hurt in 1980 will receive monthly payments for life). In these arrangements, an insurer pays us a large upfront premium, but one that is less than the losses we expect to pay. We willingly accept this differential because a) our payments are capped, and b) we get to use the money until loss payments are actually made, with these often stretching out over a decade or more. About 80% of the $6.6 billion in asbestos and environmental loss reserves that we carry arises from capped contracts, whose costs consequently can?t skyrocket.

When we write a retroactive policy, we immediately record both the premium and a reserve for the expected losses. The difference between the two is entered as an asset entitled ?deferred charges ? reinsurance assumed.? This is no small item: at yearend, for all retroactive policies, it was $3.4 billion. We then amortize this asset downward by charges to income over the expected life of each policy. These charges ? $440 million in 2002, including charges at Gen Re ? create an underwriting loss, but one that is intentional and desirable. And even after this drag on reported results, Ajit achieved a large underwriting gain last year.

What I am trying to point out here is that Buffett had significant knowledge of the retroactive (finite) deals at Berkshire Hathaway. He was even somewhat proud of them, though perhaps that is a matter of interpretation. He liked the almost riskless profits that they provided.

Before I move onto my last point, I’d like to digress, and simply say that not all finite reinsurance is a matter of accounting chicanery. The key is risk transfer. Without risk transfer, regardless of what the technical accounting regulations might say, there should be no reserve relief granted, regardless of the amount of money given to the cedant by the reinsurer; that money should be treated as a loan, because it will have to be paid back with interest. With full risk transfer, the company ceding the risk should not have to hold any reserves for the business. In between, the amount of reserve credit is proportional to the amount of risk shed; excess money given to the cedant by the reinsurer should be treated as a loan. Economically, that’s what it should be, even though that is not what always happens in the accounting. (Side note: yes, I know that it is difficult to determine the amount of risk shed, and different actuaries might come to different conclusions, but can’t we at least agree on the underlying theory?)

What has happened is that in many cases, little risk is shed, and a full credit for risk reduction is taken. Sometimes FAS 113 would be followed, with its 10% chance of a 10% loss as a miserably low tripwire for risk transfer. Sometimes FAS 113 would get bent, and other times, badly bent. That brings me to point 4.

4) Berky had a lot of experience with many different types of finite insurance. I remember a notable asbestos contract they took on for White Mountains where they would bear a large amount of risk. (On that one, I think White Mountains got the better end of the deal.) There were others, like the finite contract with Australian insurer FAI, which made them look solvent while experience was deteriorating. HIH bought FAI, and later went bankrupt, partly due to the acquisition. There were other finite reinsurance deals, like Reciprocal of America, where it made a company that was insolvent look solvent.

I can argue that in many cases, Berky’s underwriters did not know the accounting treatment that the cedant would use, and could not be responsible for the troubles that followed. In many cases, Berky bore significant, if limited, risk. That’s fine too. The greater question is if they were a large writer of finite coverages, which they were, they would have to have some knowledge of the cedant’s goals if they were to underwrite properly. Also remember that Buffett watches the “float” that his insurance businesses generate like a hawk. If there was a large amount of float that would come from a new contract, he likely would have known about it.

The AIG contract was big. AIG is a tough reinsurance negotiator. AIG and Berky have been rivals (Greenberg insulted Buffett on at least one occasion). Buffett watches underwriting carefully, even that of his trusted lieutenant Ajit Jain (a nice guy, really). That makes it really hard for me to believe that Buffett did not have any significant knowledge of the AIG finite reinsurance contract. In the end, I really don’t know; I’m only guessing. My guess is this: Buffett had general, but not detailed knowledge of the deal with AIG. In my estimation, he probably checked to see that there were adequate risk controls to make sure that AIG was not getting too good of a deal.

I admire Buffett. I have learned a lot from him. In general, compared to most businessmen, he is an honest and open guy who speaks his mind. If he said that he never had any significant knowledge of the contract with AIG, we should give him the benefit of the doubt, maybe. But from my angle, it is inconsistent with the way he has done business generally.

Tickers mentioned: AIG, BRK/A, BRK/B, WTM

PS — If you ask me how I feel about writing this, I will tell you that I am not crazy about what I have written. I’m not after publicity for criticizing a man that I admire greatly. I think that Buffett should be more forthcoming on the topic, and be willing to be a witness in the trial. Five people are facing ruined lives, and if Buffett really knew about it, and is saying nothing now because he is powerful enough to get away with it, well, shame on him. If he didn’t know anything about it, well, his testimony would clear the air, because it is a distraction at the trial.