For whatever reason, American International Group (NYSE:AIG) doesn't generate that much interest among individual shareholders. It's one of the largest companies in the world, with a market capitalization that has exceeded $200 billion; it is certainly the largest insurance company in the U.S.; and it turned in simply phenomenal rates of growth over the last 40 years. Until recently, AIG's CEO was Maurice "Hank" Greenberg, every bit the full-contact capitalist that his more celebrated insurance executive brethren, Warren Buffett of Berkshire Hathaway (NYSE:BRKa) (NYSE:BRKb) fame, is.

And though Greenberg's public profile has been substantially lower than Buffett's (and whose isn't?), the returns that AIG generated for shareholders since he took over in 1967 had been, at a minimum, in the same ballpark. That's why it ought to be more shocking and alarming than it seems to be that AIG's board stepped in and compelled Greenberg to resign as CEO of the company. This wasn't done over his failure to sign expense reports or something procedural -- the problems facing AIG are enormous. They can't not be.

Still, the amount of navel-gazing and rune-reading that's taken place over the last week regarding AIG has been minimal. So make the same translation: What would the hue and cry have been had the Berkshire board felt compelled to remove Warren Buffett as operating head of the company? When you get down to brass tacks, the level of control each one held over his company was not dissimilar. Greenberg's defenestration, while surprising, left neither investors nor the general public dumbfounded the way such an outcome would with Buffett.

I have some theories here, and they go back to something I've said for years: That which cannot go on forever, won't. It's hardly an original thought (and actually, I'll send an Apolo Anton Ohno fanzine I found at my house this weekend to the first person who emails me correctly identifying the source of this quote). AIG turned in year after year of above-market growth, with below-market volatility of results. Everything at AIG seemed to come in right at prediction, without much volatility.

Surprise! Everything's as we said it would be
But this is insurance we're talking about. If there is a business out there with more volatility, more variation of results from year to year, I'm having a hard time thinking what it might be. Pixar (NASDAQ:PIXR), with its feast-or-famine history with movie releases, maybe. But you knew that was coming: No movie release equaled lower revenues. Insurance works differently: You have a steady stream of revenues coming in, but surges and swoons in expenditures and investment results are unpredictable in timing, duration, and severity. During all of that, AIG looked to all the world like a duck on a pond.

As we are now finding out, there may have been plenty of paddling underneath to deliver that level of smoothness and predictability. The possibly sham finite insurance deal involving Berkshire subsidiary General Reinsurance, which allowed AIG to deliver its predicted results to Wall Street in 2000 and 2001, may have been the last straw, particularly given the fact that Greenberg himself was the instigator who contacted General Re to make the deal, which allowed AIG to report $500 million more in insurance reserves. This comes on the heels of several other scandals, including AIG's helping both Brightpoint (NASDAQ:CELL) and PNC Bank (NYSE:PNC) with accounting woes of their own. But this couldn't possibly have been it.

An aside: As financial commentators and (yes, it's true) politicians have tended to do in the past, we have seen a witch hunt surrounding finite insurance in the last several weeks and the apparent ease with which it can be used to diddle corporate financial statements. Let's be very clear: While the use of finite insurance may be a tip-off that financial shenanigans are going on, this is not a one-to-one correlation. Finite insurance where there is a definitive transfer of risk is perfectly legal. What's going on now is a search into the internal communications at AIG and GenRe to determine whether people on either side knew or believed that this particular policy was a sham. Heaven help them if prosecutors find such evidence.

I've started to receive mail from people discussing a column I wrote three years ago that delved into the fact that there are two separate private entities that are controlled by AIG executives that in turn hold large stakes in AIG. I pointed out that such a structure meant that AIG shareholders had no way of knowing how much executives received in compensation for doing their jobs. At the time I wrote the article, I received about five comments, the majority of which were from people at AIG defending the structure.

But it wasn't the structure that caused me fits; it was the inscrutability that the structure and its lack of disclosure represented. And the moment I figured this out, the combination of the lack of disclosure on executive salaries and the company's preternatural tendency toward predictability informed me that the odds were high that governance problems would eventually take a big bite out of AIG. And now we are there, and evidence suggests that what has been disclosed so far is the outer layer of the onion.

That smoothness comes at a cost
The inner layers of the onion look a whole lot like Coral Re from a decade ago. Coral Re was a reinsurer based in Barbados that three state regulators claimed in 1996 was little more than a shell company to which AIG had sent more than $1 billion in business. Regulators claimed that Coral Re was controlled by AIG, and as such, any insurance liabilities that AIG transferred to Coral Re should still be reflected on its balance sheet. AIG quit doing business with Coral Re, which disappeared soon after.

But an absolutely fascinating article in this week's Barron's invokes Coral Re's name once again, disclosing that investigators have uncovered material that suggests AIG had simply transferred its Coral Re business to two new reinsurance companies that it actually controlled. The same article describes how Goldman Sachs (NYSE:GS) came to the original Coral Re investors, got them to sign up as the investors in the company, got them risk-free non-resource loans, and paid them an annual profit that just exceeded the level of debt servicing. It was risk-free for them, a sweet deal. None of these investors knew a thing about reinsurance, nor did Coral Re ever have any shareholder meetings of any kind.

How hard would it have been to replicate such a deal? Apparently not hard at all, as the Barron's article details that investigators are looking into two offshore reinsurance companies that apparently look dramatically similar to Coral Re. So while regulators and investigators are looking into transactions they consider shams, it could be that the level of deception runs a great deal deeper than that.

This was my overriding fear with AIG, the thing that started me down the road of being highly skeptical of the company, though I was dealing in hunches and hints rather than hard evidence. The impossible is just that: impossible. Insurance isn't a smooth business; it isn't overly predictable; it has substantial pricing cycles that can be brutal. The very smoothness that AIG prided itself on told me that there was something very, very wrong. I just didn't quite know what. Sounds like we're all going to find out soon enough just what it was.

Bill Mann holds a stake in Berkshire Hathaway. He attributes his own smoothness to a longtime predilection toward listening to Parliament/Funkadelic. The Motley Fool has a disclosure policy.