Arnold Kling  

Downgraded into Bankruptcy?

I Refused to Shake His Hand... Jenkins on CAFE...

The final chapter in the Washington Post's telling of the AIG saga is the most exciting--and the most frustrating. In the end, it is still not clear whether there was anything fundamentally wrong with AIG's portfolio. After all their research, and all of their writing, the reporters have failed to get to the bottom of the story. They end up saying,

It seems that as Financial Products ramped up its credit-default swap business, its leaders assumed that its parent, AIG, would always be as strong as it was the day it backed the firm's first big trade in 1987...they had failed to prepare for the possibility of a downgrade in AIG's credit rating.

If that is the story, then it is possible that the actual losses on the AIG credit default swap portfolio could turn out to be zero. Instead, what did them in was the collateral posting that was required because their credit rating was downgraded. The collateral demands multiplied as their credit default swaps went from being deep out of the money to being slightly out of the money.

If that is the story, then the stern-sheriff metaphor, that I first introduced here and later included in my Congressional testimony, looks really apt. Instead of putting money into AIG, the Treasury and the Fed should have told Goldman Sachs to stop grabbing for collateral. Make Goldman wait for actual losses to occur before they make claims against AIG.

Early in the article, the authors write,

Financial Products made its money by selling credit-default swaps only on the super-senior tier. It seemed a safe bet: Cassano once defined super senior as the portion of the deal that was safe even "under worst-case stresses and worst-case stress" assumptions.

With risk-layered mortgage securities, the super-senior tier is protected from the first X oercent of losses, where I believe that X is usually at least 5 percent, perhaps higher. Part of the reason that it has become difficult to calculate exposure is that when you have a pool of mortgages with, say, 3 percent defaults, you don't know whether what remains in the pool are good loans that have demonstrated an ability to survive or just more bad stuff that has not yet surfaced.

The article says that AIG stopped writing credit default swaps on mortgage-backed securities late in 2005. If so, then: (a) they were certainly a lot wiser than Freddie and Fannie, which plunged in to the subprime market right around that time; and (b) the rate of defaults on the loans in the mortgagte securities ought to be a lot lower than the default rates we are seeing on the 2006 and 2007 books, because the earlier books had less inflated house prices.

But I can only speculate on the extent of actual losses in AIG's credit default swap book. And, in spite of all of their digging, the reporters appear to be in the same position. That makes for a story that is entertaining but not satisfying.

Of course, Roger Lowenstein's entire book on Long Term Capital Management, When Genius Failed, left me feeling the same way. Did LTCM make bad bets, or did they make good bets that just took too long to come in? Maybe I'm just a geek, but that was the main question I was looking for Lowenstein to answer, and he didn't.

Remember that Tyler Cowen thinks that the bailout of LTCM lulled folks into a false sense of security. But it seems to me that at AIG the sense of security came from the fact that the folks there thought they had a safe portfolio. Whether that sense of security was false or not is the question that is still unanswered, in my view.

Comments and Sharing

TRACKBACKS (1 to date)
TrackBack URL:
The author at PointOfLaw Forum in a related article titled Around the web, January 3 writes:
    More on EFCA's real kicker, compulsory arbitration [Jackson Lewis, h/t Eric B. Meyer] In L.A. and even Boston, billboards now part of the historic-preservation legacy [Katherine Mangu-Ward, Reason] Big Washington Post what-went-wrong-at-AIG series, pa... [Tracked on January 2, 2009 11:03 PM]
COMMENTS (25 to date)
Robin Hanson writes:

This seems a pretty important detail to get straight - I hope you dig to the bottom of it.

Brian writes:

As an employee of AIG, this is the story that management has been telling us. It seemed more like a run on the bank, than actual losses.

Patrick writes:

Spot on Arnold. Let's have the government intervene by setting collateral levels. Because that wouldn't dry up the credit markets.

8 writes:

If I'm an LTCM creditor and the government tells me LTCM is technically insolvent and I have to wait for their bets to pay off, why is that better than a bankruptcy where I'm forced to trade my debt for equity?

And how do we avoid zombie situations? Does the government appoint Dwight Schrute as regulator?

"There are several ways to kill a zombie, but the most satisfying one is to stab it in the brain with a wooden stick."

Melancholy writes:

"Instead, what did them in was the collateral posting that was required because their credit rating was downgraded."

This is exactly what happened. However, the contracts called for such posting - in reality AIG never put any money up-front for its CDS and never reserved any for them - why should it when its rating was not endangered until Lehman became permanently troubled?

Goldman et al had its own deals for which it likewise had to post; if it didn't call for AIG's collateral, it too would have gone down; and in the end of course it took refuge in converting to a bank holding company.

The reason this seems confusing to people is that they forget the intertwined nature of the companies via these swaps. If one goes down, they're all dead due to collateral requirements and the guarantees. Which is what happened. And which was why everyone was screaming for a Lehman rescue.

Didn't you think it odd Lehman's competitors weren't rubbing their hands going "Goody goody goody, another one bites the dust!"? That didn't happen because everyone understood the nuclear destruction that would ensue - except for Treasury, apparently.

This is the problem with putting stupid people in charge in important government positions.

Greg Ransom writes:

Get real. Paulson is a Goldman guy, and he looks at things as a Goldman guy.

This isn't Jesus in charge of the Fed.

Greg Ransom writes:

A bet that takes too long to come in is a bad bet. QED

"Roger Lowenstein's entire book on Long Term Capital Management, When Genius Failed, left me feeling the same way. Did LTCM make bad bets, or did they make good bets that just took too long to come in?"

Melancholy writes:

Greg, surely you are not arguing that Paulson "as a Goldman guy" saved AIG because he knew that killing it would also kill Goldman? That he was maliciously & corruptly regulating on behalf of Goldman?

If he had known such, and helping Goldman was his motivation, he surely would have also understood that the death of Lehman would be bad for Goldman as well.

So one way he's malicious & corrupt, the other way he's just dumb. Until proven otherwise, I'll stick with choosing stupidity over malice, in line with the old saw. What exactly do you mean here, please?

Jeff writes:

How do you square your hypothesis with the huge losses AIG has been reporting in it's earnings statements? In 2008Q3 alone they report $7.5 billion in losses on "AIGFP unrealized market valuation loss, credit valuation adjustment, net of deferred compensation reversal". It appears that much of the collateral AIG has had to post it will not be getting back, ever. That's not due to a fall in AIG's credit rating, but to the detiorating value of the securities the CDS were insuring.

babar writes:

how much of the problem is that they wrote these contracts in an intentionally obfuscated way and then hid what they were so that nobody could tell how much they were worth?

Arnold Kling writes:

As the article points out, the losses AIG reports are mark-to-market accounting.

Suppose I sell deep out-of-the money options for a penny. A year later, they are still out of the money, but now closer to in the money. Their mark-to-market value is now $1.00, even though it is still likely they will expire worthless. Since I sold you those options, on a mark-to-market basis I lost money (it would cost me $1.00 to buy back what I sold you for a penny). But I may not have lost anything.

Not that I'm against mark-to-market. But I think it does not necessarily prove that AIG is under water. They could be mark-to-market broke but ultimately turn out to be reasonably profitable.

Melancholy writes:

"the losses AIG reports are mark-to-market accounting"

Once again correct, which is why in the week after the extinction event, everyone - including Dr. Doom, Nouriel Roubini - immediately began pushing for a policy of temporarily suspending mark-to-market rules. I attended a webinar with Roubini and Ritholz where they discussed this for an hour.

Once you have to sell to make your collateral, and everyone has to sell at the same time to make their collateral, prices drop, and by mark-to-market you're worth far less, your rating then takes another hit. That means you have to post even more collateral. So everyone sells again. . .

Mark-to-market's a screaming death spiral - flames bursting out of the Fokker engine - in cases like this. I mean, completely Dawn Patrol.

Several banks were brought down this way and forced into merger during the crisis so far.

Unit writes:

How much of what banks like this do can be characterized as Ponzi-like? As long as more and more people sign up the flows can be sustained but then at the first sign of a slow-down, well it all goes south very fast.

Or is this too naive?

Whit Stevens writes:

"Did LTCM make bad bets, or did they make good bets that just took too long to come in?"

Estimating the potential liquidity demands (including margin calls) of a credit ratings downgrade is a very standard credit risk management practice. S&P requires companies involved in trading to perform this type of analysis on a regular basis. Even little old gas utilities are doing this kind of work.

A more likely explanation to me is that AIG vastly underestimated the size of the potential collateral calls by underestimating the volatility of the CDS products themselves. So they figured they’d post Y collateral in the event of a downgrade, but they ended up having to post Y*10.

Lastly, remember Keynes: "The market can stay irrational longer than you can stay solvent." These blow ups were do to the hubris that comes with thinking you have the best people, the best analytics, and the best practices. Risk managers need to remain humble to their ignorance.

William Dunn writes:

Failing to anticipate a collateral call - no matter how remote it appears - is a fundamental mistake; let's call it what it is, speculation. This seems very similar to LTCM's failure to plan or reserve for the one in a ___ chance that everybody would try to unwind their contracts at the same time. I think Lowenstein "got it" and explained it well, however the explanation has no sizzle. Failure to anticipate adversity just doesn't seem like it should be a capital crime. There's no fraud, no dramatic event. Sort of like playing Russian roulette for money with a revolver with a 1000-bullet cylinder: it is profitable until it isn't.

With leverage it all comes down to liquidity; either you have enough to withstand an n-sigma event or you don't. Since nobody can profitably hold enough guaranteed liquidity to weather all such events, any leverage that exceeds the amount of liquidity held is a gamble. In hindsight, it looks like AIG was gambling. The interesting question is whether management knew they were gambling or whether they drank their own Kool-Aid. Asserting that the odds were massively stacked in their favor is not evidence of prudential behavior -- only of delusional thinking.

The failure, it appears, was in risk management, in the "hold" limit. As with so many of the blowups of highly-leveraged entities, there were no limits, or the limits were too high, or they were waived or ignored. Looking for a more dramatic storyline is, I think, missing the point. The old simple story of greed and social conformance is enough to explain the behavior, and modern capital markets and instruments are enough to explain the outsized effects.

Jay writes:


If I remember correctly, LTCM's trading books were taken over by the banks that bailed them out, and ended up being solidly profitable for those banks by the time all their trades were unwound. In fact, that might have contributed to the financial services industry's willingness to give Meriwether money to manage (and blow up) again.

Dave writes:

One of Arnold's themes has been that the suits vs geeks divide harmed these financial institutions b/c of the division of knowledge and power. The geeks understood the risks, the suits were after market share, short term profits and bonuses (I may be taking some creative liberty in explaining his position as I view it). He brought up the question about part 2 of the Post story as to whether the geeks were wrong or if the suits just ignored them.

I would like to see Arnold respond to Taleb's claim that the geeks had way too much faith in their mathematical models and were insufficiently modeling risks they could not possibly understand. Example here.

I think a lot of Arnold's thinking about the crisis has been clear, but this is one area in which there is a lot of disagreement among knowledgeable pundits. I don't know who is right, and it's quite possible that both sides are correct to some extent; that is to say that the quants better understood the risks than the suits, but did not fully so, and the suits threw even understated caution to the wind.

Melancholy writes:


"that is to say that the quants better understood the risks than the suits, but did not fully so, and the suits threw even understated caution to the wind."

No, I disagree, in general. And we have the documents to prove it. It seems that you all here are not familiar with the famous "smoking cow" IM - you have to read all 2 pages to believe it - or with the S&P emails.

The quants had no power in the situation. For example, in this famous email, an S&P executive just tells the quant team just to make the evaluation up outta whole cloth pronto, and no, you can't see any underlying loan information.

Here is the famed cows IM, where 2 quants express their unease at being forced to rate deals against their own guidelines and flatly say the models are completely wrong.

These are both reflective of standard problems in the entire sector, not just as the ratings agencies - risk management was never truly independent of business pressure.

As to Wm Dunn's point about AIG's collateral - most of their MBS/CDS deals were with Morgan, Lehman & and itself, as Nouriel Roubini made public last May at his NYC conference.

They posted no collateral of any kind, and allocated no reserve to ever post any collateral. Because, of course, we all knew under no circumstance short of nuclear war could both Morgan and Lehman suffer, right?

Of course such things were obviously impossible! If you had walked into the manager and said you needed to put capital aside - take capital out better use in other deals - in the event Lehman went down - seriously, you would have been laughed out of the room.

You're compensated on maximizing profits, you know, and so's your boss, not worrying about impossibilities!

My take on the credit default swaps is written up in my blog --

As far as whether m2m caused AIG's failure, it's hard to say, but it made certainly made it quick.

As far as collateral calls on derivatives, the infamous Berkshire puts were custom written by Warren himself and contain no provision for collateral.

BJ Feng writes:

Some of you make it seem like Lehman's failure caused this crisis, if not for Lehman, we'd all be celebrating the end of an overhyped "crisis". Lehman changed nothing in the long run. It woke people up, but people were going to open their eyes regardless and reassess their willingness to take risk.

Paulson did the right thing, bailing out both Lehman and AIG was not politically possible at the time. Congress responds to Main Street and Main Street was not aware that ANYTHING was amiss, look at the initial failure to pass TARP.

When it's all said and done, Lehman was and will not be a factor in the overall outcome. Its failure can mark a point in time where risk tolerance changed, but saving Lehman would have only delayed this change by a week at most.

El Presidente writes:


These are both reflective of standard problems in the entire sector, not just as the ratings agencies - risk management was never truly independent of business pressure.

You're compensated on maximizing profits, you know, and so's your boss, not worrying about impossibilities!

I agree, and I elaborate. At some point, elevated profit expectations became unsustainable for the whole sector. But, nobody wanted to admit it because the fight for capital, which compels one to be one step ahead of the competition and drives compensation, was too fierce. When nobody will be the first to lose honestly, everybody is made a liar and we all lose eventually. Clearing any particular profit threshold is subject to the markets and real variables, not merely the indomitable or delusional or antisocial manager. How is it conscionable that these individuals put everybody else at risk for their own benefit the way they did? That seems like a massive abdication of fiduciary responsibility to shareholders, to say nothing of patriotic obligation to their nation, or humanitarian concern for the whole world. I think it illustrates that if you pay somebody enough money they might promise to do just about anything you ask, whether or not it's possible. Telling particular lies seems to be a valuable skill, according to market rates.


We're all paying now for decisions that some of us made. Lots of externalities here. The real loss isn't the evaporation of promised profits, for they never really existed, but rather the lost resources spent chasing things that would never materialize. We can't get back our time to invest and improve our mutual and individual wellbeing. Many retirees have watched their future grow dark, powerless to intervene on their own behalf. Many would-be retirees will now have to continue working whether or not they are healthy enough to do so. Many mid-career professionals will have to ride out this storm waiting for the waters to calm before they can steer their personal portfolios to a better course and pursue their best alternative. Many young professionals now have vastly reduced prospects for advancement because their senior counterparts are sitting on top spots and will not relinquish them to make advancement possible further down the organizational and industrial hierarchy. And this is in a place that will weather the storm. What about the third-world nations that are starved for investment and subjected to the paternalistic demand that they form better institutions that can lie as effectively as ours before we will lend them some money on reasonable terms? The lies have guided us to invest in ways we would not have chosen had we known the truth, and not to invest in ways that we should have.

If mark-to-market was the real culprit, and if we would do just as well rescinding it, then is the difference anything more than ignorance on the part of investors? Is it that they received more information and thus made different decisions? Is this an ignorance-is-bliss argument where we want to trust the supposedly omniscient and inherently wise market with everything except setting prices? Seems like an odd thing for libertarians to be supporting. The problem isn't so much that individual and institutional investors withdrew their money. The problem is that they put it in the hands of people who shouldn't have had it in the first place because they promised a certain combination of risk and profit they could not deliver. To argue in the alternative is to say that lies are a necessary part of business and growth. If that's true, then I believe we have returned to the zero-sum economics of mercantilism.

Melancholy writes:

@El Presidente

Yes, we largely agree. The system we had - "Wall Street," meaning ratings agencies and investment banks - worked great for a long time through many strong global changes. Why?

Because for most of this time, these firms were private. They worked as partnerships, and an employee could be there 30 years. You had no interest in messing up the agency, ripping off the bank.

Once they went public - Moody's went in 2000, which was also when Goldman finally let some public in - employees lost all sense of loyalty. By 2005 the IBG YBG mentality had fully taken hold. Just make your 5 million and get out.

Who cares if the deals are any good? You'll be gone in 2 years with your bonuses, so it's fine if they explode in year 3 (a.k.a., 2008). The ratings agencies clearly became pay-for-play - public ownership brought onerous revenue pressure, so it's easy to just start selling ratings outright. I think this is called "fraud." And those guys should go to jail.

Maybe our old system just depended on the investment banks and agencies being private to make the incentives work long-term.

But as Robin Hanson will tell you, giant catastrophes like these require failure at points throughout the whole system. Don't we have 5 points of failure here?

1 - Glass-Steagall repeal makes investment banks & insurers think they need to be hedge funds, and take on hedge-fund levels of leverage, without hedge-fund-style discipline;

2 - Investment banks and ratings agencies don't realize going public has turned their employees into their worst enemies (effed-up incentives);

3 - Investment banks and ratings agencies fail to fully insulate risk managers from the business and don't have mechanisms for employees to blow the whistle;

4 - Dealmakers in all entities don't keep good enough information and refuse to offer any transparency;

5 - Regulators, who were abundant (Fannie had its own special bureau of regulators; Lehman et al. had their own in-house regulators, who were either stupid or captured through Stockholm syndrome), and Treasury didn't know enough about how the businesses they oversaw worked, didn't understand the products, and when finally alerted to the danger, both moved too slowly and moved in ways that destroyed more market confidence than they shored up.

This last is to BJ Feng's point. Letting Lehman die in a mysterious fit of principle was the mistake that froze the credit market tight - no one knew who was solvent, who the government would save, or why or who they would likewise let die in a similar fit - so they sat, and are still largely sitting, on their capital.

You couldn't even feel safe doing normal intrabank overnight type lending for a while because you had idea who would survive to first light.

Saving both AIG and Lehman would have been politically possible if Treasury had bothered to launch even a half-hearted PR effort. Yet another sign of gov't incompetence here.

Bill writes:

I have no inside information, but I suspect that one way to understand the AIG fiasco is as follows: The firm's financial products group, which earned billions in bonuses for a few hundred employees over five years, got into the business of selling financial insurance in the form of complex derivative products that did not require the type of reserves typical in the insurance industry. Then a Category 5 financial hurricane hit. As a result, AIG appears to have experienced explosive growth in contingent liabilities that the firm simply was not prepared to handle.

In essence, it appears that they were exploiting their AAA rating by, in effect, selling naked puts to the rest of the financial world. I suspect that when AIG's finacial autopsy is completed, it will turn out that they were systematically "short volatility" at a time when the VIX index shot up from 20% to 60% annualized volatility and then as high as 80%.

Andrew Lo of MIT had warned in a 2005 paper on "Systemic Risks in Hedge Funds" that naked put selling could create illusory profits and massive risks. He illustrated his point with a hypothetical firm called "Capital Decimation Partners, LLC" whose only strategy was to continually sell out of the money puts. It appears that AIG's financial products group became AIG's own in-house "Capital Decimation Partners" by being massively short volatility ahead of the Lehman crisis.

When all of the blue-ribbon commissions have finished their reports on the financial crisis, it is almost certain that a major conclusion will be that the shadow banking system of investment banks, insurance companies and hedge funds were allowed to function with too much leverage. That meant they were being permitted to enjoy "heads I win, tails the taxpayer loses" incentive structures -- structures that permitted enormous private profits but which would eventually require socialization of even more enormous losses.

But another conclusion should be that not only was there too much conventional leverage; there was also too much hidden leverage in the form of contingent liabilities that even the senior management teams of the top financial institutions did not really understand. The 10-fold plus growth in derivatives exposure over the last decade (to more than $600 trillion in notional contracts) is what facilitated the huge increase in hidden leverage. It was these contingent liabilities that turned into a type of financial doomsday machine when the Category 5 storm hit.

Here is Warren Buffett's perspective from a recent CNBC interview on the "suits vs geeks" issue.

"... managing complex financial institutions where the management wants to deceive you can be very, very difficult. Or even when the management doesn't know what's going on, and--just take Bear Stearns. Bear Stearns had--I read it, anyway--750,000 derivative contracts. Now, you know, I could clone Albert Einstein, you know, and--many, many times and have him work 12-hour days for me and he would not be able to keep track of what's going on in an institution like that. It's--the ones that are too big to fail may be too big to manage"

It appears to me that the suits at AIG did not understand the contingent liabilities the derivative geeks had created. Alternatively, even if they were vaguely aware that they were taking on risks that were almost impossible to quantify, they could not bring themselves to ask the hard questions that effectively would have killed the goose that was laying massive golden eggs.

Since our regulatory authorities have obviously not been able to clone an army of Einsteins to understand the $600 trillion of derivatives exposure in our current system, we should all be prepared for further surprises as financial de-leveraging continues. If the FASB begins to require more frequent and more extensive disclosure of contingent liabilities in the future, as it probably should, it would not be surprising to see other large victims that will be unable to survive the scrutiny.

Peter Schaeffer writes:

I read all of the books on the fall of Enron (I live in Houston) including the rather good book by Kurt Eichenwald (Conspiracy of Fools). One of my goals was to determine if Enron crashed because of a liquidity crisis or insolvency. I have also discussed this topic with local bankruptcy lawyers.

The answer is that no one really knows. I suspect that the AIG situation is similar. In the short term, a lack of liquidity brought down AIG. However, over time it will probably become very clear that AIG was highly insolvent.

Note that in bankruptcy Enron has not been able to pay more than a fraction of the claims against it. Nor has Lehman. I think that CDS claims on Lehman bonds settled for around 8 cents on the dollar. This implies very deep insolvency.

My guess is that AIG will ultimately impose vast losses on taxpayers.

The broader point is that the crash of Lehman didn't create the crisis, but simply revealed it. Bailing out Lehman would not have stopped the day of reckoning, but simply postponed it (by a few weeks at most).

And yes, I do regard the AIG bailout as crony capitalism, by and for Goldman, by an ex-Goldman guy. Note that Lehman wasn't bailed out (and didn't threaten Goldman directly). AIG was.

Robert Arvanitis writes:

Bill, even without inside information, is quite the closest to the mark. It is useful to contrast FP with ILFC (the lease finance company) as an alternative re-leveraging of the triple-A. FP sold protection far too cheaply. ILFC more wisely was able to exploit the weakness of airlines.


I was at the firm, worked for Greenberg.

I left when AIG began its push into "finite" insurance, the first of many efforts to grow revenues in a low-RoE, low-growth industry.

Greenberg's first game was arbing reinsurers. He sold underpriced primary insurance, bought even-cheaper reinsurance, and pocketed the difference either way. The greatest effort was in monitoring reinsurer performance.

When that ran out, he developed new products like directors & officers coverage. Sold it before there were any such lawsuits, by creating fear. Once clients learned how to file a claim, he'd move on to another new policy cover.

When the new products game ran out, he grew sideways into new sectors, in what the Peter Principle calls "the lateral arabesque." On the upside, ILFC was able to borrow cheap with the triple-A, then lease to airlines dearly, and keep the depreciation to boot. The push into life insurance was a far less clear strategy.

Even that was not enough growth. AIG has always pursued Soviet-style "finite" insurance (you pretend to pay a premium, they pretend to pay a claim). The aim was to show revenues without taking risks. NY State and other insurance departments tried mightily to define the accounting flaws well enough to curb the practice, with limited success.

With FP, Greenberg's affinity for "finite" showed. Whatever quant work earned FP's early profits, by the late 1990s, FP was trying to do riskless accounting/regulatory trades, the capital markets equivalent of "finite." One area in particular created a huge interest. FP-Tokyo was all over the Japanese banks, offering ways to put off taking losses on bad assets. (This is exactly what drove the PNC transaction.)

When CDS came along, Greenberg got the idea it was likewise free money, no risk, just a fee for "helping" banks with their own capital requirements. Except not.

That's why Greenberg, properly renowned til then for his vigilance, erred.

Comments for this entry have been closed
Return to top