Sunday, April 17, 2011

Fatal Risk – the must-read book on AIG's failure

There are dozens of books on the financial crisis: I have read many of them and the Kindle samples for just about all of them. There are only two I would recommend: those are Bethany McLean and Joe Nocera’s excellent All the Devils are Here and the much more specifically detailed Fatal Risk from Roddy Boyd. Roddy's book is solely concerned with the failure of AIG.

Both books start without any strong ideological preconceptions and let the facts woven into a good story do the talking - and both wind up ambivalent about many of the major players - with many players having human weaknesses (gullibility, delusion, arrogance etc) but committing nothing that looks like a strong case for criminal prosecution. Reading these you can see why there are so few criminal prosecutions from the crisis. And you will also see just how extreme the human failings that caused the crisis are.

If you are not familiar with the saga that led up to the mortgage collapse, the rise of securitisation, the depth of the repo market, the lowering of credit standards start with the McLean/Nocera book. If you have to give a book as a gift to someone who is not a financial professional you could do little better. That is the best general book yet written on the crisis.

But for me (and because I was familiar with the broad details of the crisis anyway) the best book of the crisis is Roddy Boyd’s Fatal Risk. It is not a good first financial book to read and I had to think quite hard as to the details that Roddy glossed over - but that was because Roddy had to make a choice - was he writing for someone who vaguely knew what a credit default swap was or was he writing for someone who had actually read a “credit support annexe (a CSA)”.

Fortunately for most people he does not want to assume you have actually read a CSA (although I have). But less forgivingly Roddy does not feel the need to define an Alt-A mortgage or a repo line. This is a fabulous book – but it deals with complex subjects without shying away from their complexity and it assumes you have enough knowledge and intelligence to cope.

Truths, generalities and people

Underlying Roddy’s books are a few financial truths that bear repeating. Firstly anything that has any chance of going wrong if done for long enough will go wrong. It doesn’t matter if your model tells you that you will be fine in any mortgage default environment short of the great depression: if you continue to bet on that model you will lose. Maybe not next year. Maybe not in ten years but you will eventually lose.

Likewise if you write large quantities of out-of-the-money puts you will eventually lose a lot of money.

Likewise if your model assumes that there is always going to be a deep liquid market in any security (with the possible exception of a Treasury bond) then one day you will wake up and the buyers will have scampered like antelopes from a waterhole at first sight of a lion. Any business that has to roll a large amount of debt at regular intervals is dangerous.

Ignore these truths and you take a risk. Ignore them on a grand enough scale and the risk will be fatal.

Whatever: if you ignore these truths you might become rich in the interim. Earnings and growth might be fine. You might even look like a genius. Maybe a “legendary CEO”.

So Roddy’s book starts a long time ago - the 1970s and 1980s when AIG did not forget those truths - and it talks about AIG as a superlative risk management machine. The first section of the book is a repeat of the AIG legend - a legend of superlative risk management mostly in the head of one man: Hank Greenberg. It is a legend that might be overstated but that doesn’t mean that it is not mostly true. Hank really did work absurd hours, pick at steamed fish and vegetables and ask sophisticated questions to six people at once. Hank knew to really understand what was going on you had to go three to four levels deep in an organisation and ask the right questions of assorted lower/mid ranked officers. They would answer truthfully because of a desire to impress or fear or even that (unlike many senior managers) they were not accustomed to spinning. He would get the raw data. He would make the assessment.

There were two things however that Hank did not assess properly: his own mortality and his declining skill in old age. There is no question of declining skill. It is very hard to imagine the Hank Greenberg of 1975 falling for China Media Express - but the Hank Greenberg of 2010 was suckered. As for mortality he had no plans.

He also did not plan for Eliot Spitzer.

By 2000 Hank was extremely concerned about what Wall Street thought of his stock. That is no surprise - AIG was the most highly valued large financial firm in the world (I remember being startled that its PE was three times Wells Fargo). And - unstated by Boyd - Hank liked that a lot because he used AIGs stock as currency to do acquisitions. He was - much to the chagrin of many investment bankers - very selective as to the acquisitions he would do (he knew acquisitions were fraught with risk) but he did some mighty big ones including the purchase of Sun America. I remember that one - and thought (correctly in hindsight) that it probably made sense primarily because AIG was paying with inflated stock.

So by the year 2000 Hank was - apart from running the business - actively manipulating the earnings of the business. As far as I can tell he ran the business particularly well (the legend of AIG was not false) but he also played Wall Street like a fiddle and gave them the numbers they wanted even if they were massaged a little (or maybe a lot).

Moreover - and this is critical for the story - AIG had no overarching operating system. It was a bunch of fiefdoms all reporting to Hank. This meant that AIG could not produce earnings results until the very last day they were legally allowed to file them.  It meant Hank could personally massage earnings. At many financial institutions some approximation of earnings are known every month. At AIG there was no system they could query and ask for their aggregate Alt-A mortgage exposure. Hank might have known - indeed almost certainly would have known - but the system is Hank and Hank being removed or dying would be disastrous for AIG.

And then along comes Spitzer. Spitzer discovers a relatively minor finite insurance transaction between AIG and General Re. (Believe me it was minor - I know of plenty of nastier transactions than that... many of which were never prosecuted.*) However it is a clear attempt to fudge the numbers - Spitzer really is onto something. And with bombast and the power of the Attorney General he makes Hank Greenberg’s world fall apart. Spitzer fights dirty (and it is no surprise that several Spitzer prosecutions later failed because of prosecutorial misconduct) but Spitzer has his clear piece of fakery and he wants and gets his pound of flesh.

Hank is forced out - which is the equivalent to AIG of his sudden death. Worse because AIG went on to repudiate many things Hank stood for including many of his risk-control edicts. If he had died the hero CEO it might have been marginally better for AIG.

The minor nature of the AIG-Gen Re transaction is laid clear when Roddy suggests that there is an “excellent chance that Greenberg gave the Gen Re issues - which cost him his job, his honor, his status and perhaps over a billion dollars in personal wealth - all of five minutes of consideration.”

Still AIG-Gen Re was a transaction designed to massage (ie fake) the numbers - and thus speaks to a relationship with Wall Street and a concern to stock price that is unhealthy.

Unstated by Roddy: Hank had forgotten a cardinal rule of risk management: you do that sort of thing for long enough then one day you will find your Eliot Spitzer. This is just as sure as the statement that if you write put options long enough you will one day get your comeuppance.

The new CEO

The new CEO - Martin Sullivan - was the best salesman AIG had. Joe Cassano (who ran the disastrous AIG Financial Products) observed that he never saw Sullivan ask a single penetrating financial question. It's a telling observation.

The place I used to work had a boss who was very suspicious of financial product salesmen because inevitably they wanted to produce what the market (ie the crowd) wanted. And in financial services if you run with the crowd you can get your comeuppance delivered abnormally sharply.

To be fair, there are financial service companies that require salesmen even as leaders. Insurance brokers spring to mind.

AIG however was not one of those companies. It was global, complicated and pervasive and it had no overarching risk management system now that Hank was gone. To replace a control freak they needed another control freak at least until they built control systems. They never got that - and only at the very end (in Willumstead) did they get a CEO that even understood there was a problem.

There is a truism about financial product salespeople: if you put a salesman in charge of a financial institution with large reach and allow him to operate with thin risk control then your earnings will go up. And up. And up. At least until they don’t. Martin Sullivan proves that truism.

Under Sullivan some small businesses were allowed to expand in new ways until they became big businesses - ones big enough to threaten AIG and indeed the world. A decent example of the Martin management style comes from a small part of AIG - United Guaranty. United Guaranty was a mortgage insurer - at least for a while the best mortgage insurer on the Street. (I remember thinking that a couple of other players, notably PMI and MGIC were much riskier.)

The right thing to do with a mortgage insurer was stop writing business about 2005 - and certainly by 2006. [Or you could sell it as GE did.]  Margins were collapsing and the risk of the loans was rising fast. The independent companies couldn’t really stop because that was their only business. AIG was under no such constraint - United Guaranty was a tiny part of AIG and stopping would not have affected the stock price. It might have even been seen by some (myself included) as a sign of discipline. Here is the quote from an AIG unit chiefs meeting in mid-2007.

As UGC posted its first losses, about $100 million, Nutt was explaining to Martin Sullivan and other senior management that while they hit a rough patch, they were writing excellent new business, and, at any rate, the competition was getting killed. Sullivan smilingly told Nutt that even if he didn’t write another dollar in business for a few months, “We would still love him”. AIG staffers had a phrase for this sort of response: “classic Martin”. It was a decent word or gesture, directed at a manager who was clearly fumbling, both publicly and on the job. But it also carried a serious message: better to be safe than sorry. The trouble is that the time for this was two to three years earlier.

To not realize that a mortgage insurance business in mid 2007 was problematic was seriously inept. UCG has now booked $3.9 billion in losses. Hank would have been on top of this at least a year earlier. Whether he would have been on top of it two years earlier is more dubious. One year earlier and UCG would still have had substantial losses.  But they might have been absorbed by profits in an otherwise functioning AIG.

The two businesses that blew up AIG

There were problems all over AIG (and there were good bits too where individual managers saw the mess coming and ducked for cover). But two businesses stand out for the sheer destruction that they wrought. The better known was AIG Financial Products (FP) credit guarantee business. The less well known was Win Neuger’s securities lending business.

The credit guarantee business for thin fees guaranteed securitisation deals - usually very high grade paper or just as often resecuritisations of high grade paper. These were deals that would be fine in any credit event less bad than the great depression. In other words they were “great depression puts” and FP was writing puts. You should know the truism by now.

But worse the credit default swaps had a credit support annexe (CSA) attached. This made it mandatory for the parent company of AIG to collateralize the deals (ie put up hard cash to guarantee payment) under certain events. Senior management of AIG did not even know of the existence of the CSA until the company was at death’s door. They believed until very nearly the end that mark-to-market did not threaten liquidity.

I understand how a salesman (Sullivan) missed the CSA. If you followed the credit enhancement business you would know - by law - that the monoline insurers were not allowed to collateralize their obligations. Why of course should AIG be any different? But even that cursory “knowledge” could be dangerous. Both AIG and Ambac had CSAs attached to their guaranteed investment contract business (a business that was run by parent companies). I did not know of these until a well known hedge fund manager sent me a copy and even read it over the phone to me.

But that is a thin defence of AIG. The above mentioned hedge fund manager knew of the CSAs at Ambac and MBIA a couple of years before the disaster - and he had to look and find it. AIGs senior management should have just asked. Their risk management department should have been over every material contract - and believe me these were material contracts. This was an epic failure.

Win Neuger’s business was similarly destructive. What he did was [get the parent company to] borrow high grade securities from the life insurance companies, repo them, buy lower grade securities and pledge those back to the life companies to secure parent company obligations to the life companies.

Two things went wrong. The life company management (and later regulators) got mighty jacked when the life companies had lent their good securities and were holding trash security. They required hard capital injections from the parent company to solve this - and along the way AIG kicked in $5 billion. At the end the Texas Insurance Commissioner was going to confiscate four insurance companies (which would have collapsed AIG).

The second thing that went wrong is the counter-parties to the repo loans just wanted cash their back. They wanted it now. To get it though the parent company would need to get back the trashy (and hence heavily discounted) security from the life company, sell them, top up the (now large) shortfall and pay the investment bank on the repo line. This turned marks on the low-grade securities into an immediate liquidity drain on the parent.  That is truly ugly.

How they got there too was a story of failure to consider fat tail risks.  It is the main story of the book.

Liquidity versus solvency

At the height of the crisis it was very difficult to see whether AIG was a liquidity problem or a solvency problem. If it is a liquidity problem then bailouts don’t cost much -indeed structured right they are profitable. If it was solvency then a bailout will be very costly and in extrema (such as Ireland) can bankrupt the nation.

I originally thought AIG was liquidity. I later thought it was solvency. But now the Government looks like it is making heroic profits on the AIG bailout - and it was surely enough a liquidity problem.

There are a couple of lessons here: sophisticated observers (if I am a sophisticated observer) can’t tell the difference between liquidity and solvency in a crisis. The second lesson is that any contract that can cause a liquidity problem will - if repeated long enough - actually cause a liquidity problem. Modelling solvency does not cut it... if you run a financial institution you better model liquidity as well – and better be ready for the closure of debt markets.

The AIG people after the failure

There is a lot of anger in the broad community about the people at AIG especially as none of them - those that caused the largest bailout of the crisis - were ever charged criminally. Roddy does not share the anger about the lack of criminal charges but he is angry about the sheer recklessness of some AIG people. This quote was revealing:
Al Frost’s [a key salesperson for AIG FP] job was to drum up deals and revenue from the major investment banks and he did. Cassano’s job was to ensure that decisions made at FP were logical and made with all available information. He failed... 
But Cassano did not fail in a vacuum.... 
That Martin Sullivan and Steve Bessinger did nothing is now well established. But neither did Financial Services chief Bill Dooley, his CFO Elias Habayeb, Risk Management Chief Bob Lewis and his head of credit-risk Kevin McGinn. Anastasia Kelly’s legal department was similarly silent. These people saw everything AIG FP did in real time and had plenty of authority to force at least a reevaluation. It was, in fact, their job to do this... 
Save for Anastasia Kelly (who retired) every other person in the line of oversight of the FP swaps book is now gainfully employed as an officer at a publicly traded company with as much or more responsibility than they had previously.

Roddy is right. The fact that the failure of these people was not criminal does not excuse it. These people were paid big multiples of average earnings and demonstrated that they can’t do their job. So they are still paid big multiples of average earnings.

Along this line special scorn needs to be reserved for Win Neuger. He ran AIGs internal asset management business - especially the securities lending business which in itself was big enough to destroy AIG.

He now runs an asset management company with over $80 billion under management.

Where else - except Wall Street - can you be that well rewarded for failure?


I don’t want to give too much away. This is the best book yet written about any specific episode of the crisis. I just think you should buy it. Buy multiple copies. Give them to your friends. They will be grateful too.


*Hint to the regulators: try and work out the large finite transaction between Unum Provident and Berkshire Hathaway. There lies a can of worms...


BlackRaven said...

John, you make the point that if you're selling tail risk, for long enough you will hit an event.

I would say the risk comes not from taking tail risk, but from increasing the size. Selling puts isn't dangerous, selling puts on a notional multiples of your capital is, ie. its the leveraging of tail risk (partly to keep it a viable strategy when premiums are low).

John Hempton said...

I was careful there. We sell volatility. And we do it all sorts of subtle ways.

We also limit the amount we sell.

Insurance is selling vol. No problem with that provided adequately diversified and no single piece is big enough to kill you.

The problem is not selling puts. Sell puts and one day you will have a bad day.

Sell too many puts and one day that bad day will be fatal.


Anonymous said...

The literal Bernanke put
by Izabella Kaminska

I wonder whether Bernanke reads this blog :-)

"Cassandra" said...

John, There was a time when the minor technical transgression was still a minor and probably manageable transgression (possibly still defensible either by sacrificing some heads like Buffet, or settling as a company with tail between legs). But Mssr Greenberg, apparently was not one to lose a battle (and a publicly visible one) in order to win the war. In amazement, Not long after, I watched Greenberg, in what could only be described as delusions of his own power, launch a vicious public ad-hominem attack upon Spitzer at a press conference. He used some highly charged language, and leveled retaliatory accusations extremely unwise in light of their relative positions of immediate power, however angry he might have been. My jaw dropped at Greenberg;'s audacious miscalculation, and I knew at that moment, with near-certainty, Greenberg was toast.
For the corollary of "Don't break the law when you're breaking the law", is Don't Mess With The Law When You're Breaking (or just Bending) The Law"....

I don't pretend to know what went on behind the scenes. Perhaps Spitzer (in private) read Greenberg an ultimatum of resign or face the consequences. Perhaps he cornered him more completely, and counter-attack was the only viable option. But I did understand that the chest-thumping show of bravado sealed his fate more completely than before.

I can't wait to read the book!

polit2k said...

More on the literal Bernanke put by Tracy Alloway FTAlphaville.

Anonymous said...

"But now the Government looks like it is making heroic profits on the AIG bailout - and it was surely enough a liquidity problem."

Sorry, John, but can you please elaborate on this a bit?

I'm very unsophisticated, but it seems to me that given the amount of money and guarantees that the government pumped into the financials, it could have made "heroic" profits on just about any individual company (although not on all in aggregate). In this case, you can say that anything is a "liquidity" problem from the government's perspective, I think.

Could you please correct me if you have the time/patience?

Anonymous said...

On liquidity vs. solvency: I questioned in late 08/early 09 why AIG was unwinding its CDS book at record spreads. Seemed like a dumb move just for the sake of "risk reduction". With perfect hindsight today, it was an extremely bad move and cost AIG a lot of money.

Hubert said...


AIG had to renounce the right to sue Goldman when it took the money from the Feds.
AIG was famous for suing and not paying. While I understand that insurance fraud is hard to state in a CDS arena, I guess there could have been several openings for AIG to get into court with Goldman. Questioning Indepence of CDO managers might have brought interesting discovery in courts. Also who sold those MBS to Neuger? Did those parties have their Clayton DD at that time?
My point is that not all the stuffees were defenseless like german landesbanks. Some, the Mortgage Insurers including AIG, Fannie, Freddie might have had a real chance in court to blow up the whole fraud scenario. It was not clear if GS would have had the same support from a McCain Government and Wall Street would have found itself in some interesting legal battles with companies who also have some political clout....

What is your view of the legal look-out in autumn 2008?

John Hempton said...


I am not aware what rights to sue AIG gave away - I doubt they gave away any - but they chose not to sue at the time.

If they had sued at the time their funding would have died and they would have died anyway.

Repo funding in particular is like that. No firm survives a repo fail.


The bit of the book I did not reveiw was relationship with Goldman Sachs. Did not want to give everything away in the review - but it is also interesting.


pater tenebrarum said...

There have been rumors (which quickly disappeared into the memory hole again) that AIG's entire CDS business was in reality a variation on the so-called 'side letter fraud'. That CDS were merely used because the old variation of that type of scheme had come to the attention of regulators and was no longer viable. Of course no-one ever looked for evidence to that effect. If such evidence had come to light, it would have been highly embarrassing not only for the financial establishment but also the authorities that were so quick to open the public purse, not even demanding haircuts from CDS holders. We will of course never find out whether there was anything to these allegations, but they were interesting in view of the sophistication of AIG's counterparties. How could they not have known that AIG would be unable to pay up if push came to shove?

Anonymous said...

Mr. Greenberg was in charge of AIG when they first started writing credit default swaps on mortgage-backed securities, and later, collateralized debt obligations. From the very beginning, the collateral calls were in the contracts. It was one of the reasons AIG wrote so much business at the expense of the monolines. To blame Martin Sullivan for the sins of omission committed by Hank Greenberg, and the sins of commission by Joe Cassano is unfair and dishonest.

Hank Greenberg's manipulation of AIG results can be seen today. In the last two year end reviews, AIG has taken $6.4 billion in reserve increases from business written in the early 2000s. That would be, in case you weren't paying attention, when Hank Greenberg was in charge. Every other company on the street has been taking down reserves for the same period.

A Sanford Bernstein analyst had done a global reserve adequacy review 18 months ago. His data showed an industry under reserve of $11 billion. He kept running the numbers, and couldn't figure out where the problem was, so we started looking at individual companies. His conclusion, the entire $11 billion in reserve inadequacies were coming from AIG. At the time, AIG attacked him, calling the analysis ridiculous. Since then, they've added the $6.4 billion.

Just a few months ago, the same analyst, now working for another company, looked at workers compensation reserves during the hard market (increasing prices) of the early 2000s. He compared business written, price increases, and claim exposures. His conclusion, major writers like Travelers, the Hartford And Liberty Mutual were adequately reserved. AIG, on the other hand, came up $2 billion short in reserves. (In addition to his earlier analysis of $11 Billion.) Look for more reserve increases in the coming years.

The final chapter on Hank Greenberg's genius risk management has yet to be written. Everybody in the insurance industry knew that AIG was playing games with their numbers. They always reported after everybody else, and always had a combined ratio 4-5 points better than everybody else.

Nobody could figure out how they can consistently write the riskiest business, business and everybody else was getting killed on, and still come out smelling like a rose. The answer is that they were making the numbers up.

John Hempton said...

Yes they started writing puts when Hank was around.

Insurance is writing puts. (Earthquake puts, hurricane puts, medical inflation puts...)

The issue is not writing puts per se - as Black Raven has said. It is the QUANTITY of puts written.

A LITTLE bit of puts was a diversification (and hence consistent with good risk management). A lot of puts - well that was a lot of puts.

Moreover when Hank was around most the puts were corporate CDOs. The company was originally very averse to mortgages and there is plenty ot document that. (Though UGC was there the whole time - so they were not totally averse to mortgages.)

Lets take UCG. It is down 3.9 billion dollars. It is a mortgage book. It was there when Hank was there.

We can blame Hank for UCG. $3.9 billion (though I think it might have been smaller had Hank stayed). $3.9 billion loss is result of diversification. $50 billion loss is not the result of diversification.


Unknown said...

I have not read the book but any discussion of AIG can't be truthful if not discussing Professor Gary Gorton and his faulty models. His models didn't take into account collateral calls. This is why AIG was so gung ho in writing them. It was the collateral calls from GS amongst others that pushed AIGFP over the edge. The sin of AIG is not selling CDS insurance, it’s the fact that they were 100% unhedged on these positions.

John Hempton said...

The good (or is it bad) professor was discussed.


yedlob said...

I recall being shown an analysis of AIG's reserve smoothing (and under-reserving) in the 1970s.

This has made me sceptical of the line that AIG was a basically good insurance business undermined by the financial products.

Anonymous said...
I gotta love how Win Neuger's profile at Pinebridge does not make any mention of AIG!

Anonymous said...

As far as those who failed today being gainfully employed, the key question is if they learned anything from their experiences. If so, the blemished managers may be better risks than those who were not implicated at all in these markets. Of course, managers who did play in these markets and made good risk decisions would be the best prospects of all. But the "one big screwup and you're unemployable" model has never been popular in the U.S. Check out Silicon Valley, for instance.

John Hempton said...

There is a big difference between Silicon Valley screw ups and Wall Street ones.

Start a business in Silicon Valley and you get to work 70 hours a week for five years for not much money. But you have a huge lottery ticket at the end. (If your business succeeds you get very rich...)

If you fail - you get to repeat the process. You work 70 hours a week for five years for not much money.

On Wall Street you work 55 hours a week for a LOT of money.

If you stuff up you get to repeat the process.


Failure is punished in Silicon Valley. You don't get paid a million bucks a year.

These guys are paid that or much more - both before and after failure.


Tom Lindmark said...

If as you say that in a crisis sophisticated parties can't differentiate between a liquidity crisis and a solvency crisis, what does that suggest should be the approach for regulators at such a time? Rescue and hope it's a liquidity issue and if it is you get the AIG outcome you refer to, but if you're wrong you get Ireland.

I suspect regulators will tend to opt for the liquidity explanation so perhaps that argues in favor of Arnold Kling's suggestion that we need a system that is easy to fix when it breaks, i.e. smaller financial institutions.

Anyway, a great post. I can't wait to read the book.

General disclaimer

The content contained in this blog represents the opinions of Mr. Hempton. You should assume Mr. Hempton and his affiliates have positions in the securities discussed in this blog, and such beneficial ownership can create a conflict of interest regarding the objectivity of this blog. Statements in the blog are not guarantees of future performance and are subject to certain risks, uncertainties and other factors. Certain information in this blog concerning economic trends and performance is based on or derived from information provided by third-party sources. Mr. Hempton does not guarantee the accuracy of such information and has not independently verified the accuracy or completeness of such information or the assumptions on which such information is based. Such information may change after it is posted and Mr. Hempton is not obligated to, and may not, update it. The commentary in this blog in no way constitutes a solicitation of business, an offer of a security or a solicitation to purchase a security, or investment advice. In fact, it should not be relied upon in making investment decisions, ever. It is intended solely for the entertainment of the reader, and the author. In particular this blog is not directed for investment purposes at US Persons.