The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FCIC Interview of Gary Gorton, May 11, 2010 United States of America Financial Crisis Inquiry Commission INTERVIEW OF GARY GORTON Tuesday, May 11, 2010 *** Confidential *** Financial Crisis Inquiry Commission 1 FCIC Interview of Gary Gorton, May 11, 2010 INTERVIEW OF GARY GORTON Date: Tuesday, May 11, 2010 Location: Milbank, Tweed, Hadley & McCloy LLP 1 Chase Manhattan Plaza New York, NY 10005-1413 United States of America Participants: Professor Gary Gorton Dorothy Heyl, Milbank, Tweed, Hadley & McCloy LLP Andrew Tomback, Milbank, Tweed, Hadley & McCloy LLP Thomas Santoro, Milbank, Tweed, Hadley & McCloy LLP Dixie Noonan, FCIC Clara Morain, FCIC Al Crego, FCIC (by phone) Mike Easterly, FCIC (by phone) 2 FCIC Interview of Gary Gorton, May 11, 2010 Tuesday, May 11, 2010 MS. NOONAN: Thank you for letting us come and use your office space to do this. So, you know who we are. You came and presented at the forum that the Commission held in February in D.C. and I know that you have been in contact with other members of our staff on various issues. So, I appreciate your taking the time to meet with us today, specifically about your role at American International Group Financial Products. And just wanted to start there and get you to tell us a little about when you got involved with AIG-FP and the nature of, I believe it was a consulting role, but if you could just fill us in on the details. MR. GORTON: That’s pretty much it. started in the spring of 1996. I mean, I I’m not sure what month and I was hired as a consultant without really much of a mandate. That’s it. MS. NOONAN: I’m assuming they brought you on for some reason. MR. GORTON: They had a financial economist from Wharton before me who had left and gone to Goldman, and they wanted someone to replace him. MS. NOONAN: Okay, and what did he do? MR. GORTON: Do you mean when he was at AIG? 3 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Yes, for the company. MR. GORTON: I don’t know what he did. There’s –- I just have no idea. MS. NOONAN: Once you got there, what sort of things did you start working on for the company? And I’m assuming you worked out of the Wilton, Connecticut, office; is that right? MR. GORTON: Well, it was at Westport, yes, at MS. NOONAN: Okay. MR. GORTON: You know, I think they had the the time. view that I was going to be a resource; but you know, it was kind of an entrepreneurial environment so you know, I asked a lot of questions, started talking to people, and, you know, that’s what everybody there did. MS. NOONAN: So, what kind of business was AIG-FP engaged in when you started there? MR. GORTON: derivatives house. FX derivatives. It was at the time completely a It did interest-rate derivatives, It was still very close to its roots. It started as a long-dated derivatives place. And it was clear that that was not going to be the future, that nobody knew what the future was going to be. They were looking at other things. MS. NOONAN: So you started in the spring of 4 FCIC Interview of Gary Gorton, May 11, 2010 1996. At the time, you were also on the faculty at Wharton, is that correct? MR. GORTON: Uh-huh. MS. NOONAN: Okay. And how were you compensated initially? MR. GORTON: I was paid by the hour. MS. NOONAN: Is that how you were compensated the entire time you worked at FP? MR. GORTON: Yes, I always got paid by the MS. NOONAN: And did -- MR. GORTON: Later, later I got a bonus. MS. NOONAN: What year would you have started hour. Much later. getting a bonus? MR. GORTON: don’t know. I don’t think I even remember. I It was certainly, I don’t know, four, five, six years after I started. MS. NOONAN: In the early 2000’s? MR. GORTON: Yes, that would be a guess. MS. NOONAN: So, I’m sort of curious. You said that you just sort of came on to brainstorm and it was entrepreneurial. MR. GORTON: Yes. MS. NOONAN: Could you give me a little bit 5 FCIC Interview of Gary Gorton, May 11, 2010 more background on -MR. GORTON: Well, yes. I mean we, at the time I sat next to a guy who had just been hired from Goldman in Credits so I talked to him at lot, and he explained -– you know, I asked him what he was working on and he would tell me. lot of questions of. Lots of people I would ask a MS. NOONAN: And -Who is person from Goldman who was working on credit? MR. GORTON: Jim Wolf. MS. NOONAN: Okay. MR. GORTON: He -- so he explained to me —- you know this was the early days of credit derivatives. So I asked a lot of questions about that, and we started thinking about that, and that’s sort of how a whole group of people started thinking about credit derivatives. MS. NOONAN: So when you first started working with FP, about how much of your time would you allocate to that work? MR. GORTON: Which work? Oh. MS. NOONAN: As opposed to being a professor, and the other -MR. GORTON: week. Oh, a day a week. It was a day a I went there a day a week. 6 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: One day a week? MR. GORTON: Yes. MS. NOONAN: And was that consistent throughout -MR. GORTON: Yes. MS. NOONAN: -- your work at FP? MR. GORTON: Yes. I mean, there was a few years where I took a leave, and I was there more than a day a week. And there was some times when we got really busy and I would do some work at home. But otherwise, a day a week. MS. NOONAN: Okay, okay. I think we now know that you developed, at some point, a model that the Financial Products Group used in its credit derivatives business. Is that right? MR. GORTON: Yes. MS. NOONAN: And when did that model -- when did you develop that model? MR. GORTON: My guess, it would be in ’98 or MS. NOONAN: And why did you develop the ’99. model? Was it your idea? MR. GORTON: a quasi academic place. models. Was it someone else’s idea? AIG Financial Products is kind of So they were very used to The idea of a model was, you know, how the 7 FCIC Interview of Gary Gorton, May 11, 2010 whole business was —- all the businesses was run. When I was hired, the CEO was a Ph.D. at mathematics. So it wasn’t the kind of place that would do anything without a model. MS. NOONAN: So wasn’t -- What was it that AIG-FP wanted to do that it needed your model for? MR. GORTON: Well, we didn’t know what we wanted to do in the beginning because it wasn’t a place that could compete with a firm like Goldman. So, we weren’t going to do the same business as Goldman. So, eventually we realized that, you know, we would have the possibility of investing in large portfolios —- pieces of large portfolios. So, I mean, AIG Financial Products was an investor. There was no clients. There was nobody -- you know, we weren’t trading for anybody or with any -— we were an investor. So -- so the model was developed to -MS. NOONAN: When you say you were an investor -— I’m sorry to interrupt you -- did you see FP as a proprietary trading organization? MR. GORTON: sure what that is. No. It wasn’t -- I mean, I’m not Normally, I think of that as a small group inside a big bank that trades for the bank. We weren’t —- we were completely separate from AIG 8 FCIC Interview of Gary Gorton, May 11, 2010 physically. business. with them. So we didn’t know anything about their We, you know, the model had to be developed You know, their risk management was always working with us. But, so it wasn’t –- it wasn’t proprietary trading in the usual sense. Anyway, so the model was developed to assess these various portfolio trades. MS. NOONAN: In English, please. MR. GORTON: It was the —- the -- a large number of obligations would be referenced and then we would invest with a derivative in a piece of that. So initially it was corporate portfolios. MS. NOONAN: Okay. So, I mean, just to give you a little —- we are really interested to understanding how your model works, what it was designed to do, what it was not designed to do, and how AIG-FP used it or did not use it in understanding and assessing the risks involved and taking the investments that it took and the credit defaults swaps portfolio, and – so that’s so that I hope you can, can elucidate, you know a little bit about -- about that as we go, just so you know that’s sort of -— I really want to understand that. And I want to understand it from the genesis of the business all the way through to the end. And if it started —- it sounds like you said it started in 1998 or 9 FCIC Interview of Gary Gorton, May 11, 2010 ’99. And it was developed initially to work on —- was it always collateralized debt obligations -MR. GORTON: No, no, no, no, no. MS. NOONAN: Okay, okay. MR. GORTON: No. No, that was -- that was a small thing that came at the end. MS. NOONAN: Okay, okay MR. GORTON: By that time I was working at commodity futures. So, originally it was a list of corporate names. And this thing would be tranched. Do you know that word? MS. NOONAN: Uh-huh. MR. GORTON: Okay. So, in the beginning we worked with the parent company risk officers to develop a model. I mean, I came up with the idea for our model, but they —- we investigated a number of models that you could buy. And in fact, they bought one. that as a way of checking our model. agreed, they would be happy. And they used If the two models So these models are actuarial models that calculate the distribution of losses. That’s what they do. And that’s what they were designed to do, and that’s what they were expected to do. The parent company risk officer, Bob Lewis, 10 FCIC Interview of Gary Gorton, May 11, 2010 set the criterion for how these models would be used. And then later, you know, different versions of the model were developed for other asset classes. And then eventually, year later, it was used for these CDO things. MS. NOONAN: Okay. So when you say it started out with a list of corporate names – MR. GORTON: Uh-huh. MS. NOONAN: -- are you saying that was those were corporate bonds? MR. GORTON: No, no. MS. NOONAN: Or were those credit default swaps on a company itself? MR. GORTON: No, no. It’s just a list of companies’ -MS. NOONAN: Okay. MR. GORTON: -- names. MS. NOONAN: Okay. MR. GORTON: Right. So a portfolio is a list of 200 companies, and there is an amount of money associated with each one, say, $10 million. MS. NOONAN: Where does the amount of money MR. GORTON: There is no money. come from? It’s just a number. 11 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Okay. MR. GORTON: This is derivatives, so there’s no money; right? So, we are going to specify this list, we are going to specify the number of dollars per name, and we will say, you know, to whomever we are talking to, “We will protect the senior 60 percent of that portfolio.” A portfolio has no money, there’s no money, right? It’s just a specification of a portfolio. MS. NOONAN: What is it based on? Their stock price or the -MR. GORTON: No, no, no. MS. NOONAN: -- performance of their bond? MR. GORTON: No, no. MS. NOONAN: But the list of names have to go It’s just a list of names. in one direction or another; don’t they? MR. GORTON: No. No. The contract says that if enough of these names default -MS. NOONAN: Default on what? MR. GORTON: There is a credit event. MS. NOONAN: Okay. MR. GORTON: Right? They go bankrupt. So there is a -- is the 12 FCIC Interview of Gary Gorton, May 11, 2010 specified number of events, changed over the years, which would trigger credit default swap. And if enough of these -– enough of the companies on this list go bankrupt and then lose a certain amount, which can depend on recovery rate and the bond price and so on, then, you know, somebody loses. In my example, it was 200 -- it was what? Two hundred -- I forget what I said -- $200 million portfolio. We took 50 percent. So somebody would have to lose $100 million before we would pay out anything. MS. NOONAN: And so, essentially there are people on both sides of this deal. MR. GORTON: No there doesn’t have to be. I mean somebody has to be receiving the money, but we don’t have to have the full capital structure, right? So, suppose you do it with –- I mean with a CDO, you have to have the full capital structure. this, you don’t necessarily have to; right? With So you could go to Goldman and say, “I will write protection on the senior 50 percent of this list, 10 million each.” And they decide whether they want to do the trade or not. We’re investing in that and they are doing something else. 13 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: And you would analyze it from the perspective of, “We don’t think that 50 percent of the companies on this list are going to go bankrupt.” And Goldman, if it is on the other side, would analyze this saying, “We think that at least 51 percent of these companies are going bankrupt.” MR. GORTON: Well Goldman, you know, has -— wouldn’t necessarily keep this risk, right? So, we don’t -- I don’t know what -MS. NOONAN: Or whoever they sell it to. I’m just trying to understand the -MR. GORTON: Well, somewhere. MS. NOONAN: -- the rationale for the product to begin with. MR. TOMBACK: I think just to be clear. In terms of the sides, maybe Gary can explain this perhaps better, but Goldman’s not of the view that if they buy this product that 51 percent are going to go bankrupt. They’re of the view that it’s worth having the protection against the possibility of that event at the price that AIG is willing to sell the product. MS. NOONAN: MR. TOMBACK: Understood. And second, I think what Gary’s trying to tell you is, is that Goldman, if they have that product, or they go out and buy that product to 14 FCIC Interview of Gary Gorton, May 11, 2010 match the portfolio that AIG is, you know, offering to them, may sell it. MS. NOONAN: Understood. MR. GORTON: So this example -- MS. NOONAN: I’m just trying to understand sort of where is idea came from. MR. GORTON: What idea? MS. NOONAN: The idea to write protection on a list of 200 names and -MR. GORTON: Oh, I don’t know where it came MS. NOONAN: -- assign -- assign $200 million from. to it and then sell protection of 50 percent of it. mean, it seems so of out of the blue. I I mean, eventually we’ll get to the -MR. GORTON: us. The thing, it didn’t start with I mean it was a whole -MS. HEYL: It was a basic credit default, right? MR. GORTON: -- there’s a whole market that developed, right, that -- I’m not sure who started it. I mean, I know, there’s people outside the industry who’ve written about this, but I’m not sure they’re right. 15 FCIC Interview of Gary Gorton, May 11, 2010 So I don’t know who started it. But they’re -- they’re —- a market requires lots of participants, so you can’t -- you can’t do these things unless there’s lots of people -- firms willing to do this. So, the way the entire market started was with single name swaps and then came portfolio trades. MS. NOONAN: So your model then, if we take your example of the early genesis of this at FP —- was you4 model designed to then figure out where the, quote, unquote, attachment point would be on the —- whether AIG is going to write protection on 50 percent or whether it would write protection of 60 percent or 40 percent. MR. GORTON: Well, the –- MS. NOONAN: What was the model designed to do in that scenario? MR. GORTON: The model was designed to calculate a distribution of losses. So a dollar number of losses and a probability that you would have that loss, okay? And it was designed to do that in a setting which actually hadn’t happened in human history -— even now, I suppose -- because it was calibrated to the worst post World War II recession so that the mean would be that. And then you would have this tail, and then where we would attach was the number that Bob Lewis came up 16 FCIC Interview of Gary Gorton, May 11, 2010 with, the chief risk officer at AIG. He said, “I want you to attach at this point.” That’s what we’re going to call an acceptable transaction given that he believe the model was telling him meaningful information because it agreed with this other model he’d purchased. MS. NOONAN: So your model would spit out -- MR. GORTON: A distribution, a picture. A distribution. MS. NOONAN: A graph? MR. GORTON: Not a number. A pic- -- yes, think of it as a graph. MS. NOONAN: And it would say -- MR. GORTON: It would say -- MS. NOONAN: -- there’s X probability of this dollar loss. There’s Y probability of another amount of dollar loss. MR. GORTON: Yes. And then Lewis said that “I want it to be the case that 99.85 percent of the time, there is no losses.” That was his -- that number -MS. NOONAN: So the attachment point was whatever dollar amount correlated -MR. GORTON: Yes, yes. MS. NOONAN: -- to that? 17 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Yes. MS. NOONAN: What did you say, it was? MR. GORTON: 99.85 MS. NOONAN: 99.85. 98.5? And how did he come up with that number? MR. GORTON: don’t know. I don’t remember really. He -- I I mean, we had -- I mean he had looked at a lot of models. I don’t know where he came up with that number. MS. NOONAN: And what was the model he bought to check yours against? Vasicek. MR. GORTON: It was KMV. MS. NOONAN: KMV? MR. GORTON: Kealhofer -- somebody, and model. It was a model. What is that? It was basically a famous It was the Merton, Robert Merton’s, that won the Nobel prize. It was a famous paper he wrote. that model and implemented it with real data. Moody’s later bought it. They took And It doesn’t really exist anymore. So, so the KMV model was a slightly —- well, a very different methodology. idea. So it was kind of a good We always built two different models to -- you know, under different assumptions just to make sure you getting sort of in the same place. 18 FCIC Interview of Gary Gorton, May 11, 2010 So the parent —- I mean, we did that ourselves, but the parent company said, “Okay, let’s run this portfolio through your model. it through this model. We’re going to run And we are going to compare.” And for many years thereafter, the two models were compared. And if they didn’t match up very well, we spent a lot of time trying to figure out why. MS. NOONAN: Were the models similar in their MR. GORTON: No. MS. NOONAN: So they spit out the same thing, inputs? No, they weren’t similar. The KMV -- but they didn’t put in the same thing? MR. GORTON: Yes. Right. Some of it -- some of it was the same inputs, but a large part of it wasn’t. I didn’t like the KMV model. I thought it was too many assumptions and -– but -- but anyway, so that’s how we -- that how the whole thing was run. So for, you know, every transaction has an approval memo that goes to the parent company. And Lewis, who was the chief credit officer at the time or one of his assistants, you know, would have to sign it and read it and sign it. They would often call up with questions so we often reported – 19 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: And you drafted this memo? MR. GORTON: No. I usually wrote -— sometimes I wrote a small section, but it was somebody else. MS. NOONAN: Who would it have been? MR. GORTON: Well, it depends. I mean, there was a lot of people in this business, you know, who -so, it depends on whose -- what office it was, who was working on it, and… MS. NOONAN: So I think it makes sense, just for my purposes to understand -- in my mind, I already understand that there was -- and maybe there was more than one -- but at least there was a model that was used to evaluate the super-senior credit default swap, multi-sector CDO portfolio. MR. GORTON: Yes. That’s different. That’s comes much later. MS. HEYL: That’s later. This is the earlier period. MS. NOONAN: Exactly. I understand and so that’s where FP ultimately gets to -- and that’s obviously the focus of a lot of what ultimately happened with AIG-FP’s losses. MR. TOMBACK: To be really fair. They ultimately get there, as you put it, but there is no connection between Gary’s model or, I take it, the FMB 20 FCIC Interview of Gary Gorton, May 11, 2010 model -- although I’m not that familiar with it -- and the valuation model you are talking about. MS. NOONAN: valuation model. I’m not talking about the I’m talking about the model that was used when AIG-FP entered into these transactions. MR. TOMBACK: MS. NOONAN: Okay. I understand that at a later point in time, there was a model that was used starting in 2007 when collateral calls came in to start trying to value these things because they had been held at par, et cetera. I understand that that’s a different model than the model that was used, that I believe was your model that was used in making the business decision to enter into each of the transactions in the first instance. MR. GORTON: Well, I don’t think that’s exactly accurate. The model wasn’t –- there’s a long process that a transaction would go through. It would have to go through quite a bit before it got to the model. MS. NOONAN: Okay, and actually let’s table that because I want to talk about that, but I want to – I think it would be helpful – I know it would be helpful for me to map out if we know that this is one line of business, the multi-sector CDO super-senior credit default swap business. We’re starting from a point of 21 FCIC Interview of Gary Gorton, May 11, 2010 the corporate names, right? So, if you could walk me through maybe just a list of the ones that you recall, the different of types of transactions that you developed models for. MR. GORTON: Okay. MS. NOONAN: And whether the model was the same throughout and just tweaked, or whether you actually built different models for these different business lines. That’s something that I – I don’t know, and I don’t want to conflate what you’re saying about, sort of the ‘98, ‘99, stuff -MR. GORTON: Right. MS. NOONAN: -- with what happened later. I just want to make sure that I understand exactly what businesses you touched and your model was used for, and if you developed new models or tweaked them so that when we talk about assumption and inputs going into the 1998 model, that we can work from there on “Okay, were those same inputs used in these other areas?” Because I’m guessing that maybe there were significant differences in the inputs, but maybe that’s not the case. MR. GORTON: I mean, broadly speaking, there was two classes of models. There’s one big group that has to do with mortgages, right? And those are all 22 FCIC Interview of Gary Gorton, May 11, 2010 European mortgages. MS. NOONAN: model now? You’re not talking about your You’re talking -MR. GORTON: No, I’m talking about -– I’m talking about another model that we built. MS. NOONAN: Okay. MR. GORTON: I’m not exactly sure that we should call it “my model,” by the way, but, you know… MS. NOONAN: The popular – the popular parlance has been the “Gorton model” or the “Wharton model,” so I mean we can give it -MS. HEYL: MS. NOONAN: I think it’s an actuarial model. MS. HEYL: An actuarial model, okay. MR. GORTON: As compared with a valuation model. I think if people of the firm heard you say that, they’d be irate. So, you know, there was a lot of people who contributed to -MS. NOONAN: Well, I’m not going to take credit for being the first person to say it, but -MR. GORTON: In any case, there’s -- there was another -- you know, the bulk of the business basically had to do with European mortgages eventually -- you know, for most of the time. And so those are quite different, depending on the country they’re in. And so 23 FCIC Interview of Gary Gorton, May 11, 2010 there was a different model for each country. And then the basic corporate model was then adapted for later, for the CDO business. MS. HEYL: Can I just try to clarify this? So initially, you have the model that finds loss distributions on corporate names? MR. GORTON: MS. HEYL: Right. Is that the same -- is that what we call sort of the ER [phonetic] model, like the beginning model; and then the others come from it? MR. GORTON: No. The basic idea -- since the business was -- you know, we want to know the loss distribution actuarially to make this underwriting decision, the output always had to be a distribution. But the mortgage models were very different because mortgages are just different than, you know, corporate names. So corporate names -- so that was, you know, in spirit, a similar kind of model, but the inputs and so on were different MS. HEYL: So if we look at the genealogy of the models, like it’s animal kingdoms or something. MR. GORTON: MS. HEYL: Yes. MR. GORTON: You have the -The corporate model. 24 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: MR. GORTON: -- the corporate model? MS. HEYL: Then you have mortgage models. MR. GORTON: Separate, next to it? MS. HEYL: Yes. MR. GORTON: Okay. And then later, this CDO stuff comes in. MS. HEYL: MR. GORTON: It descends from which one? It descends from the corporate one. MS. HEYL: All right, so let’s not talk about the European one. MR. GORTON: Okay. MS. NOONAN: So you have the corporate model, you have the mortgage model, and the CDO model descends from the corporate model? MR. GORTON: Right. MS. NOONAN: Even though CDOs often contain mortgage securities? MR. GORTON: though. Or referenced mortgage securities? They -- that’s the difference, That’s exactly the difference. In Europe, you’re referencing actual mortgages, right? In CDOs, you’re not referencing mortgages; you’re referencing bonds, some of which might be backed by portfolios of mortgages. 25 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: So to answer her question, how does it change from the one that’s for corporate bonds -MR. GORTON: MS. HEYL: Well, the corporate --- to the one that’s for CDOs? MR. GORTON: based on ratings. ratings. The corporate bonds model is But we never used, you know, their We had our credit officers rate them. The CDOs were based on ratings, but we never believed the absolute meaning of the rating. We just believed the relative ranking of the rating. And then we stressed those ratings. MS. HEYL: Okay, so did the model change between the time you used it for the corporate bonds and the time you used it -MR. GORTON: MS. HEYL: Yes, the model changed. Okay, so how did it change? MR. GORTON: It changed by adding all these additional stresses because we were -- so the ratings -the rating -- the part of the ratings that we liked, was the fact that the agencies could tell you Company A is riskier than Company B. So that ordering we used. Whether Company B was ten times riskier than Company A or not, that’s the part we didn’t believe. So we -- when we got to the CDOs, we added this big layer of stressing that, again, was worked out 26 FCIC Interview of Gary Gorton, May 11, 2010 with the parent company. MS. HEYL: So why don’t you tell her how that stressing works? MR. GORTON: Okay, so the way one of these CDOs would work would be, come -- an investment bank would say, “Here’s the proposed portfolio rules,” okay? So they’d say, you can have 10 percent in prime mortgages, 10 percent in auto loans, 10 percent in this. You know, they have a whole, big list of rules. And the rules would have -- they’re just a whole, big page of things -- pages, sometimes, of rules, right? MS. NOONAN: Investment criteria? MR. GORTON: Yes. And the reason there’s rules is because they haven’t purchased all the bonds yet. So they don’t necessarily know what bonds they’re actually going to buy, but they know they have to satisfy these rules. And moreover, sometimes there’s a manager who can actually trade things, right? And so we had lots of rules that had to be worked out and, oftentimes, deals wouldn’t happen because they didn’t like our rules. Our rules were much tougher than the rating agencies. For example, we insisted that managers own 27 FCIC Interview of Gary Gorton, May 11, 2010 half the equity, and we insisted on all sorts of triggers that the rating agencies didn’t want to include. And then we would stress the ratings. So when we would simulate, we’d go to the first bond and we’d look at it, and it says that it’s a student loan -- backed by student loans and it’s rated A. So when we simulate it, 20 percent of the simulated paths would bump that bond from A to Bbb. MS. NOONAN: The next notch down? MR. GORTON: The next full notch down -- MS. NOONAN: Okay. MR. GORTON: -- right? So this -- this -- these stresses -- so people refer to that as a 20 percent stress-down. A full notch. So basically you’re saying, “I’m not sure the agency’s got this right when they called it A. think they really understand the stuff. I don’t We’re going to stress it down.” So there’s a whole series of stresses that were worked out, and then these additional criteria that would have to be adhered to. So many transactions would fall apart because counterparties would complain that, you know, we -- this thing had already been rated, and what was wrong with us, and so on and so forth. 28 FCIC Interview of Gary Gorton, May 11, 2010 So -- so all these -- all these things were added, you know, in consultation with the parent company, because the parent company wasn’t going to let us do this business unless we explained and they agreed with all the criteria. MS. NOONAN: one second. Okay, that’s -- going back just The mortgage model, that was used for European deals? MR. GORTON: Right. And they’re very different because there, you’re dealing directly with the bank, and you know there is these 80,000 mortgages, and you know lots of information about each borrower. And they already have the mortgages. Whereas with the CDOs, they have to go buy the bonds. MS. NOONAN: Okay, okay. So the mortgage model was used for the European business -MR. GORTON: Correct. MS. NOONAN: -- which I believe is referred to as a “regulatory arbitrage business”? MR. GORTON: Yes. MS. NOONAN: Is that the same, or is that MR. GORTON: No, that’s the same, that’s the MS. NOONAN: Okay. different? same. 29 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: This was a bank equity business in economic substance. MS. NOONAN: Okay. MR. GORTON: Which everybody understood. MS. NOONAN: Would you mind, actually, if we took just a brief break? I think that we actually have a colleague who would like to dial in. Is that possible? MS. HEYL: MS. NOONAN: Sure. Go ahead. Let me -- should I use that phone over there? MS. HEYL: MS. NOONAN: Who is that going to be? Al Crego. -- BREAK -- MS. NOONAN: MR. CREGO: Hi. MS. NOONAN: Yes. Is this Al? Hi. Hi, Al. Good morning. This is Dixie and Clara. We’re here with Dr. Gorton and Andy Tomback and Dorothy Heyl and Tom Santoro of Milbank. Is there somebody else there with you? MS. HEYL: MR. CREGO: Hi. MS. NOONAN: I’m waiting for folks to come in. Okay. 30 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: So what’s Al’s full name? MS. MORAIN: Alberto Crego. MR. CREGO: Do you want to give us a couple more minutes? MS. NOONAN: MS. HEYL: Well, actually, we’ve already -Maybe we should take a break. MS. NOONAN: Yes, we can give you a couple more minutes. MS. HEYL: We’ll take a break. MR. CREGO: Thank you, Dixie. -- BREAK -- MS. NOONAN: model. -- talking about Dr. Gorton’s And so I’m not going to back through that. We can fill you in later. MR. CREGO: MS. NOONAN: MR. CREGO: MS. NOONAN: That’s fine. Okay, cool. Thanks. Thank you. So, back to the models. You said that there was criteria that the investment bank would have in selecting, I think, the underlying assets for the deals, and that AIG had criteria that sometimes or maybe often differed, and that sometimes that worked to derails the transactions. 31 FCIC Interview of Gary Gorton, May 11, 2010 Could you just give us some examples of criteria that you might have had that would have conflicted with -- or even specific examples of deals you know that didn’t happen because of criteria that conflicted? MR. GORTON: Too many deals were rejected. I can’t remember a specific deal. I think the main -- one of the main problems was this requirement that the manager owned half the equity. And that trading stock -- you see, we had -- they’re allowed to trade certain amounts for a certain period of time, and we didn’t like them to trade. So we required that they hold equity and we put in many more triggers to stop the trading. They could sell, but they couldn’t buy. MS. NOONAN: So this is the difference between what’s called a dynamic or a static pool? Is that what you’re referring to, or -MR. GORTON: MS. HEYL: Yes, the static is -Actively managed. MR. GORTON: There’s actively managed and MS. NOONAN: Okay, okay. static. I’ve heard some people refer to actively managed as a dynamic pool, but -MS. HEYL: Actively managed in the sense of a 32 FCIC Interview of Gary Gorton, May 11, 2010 reinvestment period. And you say you didn’t like that. MS. NOONAN: And so, do you know if most of the deals that FP did were static pools? Is that your understanding, or… MR. GORTON: I don’t remember. MS. NOONAN: Okay. Would you have been involved in the negotiations of -- in that part of the negotiation? MR. GORTON: No. I mean, I basically, you know, from roughly 2002 or 2003 to the end, I didn’t really work on this anymore. I mean, I was working on commodity futures, and I had another big project, but, you know, this was done by the credit team. MS. NOONAN: Okay. Okay. So going back to -- so the corporate model is used for the CDOs. So I’m interested in if you could walk me through, and I know this may be slightly painful given my lack of experience in quantitative models, but I would like to have an understanding of what inputs you put into the model. What was the criteria that went in that affected the distribution? MR. GORTON: So the -- I mean, very roughly speaking, there’s just two pieces of information that you need for each bond. You need its rating and you need to have an assumption about its recovery rate. 33 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: What do you mean by recovery MR. GORTON: If there’s a default, how much do rate? you -- are you going to get back? MS. HEYL: MR. GORTON: In other words -- MS. HEYL: So, for --- it’s like in a bankruptcy proceeding. MR. GORTON: It’s a bond, so if the bond, yes, if the bond suffers losses, if it gets to the point where it suffers losses, how much are you going to recover? MS. NOONAN: And how would you go about making that assumption? MR. GORTON: punitively. Well, you make them very You just say you’re going to recover zero, unless it’s so -- for most ratings, it was a zero. And then for, you know, I forget the exact numbers, but for Aaa and Aa, it was positive, but it was like 20 percent. So it was way below anything that we’d ever seen before. MR. CREGO: MR. GORTON: MR. CREGO: Professor Gorton? Yes. Did I hear you say that the first primary element of the model input was the credit rating from an NRSRO? 34 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Yes. MR. CREGO: Okay. Is there any -- is there any sensitivity given the inherent lag in some of the ratings? And at the time that they had not fully updated their rating models, do you account for their rapidly deteriorating underlying collateral? MR. GORTON: So you missed the first part of this, so let me -MR. CREGO: I apologize. MR. GORTON: --just quickly summarize. First of all, these are new deals, so… MR. CREGO: Okay. MR. GORTON: In fact, many of them haven’t even been issued yet when you’re analyzing the transaction. The second thing is, we didn’t believe the ratings. We never, in the entire history of this business, believed the ratings. So, either we had our credit officers give their ratings in the case of corporates, or we stressed downwards the agency ratings in the case of structured products. So the thing, the piece of information that we take from the agencies is the relative ranking. Not the rating, the relative ranking. So we think -MR. CREGO: Okay. 35 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: -- that they do get it right, that Bbb is in fact worse than A. MR. CREGO: MR. GORTON: Gotcha. That’s the only piece of information that we take. MR. CREGO: MS. NOONAN: Okay. Thank you. So, on the second part, the assumption about the recovery rate if there is a default? And you said for most it was zero, and then if it was an Aa or an Aaa-MR. GORTON: Yes, I don’t remember exactly--I can’t remember what the numbers were--it’s like 10, 20, 30, I can’t remember, but it’s something like that. MS. NOONAN: So, I’m just -- and I apologize, but that seems pretty stark that the assumption would be zero. I mean, doesn’t it then make the assumption about if and when there’s going to be a default -- it seems like there are two assumptions embedded there, right? What would the recovery rate be if there was a default, and if there’s a default is a second assumption really, right? I mean, how do you -MR. GORTON: Yes. I mean, if it’s a default, then what’s the recovery rate? That’s the -- it’s conditional on default that you would recover. MS. NOONAN: Right. 36 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: I want to make sure I understand the question -MS. NOONAN: So -- so, is it an input into the MR. GORTON: Yes. MS. NOONAN: -- is there an assumption input model -- into the model about the probability of a default? MR. GORTON: No. That’s what you’re going to simulate from the ratings, right? So, you have the history of ratings, and you know this is the information you’re going to use to simulate, right? So ratings can go up and they can go down and you simulate that under this assumption that you’re in the worst post-war recession in the U.S. for the entire life of the transaction. MS. NOONAN: So when you say you assume the worst post-war recession in the Unites States, does that differ depending on the asset class? MR. GORTON: No. MS. NOONAN: So what is it? What time period do you use? MR. GORTON: I can’t -- I think it was ‘73- ’74, it was early seventies. I don’t remember the exact dating of the recession, but… Anyway, so the model was calibrated to that, 37 FCIC Interview of Gary Gorton, May 11, 2010 and with the parent company’s agreement, we stuck to that the whole time. MS. NOONAN: For these U.S. things. Just trying to figure out how that gets fed into the model. I mean, I understand sort of qualitatively looking that time period, but is it the performance of bonds at that time period, or is it -MR. GORTON: go up or down. So, it’s -- it’s -- ratings can And whether they go up or down is a function in part of what the overall economy is doing. So you pick periods where most of the ratings go down, until you get a loss rate, an average loss rate that’s equal to the worst post-World War II recession. Then, in the distribution, the average is that, but the tail, the really unlikely events, are huge losses that we never experienced in history, but which is what you’re using for the underwriting standard. MS. NOONAN: entirely understand. Okay. Okay. I’m not sure I I mean, is it basically that you take something that was rated Aaa in, say for example, 1970, and in 1973 it was rated Bbb and it experienced losses of X amount and then you -MR. GORTON: You track -- you track what ratings actually did, right? So you have millions of these ratings transitions, right? So you track these paths, and you simulate these paths. So you draw a 38 FCIC Interview of Gary Gorton, May 11, 2010 path, you see where it went; and if it hits -- you know, it defaults, then you figure out how much you lost and you accumulate that for all the however many bonds there are. MS. NOONAN: Okay. Okay. So the inputs are the rating, the relative rating with some stresses -MR. GORTON: Uh-huh. MS. NOONAN: -- and assumptions about recovery MR. GORTON: Uh-huh. MS. NOONAN: -- and then the exports, I think rates -- you’ve already said this, but it’s the probability of loss –MR. GORTON: It’s the – MS. NOONAN: -- it’s the dollar amount of MR. GORTON: The distribution. MS. NOONAN: -- distribution of loss. loss -- Which would tell you the dollar amount at a given probability. MR. GORTON: Yes. And then for our piece, 99.85 percent of the time, we have to have zero losses. MS. NOONAN: And that’s where the attachment point comes in? 39 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Well, that’s not the attachment point, but that’s the minimum, right? So we always asked for a buffer, right? Remember the model -- you’re attaching way too much weight to the model. The way the business actually worked is, I go, the model says to Al Frost: 20 percent. Attach at Al calls his counterparty who says it’s 30 percent. MS. NOONAN: Okay. So we’re going to get to that. So you said that the model comes in at the end, so now I would like to go back to that process, sort of the whole deal process. What you know of it -- I understand that you weren’t involved in all of it -MR. GORTON: Yes. MS. NOONAN: -- but it sounds like you have a general familiarity, and -MR. GORTON: Yes. MS. NOONAN: -- that’ll help us as we go throughout the course of the rest of the day, I think. If you know, if you can give us sort of a walk-through of the steps. MR. GORTON: Yes. I mean, as you say, I’m not sure I have all the details, but my general understanding is, you know, a trade would come in to 40 FCIC Interview of Gary Gorton, May 11, 2010 somebody in some office, some bank would call up. And they would, you know, e-mail the criteria, and you know, maybe they had a term sheet or something, and the person, you know, in that office, is going to look at this and if they’re not a credit person, they’re going to take it to the credit person. The credit person is going to say “This is not going to work. these things.” Tell them And there may be -- MS. NOONAN: So all deals would eventually come to the credit person to -MR. GORTON: person, right? All deals pass through the credit So there’s going to be a credit person who’s going to, you know, be the point person on the trade. MS. NOONAN: And who were the credit people MR. GORTON: Well, it’s a changing group, you involved? know, over the years, I mean-- name. MS. NOONAN: So starting in ‘99? MR. GORTON: Yes. Jesus. So -- I can’t remember his The guy who –- Few- -- no, Forster replaced a guy who -MS. HEYL: MR. GORTON: Tom Fewings? remember. No, not Fewings. I can’t Anyway -41 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: We can find that out separately. MR. GORTON: Yes, I can’t remember. MS. NOONAN: So it would have been -- it would I can’t remember. have -- Andrew Forster would have replaced this person? MR. GORTON: Yes. Forster eventually replaced this guy, I can’t remember his name. MS. NOONAN: Okay. MR. GORTON: But anyway, so that -- he was the head of credit trading. officers. And then you have credit So the credit officers were always involved, and then you have –- so, the credit group was, you know, a group that had people in every office. And, so there would be, you know, a whole series of iterations about their criteria versus our criteria. MS. NOONAN: On the underlying -- MR. GORTON: The rules. MS. NOONAN: -- on the type of assets. MR. GORTON: The type of assets of the rules. And, you know, I don’t know how many trades died at that point because I wasn’t, you know, typically involved in that process, but if they got a trade that would satisfy their rules, then they would put it through the model. 42 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: And so the credit department really was the one who issued the rules on the criteria? MR. GORTON: No, the rules were developed -- so there was a group that met every quarter to review the portfolios, to talk about issues in the business, and to discuss every aspect of this. And this group, you know, originally was maybe 12 people and it grew to, you know, it would be like 25, 30 people including people from the parent company and so on. And -MS. NOONAN: Were you on that group -- in that MR. GORTON: Yes. group? This was the quarterly meetings, I mean, I wasn’t always there for the meeting, but I would try to go to the meetings. review. So this was this There was this continuous review process, and this group also, you know, considered all sorts of things about the business. So I don’t really remember exactly how these rules were, how they came into being. MS. HEYL: What was this group that held the meetings? MR. GORTON: This was the quarterly -- it was called the quarterly credit meeting -MS. HEYL: MR. GORTON: I see. Okay, thank you. -- was the name of it. 43 FCIC Interview of Gary Gorton, May 11, 2010 So I’m not sure where the rules came from, but we had these rules. So if you -- MS. NOONAN: Who led those meetings? MR. GORTON: It was Forster, mostly. would be a big conference call. But it And sometimes the parent company people would come to Connecticut and join us for that call. MS. NOONAN: Who from the parent company would MR. GORTON: Well, it was typically Kevin have come? McGinn in the later parts, and then in the earlier parts, before – before -- in the early days, Cassano and I would actually go to the parent company and sit there for like half a day and discuss the whole business with them. MS. NOONAN: With “them” meaning still Kevin McGinn, or… MR. GORTON: No, then it was Lewis and Weibel. MS. NOONAN: So this is the chief risk MR. GORTON: Yes, Bob Lewis, at the time I So Bob -- officer? started, was the chief credit officer. he becomes the chief risk officer. And then later, And he’s replaced by Kevin McGinn. 44 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: As chief credit officer? MR. GORTON: As chief credit officer. But that was probably 2004, something like that. Anyway, so, you know, the transaction gets to the model stage and sometimes, you know, our attachment number would be so far off from where they wanted us to attach that a trade would die also. So, you know, my guess is most trades died. MS. NOONAN: So it sounds like there were – there was maybe a stage, at least one stage outside of AIG-FP where the investment bankers are putting together a deal, it gets somehow in the door, and it makes it way to the credit department? MR. GORTON: firm, right? Yes, I mean, this is a small So, you know, it’s pretty easy. MS. NOONAN: It’s fluid? MR. GORTON: Yes, I mean -- MS. NOONAN: Okay. MR. GORTON: -- you’re all sitting next to each other. MS. NOONAN: Okay. And then they decide up or down, yes this works, no it doesn’t. If it does work, then they bring it to you. MR. GORTON: No. I -45 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Okay. MR. GORTON: --I -- there’s -- the model was built from scratch in every office. And there’d be some -MS. NOONAN: In every office? MR. GORTON: Yes. MS. HEYL: He developed the model. He didn’t necessarily run the model. MR. GORTON: MS. HEYL: I didn’t run the model. MR. GORTON: Others ran it. Right. I’m not doing transactions. MS. NOONAN: Okay. MR. GORTON: So there’s U.S., London had the model, and Tokyo had the model. MS. NOONAN: Okay. MR. GORTON: Right? So there’s people who run the model in these various offices, so -MS. NOONAN: But we’re still working on the two basic frameworks for the models? MR. GORTON: MS. HEYL: Yes. Two being what? MS. NOONAN: The corporate and the mortgage. MR. GORTON: No, no. We’re talking about CDOs. 46 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: Right now, we’re just talking about CDOs. MS. NOONAN: Oh. Oh, okay. Okay. So we’re just talking about the CDO model and there were different CDO models for the different offices. MR. GORTON: No, no, no. They’re all the same MS. NOONAN: Oh, I thought you said that they model. were built for each office. MR. GORTON: They were built -- the model was built, so you have to make sure you don’t have coding errors; right? once; right? So you always build models more than So you don’t want the computer program to contain a mistake; right? So, in Wilton, two people could separately write the program. London writes their own program. Tokyo writes their own program. MS. NOONAN: And then we compare. So you basically write the instruction booklet on how to build the model -MR. GORTON: Yes. MS. NOONAN: -- and then they go build it. MR. GORTON: Yes. MS. NOONAN: And then are you involved in Right. testing it? MR. GORTON: Well, we tested it -- you know, 47 FCIC Interview of Gary Gorton, May 11, 2010 that’s the part that involved this interaction with the parent company for a long time; right? So nothing gets done until the parent company agrees, and then, you know, then you still have to go through this whole process, write the approval memo, and the parent company has to agree. MS. HEYL: I’ll ask something. Did you write an actual instruction manual on how to do the model? MR. GORTON: I wrote lots of things over the years. MS. HEYL: MR. GORTON: What, like memos? MS. HEYL: Yes, memos. MR. GORTON: Right, various memos -Big memos. Lots of big memos, mostly, you know, for the parent company and internally and, you know, you had to have something to send to the other office, and the people downtown wanted to build a model, they needed to see it -MS. HEYL: So various people built models based on these memos you wrote? MR. GORTON: Right. They built the same model, right? MS. NOONAN: Do you still have copies of those MR. GORTON: I don’t -- I don’t know. memos? I don’t 48 FCIC Interview of Gary Gorton, May 11, 2010 know. I’m sure AIG has them. MS. NOONAN: Right. MR. GORTON: I don’t know if I have them. MS. NOONAN: I just am wondering if it would be more expeditious on our part if you had them -MR. GORTON: I can look. The problem is my whole e-mail -- you know, I just left at AIG all my files, I didn’t take that with me. I just, you know, one day I stopped going there, and that was just it. MS. NOONAN: When did you leave AIG? MR. GORTON: Well, I didn’t -- I mean-- MS. NOONAN: When did your relationship with AIG formally end? MR. GORTON: Well, it ended after the second retention payment, but I actually didn’t go to the office some point, like, I probably went in 2008 and I didn’t go any further. MS. NOONAN: payment paid? And when was the second retention That was in two thousand -- MR. GORTON: Well, that was like, just, like MS. NOONAN: Of this year? MR. GORTON: Of this year. MS. NOONAN: But you didn’t do any work -- January. MS. HEYL: You haven’t been to the office 49 FCIC Interview of Gary Gorton, May 11, 2010 since -MR. GORTON: I haven’t been to the office in, I don’t know, probably two years. MS. HEYL: And when’s the last time you did work for them? MR. GORTON: Three years, or something. mean, I would go into the office. should I show up? I I asked –- I asked, What should I do? MS. NOONAN: Well, can we be a little more specific, because I mean, three years, there was a lot going on, sort of, three years ago. MR. GORTON: So there’s -- Yes, I don’t really remember. I mean, I was there -- in 2008, in the beginning I was certainly there. I probably, I think I went in through 2008, but I didn’t have anything to do for a good part of that time. And then I just stopped going in. I talked -- they put in this new head of the company, and I said, “What do you want me to do?” And he said, “Well, you know, we’ll call you if we need you.” said, “Fine.” That was it. I So… And then I sat there -MS. HEYL: But as far as the documents, we can talk about that. MR. GORTON: MS. HEYL: Yes, the documents -I mean -50 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: MS. HEYL: -- I can look, but I --- I don’t have copies of memo, or I’d go and get you one this week. MS. TOMBACK: I think in fairness to AIG, we would take the position that the documents belong to it, not to Gary. So I think that we’ll all end up going back to AIG to get the paper. MS. NOONAN: Well, we could certainly get their permission for you to give it to us. MR. TOMBACK: MS. NOONAN: Right. But I’m just -- they’re actually collecting quite a few documents for us, as you can imagine. And if you happen to have them in a way that was easily accessible and we could get them more quickly, it would be very helpful to us. MR. TOMBACK: MS. NOONAN: Absolutely, we’ll look and -And so, and I don’t think it would be an issue with them to get that permission, as long we worked through you and they got copies, and things like that. MR. TOMBACK: We’ll look. They authorize. You’ll get. MS. NOONAN: MS. HEYL: That would be great, yes. That’s not clear that -- I mean, did you have -- you see, part of the issue has to do, 51 FCIC Interview of Gary Gorton, May 11, 2010 because I know we’ve been through this before with him. The move from Wharton to Yale is also relative -- very relevant to documents, I think. MR. GORTON: Right. So a lot was lost -- MS. NOONAN: When did that happen? MR. GORTON: August 2008 -- MS. NOONAN: Okay. MR. GORTON: -- I moved -- MS. NOONAN: Okay. MR. GORTON: -- and I lost a big chunk of e-mail files. MS. HEYL: Meaning, you left them at Pennsylvania? MR. GORTON: Yes. I mean, I, you know, I mean I didn’t take the computer from my home computer. somehow -- I don’t know what happened. So Anyway, I lost a bunch of files and I lost a bunch of paper things in the move somehow. I don’t know where they are. But I have -- I have files from those days still, but I don’t think there’s anything in there, because I’ve looked for other things that people have asked, but I can go look again. MS. NOONAN: Okay. MR. GORTON: But most of it should be -- I mean, all my old, you know, files are on some computer 52 FCIC Interview of Gary Gorton, May 11, 2010 at AIG, and certainly these memos, I mean, I don’t know how many times I sent the same memo downtown. MS. NOONAN: But you would have drafted it on your computer at AIG? MR. GORTON: Yes. MS. NOONAN: Not on your personal computer? MR. GORTON: Right. Exactly. And there –- there are descriptions of, you know, what the model does, basically. And I didn’t write all of them. Sometimes there was somebody I was working with who wrote part of it. MS. HEYL: So as far the CDO modeling, when would you have written a memo about that? MR. GORTON: I don’t think -- I’m not sure I ever wrote a memo about CDO modeling. MS. HEYL: All right. Did you ever send another memo you wrote about modeling in general to people who were working on the modeling of the CDOs? MR. GORTON: No, because it was the corporate model and then we agreed on these stress and recovery rates, and we didn’t need a memo for that. MS. NOONAN: So the only change to the corporate model when you started -- when AIG started using it for the CDOs, was different stresses and 53 FCIC Interview of Gary Gorton, May 11, 2010 different -MR. GORTON: Right. MS. NOONAN: -- assumptions about the recovery MR. GORTON: Right. rates? happened outside the model. And all the other stuff So much more happened outside the model for the CDO business -MS. NOONAN: In terms of the criteria MR. GORTON: Yes, right. selection? So that -- I don’t know that -- I don’t know if that was ever written down frankly. It wasn’t a model -- it wasn’t a technical thing, so… MS. HEYL: Okay, but you’ve written a lot of memos about the model? MR. GORTON: MS. HEYL: Yes. And so, the question is, you know, are there e-mails -MS. NOONAN: Or if you could give us a timeframe when these –- when you would have been writing these, sending them, because -MR. GORTON: I mean, the basic memo, I mean, it’s just a guess, but my guess is that it’s prior to 2002 or 2003. So there’d be, you know, a basic memo about the corporate model, and a memo about the mortgage 54 FCIC Interview of Gary Gorton, May 11, 2010 model. So those are the two memos. But then over the years, for various purposes, I wrote, you know, lots of memos about the business or, you know, at least how the models work for various people -MS. NOONAN: And would these have come out of those quarterly meetings? MR. GORTON: No, no. These were things like, for example, the parent company hired a more quantitative person who said, “Can you send me the memos?” MS. NOONAN: Okay, and his or her name was? MR. GORTON: His name was Paul -– what’s his name? MS. HEYL: MR. GORTON: Narayanan? Narayanan. Right. So at some point they hire this guy -MS. HEYL: MR. GORTON: N-A-R-A-Y -- Narayanan. So at some point, we send him all the memos and we send him all the inputs, you know, we get him tooled up. MS. NOONAN: Okay. MR. GORTON: So just as an example, so there were many such examples. MS. NOONAN: Okay. So, but basically by 2002, 55 FCIC Interview of Gary Gorton, May 11, 2010 2003, the models were designed, written, up and running for the CDO business? MR. GORTON: Yes. MS. NOONAN: Okay. Somewhere in there, yes. And that’s about the timeframe when they started doing the CDO business? MR. GORTON: I think that’s right. MS. NOONAN: Okay. Okay. Were you involved when -- wait a minute, let me make sure -- I may come back to a few things about the model, but maybe we’ll take a break and I’ll look at my notes. Were you involved in -- when would you -- when did AIG-FP get involved in the multi-sector CDO business? MR. GORTON: Well, that’s what we were just talking about, so -MS. NOONAN: So that’s the 2002, 2003? MR. GORTON: I think that’s right, yes. I mean, Frost would probably know better than I -MS. NOONAN: Okay. MR. GORTON: -- but that’s my guess. MS. NOONAN: And when, if you know, would those CDOs have started containing assets that were backed or related to subprime or Alt-A U.S. residential real-estate mortgages? MR. GORTON: I don’t know. I mean, you know, 56 FCIC Interview of Gary Gorton, May 11, 2010 I wasn’t looking at all these transactions. And then there comes a point where Gene Park takes over. And Park comes over to my desk one day and says, you know, “What is this stuff?” MS. NOONAN: Takes over what? MR. GORTON: He takes over Frost’s -- Frost was running this business in the U.S. Al Frost. And Frost was going to go work on some other stuff, and Gene Park is going to do this business. MS. NOONAN: And what do you -- roughly a MR. GORTON: Oh, gosh. MS. NOONAN: Was it baseball season? timeframe? MS. HEYL: MR. GORTON: January 2006? So, I think it must have been, yes, January 2006 or before that, it must have been -MS. HEYL: MR. GORTON: Late 2005? Late 2005. Late 2005. Right. Park takes over, and Park comes over and he says, you know, “What is this stuff?” So I’m explaining to him what it is. And he says, you know, the trades we’re seeing have a lot of this stuff. MS. NOONAN: MS. HEYL: What was Park -“Trades we’re seeing” meaning? 57 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: The stuff that were being shown to us. MS. HEYL: “Stuff we’re seeing,” new trades? MR. GORTON: Yes, new trades. MS. NOONAN: I’m sorry. MR. GORTON: The new transactions that the I didn’t quite follow that. investment banks were showing to us are increasingly these kinds of assets. MS. NOONAN: Which kind of assets? MR. GORTON: Subprime and Alt-A. MS. NOONAN: Okay. Okay. What was Gene Park’s role before he took over Al Frost’s role? MR. GORTON: Well, he was part of the credit team, but, you know, he did a variety of transactions, different kinds of transactions, not only credit. I think officially he was in the credit group. But I mean, the groups are very fluid, so many people worked on multiple things. MS. NOONAN: So, let’s step back a little bit. If you developed the model for CDOs around 2002, 2003, and if you ever remember a more specific time period, please -MR. GORTON: Okay. MS. NOONAN: -- just let me know, and we can 58 FCIC Interview of Gary Gorton, May 11, 2010 be more specific. What was your involvement after building the model when the deals came in? MR. GORTON: involved. I typically wasn’t really I mean, I was -- I would go -- so I’m basically working on commodity futures, and -- but we sit on these trading floor, so you know, I would talk to, you know, all sorts of people, and I would try to go to the quarterly meetings. I think I probably did go to every quarterly meeting or I called in. And the quarterly meetings were to review -you know, it was a big preparation for the quarterly meeting, because you would re-run the model and see where the transaction stood now, you know, one year, two years, five years later. And there would be all this discussion about it, and there’d be this intense focus on transactions that looked like there’d been a little deterioration, and, so I would go to that. But, you know, by this time, you know, Adam Frost -- Al Frost and Adam Budnick -- Adam Budnick had been hired at some point -- and they worked on this. They -- you know, I didn’t really -- they didn’t need me. So I didn’t have any input on it. MS. NOONAN: So you never ran the models, say, from January 2004 after? 59 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: I don’t think I ever ran a model. I never –- I never actually physically ran the model. MS. NOONAN: Like, gathered the data that would go into it, and -MR. GORTON: Well, I gathered the data, and I wrote memos saying, “Here’s how to build the model.” And then somebody would go -- we’d run all these experiments, and somebody would go run it and report back. And then we’d go to the parent company and discuss it, and so on. But in terms of, you know, literally running a model -- like I type on the computer, input the portfolio, get the results -- no, I never did that. You know, I don’t even know how to do that. So, but beyond that, I mean -MS. NOONAN: It’s interesting. How do you -- I mean, really, I’m just curious, how do you know how to write a model but not run the model? MR. GORTON: It’s very simple, right? half of my academic work is math. I mean, I do math, right? You write math down. So a model is a lot of math, and that’s what you write down. You write down the math, you write down how you use the data, you write the memo, you hand it to a programmer, and you hand it to another programmer, and 60 FCIC Interview of Gary Gorton, May 11, 2010 you say, “Go do this.” MS. NOONAN: Aren’t you curious about how -- MR. GORTON: It’s not -- it’s very easy for them, right? They’re computer programmers. They take the math, and they write it into computer -MS. NOONAN: I’m not talking about the computer programmer part. I’m talking about once you actually get to the deal, and you put in the ratings, and you put in the assumptions about the recovery rates -MR. GORTON: Right. No, there was in the beginning, there was a lot of discussion about this, right. And what these numbers looked like, and whether they were reasonable and what the stresses should be, and what would it look like if we had bigger stresses, and… So the sensitivity of the model to various inputs was something that we studied, right? But we had to study it and we had to report to the parent company all these experiments so that we all sort of understood, you know, how this things behaved in order to finally get to the point where you were willing to write business on the model. And then once that happens, I typically wasn’t involved, although, sometimes I would write the little paragraph in the approval memo saying, 61 FCIC Interview of Gary Gorton, May 11, 2010 you know, “Here’s the various scenarios we ran. Here’s the model numbers.” MR. TOMBACK: And just to be clear, in terms of stresses, was one of the stresses that you ran against the model or ran through the model, significant economic downturn? MR. GORTON: No, the model was in a significant economic downturn. That’s what, you know, you didn’t have to model that because the model was in a significant economic downturn the whole time. MR. TOMBACK: Right. So, the model itself has integrated into it -MR. GORTON: Yes. It’s built in that the entire life of the transaction is in a significant economic downturn. MS. NOONAN: Have you gone back and compared what the early seventies looked like to 2008? MR. GORTON: No. I mean, I -- I’ve asked, you know, how -- you know, are there any losses, and the answer was: No, we don’t have any realized losses in any of these positions. I mean, the company had a loss in Maiden Lane III, I’m told, because of the way the accounting worked, how they priced it or something. I don’t really know the details. 62 FCIC Interview of Gary Gorton, May 11, 2010 But a few months ago is the last time I asked. And I said, “Do any of our positions have any realized losses?” And the answer was no. So, you know, and I was curious about that, because, you know, the model had been predicated upon: happened? Could you survive exactly what’s And at least so far, the answer appears to be yes. I mean, there’s no -- no dollar has been paid out to a counterparty under one of these swaps. MS. NOONAN: Not under the actual terms of the swap as opposed to the collateral -MR. GORTON: Yes, the collateral was something -MS. NOONAN: -- provisions. MR. GORTON: -- completely separate. MS. NOONAN: Right. Right. Did you know that the model for the CDOs -did it make any difference to you what assets were in the CDOs? Whether it was -MR. GORTON: a difference, right? Oh, yes, yes. Absolutely it made I mean, we -- we, you know -- MS. NOONAN: I mean, as long as they have a MR. GORTON: No, no, it’s not as long as they have a rating, right? The criteria, you know, required rating. 63 FCIC Interview of Gary Gorton, May 11, 2010 various category -- different categories of, you know, asset classes and so on. MS. NOONAN: The criteria for the model or the criteria for getting the deal done for -– for -MR. GORTON: The criteria -- the model just takes in the ratings, so this is outside the model. MS. NOONAN: So what’s -- right. So putting aside what’s outside the model -MR. GORTON: Right. MS. NOONAN: -- for a moment, do you -- does the model -MR. GORTON: No, no. MS. NOONAN: -- care about the type of asset? MR. GORTON: No. MS. NOONAN: Okay. MR. TOMBACK: So -- You’re asking like, if you put in Fff as opposed to Aaa assets-MS. NOONAN: No, actually that’s not what I mean, because I think that actually would be taken -MR. GORTON: Right. MS. NOONAN: -- into account in the model? My questions is really do you care if these are corporate bonds versus subprime mortgage-backed security bonds? MR. GORTON: Oh, we would care -64 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: Versus cars or credit loans? MR. GORTON: They do care about -- MS. NOONAN: Versus cars or student loans, things like that. MR. TOMBACK: Yes, let’s be clear, because they do care about the type of instrument. MS. NOONAN: Well, as I understood it, the group cares about -MR. GORTON: Right, the model –- the model -- MS. NOONAN: I’m asking if the model has -- if the model differentiates at all between the different types of assets. MR. GORTON: No. MS. NOONAN: That may all be rated Aaa. MR. GORTON: Yes. MS. NOONAN: That may all be bonds that have cash flows. MR. GORTON: Yes. MS. NOONAN: But the model doesn’t treat them differently? MR. GORTON: MS. HEYL: No. So the model doesn’t take diversification into account -MR. GORTON: No, that’s outside the model. MS. NOONAN: That’s sort of -- thank you, 65 FCIC Interview of Gary Gorton, May 11, 2010 that’s really -MR. GORTON: That’s outside the model. But the model assumes that these things are going to be -- have highly correlated performance because you’re in this significant economic downturn the whole time. And so -- MS. NOONAN: So could you explain how the model takes -- because the correlation was actually the next place I was going -MR. GORTON: Uh-huh. MS. NOONAN: -- and I’m curious to know what correlation assumptions that the model had -MR. GORTON: Right. So -- MS. NOONAN: -- or indirectly or directly. MR. GORTON: So the -- the model assumes a very high correlation implicitly, because it assumes that you live your entire life in this significant economic downturn. So when things go bad, everything’s highly correlated. You know, and so, we could build this because we had an actuarial model. weren’t doing what Goldman does. We didn’t -- we We weren’t, you know, running a big book of business and have to worry about market prices and all that. So they had, you know, insolvable correlation problems and -- the whole street did. They didn’t know what to do about it. They didn’t 66 FCIC Interview of Gary Gorton, May 11, 2010 know, you know, how to treat it, and so on. We had a completely different business, so we were able to take it into account by assuming it’s always very high, because you’re assuming that we’re always in this worst economic scenario. MS. HEYL: So why don’t you spell it out--why a terrible economic scenario makes high correlation? MR. GORTON: Because everybody gets downgraded much more frequently regardless of your asset class. MS. NOONAN: true, though? Well, I mean, is that really I mean, weren’t there -- were there bonds that weren’t downgraded in 2008 and 2009? MR. GORTON: No, no. It’s a probabilistic statement. Of course -MS. NOONAN: Because there –- there -- MR. GORTON: Of course, there are bonds that have been upgraded in the crisis, right? But most bonds get downgraded in a crisis. MS. NOONAN: And so I’m -- I mean, I’m assuming, and you can explain, that the data that you put into the model to replicate the 1970s recession would have included a wide array of different types of assets that got downgraded? MR. GORTON: Yes. I mean it, it –- what do 67 FCIC Interview of Gary Gorton, May 11, 2010 you mean by wide array? We had the entire -MS. NOONAN: Diverse. MR. GORTON: -- we had the entire history of MS. NOONAN: For every -- MR. GORTON: -- to choose from, right? ratings -- So we selected -MS. NOONAN: Right. MR. GORTON: -- these periods where everybody gets -- is more likely to be downgraded, in order to model this worst economic scenario; right? So we picked periods that had exactly this feature. MS. NOONAN: And within that period is every MR. GORTON: No, not every bond. MS. NOONAN: Okay. MR. GORTON: Nothing -- it’s never everything. MS. NOONAN: Right. MR. GORTON: Right, it’s -- MS. NOONAN: So that’s -– so -- MR. GORTON: It’s probabilistically much more MS. NOONAN: But did you pick a diverse -- bond? likely. 68 FCIC Interview of Gary Gorton, May 11, 2010 did you pick diverse assets to go in there, so for example -MR. GORTON: In the history of ratings since 1970, you know, you don’t have. securitization. This is before So again, we’re taking a relative ranking of the ratings, right. That’s the piece of information, so -- that we’re using. We’re asking, “Let’s pick periods where everybody’s more likely to go down,” and, you know, whoever has these ratings, the basic -- you take the ratings and you trace sort of what happens to them. And in that period that we focused on, we wanted to have the worst case. We didn’t want to have the worst case during -MS. NOONAN: Businesses that had done well in this recession? MR. GORTON: Well, we didn’t want to have that, but we also don’t want to have just the recent period, just because they’re certain asset classes. don’t care about that. We What we care about is the relative ranking; right? So if the rating agencies can get the relative ranking right, the more data you have the better to pick this period that you want. MS. NOONAN: But even within the period, as you say, you don’t pick every rated asset in the period? 69 FCIC Interview of Gary Gorton, May 11, 2010 You pick certain rated assets within the period? MR. GORTON: We picked what happened on average in the bad scenarios. MS. NOONAN: So that makes it sound like you picked everything. MR. GORTON: So everything, you know, everything over a limited period of time. But everything -- just the point I’m trying to make is that everything that exists today didn’t exist, you know, always. MS. NOONAN: Right. So, there’s a handicap MR. GORTON: No, I’m not sure that is a there? handicap. I think going the other way would be the handicap. Emphasizing picking diversity today, but having a shorter history would be mistake. MS. NOONAN: I’m just trying to figure out how diverse the assets were in the underlying data. MR. GORTON: It doesn’t -- that’s not the right way to think about it, right? Because the correlation issue is what you care about, right? If correlation goes up, there is no diversification. So you want to pick a period where you know correlation is high, and assume you’re always living in that period. 70 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: So do you know, for example, the percent of correlation in the period that you picked? I mean, was it -- were the ratings and downgrades -MR. GORTON: No, no, I mean, you -- I don’t -- I mean, we never actually computed the actual correlation coefficient. periods, right? We compared it to other So we’re looking at these transitions, we want to pick periods where everything tends to go down on average. MS. NOONAN: But then in terms of -- so if we move forward to the -- not underlying the model, but the model itself and how it was used at AIG… Maybe, can we just speak in hypotheticals just for purposes of testing my understanding, or maybe explaining how this worked? If you had a CDO in 2003 that was 10 percent subprime RMBS and 90percent a whole bunch of other stuff -MR. GORTON: Uh-huh. MS. NOONAN: -- and then in 2005, you have a CDO that’s 95 percent subprime and 5 percent something else, I mean, would the correlations not be different in those two CDOs? MR. GORTON: answer to that. I don’t think anybody knows the I mean, the key question is, how many 71 FCIC Interview of Gary Gorton, May 11, 2010 different bonds did you have? And then in our scenario, where everything is in the worst case, the only thing we know so far is that there haven’t been any losses. So, so far, it seems like the answer is no. I mean, most Aaa CDO tranches don’t have any losses -- subprime CDO tranche -- subprime securitized transactions don’t have losses. MS. NOONAN: Maybe this would -- this may be related, it may be not, so, I apologize if it’s not. I’m going to show, Al, I’m going to show Dr. Gorton just an excerpt from the December 5th, 2007, AIG investor presentation and conference-call transcript. MR. CREGO: Okay. Also -- MS. NOONAN: Oh, do you have -- MR. CREGO: -- Mike Easterly -- MS. NOONAN: MR. CREGO: Okay -- MS. NOONAN: -- has entered the room. Okay, great. Thanks. So the part I just want to understand is actually, it’s a little hard to tell. This copy is not very good. One, two, three, four. MS. HEYL: It’s on the-- You’re showing him the PowerPoint, not the transcript, right? 72 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: The Power- -- well there’s an excerpt from the transcript on the end of this. MS. HEYL: Okay. MR. SANTORO: Could you tell me the Bates numbers that you’re looking at? MS. NOONAN: Bates. It’s on page five. There is no This is on AIG’s website. MR. SANTORO: MS. NOONAN: Okay. It’s the December 5th, 2007, AIG investor meeting presentation. MR. GORTON: We have it. MS. NOONAN: And it’s slide 17. It’s actually titled “Subprime RMBS Models Versus Reality, Moody’s 2005 and 2006”-MR. GORTON: Uh-huh. MS. NOONAN: And basically what it seems to show, it has the 2005 vintages and then the 2006 vintages. It has tranches on the left side, Aaa, Aa, A, Baa, and Ba. And then it has the percent of Moody’s rated subprime, AIG-FP model stress, and AIG-FP experience columns. And then it does the same for 2006. Could you maybe just explain this? And I specifically want to focus once we sort of get like a baseline explanation, just for the record, of the 2006 73 FCIC Interview of Gary Gorton, May 11, 2010 vintage and sort of like the three bottom rows there. MR. GORTON: So for each vintage, there’s three columns. The first column is the actual percent of the 2005 vintage bonds that have been downgraded. So none of -- so the first column -- the first number zero says -- means that of the transactions that Moody’s rated in 2005, at the time of this computation, none of them had been downgraded, is what that means. And similarly for the numbers below it. The model, which is very noisy over this kind of horizon, says that we would have predicted, based on the model, that, you know, between 32 and 38 percent of the Aaa’s would have been downgraded. That’s what the model would say. And then our experience with transactions, the bonds underneath in our transactions, none had been downgraded. 2006, so now you have a very short period, the first column is the same. You can see that bonds are rated Ba, you know, as you’d expect, lots of them have been downgraded -MS. NOONAN: Right. I think it’s 93.7 percent MR. GORTON: -- but many more have been downgraded compared to what the model would have 74 FCIC Interview of Gary Gorton, May 11, 2010 predicted over a year. And the experience is that of the Ba bonds, I guess we had some Ba -- probably had like two or something, I don’t know. MS. NOONAN: So I was wondering about because -- and tell me, there are a few questions here, and so, we can take them in turn. It looks to me like on the 2006 vintage, once you get to the A tranche and below, that the model is no longer accurately predicting the losses that were experienced. Am I reading that correctly or not? MR. GORTON: Yes, yes. MS. NOONAN: Okay. Okay. And, I mean, can you -- I mean is there an explanation for that? MR. GORTON: Yes. We stopped writing the business for exactly this reason. We were finished. We were out of the business by the time of this call. MS. NOONAN: MR. TOMBACK: MR. GORTON: MR. TOMBACK: And so, we’ll get there -Just to be clear -We were concerned about --- this call in December 5, 2007, you’re no longer in this business and you’re not writing -- you’re not writing CDSs. MR. GORTON: Right. I think there were -- I 75 FCIC Interview of Gary Gorton, May 11, 2010 think Frost had agreed to some transactions that hadn’t closed yet, that I think closed in 2008. MS. NOONAN: You mean ‘06? MR. GORTON: Or 2006. yes, closed in 2006. I can’t remember, so, So -- MS. NOONAN: Because there must have been MR. GORTON: So there was some -- MS. NOONAN: -- 2006 vintage -- MR. GORTON: Right, there was some. MS. NOONAN: -- for there actually be losses MR. GORTON: Right. some -- here. But the whole point of us getting out of this business was exactly this kind of concern. MR. TOMBACK: And just to be clear this, because Dixie uses the word “losses” to describe what’s being -MR. GORTON: MR. TOMBACK: MR. GORTON: MR. TOMBACK: Yes, these aren’t losses. -- these are downgrades -Yes, these are downgrades. --not losses, right? So this doesn’t reflect the loss by AIG of one cent. This-MR. GORTON: No. There haven’t -- there 76 FCIC Interview of Gary Gorton, May 11, 2010 haven’t been any losses. That’s a good point. These -- MS. NOONAN: I appreciate that point. Thank MR. GORTON: These are -- these are downgrades you. over a short period of time, but, you know, we were concerned about what was going on, you know, again, outside the model. It has nothing to do with the model. And the decision was taken to exit the business. And you know, the 2006 and 2007 vintages, you know, just were horrible compared to earlier vintages. MS. NOONAN: I guess, I thought that the model didn’t care about the asset class. MR. GORTON: That’s right. care about the asset class. MS. NOONAN: The model doesn’t That doesn’t mean -- But does the model have limitations, then, on -- I mean -MR. GORTON: Well, all models have limitations. MS. NOONAN: Okay, so what were the limitations of this model? MR. GORTON: Well, you know, it depends on what your standard is, right? You know, I think the relevant question is, could you have built a better model? And I think -- I don’t think that we could’ve. I don’t think -- you know, so, all models are models. 77 FCIC Interview of Gary Gorton, May 11, 2010 They’re -- all models are wrong. All models have limitations. So the purpose of a model is to organize very complicated reality in a way that makes it kind of understandable and comparable across time for different transactions. MR. TOMBACK: And just to be clear, right. I mean, I think Gary’s explained that given the model and its reliance on ratings that exogenous to the model. but impacting any output that the model had, was, I don’t even know if all exogenous in the model, but both –- but you factored in sort of a haircut where you made more negative the ratings when using the model. MR. GORTON: Yes, the stresses. Is that -No, that’s right. But the thing about a model is one of the reasons that AIG Financial Products always built its own models, and never used purchased models, is so that we would always understand the model. And understanding the model means exactly knowing its limitations. Like you keep saying, well, what about the underlying asset class, and so on and so forth. you know, you tell me your model. you why my model is better. Well, I’m going to tell So you can’t -- you know, at the time, the question is what would we have done 78 FCIC Interview of Gary Gorton, May 11, 2010 differently, and I think the street did things differently, and they made -- I think they made mistakes, they did things that were -- I wouldn’t have done. I mean, I think I might have done those had I had their business, but, you know, the decision to stop the business was in large part that we thought that, you know, the world had changed and the model wasn’t going to be able to pick that up. That’s why we exited this business. MS. NOONAN: So why wouldn’t the model have been able to pick it up? MR. GORTON: Well, I think it’s because we were concerned about the underwriting standards of subprime brokers and originators and whether the agencies -MS. NOONAN: The rating agencies? MR. GORTON: -- the rating agencies were taking that into account properly. We didn’t know. We spent a long time investigating this, like two months. meetings. We had millions of And you know, we were pretty sure that somehow things had changed. We weren’t exactly sure at the time how bad it was or what exactly had changed. But everybody, you know, in the business noticed this kind of thing about the 2006 vintage. But by that time, 79 FCIC Interview of Gary Gorton, May 11, 2010 we had already taken the decision to exit. And I don’t remember the exact date, but, so we ended up with some 2006 transactions that turned out to be really bad, because they’d been agreed to in 2005. MS. NOONAN: So just to get, just to get back here, on the A tranche of the 2006 vintage, the percent of the Moody’s rated Aaa tranches that were downgraded was 56 percent. AIG-FP beat that with 47.8 percent. MR. GORTON: I think the thing you want to keep in mind here is that, you know, this is a completely honest representation of things, but the point is exactly the point you’re making; right? sucked, right? And we didn’t have a lot of 2006. 2006 We had some, but -MS. NOONAN: Well, I mean, I was wondering if the denominator being small is actually part of what highlighted -MR. GORTON: No, no -- MS. NOONAN: No? MR. GORTON: -- because it’s not a -- it’s, you know, you put in a dollar and you simulate, and you get a percent out. So it’s not -- it’s apples to apples here -MS. NOONAN: Okay. MR. GORTON: -- but it’s exactly the point 80 FCIC Interview of Gary Gorton, May 11, 2010 you’re making. We had the same view; right? MS. NOONAN: That there are limitations -- MR. GORTON: Yes. MS. NOONAN: -- and that’s why you exited. MR. GORTON: Exactly. MS. NOONAN: And it’s actually reflected by the fact that the model isn’t -MR. GORTON: Yes. MS. NOONAN: -- really sort of working in this MR. GORTON: Yes. MS. NOONAN: Okay. period. Yes. [Knock on door.] MS. HEYL: Come in. MR. GORTON: MR. TOMBACK: So -– You know, just to summarize, I mean, Gary, tell me -MS. NOONAN: No, this is really helpful. I mean this is -- I appreciate your -MR. TOMBACK: right. Gary, tell me if I’ve got this The model is dependent, at least in significant part, on ratings being reasonably accurate, right? MR. GORTON: MR. TOMBACK: Right. Yes. Once ratings become, you know, less than reasonably accurate for whatever reason, it 81 FCIC Interview of Gary Gorton, May 11, 2010 becomes more and more difficult to tweak the model to take that into account. MR. GORTON: should be right? instruments. We don’t know how big the stress You’re just flying, you know, without So nobody wants to write, you know, huge numbers of trades on that. MR. TOMBACK: So one -- if you believe underwriting standards or other reasons that should be other things that should be taken into account by the raters are not being taken into account by Moody’s, et cetera, that causes the model to be less and less reliable as a predictive device. MR. GORTON: Right, and then we stopped using MR. TOMBACK: And that’s -– and that –- the it. gap -MS. HEYL: Well, you didn’t stop using the model? MR. GORTON: We stopped -– we stopped writing new business, so… MS. HEYL: But you used the model to monitor and stuff like that. MR. GORTON: MR. TOMBACK: Yes. Yes, right. And the –- and the gap that’s reflected here in a document that’s -– what is it -82 FCIC Interview of Gary Gorton, May 11, 2010 December 5, 2007, is, in fact, I think you’re saying that the difference between the column AIG model stress versus AIG experience is reflective of the model’s inability to keep up with the adjustments needed in the ratings that are immediately spat out by the agencies -- rating agencies. MR. GORTON: Yes. MS. NOONAN: Did you agree with that decision at the time? I mean, were you on –- MR. GORTON: Oh yes. I absolutely agreed. I was a big advocate for a huge short position. MS. NOONAN: Was there any reason why you couldn’t have just raised the attachment point? MR. GORTON: We didn’t know how much to raise it by, right? MS. NOONAN: Okay. MR. GORTON: So we, you know –- you have to –- I mean, it wasn’t – it wasn’t the way we did things, to just, you know –- we needed to have some structure to know by how much to raise the attachment point. So to decide –MS. NOONAN: I see, so if the model tells you that the attachment point at a minimum should be –MR. GORTON: 30. 83 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: It – it goes to the 99.85 -- MR. GORTON: Right. MS. NOONAN: -- or whatever. And then, beyond that it’s, sort of anyone’s guess how much above that you would actually need. MR. GORTON: Right, and we didn’t want to make MS. NOONAN: Although, I think you said a guess. earlier, that the way this business actually worked was you came in and said it should be 20 and then Frost would go tell somebody it was 30. MR. GORTON: Right, but that was -– that was a way to get an additional buffer, right? MS. HEYL: It’s added protection. MR. GORTON: It’s added protection, right? So if –MS. HEYL: It’s not arbitrary, right? MR. GORTON: It’s not arbitrary. It’s a negotiating –MS. NOONAN: Oh, I thought that was making – I thought that was making AIG’s position bigger. MR. GORTON: No, smaller. MS. NOONAN: Smaller. MR. GORTON: Right? MS. NOONAN: Okay. 84 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: So, so when you say its attachment is 20, that means 20 percent has to be lost before AIG gets hit, and if you -MS. NOONAN: 20 percent of the super senior, or of the triple -MR. GORTON: No, 20 percent of the portfolio. MS. NOONAN: -- of the portfolio. MR. GORTON: If you make it 30, then 30 Okay, okay. percent has to be lost. MS. NOONAN: Okay. MR. GORTON: So, by making it 30 instead of 20, you’re buying a lot more protection. So –- MS. NOONAN: AIG is buying a lot more –- MR. GORTON: Yes. MS. NOONAN: Just to be clear. MR. GORTON: Yes, yes, AIG’s getting a lot more protection, so the, you know, the benchmark –- the model was the benchmark, and then if you could get further protection, we would prefer that. MS. NOONAN: And that was the negotiation MR. GORTON: Yes. MS. NOONAN: -- and that would have been Al point –- Frost, too. MR. GORTON: Yes, Frost would take less money 85 FCIC Interview of Gary Gorton, May 11, 2010 and more protection, that was basically the philosophy. MS. NOONAN: Did you -– were you aware that the model –- the CDO model was starting to be used for deals that included subprime. MR. GORTON: Yes, yes. Because when Gene came over, that was what we talked about. MS. NOONAN: over? When Gene – and when did he come But that was later –- I’m sort of thinking more like the ‘04 timeframe. MR. GORTON: subprime. Yes, yes. It included some Subprime wasn’t always bad. Subprime isn’t always bad, you know, I mean –- it’s -– it’s -– the reality is a lot different than what you read in the papers, so… But anyway… MS. NOONAN: So you didn’t –- there was no thought on your part that the model wouldn’t have been suitable -MR. GORTON: No. MS. NOONAN: -- for –- MR. GORTON: No. MS. NOONAN: -- for that asset class in MR. GORTON: Right. general. MS. HEYL: Do you think we could take a break? 86 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Yes, that’s fine. -- BREAK -- MS. NOONAN: MR. EASTERLY: Okay, Mike, go ahead. One question –- there are basically three questions that I had. One was, my understanding is that the model was based on the 1974 recession, and that this was before, you know, we had Alt A and subprime, so I was wondering, how did you adjust the models for, you know, those types of things? MR. GORTON: So the –- as I was explaining earlier, the model takes the ordering associated with ratings, and it assumes that the rating agencies can relatively order risks correctly. So, we chose to use as much data as we could to find the worst case and because we calibrated to the worst case, the implied correlations are very high, and the only thing we did later was to increase for change the stresses for different asset classes. MR. EASTERLY: Okay. I apologize if these questions have been asked before. I’ve been in and out because I was at other meetings. 87 FCIC Interview of Gary Gorton, May 11, 2010 In terms of the model, my understanding is that it also –- it was a credit model and not a valuation model and so therefore it didn’t take into account some of the things that went into the collateral calls; am I correct? MR. GORTON: Yes, it didn’t -– it was an actuarial model that was designed to produce a distribution of losses. MR. EASTERLY: It had –- MR. GORTON: Right, and –It had nothing to do with valuation; it had nothing to do with collateral. MR. EASTERLY: Right. Given that those were in the contracts, CDS contracts, why wasn’t there some sort of -- or was there some separate initiative to make sure you were covered there? MR. GORTON: I don’t know the answer to that. I didn’t know -– basically I didn’t know –MR. EASTERLY: Would you know who would be making the call on that, or –MS. NOONAN: Who would have looked at the contracts for the –- who would have negotiated the initial contracts? MR. GORTON: I mean -– the contracts would have been -88 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: That included the collateral provisions. MR. GORTON: It would have been Al Frost, for most of the deals –MR. EASTERLY: MR. GORTON: Right. -- I presume. I mean, I don’t know, to tell you the truth, because I was -– I didn’t know anything about this collateral until I walked in, and people were on the phone, you know, answering collateral calls. So I’m not sure I can really help you on that -MR. EASTERLY: MR. GORTON: Okay. But Frost would know, you know, whether he did it, or who -– I don’t know, but he would be the person to ask. MR. EASTERLY: MR. TOMBACK: All right. Let me just –- let me just say one thing –- it’s Andy Tomback. I’m with Gary. Just briefly, and you can ask Alan on this –I don’t think Alan is going to know. I think you’re going to have to ask other clients of ours or other people, but I don’t think that any valuation model took into account collateral calls, per se. In other words, those just weren’t -– they weren’t inputted, they were external to the valuation process. 89 FCIC Interview of Gary Gorton, May 11, 2010 MR. EASTERLY: Uh-huh. Okay, but -- all right. And then, I had thought that Al Frost was in on this meeting? MS. NOONAN: No, he is this afternoon at 3:00 p.m. MR. EASTERLY: Oh, at 3:00 p.m.? Okay, so that wipes out my next question. MS. NOONAN: MR. EASTERLY: All right, are you done? MS. NOONAN: Yes, yes. Okay, thanks Mike. Does your -– did the model have any assumptions about –MR. TOMBACK: MS. NOONAN: Hang up? Hang on one second -– no, no, no, no, no, no, no, no –- I mean, I think he -– Mike said that someone from their research staff might be joining us. MR. TOMBACK: MS. NOONAN: Go ahead. And, Mike, would you just let them know to tell us when they’re there, and who it is? MR. EASTERLY: MS. NOONAN: Absolutely. Great, thanks. Did your –- did the model have any assumptions about housing prices? 90 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: No. MS. NOONAN: So the -– it really would have been sort of, an indirect based on the -MR. GORTON: Right. MS. NOONAN: -- the recession data built into MR. GORTON: Yes, and also, it was a large the –- part of the discussion when we were deciding whether to close the business or not. MS. NOONAN: Okay, and I want to get to that, but I wanted to get that out of the way. So, it’s interesting, actually what -– a little bit of what Mike raised about the contractual terms on what we know, ultimately caused the problems that AIG- had, were, not, as you point out, the protection that AIG-FP wrote, which you said that you recently heard that there weren’t any actual losses on that protection -- but all these liquidity issues that it had based on the collateral calls -MR. GORTON: Right. MS. NOONAN: - and as I understand it, and I don’t know how much of this you know or don’t know or knew at the time, even if maybe you know it now, so I’ll just -– my understanding is that there are two provisions: One required AIG to post collateral upon a 91 FCIC Interview of Gary Gorton, May 11, 2010 downgrade of the parent company. I think it was AA at the time, and so if it was downgraded, then it required AIG to post collateral on the various swaps that it had. And the other trigger was, if the market value of the bonds declined –- of the underlying bonds declined. Are you familiar –- were you familiar with that at the time? MR. GORTON: Not really. I mean I knew that when you trade a derivative, you know, you have to be able to, kind of make good when you owe somebody money and it looks like you’re going to have to pay. derivatives require some kind of backing. So, I didn’t really think about it. I mean, I know that we sometimes were in a position to have counterparties say, “Well we can’t do any more trades with you because we have too much exposure to you, unless you’re willing to post,” right? So, you know, why we said, “Yes, we’re willing to post,” or who we said that to, I don’t know. But when all of this happened, you know, I was very surprised that we, you know, had so much that we had to post and that so much of it was based on prices. I mean, basically, it was news to me as well as you. MS. NOONAN: Right, right. It sounds like Al Frost negotiated the deals, and as I understand it –92 FCIC Interview of Gary Gorton, May 11, 2010 and this also may go beyond -MS. HEYL: That’s okay. He doesn’t know, really. MS. NOONAN: No -- I understand, I understand, but I’m sort of curious. I mean, this gets to, I think where I’d like to focus is, who at AIG would have known, not only how your model worked and how the deals were structured, per the model, and also would have had the knowledge about how the other components of the contracts might affect AIG independently? trying to figure out, is it Al Frost? I mean, I’m Or are there other people that would have had that knowledge of –that would have sort of, the universe of knowledge, not discrete pieces of –MR. GORTON: Right. MS. NOONAN: -- knowledge? But is there a person or people who have sort of the overall picture that we can talk to that should have or would have understood this at the time? MR. GORTON: I frankly don’t know the answer. I mean, there was a risk officer, there’s Cassano. I don’t know who knew what, frankly. MS. HEYL: Maybe that’s really a better question for Alan. MS. NOONAN: Okay, okay. 93 FCIC Interview of Gary Gorton, May 11, 2010 Was Alan at the quarterly credit meetings? MR. GORTON: I think he came to some of them. MS. NOONAN: Okay. MR. GORTON: I’m not sure he came to all of MS. NOONAN: Was this discussed at those them. meetings at a later point in time when –MS. HEYL: What is “this”? MS. NOONAN: The collateral calls, and the issue of needing to post collateral on these two triggers that I mentioned. MR. GORTON: You mean, in the crisis? MS. NOONAN: Yes. MR. GORTON: Oh. there was a quarter. I’m trying to remember if I mean, you know, what happened was, you know, bond prices plummet, right? crisis. That’s the When bond prices plummet, you have all these marks, to these low prices, there’s no more trading, there’s no more prices, there’s all these disputes and all hell breaks loose. had a quarterly call. I frankly don’t remember if we Because everything – everybody, you know, it was like a 24-hour fighting-the-fire kind of thing, so I –- frankly, I don’t remember if there was a call. MS. NOONAN: Okay. 94 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: That was a yes or no question. MR. GORTON: Yes, I don’t remember. I don’t MS. NOONAN: I appreciate your –- MR. GORTON: I just don’t remember, I mean -- MS. NOONAN: -- elaboration. know. MS. HEYL: Yes, I mean, I’m just responding that way because I think -- I don’t know if there’s a quarterly call either, and I think that that needs to be figured out before anybody know what was discussed at them. MS. NOONAN: Oh, okay, so maybe if you remember when the quarterly calls may have quit taking place? MR. GORTON: Again, there’s a -– after a quarterly call, there’s a report written, so the parent company, and somebody at FP would have the report. MS. NOONAN: And what’s the report called? MR. GORTON: It’s called a “quarterly credit report,” or something like that. MS. NOONAN: And, what sort of things would the credit report -- the quarterly credit report have in it? MR. GORTON: It would give overall numbers for exposure, it would discuss particular problems that came 95 FCIC Interview of Gary Gorton, May 11, 2010 up, and it would summarize some of the discussion, and it would list who was at the meeting. MS. NOONAN: Okay, okay. So sort of like MR. GORTON: Kind of like minutes, I mean, it minutes? was the same format every meeting, and, you know, parts of it would always be the same with different numbers, and then there’d be this little part at the end, kind of discussed, you know, other issues that came up, and those might vary from meeting to meeting. But it would also say who was there, so, I mean, you want -– you can ask Al whether he was there, but it would also say on all these things –MS. NOONAN: Okay. MR. GORTON: -- who was there. MS. NOONAN: Okay, that’s helpful. MR. GORTON: And it would divide them up, too, like credit officers, risk people, you know. MS. NOONAN: So, I want to make sure I understand exactly your level of involvement after, sort of 2002-2003. I understand you didn’t put in the inputs into the models, for the various -– MR. GORTON: Oh, I’d never put in the inputs. MS. NOONAN: -- CDO deals that were done. You did participate in these quarterly credit 96 FCIC Interview of Gary Gorton, May 11, 2010 meetings up until some point. MR. GORTON: Yes, I think went to -– I tried to go to all of them. MS. NOONAN: MS. HEYL: Okay. I think he told you, was that he wrote the section of the approval memos that had to do with the model. MS. NOONAN: MS. HEYL: Each approval memo. MR. GORTON: do that always. That’s right –- Well, I did –- I mean, I didn’t I did that in the beginning. At some point I stopped doing that and Adam Budnick started doing it. MS. NOONAN: Okay, okay. Were you involved at all –- so we’ve talked a little bit about, sort of, how these deals worked, and the due diligence that was involved on the front end, and the criteria, whether it met the criteria, or not, and then it goes into the model, and then we come out with the result on the other side, and we figure out the attachment point, and at that point, presumably, if AIG decides to enter into the deal and somebody else on the other side wants to do it, AIG prices the deal. Were you involved at all in setting or evaluating the price at which AIG sold the protection? 97 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: No. MS. NOONAN: Okay. And who -– do you know who would have done that? MR. GORTON: I think it –- again, I think it would be, you know, Frost and Forster. So Forster was in London, so, most of the credit team was in London, and then in the U.S. you had Forster, Budnick, and Park. MR. TOMBACK: MR. GORTON: and Park. Frost, Budnick, and Park. Oh, I’m sorry –- Frost, Budnick, So –MR. TOMBACK: And those are all time-shifted. Mainly Frost and eventually Budnick and Park. MR. GORTON: Exactly. MS. NOONAN: Right. MR. GORTON: So, some combination of those people decided on the pricing. MS. NOONAN: Okay, Okay. So you were never involved in that part of it? MR. GORTON: I was involved in the boring MS. NOONAN: Has the model that you developed stuff. for AIG been used for other companies, or have you kept an interest in it? I mean, do you own the -– I don’t know if there’s a copyright or I don’t -– I have no idea -98 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: No. MS. NOONAN: -- what particular law –- trademark, I guess. MR. GORTON: It’s -– no, it’s owned by AIG. I mean, when I -– I signed a consulting agreement, so it’s owned by AIG. But, I think the model is extremely specific, for a very specific kind of business, and the universe of people who, you know, would, you know, maybe be interested in it would be big institutional investors, and I’m not sure they –- I’m not sure that’s really how they think about the world. Because they don’t buy super-senior stuff, they buy cash bonds, so I don’t –- I’m not sure there’s really a clientele for it. I don’t know, I haven’t really, you know, pursued it. MR. TOMBACK: The likely clientele would be other entities that sell protection in –MR. GORTON: MR. TOMBACK: Yes, on portfolios -- like CDS. And it’s unlikely that AIG would want to sell that to their competitors, and as you’ve testified or stated, most places like to build their own. MR. GORTON: Yes, so I don’t… MS. NOONAN: To your knowledge, did AIG’s 99 FCIC Interview of Gary Gorton, May 11, 2010 Financial Products group hedge any of the positions that it did enter into before it decided to leave this business in late ‘05? MR. GORTON: at various points. Over time, in the past, it did, The parent company would want certain things hedged, so that you’d have certain positions that -– they were super-senior, but then they’d have a little sliver that would be, say Aaa, or Aa, and the parent company would want things to be hedged, which was really hard to do, I gathered. I mean, to find somebody who would, you know, but I don’t know –- I don’t know over the years how much was actually hedged. The parent company asked them for it to be hedged. MS. NOONAN: But the parent company wouldn’t have cared about having the super senior part hedged. MR. GORTON: No, no, not – not typically, they, you know, never were worried about that. MS. NOONAN: So, if we move to the point in time in –- in some point in 2007, when Goldman-Sachs, and other counterparties start calling up and asking for collateral – MR. GORTON: Right MS. NOONAN: It’s my understanding that AIG developed different models, not the model that you 100 FCIC Interview of Gary Gorton, May 11, 2010 worked on. That it used what’s called a “value-at-risk model,” and that it used a binomial-expansion-technique model to start valuing these positions. Were you involved at all in helping on those? MR. GORTON: No. I was very surprised that suddenly this thing appeared one day. I had no idea –- I have no idea where it came from, or who worked on it, or anything about it. MS. NOONAN: Why were you surprised? MR. GORTON: Well, I was –- I was –- I mean, usually, these things were sort of, you know, discussed. But I think they were discussed, I just wasn’t, you know, involved in discussing it. I was, you know, I mean -– you know, people had the frame of mind that I was off doing commodity futures at that point, so I was spending most of my time working on that. So, I mean I’m not saying I should have been involved, I’m just -– you know, I was a little amazed at the kind of -– it seemed like it just suddenly appeared, but that was only, it suddenly appeared to me. MS. NOONAN: Do you know how those models MR. GORTON: I mean, this binomial expansion worked? thing, is a –- comes from Moody’s –MS. NOONAN: Moody’s? 101 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Moody’s, yes, and I –- I remember thinking, well, this is a -– this was a very good choice, because it’s very transparent. And so, people want to know how you value, and you would like them to know the model. So, you know, you don’t want -– that’s a –- that’s a case where you don’t necessarily want to build your own very strange model. You want to have a model that people are very familiar with. of it, I knew. So that part The rest of it, how they got inputs, you know, from prices and stuff, I had no idea what they were doing. MS. NOONAN: And you weren’t involved not only in –- in developing it, but in implementing it in any way? MR. GORTON: No. MS. NOONAN: Where am I? Do you have –- let’s see – So just -MR. EASTERLY: MS. NOONAN: Can I jump? Yes, go ahead Mike. MR. EASTERLY: I was just wondering about Gary’s opinion of the binomial-expansion model. All models as we’ve discussed have limitations. What limitations do you see in the binomial-expansion model? Or technique? 102 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: Someone is asking you your opinion. MS. NOONAN: MR. EASTERLY: Yes, that’s –- that’s right. MR. GORTON: Yes, that’s your opinion. I mean, you know, it’s a -– it’s a -– in many ways, a kind of primitive model. You know, Moody’s made this model public, and we –- we never really used it, so, I think –- I mean, I think the biggest problem from –- from our –- for our -– from our –- for our business, and our uses, this model, you know, it uses an average transition matrix that’s fixed, so it’s just not –- it’s too primitive and it –- you know, for our purposes, it would -- you know, for the business, it was too primitive, and it wasn’t in the worst case. For the -- for purposes of valuation, you know, I -- as I was saying, I think it has the advantage that it’s transparent, and that everybody understood it, but it requires these inputs that they somehow got from market prices, and I don’t -– I don’t know what –- I don’t how they did that part. I just don’t have any idea. MS. NOONAN: you done? So, let’s see. Mike is up -– are Okay. So let’s see –- we sort of -– I don’t know which –- I have a few of these, I think you mentioned 103 FCIC Interview of Gary Gorton, May 11, 2010 the memos about getting a deal approved, and it had to go up. MR. GORTON: Right. MS. NOONAN: Et cetera. And I just have a sample, and this one is the –- can you get a copy. It’s this one? I’m sorry, I don’t have, like, bunches of copies of these. I apologize. MS. HEYL: I’m sorry. MS. NOONAN: What’s the -- It’s 312177 is the Bates. I just wanted to get you to walk us through it a little bit, so that when we go back and we look at all of them, we sort of understand. See, it has an attachment. Ah, thank you. Oh, these two go together, yes? MS. MORAIN: It’s the same -- MS. NOONAN: Kind of? MS. MORAIN: -- memo? MS. NOONAN: Well, we’ll see. MR. SANTORO: 1/17/05. I’m sorry. What’s the number for the one we’re actually looking at? MR. TOMBACK: MS. HEYL: This is –- It’s a different Bates number from -– oh, we have it 104 FCIC Interview of Gary Gorton, May 11, 2010 MR. SANTORO: MS. NOONAN: Oh, I’m sorry –- MR. TOMBACK: 4-3-4-3-6. We -– we should have it. It’s AIG-FCIC, bunch of zeroes, And the other one is -MS. NOONAN: No, no, no, that’s –- MR. TOMBACK: -- is AIG-FCIC -- MS. NOONAN: Switch that one out. MR. GORTON: Oh, this one? MS. NOONAN: Yes, yes, yes. MR. TOMBACK: MS. NOONAN: --3-1-2-1-1-7. Thank you. That’s okay. MR. SANTORO: We’re looking at 1-1- -- -1-7-7? MR. TOMBACK: 3-1-2-1-1-7, which is Budnick to Gorton, et al., January 17, 2006, at 3:26 P.M. MS. NOONAN: So the e-mail is 117 and 118, and then the memo is 119 through 128. So I just –- it’d be really helpful for me if we could sort of walk through this, if you could walk us through this, and so –because there is some stuff in here that’s -- and you can flip past the e-mail, and -MS. HEYL: Can you just make a copy of this, so we can –MS. NOONAN: Yes. Sure. -- BREAK -105 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Back on. Okay, so -- so, yes, as we were talking about, I’m starting -- there’s a cover e-mail, AIG-FCIC00312117 to -118; and then the memo, AIG-FCIC-00312119, that is a memo from -- to Kevin McGinn, pardon me -copying Joseph Cassano and Douglas Poling from Eduardo Diaz Perez. And, I mean, I guess I should note that on the cover e-mail, that you’re cc’ed. MR. GORTON: Uh-huh. MS. NOONAN: So you would have gotten this. MR. GORTON: Yup. MS. NOONAN: Referencing transaction requiring approval, credit-risk committee, asset purchase, or credit derivative with Société Générale, New York Branch, Soc. Gen., in respect to the senior-most tranche of TABS 2005-4 Limited, the CDO, dated January 17th, ‘05. So is this the standard form that would have been used to gain approval of the CDO transaction? MR. GORTON: Yes, pretty much. MS. NOONAN: And would this have been both for multi-sector CDOs as well as for the other types of CDOs? The European regulatory arbitrage, et cetera? MR. GORTON: Yes, the basic form would be the 106 FCIC Interview of Gary Gorton, May 11, 2010 same. MS. NOONAN: Okay, okay. And who’s Eduardo Diaz Perez? MR. GORTON: He was the chief credit officer of AIG Financial Products. MS. NOONAN: Okay. From what time period, if MR. GORTON: So he came when I was there. you know? I’d say -- I don’t know, maybe 2003. So A guess. MS. NOONAN: Would you have helped write this? MR. GORTON: At some point -- at some point, I would have written this little part here, where it says, “Constructed worst-case scenario existing,” that stuff. MS. NOONAN: Give me one second. MR. GORTON: And this one, probably I did write that little section, since my name is on here. MS. NOONAN: Okay. MR. GORTON: So the numbers -- so -- yes, I would have put in the numbers, and I probably would have written, you know, a couple of these paragraphs. MS. NOONAN: So if you wouldn’t mind sort of flipping back to the first page of the executive summary. MR. GORTON: Uh-huh. MS. NOONAN: Just 121 are the last three 107 FCIC Interview of Gary Gorton, May 11, 2010 numbers of the Bates. MR. GORTON: Yup. MS. NOONAN: So it’s a transaction summary of a $400 million CDO deal. And I believe it says that the Class A notes would be $264 million, and that AIG-FP would either invest in such a -- either purchase the Class A notes or enter into a credit-derivative transaction, or to effect a combination of the two, depending on which approach would be most advantageous. Was there -- was it typical to -- what was AIG’s standard practice? Would it normally purchase the bond or would it normally do the credit-derivative transaction or the combination? Was there a standard practice? MR. GORTON: It changed because AIG had a number of CDOs, and so they were buying assets for that -– AIG-FP. MS. NOONAN: It managed CDOs? MR. GORTON: Yes. I think it managed a couple -- at least two or three. Horizon. They were called You have to ask Forster about it. So at some point, these -- these things had to be populated with bonds. So sometimes they would buy the bonds MS. NOONAN: Okay, so that would have been -108 FCIC Interview of Gary Gorton, May 11, 2010 okay, for CDOs that AIG managed. MR. GORTON: Right. MS. NOONAN: But the credit -- the CDS business, so if we refer to the super-senior credit-default swap business, that would have been a transaction where you entered into the credit-derivative transaction? MR. GORTON: Right. I mean, I don’t know what happened on this particular deal but, you know, they might have purchased 264 million of bonds or written a derivative or some combination. MS. NOONAN: Okay. MR. GORTON: I don’t know –- I don’t know what actually happened in the end. MS. NOONAN: Okay. And then if you -- let’s see here. And then the next page, it has the structured asset classification table. MR. GORTON: Yup. MS. NOONAN: And it has ABS CDO, autos, CMBS, credit cards, RMBS prime, mid-prime, and subprime. MR. GORTON: Uh-huh. MS. NOONAN: And who would have -- this would have been the deal terms that would have been brought to AIG-FP by the investment bank or -109 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: No, no, no, this would be the outcome of some negotiation that ended up with this. MS. NOONAN: Okay. MR. GORTON: Right. MS. NOONAN: The negotiation being between… MR. GORTON: Between Al and Soc. Gen. MS. NOONAN: Okay. MS. HEYL: Whoever it was between. It might be Al. MS. NOONAN: Well, this was Soc. Gen. MR. GORTON: Well, this one has Al’s name on it, so I assume it was Al. And then the counterparty is Soc. Gen. MS. HEYL: Right. The only thing I’m objecting to is “negotiations,” because I think that wasn’t just Frost who did all that. There might be other people that -MR. GORTON: Yes, there might have been other people, that’s true. MS. NOONAN: And it talks on here, in the first paragraph after the chart -- let’s see here, I’m trying to –- “All of the” -- so if you go down, sort of seven or eight lines. It says, “All of the securities will be rated Ba3 or better upon purchase. Furthermore, in no event will any collateral security have an average 110 FCIC Interview of Gary Gorton, May 11, 2010 weighted life, or WAL, greater than 12 years.” What is -- were you involved in any of this or this was all just part of the -MR. GORTON: This is all just standard. I mean, part of the criteria has to do with the maturity. And for structured bonds, the maturity is a little bit complicated because it depends on how fast cash is paid back. You know, so you have to have some measure of, you know, how long you think this bond is going to live. And the street, you know, typically uses different measures. And so those measures would be reported. So the part that you indicated is basically saying, “We’re not going to have -- we’re limiting how long these things could be by this measure.” MS. NOONAN: come up with that? And why -- I mean, how would you Or why would you limit? MR. GORTON: Well, it’s a risk mitigation -- MS. NOONAN: And this is a limit, by the way, not of the bonds, the life of the bonds, but of the life of the credit protection; is that right? MR. GORTON: No, it’s the life of the underlying bond. MS. NOONAN: Oh, it is the life of the underlying bonds? 111 FCIC Interview of Gary Gorton, May 11, 2010 MS. HEYL: The CDO. MR. GORTON: Yes. MS. NOONAN: Okay. MR. GORTON: And that’s the same as the credit protection, right. So your credit protection is going to be shorter if the bonds are shorter, right? So you want -- you want -- you don’t want to be in a position where you’ve written protection on this thing and it lasts for 40 years, right? So you want to have some, you know, structure or rules that keep that from happening. For example, mortgages, you know, could last 30 years. And in Europe, they could last 50 years. you don’t want to be in a transaction that long. So So you have to structure it so that this thing kind of winds down much quicker somehow. So here it’s part of the rules. MS. NOONAN: Okay. And was AIG -- did AIG-FP always write credit protection that was -- that had the same length of time as the underlying bonds? Because -- the reason I ask is because I’m familiar with some credit-default swap contracts that would be limited to, say, three or five years, and not necessarily be the same -- have the same life span that the underlying bonds would. 112 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Right. See, you can do that when you write protection on, you know, IBM, right? But you can just say, “I’ll write protection for three years,” right? With these transactions, you can’t do that. You -- because you’re writing protection on a piece of the portfolio. So the portfolio life, you know, is what the other side is concerned about. So, you know, if we said to them, “We’ll write protection for three years but not for the life of the deal,” nobody would, you know, want to buy it. We could -- you know, in principle, we could have done that. But, you know, the market for this stuff wasn’t demanding that kind of protection. But we could limit it by saying, “Well, the bonds that you put in have to be limited in their lives.” MS. NOONAN: Okay, I see. So then if you go to the next paragraph, it talks about the ten-year weighted-average rating factor -MR. GORTON: Yup. MS. NOONAN: -- of the portfolio being approximately 475. MR. GORTON: Right. 113 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Is this -- could you explain that a little bit and how that -MR. GORTON: Right. So that’s how -- so these are Moody’s things. So you have this portfolio, and Moody’s wanted ways to summarize the portfolio. So, you know, the problem was that, you know, if you have some bonds that are Aaa and some are Aa and some are A, how do you -how do you -MS. NOONAN: You can’t add A and Aa, and Aaa. MR. GORTON: Yes, you can’t add it. And you also don’t want to just say Aaa is 1 and Aa is 2. So you want to have numbers that reflect -- are more reflective of the ordering, or the likelihood that they’re going to default. So Moody’s came up with these ratings factors, right. So, you know, it says here, where 360 corresponds to Baa2, and 610 is Baa3. And they have this whole, big chart, right? So 16.35 -- so the weighted average rating factor is 475. And what they’re saying is, it falls between Baa2 and Baa3. So that’s the average quality of the bond by this -MS. NOONAN: Underwriter bond? 114 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: -- by this measure, right, of the underlying bonds by this measure. MS. NOONAN: Okay. MR. GORTON: So a lot of this is just sort of summary information that the parent company wanted to see. So in addition to the model numbers, they want to have a picture of the risk just by these kinds of kind of rules-of-thumb measures. MS. NOONAN: So is it right for me -- is it right to say then, the CDO -- the assets of the CDO are bonds whose average rating -MR. GORTON: Right. MS. NOONAN: -- is between a Baa2 and a Baa3? MR. GORTON: Right. MS. NOONAN: So -- MR. GORTON: Roughly. Average, roughly, by this method of averaging. MS. NOONAN: Okay, okay. MR. GORTON: Yes, that’s right. weighting it. So you’re You’re weighting it you by these numbers; but that’s right. That’s the weighted average. MS. NOONAN: Okay. MR. GORTON: Yup. MS. NOONAN: Is there anything about a Baa2 that sounds particularly risky or not risky to you? I 115 FCIC Interview of Gary Gorton, May 11, 2010 mean, is that like a standard -- is that sort of a typical -MR. GORTON: No, there’s different kinds of deals, and they have different attachment points, and they have -- I mean, you know, this is why you need a model, right? right. They have all these dimensions of risk, You have the quality of the bonds. You have the maturity. You have the number of bonds. You have what they are. So it’s very hard to get a good sense of the risk just by looking at these one-dimensional things. So we’ve put all the one-dimensional things in here, but then all that’s going to be taken into account by the model. MS. NOONAN: Okay, so this was something that -- this was part of that first input into the model? MR. GORTON: No, it’s not an input. MS. NOONAN: Oh. MR. GORTON: This is not an input into the model. These are numbers which are just summarizing, you know, in this -- in this -- with these, you know, kind of rules of thumb what this portfolio looks like. MS. NOONAN: Okay. MR. GORTON: So that the box tells you some 116 FCIC Interview of Gary Gorton, May 11, 2010 stuff, and then we’re telling you some other stuff. MS. NOONAN: But this is more -- just to understand what the portfolio is made up of -MR. GORTON: Yes. MS. NOONAN: -- than it is to actually put into the model? MR. GORTON: Right. MS. NOONAN: Okay. Although the ratings are separately taken into account in the model? MR. GORTON: Right. MS. NOONAN: Just not using these particular MR. GORTON: Right. MS. NOONAN: Okay. numbers? And then it goes on to say, “The correlated diversity score, the correlation factor is 16.5.” MR. GORTON: Again, another Moody’s thing. MS. NOONAN: And, I mean -- so we talked a little bit about correlation earlier. MR. GORTON: Uh-huh. MS. NOONAN: Is this a similar concept? Is it based on the types of assets that are in the CDO? MR. GORTON: Moody’s -- Moody’s came up with this methodology, right? So it’s -- you know, a lot of these things, whether we like them or not, they became a 117 FCIC Interview of Gary Gorton, May 11, 2010 sort of language that people spoke in, right? So you would call up -MS. NOONAN: And I’m trying to understand it. MR. GORTON: You would call up and say, “Oh, I have a deal,” and I would say, “What’s the” -- you know, “What’s the WARF? What’s the correlation factor? So, you know, it’s just a language. MS. NOONAN: So -- So what would a correlation factor of 16.5 said to you? What would that have meant to you? MR. GORTON: frankly. I don’t know what that means, Either, I can’t remember or I never knew. I don’t know. I don’t know. I don’t remember the methodology. there’s a little box and you compute it. I mean, I don’t know. MS. NOONAN: Okay. MR. GORTON: I don’t even think I remember. MS. NOONAN: Would it have been something that was important to you at the time? MR. GORTON: No, not really. I mean, I think it’s important to report all of this information because, you know, lots of people are going to read this memo, and many of them aren’t going to really understand the model. So, you know, they need to understand a lot about the transaction in a 118 FCIC Interview of Gary Gorton, May 11, 2010 way which makes sense to them. So we want to report lots of information, especially the kind of common terminology, right. Because you have lots of credit officers at AIG who don’t necessarily work on this but may end up reading this memo. So -- you know, so when Moody’s came out with these things, you know, the parent company would say, “Could you include this in the memo?” “Okay, we’ll include this.” MS. NOONAN: So it would just require sort of going to Moody’s, maybe, to figure out what -MR. GORTON: No, I think we knew how to commute it ourselves. MS. NOONAN: Okay, who would -- would Al Frost have worked on this? MR. GORTON: No, no. There would be some programmer or assistant-type person who, you know, you would just ask, “We need a correlation factor for this portfolio. I just e-mailed it to you.” MS. NOONAN: Okay. MR. GORTON: Something like that. MS. NOONAN: Okay, but you’re not sure sort of how it’s computed or -MR. GORTON: I don’t remember. MS. NOONAN: -- what goes into it? 119 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Yes. MS. NOONAN: Okay, okay. (Speaking to Ms. Morain): Would you make a note of that, so that we can sort of follow up on that? MR. GORTON: I mean, it lists the Moody’s document here to look at. MS. NOONAN: Right, yes, that’s what I was -- yes. So then we go to the next page. Is this table something that you worked on? MR. GORTON: I don’t remember this one; but certainly, this is typical of the kinds of criteria that was important. MS. NOONAN: Okay, okay. And then where would we have -- where would sort of your section have started? Would it have started with the table –MR. GORTON: MR. TOMBACK: I think –24. MS. NOONAN: -- or the paragraphs above the MR. GORTON: I think it would start with the table? “Next.” The paragraph starting, “Next, we constructed…” MS. NOONAN: Okay, so if you could just sort of take me through -MR. GORTON: Or maybe the paragraph before 120 FCIC Interview of Gary Gorton, May 11, 2010 that. So here’s the basic issue -MS. NOONAN: Okay. MR. GORTON: -- when they -- when we’re writing this memo, they have only purchased 90.2 percent of the portfolio. So they have -- MS. NOONAN: The CDO manager? MR. GORTON: The CDO -- the investment -- MS. NOONAN: Soc. Gen.? MR. GORTON: Right. Soc. Gen. They’re warehousing this, these bonds. So they -- so we have three things we can do. One thing is, we can say, “Okay, let’s just assume that’s just the entire portfolio,” right, and, “Let’s look at what the number is there. is this worst-case VAR.” And the number So that says, we would have to attach a 22.58 percent for that 90.2 percent of the portfolio, okay. Now, the problem is -MS. NOONAN: And when you attach at 22.58 percent, that means, roughly, you could write protection on 77 percent? MR. GORTON: Yes, yes. MS. NOONAN: Okay. 121 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Exactly. Now, the problem is, for the remainder of the portfolio -- roughly 10 percent -- they have to live by these rules. But within those rules, they could very cleverly come up with the worst-possible set of bonds. So we can figure out what the worst-possible set of bonds is, too. And we’re going to assume that that’s what they do, okay? So it says, “Constructed worst-case scenario,” right? So that says, “Let’s assume the whole portfolio is this -- the worst thing they could do under the rules.” And that comes up with this number 26.33 percent. MS. NOONAN: For the attachment point? MR. GORTON: For the attachment point. Now -MS. NOONAN: Okay, so this refers to the attachment point? MR. GORTON: Right, that’s the attachment point. So then -- then we said -MS. NOONAN: And can I just pause, just for a MR. GORTON: Sure. MS. NOONAN: -- and say, “When AIG is defining moment -- 122 FCIC Interview of Gary Gorton, May 11, 2010 the worst thing they can do, does that mean it’s investing in the riskiest assets? MR. GORTON: No, this is the -- yes, the manager -- the manager -- we’re saying, under the rules, the manager could pick the riskiest, worst assets. MS. NOONAN: Okay, because some people in this world would say the risky ones are the good ones because they return -- they have more -- a higher yield, or a higher return. MR. GORTON: No, no, no. So we’re thinking they’re -MS. NOONAN: I just want to make sure. MR. GORTON: -- they’re like cherry-picking us, right? So they’re saying, under the rules, we’re allowed to put in this really toxic, horrible thing. And we say, “Gee, you know, under the rules, they are.” So let’s assume they do that, all right, and let’s assume they do that with the whole trade, never mind that they purchased 90 percent. entire trade. They do that with the So that gets this 26.33 number. So then we say, “Okay, what if we -- what if it’s the case that the average life -- this arbitrarily goes longer by a year,” right, or goes longer by two years, then you can see the effect of that, right? 123 FCIC Interview of Gary Gorton, May 11, 2010 So it could be 32.75 if this thing goes longer by two years. Now, this case, you know, can’t actually occur because they’ve already purchased 90 percent of the portfolio. But it is informative to know, you know, what they could -- what the worst case under the rules actually is. Now, so the rules -MS. NOONAN: And that would then get at -- for example, if it was an actively managed pool. MR. GORTON: Well, then, you’re only allowed to reinvest over a certain period of time, and only a certain amount per year. But you’re right -MS. NOONAN: But this would catch that? MR. GORTON: This would catch that. Because we would say -- let’s suppose you actively manage to the worst case, to the extent you can under the rules, and that’s the -- that would be what we actually model. So this -- this -- in this -- for this trade, this constructed worst-case scenario is -- you know, can’t actually happen. MS. NOONAN: Uh-huh. MR. GORTON: But we put it in, anyway, so we can see what it looks like. 124 FCIC Interview of Gary Gorton, May 11, 2010 The realistic case is the bottom one, where we say, “Okay, you have 90 percent of the portfolio. Let’s assume the other 10 percent is the worst case.” And, again, you see that the base is 24.69 compared to 22.58. And, again, it goes way up if you add one or two years. MS. NOONAN: So this was -- was this -- MR. GORTON: So we ended up with a transaction where we asked for a 34 percent attachment point. See, that’s -MR. TOMBACK: So assuming the worst on the average life. MS. NOONAN: Where do you see that? MR. GORTON: That’s on the first page. MS. NOONAN: Of the executive summary? MR. GORTON: No -- yes, where it says 34 percent is the buffer. So that’s how much we have junior to us. So the transaction is structured to be -- have more of a buffer than even this case I said could never happen, plus two years, right? That’s why -- that’s why, you know, strictly speaking, the model number should be something in this bottom table, right? But we asked for significantly more than that. 125 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Just to sort of test the limits, was there a limit to how far beyond the weighted average life the deal could actually run? So here, we’re assuming a plus-two years, which is bad -- bad. MR. GORTON: right? Well, these deals amortize, So they amortize according to schedules. So it could, it could, but it would be very unlikely for all of them to do that. You know, one or two might, for some odd reason. MS. NOONAN: Say, for example, the United States government comes in and guarantees all mortgages and outlaws prepayments. MR. GORTON: You know, I guess if they outlaw prepayments, then the mortgages are going to, you know, the end. And in that case, you know, the categories of mortgages that we have here, you know, it’s about, you know, I don’t know, 75 percent of this deal, they would go to the end, yes. MS. NOONAN: Okay, okay. I mean, just really testing the limits -MR. GORTON: No, no. MS. NOONAN: -- from a sort of contractual? MR. GORTON: Yes, that’s right. MS. NOONAN: The same point? 126 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: That’s right. MS. NOONAN: Okay, okay. MR. GORTON: That’s what would happen. MS. NOONAN: Okay, okay. And this -- so the VAR models that are here, the attachment points that are calculated over here on the right on these tables, was this a separate model? MR. GORTON: No, this is the one we were talking about before. This WVAR number is the 99.85. And then the “W” just means it’s in this worst case that we were talking about before. MS. NOONAN: Okay. MR. GORTON: Somehow, this became the terminology. MS. NOONAN: Okay, okay. The only reason I ask is because the model that I think AIG ended up using in part of 2007, before it went to the binomial expansion technique model to value the CDS portfolio, was a VAR model. MR. GORTON: doing. Oh, I don’t know what they were I mean, I don’t think it was this model. MS. NOONAN: No, no, I don’t think it was, either; but I just… MR. GORTON: Right. Yes, I don’t know. 127 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Okay, okay. So this says -- and then on know the next page, “These results are consistent with the risk of AIG-FP’s position being super-Aaa risk.” MR. GORTON: Yes. In this case, it was actually rated. MS. NOONAN: By the rating agencies? MR. GORTON: Yes. If it’s cash, it has to have a rating. MS. NOONAN: So when you say it’s cash, does that mean that AIG -MR. GORTON: No, no, no. MS. NOONAN: -- paid money to buy it? MR. GORTON: No. So you have this special-purpose vehicle which is going to buy a portfolio, and it’s going to finance that portfolio by issuing these bonds that were listed in the table, right? MS. NOONAN: page. Wait, we’ll get to the right Hang on one second. MR. GORTON: So it’s 1, 2, 1 [one to one] –- MS. NOONAN: The class notes? that’s -The Class B notes, the Class C notes -MR. GORTON: Yes. 128 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: -- right? MR. GORTON: This tells you the notes that are going to be issued in the capital market, except for equity, which is going to be retained by the manager here, which was -- I saw it somewhere in here, who it was. But, anyway -- so somebody is going to buy these bonds. Now, one possibility is that Soc. Gen. is going to buy the bonds and we’re going to write protection on it. MS. NOONAN: Okay. MR. GORTON: Right. MS. NOONAN: On the Class A or on all of them? MR. GORTON: On just the Class A notes, right. So because you’re buying actual bonds, you have to have the money to buy the bonds; so you have to actually issue and get the money. So this -- this is not, you know, a synthetic deal. In other cases, in Europe, we could write protection to, you know, commerce bank on, you know, the senior 80 percent of a particular portfolio. And for that part, there’s no money that changes hands because there’s no SPV and so on. MS. NOONAN: Right. It’s unfunded? 129 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: It’s unfunded, yes. MS. NOONAN: But this was you be funded. And the way it was -- so unless AIG is going to buy the Class A notes because of a CDO it’s managing on its own behalf -MR. GORTON: Yes. MS. NOONAN: -- then the deal here is for AIG-FP -- for someone, whether it be Soc. Gen. buying it -MR. GORTON: Or someone else. MS. NOONAN: -- in the first instance or buying it and selling it to someone else, or immediately selling it to someone else, the deal is for them to actually pay in the 264 million in cash into the SPV? MR. GORTON: Uh-huh. MS. NOONAN: And then AIG-FP enters into a separate contract to write credit protection on that class? MR. GORTON: Yes. MS. NOONAN: Okay. Correct. And does the swap follow the owner of the bond? MR. GORTON: It’s written -- it’s -- that’s a good question. MS. NOONAN: You know what I mean? MR. GORTON: Yes, I know what you mean. 130 FCIC Interview of Gary Gorton, May 11, 2010 I think the answer is no. It’s not like insurance guarantees on municipal bonds, that follows the bond around, right? So I think -- it’s going to be written to Soc. Gen. referencing these bonds. should check with Al. But that’s something you I’m pretty sure that’s right, though. So, it’s not attached to the bond, like an MBIA wrap. MS. NOONAN: contract? Okay, so it’s a separate It’s not actually within the deal? MR. GORTON: Right. I mean, it’s a good question, you should ask Al. MS. NOONAN: Okay. MR. GORTON: I don’t think that AIG-FP was writing protection to the SPV. I think AIG-FP was writing protection to Soc. Gen. and referencing the bonds, but I’m not positive about that. MS. HEYL: Could Soc. Gen. transfer its rights under the -MR. GORTON: I think you’d have to novate it. Again, these are questions – these are not questions for me. MS. NOONAN: Or Soc. Gen. to keep it but enter into a mirror swap permanently? 131 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: Something like that, yes. I don’t know. MS. NOONAN: MR. TOMBACK: That’s fine. That’s fine. I imagine if protection is worthwhile, it would be retained upon transfer in some way. MS. NOONAN: Well, I mean, the one thing that’s interesting, I don’t know if it’s this one, I know I read somewhere that the swap calls for -- swaps called for -- some swaps called for physical settlement. MR. TOMBACK: MS. NOONAN: Right. So if AIG -- if the subordinate tranches were ever eaten up such that the losses seeped into the portion of the structure that AIG had written protection on, then AIG has to pay out the value, right? But they get back the bond -MR. GORTON: Right. MS. NOONAN: -- because, presumably, the bond still has some value, and that required a physical settlement. MR. GORTON: Right. MS. NOONAN: And it would be really hard for Soc. Gen. to physically settle if it had sold the bond that kept the default swap protection. MR. GORTON: Right, then has to go buy it or, 132 FCIC Interview of Gary Gorton, May 11, 2010 you know. MS. NOONAN: Right. MR. GORTON: This way, single names work. You have to go buy it. MS. NOONAN: Right, right. But this would presumably be harder to go buy it in this instance? MR. GORTON: Yes, but, I mean, it surely is a sort of technical matter. MS. NOONAN: I don’t know. Okay. Hold on one second. Thanks. I mean, are there other things about this that I sort of should focus in that I haven’t so far? I know -- I apologize for the open-ended question. But if there are -- I feel like I have a better understanding now, and I appreciate your sort of taking me through this. MR. GORTON: I mean, I think that’s pretty MS. NOONAN: Okay. MR. GORTON: I mean, the actual bonds that much it. they had purchased were always -- would always be included. MS. NOONAN: bonds? And who would diligence these Like, who would be in charge of sort of going through, looking up the QCIPs and… 133 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: The library. MS. NOONAN: If that happened, assuming that that happened? MR. GORTON: The library. Oh, yes, absolutely, that would happen. The library would do that. So there’s a little -MS. NOONAN: The library? MR. GORTON: Yes. We had a group of three people who was called the Research Library. And they were the people you’d go to for a whole range of different things, including finding the QCIPs and checking the ratings and all that. MS. NOONAN: Do you remember their names? MR. GORTON: The head librarian was a woman named Deonna Taylor, and then there was Frank Levantino and Howard Zhang. These were the three people at the end. There was a few people who were, you know, kind of at the beginning, they were a little bit different. MS. HEYL: Zhang is Z-A-N-G? MR. GORTON: Howard Zhang is -- I can’t remember if it’s -- I think it’s an “H.” Z-H-A-N-G. So -- I think a And Levantino is L-E-V-A-N-T-I-N-O. And then Deonna Taylor. 134 FCIC Interview of Gary Gorton, May 11, 2010 So they were the library MS. NOONAN: And Deonna is D-E-O-N-A? MR. GORTON: Yes, D-E-O-N-A, right. MS. NOONAN: Okay, okay. That’s right. So, actually, if you look at the cover e-mail, I just think it’s -- it just happens to be on this one. I think it’s sort of interesting. It appears to be a discussion of what constitutes the definition of subprime versus Alt-A. MR. GORTON: Right. MS. NOONAN: Was there a common understanding within AIG-FP? MR. GORTON: There was not a common understanding in the world. There were not, for a long time -- I’m not even sure there is now -- but the agencies differed from the iBanks differed as to what they called Alt-A, subprime. So, you know, it was -- if you’re going to report it or have rules, you have to have some understanding of what it is. You have to define it. So that’s what this e-mail is about. MS. NOONAN: And do you remember of there being a resolution of that within -MR. GORTON: I don’t remember what -- I don’t 135 FCIC Interview of Gary Gorton, May 11, 2010 remember what happened. I mean, I don’t remember whether there was some agreement, you know, in the world, either. So I don’t know how that finally played out. MS. NOONAN: Hold on one second. There are a couple of other things I just wanted to ask you about. Sorry, just give me one second. Could you get this one Clara? I think this -- actually, I think there are just two that I’m just sort of curious about. And this is the other one. If you want to make copies, then let me let her get this other one. There are only two that I really think are really going to -- thanks. Okay, what can I ask you about in the meantime? Oh, I know. Could you tell me a little bit about your compensation at AIG? MS. HEYL: Andy? MR. TOMBACK: Actually, our view on this is because of all the threats, a number of clients jave actually been threatened with violence on the payments. A number of clients actually did not -- they turned back some of their payments. 136 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Uh-huh. MR. TOMBACK: But -- MS. NOONAN: MS. HEYL: Do you represent Jake DeSantis? No. MR. TOMBACK: MS. HEYL: No. He had a lot of problems. He’s the one that wrote the letter? MR. TOMBACK: MR. GORTON: Yes, right. I had a lot of death threats from the Wall Street Journal article. And I’m the one who asked for there to be more security at the FCIC hearings. MS. NOONAN: Really? MR. GORTON: Yes, really. I also had Yale security -- it was really a terrible thing, right? You know, I had to tell people in my building not to open packages. MS. NOONAN: Huh. MR. GORTON: You know, and then over Cuomo, I made plans for my family to move to Boston. No, I mean, I’m serious -MS. HEYL: So bottom line is, we don’t want to get this on this tape. MR. GORTON: MS. HEYL: You only need one nut. We don’t want this out in the public record at this point. 137 FCIC Interview of Gary Gorton, May 11, 2010 MS. NOONAN: Okay, okay. Well, this will not go in the public record at this point. MS. HEYL: No, but it might go in the report. And at this point, we just don’t want to have it given to you in a form that it would end up in the public unless we talk about it a whole lot more than we’ve already talked about it. MR. GORTON: My problem is, I’m sort of very publicly visible, right? Most people work at FP, nobody knows who they are, where they live, unless their names were leaked by Cuomo. But that’s not my situation. Everybody knows where I am, right, which is why it’s easy to write me a death threat in a letter. MS. NOONAN: Well, and just so you know, this is not unique to you. This is something that we, quite frankly, ask everyone. MR. GORTON: Yes. MS. NOONAN: And the issue of incentives and compensation are something that we’re looking at. Something that is actually, I’m pretty certain, is in the statute, that we’re supposed to look at the issue of how people were incentivized and how they were compensated and how has that sort of played a role, if 138 FCIC Interview of Gary Gorton, May 11, 2010 at all. And so this is not something that is – this is just standard. I just wanted to put it on record. We can discuss it at some point offline. MS. HEYL: Yes, we can do it offline. MS. NOONAN: MS. HEYL: Okay. That’s fine, that’s fine. Because whether he has a good memory of it or not and has all the numbers -- certainly Weill does, but my understanding is that Weill hasn’t given you that information yet. MS. NOONAN: MS. HEYL: No, they haven’t. So -- MS. NOONAN: Or at least to my knowledge. I haven’t been through every last document that they’ve given us. MS. HEYL: I think they were under the impression they had not given out information about our client’s compensation, either to you or to anybody. MS. NOONAN: Right. Although I think it has been requested. MS. HEYL: Okay. But I’m just saying that given the fallout that was so -MS. NOONAN: MS. HEYL: Sure. MS. NOONAN: -- massive last time. Right. No, I understand it’s a 139 FCIC Interview of Gary Gorton, May 11, 2010 sensitive issue, and we’ll – MR. TOMBACK: MS. NOONAN: MR. TOMBACK: We can come back to it. We can deal with it later. We can come back to it. I mean, I think we can agree that he earned more per hour at AIG than he did, first, at Wharton, and then at Yale. And let’s work on the details later. MS. NOONAN: Okay, okay, that’s fine. Let me see if there’s anything else on my list. So I don’t know if you’ve read the Michael Lewis book or not. MR. GORTON: No, I have not. MS. NOONAN: “The Big Short.” MR. GORTON: Yes. MS. NOONAN: So there’s a passage in it where, I believe, it’s -- I believe the way that the story is written, it’s that Gene Park comes to you and some other people at AIG-FP and asks what percent of the deals you think are subprime. MR. GORTON: Right. MS. NOONAN: And in the book, it says that you responded, “Maybe 10 percent.” MR. GORTON: Right. MS. NOONAN: Now, the book also goes on to -140 FCIC Interview of Gary Gorton, May 11, 2010 I can’t remember whether the book says what the actual percentage was, and I’m not clear on whether that’s right or not. But do you remember that -MR. GORTON: I remember that -- MS. NOONAN: -- happening? MR. GORTON: -- from the Vanity Fair article. I don’t remember that happening -MS. NOONAN: You don’t remember –- well, my question is, do you remember it happening? MR. GORTON: No. But I can easily imagine that, you know, because I wasn’t involved in the business for the last four years, that I was thinking, before I moved over to commodity futures, that’s probably what it was, 10 percent. And then later, it evolved somewhere else, so… MS. HEYL: Well, do you remember coming up with that –MR. GORTON: MS. HEYL: I don’t know. -- number for Gene Park? MR. GORTON: I don’t know. MS. NOONAN: Okay. I don’t remember that, no. MS. HEYL: And did anyone ask you to verify that for the book? 141 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: No, I didn’t talk to Michael MS. NOONAN: No, these are extra copies. Lewis. MR. TOMBACK: With respect to the available people at AIG FP, both in London and Connecticut at that time, were you a good choice to ask for current information on the percentage of subprime in the book? MR. GORTON: No, I wasn’t good person. I refused to talk to Michael Lewis; and Gene later apologized. MS. HEYL: MR. GORTON: Apologized for what? Well, he was very -- he -- I mean, he just felt bad about the whole thing. But he was kind of naive. MS. NOONAN: I just wanted to get it on the record. Okay, so two e-mails. One is -- the first one that I want to talk to about is dated February 16th, 2006. It’s Bates number AIG-FCIC-00109416 to -419. And it’s sort of a long chain of e-mails. If you flip sort of to the last -- page 3 of 4, at the very bottom; and then on the fourth page, it starts out with an e-mail from Andrew Forster to you and Gene Park and others. thoughts.” And the subject is, “Subprime And it’s just talking about concerns the way 142 FCIC Interview of Gary Gorton, May 11, 2010 that the market is developing. Specifically, declining underwriting standards, et cetera. So then if you go to the bottom of page 3, you respond, and you said, “I think the really bad scenarios have Bbb-rated bonds being hit and possibly suffering significantly. Are we being a bit paranoid about the other situations?” So I just wondered, what was your view at the time of the idea that AIG-FP should get out of this business? MR. GORTON: I mean, I don’t remember exactly what it was at the time. I also -- I vaguely remember this happening earlier; but, you know, there was a -there were deals -MS. NOONAN: You remember a similar discussion taking place earlier? MR. GORTON: Yes, I thought this whole discussion happened earlier, but, you know, maybe just my memory’s wrong. I think the question -- there was a long discussion in the group about the whole issue of the business and subprime and CDOs. And we all thought that, you know, house prices aren’t going to keep going up; they’re going to go down. And the question is, how bad is it going to be? 143 FCIC Interview of Gary Gorton, May 11, 2010 And there were deals called “high-grade deals,” where the underlying bonds were mostly Aaa and Aa. So the question was -MS. NOONAN: Okay. MR. GORTON: -- would we be willing to do those deals and not the deals that had all these Bbb’s. I think everybody agreed that the Bbb’s were in trouble. And in the end, we all agreed to just get out completely. And this is sort of somewhere in the middle of this big discussion. MS. NOONAN: Were there others that -- because it seems at this point that you’re sort of trying to make some case for, “Maybe we could stay in part of this business.” MR. GORTON: Yes, I mean, I was thinking about it. In the end, as a group, we were unanimous and recommending exit from the whole thing. MS. NOONAN: Okay. MR. GORTON: But, you know, we went through a long discussion and a lot of meetings and research to get to that point MS. NOONAN: So is this -- would that have reflected sort of a change of opinion on your part or 144 FCIC Interview of Gary Gorton, May 11, 2010 was that just a -MR. GORTON: I don’t know. I don’t remember, MS. NOONAN: I mean, Alan Frost, for example, really. if you look on page 2 at the bottom, like the third sentence, “I think we’re being a bit extreme about theses risks, and the consequence of the changes in our approach will fundamentally shut us out of the entire market very quickly.” MR. GORTON: Yes. MS. NOONAN: I mean, it sounds like a consensus was reached at the end, is what you said. I mean, were there others that sort of had the view that Frost sort of makes here? MR. GORTON: I mean, I don’t remember everybody’s view. I remember there was -- you know, there was a significant investment in spending –- in getting a lot more information about this market, and a lot of discussions. And when we finally talked to Cassano, you know, nobody -- there was no minority view. All right, everybody said, “We’re out.” MS. NOONAN: Was Andrew Forster the person that really prompted this discussion or was someone 145 FCIC Interview of Gary Gorton, May 11, 2010 else? MR. GORTON: No, it was Park. It was Park. MS. NOONAN: Park? MR. GORTON: Park was really the trigger for Okay. really thinking a lot more about it. And, you know, I had been working on commodity futures, so I didn’t realize that we were being shown a deal -- in fact, I was even surprised at some of the deals that we did. I don’t know how -- you know, I don’t know. MR. TOMBACK: Let me just mention because Alan’s not here but he’s quoted in Gary’s interview. But that same e-mail, Alan goes on to say, and I quote, “I don’t think we can just look at this as just an exiting the mezz – mezzanine -- subprime CDO market even though that’s the driver. Maybe we should spend some more time working on some revised rules so that we don’t shut it down completely.” “Having said all that, I do not have a very robust rebuttal to the concerns that people have, although I am probably more of a buyer of the soft-landing expectations -- expectation.” So -MR. GORTON: MR. TOMBACK: I mean, it was a very --- he’s not a wide-eye, “Let’s 146 FCIC Interview of Gary Gorton, May 11, 2010 keep going” guy, that first quote would lead you to conclude. MS. NOONAN: whole document. Right. I mean, I’ve read the I appreciate that. MR. TOMBACK: Okay, I just wanted it tape recorded again. MS. NOONAN: I appreciate that. That’s just fine. I -- I mean, this -- this string of e-mails appears to be a robust conversation about whether or not AIG-FP should stay in the business or not. I’m trying to get a little bit of more sense of what was taking place offline, off-e-mails, in discussions and phone calls. MR. GORTON: It’s no different from this. This is -- this is a kind of accurate sampling, right? I mean, we went to see a number of economists at investment banks and I think the rating agencies and a lot of people, you know, asked them a whole bunch of questions. So we -- this was -- you know, the guys came from London and New York, we had all these meetings. And, you know, over this time, so there’s this discussion. And, you know, in the end, there was this conclusion to get out. MS. NOONAN: Were you involved in any of the 147 FCIC Interview of Gary Gorton, May 11, 2010 discussions with the external people? MR. GORTON: meetings. Yes, I went to some of these I mean, I’m not sure if I went to all of them, but I went -- I remember spending, you know, a couple days in New York, going from meeting to meeting to meeting. MS. NOONAN: So what investment banks? Would it have been a lot of AIG’s counterparties on deals or… MR. GORTON: I’m not sure how they were picked. I mean, one of the ways they were picked was, we picked firms where they had research analysts or economists who seemed to write the best stuff about the housing market. So that was one criteria. I mean, our counterparties -- you know, the actual people on the other end of the phone would have no idea about this. So you don’t need to talk to them. They want to keep going. So, you know, the question is, what are the research people, in the research departments, those are the people you want to go talk to. MS. NOONAN: Okay. So then if you go up to Andrew Forster’s response at 9:03 a.m., the last sentence of the first full paragraph -- or just the last part of it -- he says, “If we think there’s a softer -148 FCIC Interview of Gary Gorton, May 11, 2010 more likely to be a softer landing, I prefer we took that view directly by investing in the underlying deals and then earning some real spreads since we are taking pretty much the same view in my mind and since we lack any feel for correlation across deals, so we have no idea if the subordination we have is worth much.” I’m just curious whether you agree with that statement, the part about, “We lack any feel for correlation across deals and don’t have any idea if the subordination is worth much”? MR. GORTON: Right, that’s exactly what we were discussing before, when you pointed to that table about 2006. MS. NOONAN: Okay. MR. GORTON: That’s the same -- he’s articulating the same thing, right? MS. NOONAN: Okay, so then the question that jumps out to me, and sort of something that I thought about when we talked about it earlier as well is, does it at all sort of -- did it make you think or revisit the information you had using the model for the prior deals? MR. GORTON: No, because there was -- the information, again, that we looked at were the 2005 and earlier vintages were much different, right? And, 149 FCIC Interview of Gary Gorton, May 11, 2010 again, as far as I know -- I may be wrong, but at least as of two months ago, we didn’t even have losses on the really terrible 2006 deals. So I think -- I mean, I think in all this regard, we made the right decisions. But the problem, as you indicated earlier, was all this collateral that was based on prices, and the prices suddenly plummet. And, you know -- you know, this part, I think we actually got right. MS. NOONAN: So then the last part is just flipping it to the front page, and then you respond, “On Aa and Aaa tranches, I think we should distinguish between taking them and controlling how much we would take. This is important because I think the distinction is between shutting down a business completely versus making an attempt to get some deals and influence the street.” I’m sort of curious about “influencing the street,” that part of the sentence. MR. GORTON: Yes, I’m not sure what I meant, other than the fact that we -- you know, we might be able to say to counterparties, “We only want deals that have Aaa bonds in them,” right? If we just said that, the question is, you know, could we -- you know, would they do that? I don’t know. But in the end, we decided 150 FCIC Interview of Gary Gorton, May 11, 2010 not to do it. MS. NOONAN: Do you know or think that -- it’s probably the wrong way to ask the question. So, first, do you know whether AIG’s decision to exit the business entirely did, in fact, have an influence on the street? MR. GORTON: No. I was -- MS. NOONAN: Or would you have been aware of any feedback from counterparties? MR. GORTON: Well, I was, frankly, pretty stunned because, you know, I thought our exit would be kind of a red flag to others; but most of our counterparties did twice as much in 2006 as in 2005. And, you know, I remember saying, “I don’t understand, like, aren’t they getting the same information we’re getting? What are they doing?” You know, these turned out to be UBS, Merrill Lynch, Citibank, and so I don’t know. MR. TOMBACK: You said counterparties. But were really people that -MR. GORTON: Yes, they were not counterparties MR. TOMBACK: Effectively, your competitors, but -- entities that are selling protection? MR. GORTON: Yes, entities that are selling 151 FCIC Interview of Gary Gorton, May 11, 2010 protection. Or what they did was they kept the bonds. MS. NOONAN: They may have been buyers of protection from AIG in previous deals? MR. GORTON: were. Some of them -– some of them But the main point is that the street, as a whole, did twice as much in 2006 as 2005, independent of us. I mean, I found that -- that just seemed kind of odd, given everything that we had seen and what we had concluded. MS. NOONAN: Were you thinking about at the time or is this sort of looking back on it, being surprised? MR. GORTON: were surprised. No, at the time. At the time we At the time we were surprised. And at the time, you know, I was very convinced when, you know, we had this meeting with Joe, and I said, “You know, we should go short.” And he said, “How much?” I said, “A billion,” right, which wouldn’t have made a difference in the end. But we before pretty sure this was going to be -- we didn’t realize how big of a blowup it was going to be. was going to be pretty bad. But it seemed like it And so we were really surprised. MS. NOONAN: So what was -152 FCIC Interview of Gary Gorton, May 11, 2010 MR. TOMBACK: Just to be clear about that, so you were really surprised about the lack of response by the rest of the street? MR. GORTON: Yes. MS. NOONAN: And so when you recommended to Cassano going short -MR. GORTON: Yes. MS. NOONAN: -- could you tell me a little bit about the discussion? Was that something that Cassano had asked you to think about or was it something that you -MR. GORTON: No, we just -- we should not only stop -- the reasons we’re stopping are so good, we should go short. You know, and with credit derivatives, it’s not like stocks, right? to pay, right? You know how much you’re going Stocks, you don’t know. So -- but this wasn’t -- this was really not the kind of thing we did. And in the end, you’d have to -MS. NOONAN: Going short was not the kind of thing you did? MR. GORTON: not very much. Yes. I mean, we did go short but I forget how much. Like, 250 million or something. MS. NOONAN: And how did you go short? 153 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: I don’t know. You’d have to ask MS. NOONAN: Forster? MR. GORTON: Maybe we did the ABX index. Forster. don’t know. I I don’t know. MS. NOONAN: But Cassano was not interested in going short or -MR. GORTON: No, he was, to some extent. But, I mean, he was happy to get out when all of us said we all agree. You know, that’s compelling. So he said, “Why?” We explained, and he said, “Fine.” Then we said, “We should go short,” and that was less -- he was sort of less interested in that. MS. NOONAN: When the decision was made to exit the business, did that cut off a significant revenue stream for the unit? MR. GORTON: Yes, you would have to ask Al. I’d say it’s 50 to 80 million a year, something like that. MS. NOONAN: Was there any discussion -- was there any pushback on that, like, losing that revenue stream? MR. GORTON: Yes. You know, you hate to close a business like that; but, you know, nobody -- there was 154 FCIC Interview of Gary Gorton, May 11, 2010 no pushback from Joe or -- I mean, Al was basically -you know, the whole business was being shut, you know, so… But he had already been moved on to something else. It was Gene’s -- Gene said that –- Gene was told, “Go take over that business.” He comes back and says, “My God, we should shut this down.” We all look at it for six weeks, and we go, “You know what? You’re right.” MS. NOONAN: So since you were in the commodities -MR. GORTON: Yes. MS. NOONAN: -- mostly working in commodities at that point, did it affect your day-to-day activities significantly for this business, too? MR. GORTON: No, no, no. MS. NOONAN: Loud. MR. GORTON: No. MS. NOONAN: And so the last e-mail that I have was this one dated February 28th, 2006, from Gene Park to Joseph Cassano, copying you and others, and the subject is, “CDO of ABS approach going forward, message to the dealer, community. 155 FCIC Interview of Gary Gorton, May 11, 2010 The Bates is AIG-FCIC-00109490. Have you -- take a minute and look at it. Do you remember this? MR. GORTON: Not really. But I remember sort of the gist of it. MS. NOONAN: Okay. Do you agree with it, MR. GORTON: Yes, I agree with it. basically? I mean, you know, in the end, it was just we’re out. It didn’t -- this idea of working with the street just never got anywhere. MS. NOONAN: Do you know if this was -- were you cc’ed or do you know if this was actually sent to the dealers? MR. GORTON: Oh, no, we’d never send this to the dealers, no, no, no. You just talk to them on the phone. MS. NOONAN: I see. So these were like talking points? MR. GORTON: to Joe, right? No, this was -- this was a memo So this is -- this is -- MS. NOONAN: The first sentence says, “The list summarizes the message we plan on delivering to the dealers.” MR. GORTON: Yes, but that would be -- that 156 FCIC Interview of Gary Gorton, May 11, 2010 would be -MS. NOONAN: Verbal? MR. GORTON: -- over the phone. Verbal, yes. MS. NOONAN: Okay, okay, okay. Were you involved in -- okay -- in delivering the message? MR. GORTON: No. I was never involved with, you know, very -- any direct contact. I mean, sometimes I would go to a meeting with, you know, somebody. But, in general, I was never -- I mean, I didn’t have a license to trade, and it wasn’t my job, so… MS. NOONAN: Give me one second. Do you have anything -MS. MORAIN: Sure. I was wondering if you could give us a little bit more information about this tour you went on when you talked to economists at the rating agencies and investment banks, and ultimately led to your -MR. GORTON: Yes. MS. MORAIN: -- your decision to stop writing protection on -MR. GORTON: I was trying to remember if –- any of the names. 157 FCIC Interview of Gary Gorton, May 11, 2010 So we -- I’m not sure -- I’m not sure -- I can’t remember any of the people; but I remember we went to New York, and Forster and Fewings came from London. And, you know, the group was something like Forster, Fewings, me, Frost, and Park. And we went to, you know, Bear Stearns, Merrill Lynch, Goldman -- I’m not sure if we went to Moody’s. I don’t remember. But the people we met with were economists, you know, that worked in the -- you know, on mortgages. And the discussion was all about housing prices and mortgages and defaults and what subprime was and, you know, these kinds of issues. And I don’t remember if maybe the other people went to a separate set of meetings when I wasn’t there. I don’t remember. MS. MORAIN: Were there any sorts of reports MR. GORTON: Yes, I don’t think we wrote a generated? report, I think -- you know, other than this kind of summary. I mean, they gave us lots of reports, you know, that they had written about, you know, the California subprime market, or whatever it was. I mean, there was a lot of pretty impressive people study this out there. MS. MORAIN: For example? 158 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: think. That’s what I was trying to You know, I was trying to think -- you know, I sometimes picture this Bear Stearns guy who was very optimistic, right? And he was a very well-known analyst in the mortgage market, but he was very optimistic. And he was so optimistic, that we came out afterwards, you know, talking to ourselves like, “This guy is out of his mind. He must be on drugs or something.” MR. TOMBACK: MR. GORTON: Remember, you’re on tape. Yes. Anyway, we just couldn’t believe -MR. TOMBACK: At this point, don’t remember the guy’s name. MR. GORTON: Well, I don’t remember the guy’s MS. NOONAN: Yes, you’re on tape. name. MR. TOMBACK: MR. GORTON: That’s okay. I’m just a lawyer. One of the things that’s always a bit of an issue is whether these people are sort of marketers, right? This is always the problem, right, and this is with my Ph.D. students who go to investment banks, they suddenly become marketers, right? need numbers. So you It’s better to have numbers, not so much your opinion -- or just ask for information, like, “How does this actually work,” and this kind of thing. 159 FCIC Interview of Gary Gorton, May 11, 2010 So, I’m sorry, I don’t have more -- I can’t remember exactly where we went. MS. MORAIN: It’s interesting, you said that there wasn’t really a question that housing prices were going to go down; it was just about how bad it was going to be. MR. GORTON: Yes. MS. MORAIN: So do you know of any discussions within AIG-FP where that discussion went to, “Well, I wonder if this is ever going to trigger one of these collateral provisions that are written into all these contracts”? I mean -- MR. GORTON: like that, no. I don’t remember a discussion I don’t remember. Because I think -- you know, we thought housing prices were going to go down. We didn’t think all bond prices are going to go down. But that’s the crisis, right? The key, as I said in my testimony to FCIC, the real question is to explain why all bond prices went down. When they went down, you know, it was truly shocking. calls. And then that triggers all these collateral And I can remember coming in and, you know, people are yelling on the phone. And I said to Tom, 160 FCIC Interview of Gary Gorton, May 11, 2010 asked him like, “What the hell’s going on?” He goes, “We have all these collateral calls.” “Collateral calls about what?” He goes, “The CDS book.” I’m like, “What?” It was like, you know -- so not to my knowledge. I mean, it may have happened with other people or with the risk people, or I don’t know. But I don’t remember any conversation that I was involved in about this. MS. NOONAN: So we know now that you have read the ISDA master agreement. MR. GORTON: Yes. It doesn’t mean I remember MS. NOONAN: And that you’re familiar with it? MR. GORTON: Yes. MS. NOONAN: I mean, were you familiar with it it. at the time? MR. GORTON: No. After the crisis -- MS. NOONAN: I’m going to go find your students from 2002. MR. GORTON: No, you can ask them. I mean, I actually read it, and I wanted to read, you know, a CSA template afterwards because I was writing these papers for these Fed conferences, and I 161 FCIC Interview of Gary Gorton, May 11, 2010 wanted to know exactly what these things said. I mean, I knew there were CSAs before all this happened, I just never looked the one. MS. NOONAN: And is the CSA where the collateral provisions are? MR. GORTON: Yes, yes. Right, so you sign a master agreement, then you trade under a confirm under the master agreement. And then there’s either a general CSA or there’s one for each transaction. So that was the -- once I was alerted that we had all this collateral stuff, you know, I wanted to just see what these documents looked like. I never looked at any of our documents, but I looked at the blank one. MS. NOONAN: And the blank one -- would the blank one be -- I mean, I’m trying to figure out -- this is sort of a separate question -- but were the collateral provisions in the AIG-FP contracts standard? Was it typical that there would be a collateral call upon a downgrade of the parent company which -- you know, which was the reason that the New York Fed was working the way it worked during the crisis, was to prevent them from being downgraded, because the downgrade would have triggered all these other collateral calls? 162 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: the answer. Yes, I don’t know. I don’t know I mean, I’ve never seen an actual one. ISDA does these surveys, but I don’t think they surveyed that. They don’t ask that question. You know, they ask, you know, “How much collateral have you posted?” But -- so I don’t know. MS. NOONAN: I don’t know. Okay, okay. Do you have one? MS. MORAIN: Well, why do you think those conversations didn’t take place? Or if they didn’t take place, why do you think they weren’t about collateral? Was this a failure of risk management? What was -- what happened? MR. GORTON: know. You know, I don’t -- I don’t I don’t know whether people just thought we would never have a crisis like this or -- I don’t -- I don’t know. I mean, I, frankly, just don’t know. I mean, I had sort of proceeded under the assumption that everything was linked to the rating. And as long as Greenberg was there, Greenberg could always raise money in China, Greenberg could negotiate. So when Greenberg went away, suddenly you didn’t have your negotiator. I still didn’t think there was going to be a problem. I didn’t think about it. When the crisis happened, I was kind of amazed 163 FCIC Interview of Gary Gorton, May 11, 2010 to have all this stuff linked to, you know, prices. I don’t know. I don’t know who knew about it. Even now, when I ask people, you know -- I mean, I never asked Al point-blank but I’ve asked, like, Fewings. You know, he doesn’t know. MS. NOONAN: What about Bob Lewis? Or who is the risk officer in FP? MR. GORTON: Pierre Micotis. MR. TOMBACK: Micottis. MR. GORTON: Micottis. I don’t know. or knew. I have no idea what they know I don’t know. MS. HEYL: Pierre wasn’t involved with the credit for a long time. MS. NOONAN: Okay. MR. GORTON: I mean, I don’t know. MS. HEYL: He wasn’t the risk manager for this business during much of the period we’ve been talking about. MR. GORTON: whether Ed Diaz knew. But I don’t know, you know, You know, I don’t know. MS. NOONAN: Okay. Do you have more? MS. MORAIN: You know, we’re supposed to be studying the causes of the crisis and the institutions that failed or would have failed but for the receipt of 164 FCIC Interview of Gary Gorton, May 11, 2010 exceptional government assistance and, obviously, AIG falls into that category. And, obviously, AIG-FP is a central part of the story of AIG’s failure. And so where would you suggest we look? Where else would you suggest we look, I mean, to help answer the questions of why the problems at AIG-FP took place? MR. GORTON: Well, I mean, other than interviewing people like you’re doing, I’m not sure what else you can do. I mean, I don’t -- I mean, this is a really good question, like who knew about the collateral. suspect that somebody knew. I don’t know. I But, I mean, all you can do is ask, I suppose. MS. NOONAN: Did you ever get the sense that people were relying on your model for the proposition that AIG would never lose money on these deals, and that perhaps that they ignored these other provisions? MR. GORTON: No, I don’t think so. First of all, they would never say “never lose money,” because I beat people over the head about that for years; right? It’s highly unlikely. But remember, the model is actuarial, right, so it doesn’t care about prices. So, you know, the model could be exactly correct -- it’s very unlikely you’re going to lose 165 FCIC Interview of Gary Gorton, May 11, 2010 money, which turns out to be pretty much true -- and yet if you have collaterals linked to prices and prices go down, the model can be right and then you’d have to pay out all this collateral, which is exactly what happened. So, you know, I was very worried about the accounting-rule change that said we were going to have to mark this very large book. But I didn’t know how much collateral was going to be triggered by these prices -- by price movements. So, you know, that was just -- that was an Achilles heel. MS. NOONAN: Did it ever cross your mind that the sheer size of the notional amount that AIG wrote protection on was a problem? MR. GORTON: No, because with derivatives, size is very, very misleading, right? I mean, if you have super-senior and you have a very, very large size, you know, that can be safer than having, you know, 500 million of the equity, right? So the dollar amount doesn’t tell you what you really want to know. This is true of derivatives generally, right? I mean, if I do an interest rate swap with you and it has a market value of zero at day one, and it’s a billion dollars -MR. CREGO: Pardon, Professor Gorton, I’m 166 FCIC Interview of Gary Gorton, May 11, 2010 going to disconnect now. call. We have another conference Sorry to interrupt you. And I thank you for your time. MR. GORTON: You’re welcome. You know, that -- what you really want to know is, you know, if interest rates move by 500 basis points, what happens. And, you know, so the dollar amount -- I mean, I think the dollar amount made a huge difference if you had these collateral linked to prices, right? But that’s not an issue that came up. It did come up about the sheer size of the book. That came up, you know, over and over again. MS. NOONAN: Well, but, I mean, aren’t credit derivatives a little bit different than interest-rate derivatives? MR. GORTON: Well, they’re not -- MS. NOONAN: When you think about the amount that’s really at risk? I mean, you had an example of the interest-rate swaps where you’re zero at day one, and the movement is really just the difference of the movement in the interest in the fixed versus the floating. But in a credit derivative, it’s a little bit different, right? MR. GORTON: Well, it isn’t –- MS. NOONAN: You’re on the hook for the whole 167 FCIC Interview of Gary Gorton, May 11, 2010 amount? MR. GORTON: No, that’s true. But the point here is that you’re attaching after this buffer and it’s on a portfolio, right? So because -- I mean, I suppose I told you it’s a trillion dollars. number? Is that a big And then before you say “yes,” I say, “But we have $10 trillion buffer.” MS. NOONAN: MR. TOMBACK: Right. But if the -- The difference for the interest-rate swap is immediate every day that goes on, you’re gaining or losing real money. I don’t know how much -- I don’t know enough about this portfolio, but I presume they even have [unintelligible] which haven’t triggered a loss in the credit-default swap world their residuals that are allowed. You know, you get the bonds. So it’s not an on/off switch. MS. NOONAN: MR. TOMBACK: MS. NOONAN: Sometimes. It doesn’t go from – There are bankruptcy -- I mean, sometimes -- under the straight contractual terms, that’s true; although in bankruptcy, in the event of an AIG bankruptcy, that would not have been true, which is sort of news to legal. MR. TOMBACK: But you’re looking at it from 168 FCIC Interview of Gary Gorton, May 11, 2010 the perspective of AIG, and so you get the bonds. I presume all of the counterparties are still alive. I don’t know. MR. GORTON: MR. TOMBACK: MR. GORTON: MR. TOMBACK: Well, Merrill isn’t. That’s right. And Lehman isn’t. Yes, but Merrill –- Merrill was merged -MR. GORTON: Yes, Merrill got merged, and so Lehman. MR. TOMBACK: But most, if not all, counterparties are still on, so you’ve got that as well. MR. GORTON: I guess my point is kind of a simple one, which is that the risk depends on where you are in the capital structure, you know, not just the dollar amount MS. NOONAN: No, I mean -- I take your point. I take your point. MR. GORTON: That’s my only point. MS. NOONAN: Yes. MR. TOMBACK: To summarize a different point, I think, that Professor -- I mean, Professor Gorton stressed that, right, the “very unlikely” as opposed to “never will happen” nature of the model. 169 FCIC Interview of Gary Gorton, May 11, 2010 And the “very unlikely,” it turns out, at least at this point -- not that the result justifies the model or the means or how you got there; but there is a point where there are no losses. MS. NOONAN: Yes, I mean, it’s an interesting fact about the AIG-FP story. MR. GORTON: Another interesting fact is that the Fed has made, like, $50 billion or so on Maiden Lane III, so -MS. NOONAN: Billion? MR. GORTON: Yes. MR. TOMBACK: Many of the CDSes -- many -- I guess virtually all the CDSes that come back -– MR. GORTON: MR. TOMBACK: The prices went up, right? So the collateral -- some substantial amount of collateral has come back to AIG. But it’s not AIG anymore, it’s you and me. It’s America. MR. GORTON: Also, I don’t know how the collateral works, either. Like, I don’t know if anybody gave the collateral back because the prices went up. I don’t know. MS. NOONAN: That’s an interesting question. MR. GORTON: Well, I think so -- I don’t know the details about Maiden Lane III; but since they bought 170 FCIC Interview of Gary Gorton, May 11, 2010 a lot of the bonds, I don’t know how -- I don’t know how the collateral -- I don’t know. MS. NOONAN: I actually think that the collateral probably isn’t at play anymore because they ripped up the –- they ripped up the swaps – MS. HEYL: Right. MS. NOONAN: -- so that the counterparties are paid in full on the swaps; and now, they just hold the underlying bond on Maiden Lane III. MR. GORTON: But do they have to give the collateral back then? MS. NOONAN: No. MR. GORTON: Why would they have to give the collateral back? MR. TOMBACK: Let’s be fair, without knowing the instruments themselves, the collateral is not for free. So if you paid out collateral to a counterparty to protect them against a loss they perceived they would have of $10 billion, and with, say, Goldman didn’t suffer any losses because the government filled them 100 percent, the collateral either contributed to filling them 100 percent or went back to AIG. But the 10 million was no longer -MR. GORTON: MR. TOMBACK: Billion. -- was no longer -- billion, 171 FCIC Interview of Gary Gorton, May 11, 2010 excuse me -- was no longer, quote, unquote, sort of out there or Goldman’s just to keep, free of charge. And -- MR. GORTON: I mean, at this point, all I MS. NOONAN: It was a pay-out on their know -- position. MR. TOMBACK: It was conceptually returned to the AIG side of the ledger, and they use it to pay off whatever obligations it had. MS. NOONAN: Right. I think that’s right. MR. GORTON: And all I know is what I read in the paper, and it’s very confusing. MS. NOONAN: This has been really helpful. I appreciate it. I should ask at the end, is there -- is there anything else that we haven’t really talked about that you think we need to know or is critical to our understanding, at least, from your point of view, what happened at AIG in the Financial Products group? MR. GORTON: Not -- I mean, I’m sure there’s MS. NOONAN: Are there big things that come to other -- mind that we haven’t really discussed today? MR. GORTON: Not that I know about, no. MS. NOONAN: Okay. 172 FCIC Interview of Gary Gorton, May 11, 2010 MR. GORTON: As far as what I know, I think we’ve covered it. MS. NOONAN: Okay, thanks very much. MR. GORTON: You’re welcome. (End of interview with Gary Gorton) --o0o-- 173