For a complete list of Beginners posts, see Financial Crisis for Beginners.
This is more of an advanced beginners topic – I already covered CDOs (collateralized debt obligations) in my first Beginners article – but I imagine that most of our readers are already familiar with structured products. At least, many people know that first a bunch of securities are pooled together, and then they are “sliced and diced,” in the common media parlance I find incredibly annoying. But Joshua Coval, Jakub Jurek, and Erik Stafford have a new paper, “The Economics of Structured Finance,” which does a brilliantly clear job of describing what these securities are and why they were so widely misunderstood, with the results we all know.
The paper is 27 pages long, not counting references, tables, and figures, and if you are comfortable with probabilities and follow it carefully you can understand everything in it. I will provide a summary to whet your appetite. I am not going to use numerical examples because the examples they use throughout their paper are so good.
The key to CDOs is that they could be used to manufacture AAA-rated securities out of underlying securities (like mortgages) that were not even close to AAA. (“AAA” is a bond rating, meaning that the security in question had about a 0.02% chance of defaulting in a given year.) This is well known. But although these new, synthetic securities had expected default rates comparable to traditional AAA-rated securities, they had other properties that were unlike their traditional brethren, having to do with (a) correlations between the underlying assets and (b) sensitivity to underlying default rates. (a) is the probability that, if one mortgage inside a pool defaults, the other mortgages will also default; (b) is the degree to which small changes in those default rates can affect the expected value of the manufactured AAA securities. This meant that these CDOs were much more sensitive both to errors in estimating their characteristics, and to macroeconomic changes, than most people realized.
If you didn’t follow that I’ll go over it again more slowly.
In a simple, “pass-through” securitization, each investor in the pool of mortgages has an equal claim to the mortgage payments. Therefore, the expected loss for each security is exactly the same as the average default rate of the mortgages.
In a CDO, the investors have unequal claims. By creating some junior tranches (“tranche” is French for “slice,” in case you were wondering) that absorb the first losses, you create a large senior tranche that is buffered and suffers no losses until all of the junior tranches are completely wiped out. This is why the senior tranche can get a AAA rating; the estimated chances are pretty low that enough people will default to wipe out the junior tranches. In a CDO-squared, you take some of the junior tranches of ordinary CDOs, pool those, and then create tranches out of that pool. The amazing thing is that you can then create not only a senior tranche but a mezzanine (middle) tranche of your CDO-squared that has the expected default rate of a traditional AAA bond, even though it is made out of junior tranches. (There are very clear examples of all of this in the paper.)
However, this only works well if the default probabilities of the underlying mortgages are not highly correlated. Assume in an extreme case that defaults among the underlying mortgages are perfectly correlated: either none default or they all default. In this cases, the tranches do nothing for you: if all the mortgages default, then the senior tranche gets wiped out along with the junior tranches.
Furthermore, the performance of AAA-rated tranches is highly sensitive to the default rates of the underlying mortgages. Conceptually, this happens because the amount of protection provided by the junior tranches is not that much bigger than the expected default rate; so if the actual default rate is just a little higher than expected, a much larger proportion of the protection will get eaten up. In the example in the paper, an increase in defaults from 5% to 7.5% can knock a AAA-rated tranche of the CDO-squared down to a BBB- rating.
The conclusion is probably apparent to many readers at this point. The underlying mortgages were more highly correlated than people thought, both because they often came from the same types of developments in the same regions (California, Nevada, Florida), but also because everything in the economy became highly correlated. And as the economy got worse, default rates climbed higher than estimated based on historical data, because all the historical data came from a period when housing prices only went up. While this would only have a “linear” impact on a simple pass-through securitization (double the defaults, double the losses), it had a “non-linear” impact on AAA tranches of CDOs, and especially of CDOs-squared.
Finally, there is one more misunderstood characteristic of CDOs. Securities with the same expected payoffs, and hence the same rating, can have different characteristics. In particular, they can differ in their degree of correlation with the rest of the economy. The authors cite catastrophe bonds (which default only, for example, if a hurricane hits South Florida) as securities that are uncorrelated with the economy. Because of their lack of correlation, they are more desirable than other securities with the same rating, and hence have lower yields (higher prices). Senior tranches of CDOs are just the opposite: they only go bad if the economy as a whole goes bad; that is, they are highly exposed to systemic risk, which almost by definition is difficult to quantify. Because of this high degree of correlation, investors should have demanded higher yields (lower prices). But because investors by and large thought that all AAA securities were comparable, they didn’t demand high enough yields, and the issuers (investment banks) made the difference.
The bottom line is that all AAA securities are not created equal – even if they have the same estimated probabilities of default. And treating AAA tranches of CDOs and CDOs-squared as if they were AAA corporate bonds played an important role in the growth of the structured finance market and, as a result, the overall asset bubble that is collapsing around us.
Update: A reader points out that CDOs generally do not contain mortgages, but instead already contain subordinated (not senior) tranches of mortgage-backed securities or of other CDOs. And the mortgages inside those mortgage-backed securities are often subprime. Because these tranches are themselves unique, it is even harder to project the cash flows coming into a CDO. A mortgage-backed security backed by a pool of plain vanilla mortgages, by contrast, would be relatively easier to value and less subject to price fluctuations.
By James Kwak
The Baseline Scenario 
http://zerohedge.blogspot.com/2009/03/exclusive-aig-was-responsible-for-banks.html
Hi James,
I used to work on Wall Street building risk models for CDOs before moving to the buyside, so I’ve seen things from both perspectives. My experience building risk models provided the ammunition needed to gravely warn anyone over the years who would listen to stay away. Fortunately, the asset management firm I worked for was smart enough to stay away from investments they didn’t understand.
During the heyday, you would go to credit derivatives conferences and see extremely bright scientists debating which CDO pricing model was superior. The thing that struck me was that NO ONE can come up with a decent pricing model and hence there can be no decent risk model. The Bayesian approach will not necessarily help much compared to what many quants were doing anyway. It is pretty futile to even try, in my opinion.
I just hope structured finance goes the way of the dinosaur.
I enjoyed the article. In fact, I just posted a few paragraphs of my own on “Structured Finance for Beginners” at my blog (same “theme” as yours, both look-wise and content-wise). It is part of a bigger plea
Economic Darwinism
http://economicdarwinism.wordpress.com/2009/03/28/letter-to-obama-our-voices-will-be-heard/
PS: I like my cashflow waterfall and champagne tower artwork better ;)
if this claim about Geithner and AIG is true regarding closing out CDS portfolios at intentional losses to benefit banks just prior to asking for more taxpayer bailouts proves true, he needs to be imprisoned immediately.
http://zerohedge.blogspot.com/2009/03/exclusive-aig-was-responsible-for-banks.html
Hi idoc,
I hadn’t thought about it that way. I was just thinking about how incompetent the AIG traders were in light of claims that their skills are crucial to the orderly unwind of the AIG portfolios. What skill does it take to dump an entire portfolio? And these people still think they deserve those retention bonuses?
Yes. I think you have a point. This smells very bad.
Permit me to try my hand at a summary. If one has a CDO with two tranches, Sr. and Jr., and there is no correlation between them and the Sr. goes bust only if they both go bust, then, if the chances are 10% for each, Sr. has a chance of 1% (10% X 10%) of going bust and is rated investment grade (almost). However, if they both go bust at the same time (100% correlation), the chance for Sr. is 10% and is rated deep in junk as is Jr.. The rating agencies assumed no correlation and rated accordingly. Thus reassured, investors paid way too much for the Sr. tranches. It gets worse for more complex instrument which greatly magnify the effects of poor models.
It’s not rocket science. It’s a mystery to me how this happened. There seems to have been a mass delusion that the good times would roll forever.
Why didn’t this research take place before the bust? Hindsight is always 20/20
Does it make sense to think about the correlation going up as the assets inflated in value? One could create a correction term that compared the asset class’s rate of value inflation to the economy as a whole. As the correlation increased, the value of the securities would decrease. This would cause the amount of capital available to that asset class to decrease. The decrease in available capital would decrease the inflation of the asset class values. Wouldn’t this create a more stable system? It seems like the core cause of the collapse was the asset class’s inflation rate exceeding the rest of the economy.
Mystery Mark? No, it’s not a mystery at all. It’s called fraud. A complete idiot could see (now, that its been exposed) that the financial modeling the ratings agencies were using simply reeked of collusion. It’s not a mystery….it’s fraud, plain and simple. And NO ONE in Washington wants to talk about THAT. Know why? Washington was a-hole deep in it too. ESPECIALLY Paulson. The guy sold you out….it was Paulson who pushed congress for the 23A Exemptions. Hank Paulson did that….Gee Hank, why was that anyway? Hmmmmm?
Have I missed your discussion of rating agencies? Or has there been no discussion of rating agencies in The Baseline Scenario. In the above piece you take the judgment of rating agencies for granted and as presumptively accurate. Is that prudent?
Have a look at the paper. It has a pretty decent discussion about the role of the ratings agencies.
Well, I just think its wonderful that you are all having these terrific debates about what happened, how it all went so terribly wrong, etc. Yep, just great. If I remember the story of the Titanic correctly, didn’t the band keep playing on the main deck of the ship after it hit the iceberg and people kept dancing…..while the ship sank? You Americans need to take the bull by the horns, arm yourselves, and march on Washington. It’s really the only thing they fear. My fear is however, you will all wait until you have been completely bankrupted before you do anything that will make any real differences.
James, since you study law, is there any way that some kind of private action on behalf of all US taxpayers, can be taken against the US government for allowing the AIG debacle? If you read WAKE UP AMERICA’s post of the zerohedgeblog, the offenses against the US TAXPAYERS are mounting…maybe something like a class action suit against our government?
Sorry if it is ridiculous, but, without the public’s participation in opposing the misappropriation of funds, nothing is likely to change generations of goodwill have been lost, and the future of America is mortgaged into bankruptcy.
If we did that now, it’d be like maybe 5 skinny (or not-so-skinny) bloggers. Not a big threat. What we’re trying to do is first raise awareness. Most people still don’t know or understand what is going on. I personally think peaceful protests are in order. First things first: education.
Good point, I’ll give you that one. But as a person outside looking in, from “across the great pond”, it just seems like the americans are watching passivly as their futures are being stolen before their very eyes. As Rome burns, they dance the night away. We get a kick (over here) out of your media in the US. CNBC is the biggest comedy of all. It’s simply amazing that no one has gone and burned their broadcast HQ to the ground. That would certainly get my attention, if I were a Politician. Understand, over here, we have experienced deep corruption in the past. What I took from the history of those events, is that if the government and their cronies had no fear of repercussion, they just continued to steal until the people revolted…..literally. Your US Constitution provides for such a thing…no? You see, the framers of your constitution experienced deep government corruption in England, and wanted to make sure the citizens of the new world had the tools in place to combat such corruption with armed militias. The founders knew THAT is really the only thing corrupt governments fear, and they are correct.
James, once again total quality work–you’re really a super smart guy. This particular take seems like a good intro to a second follow-up article outlining the key math algorithms that supported the various risk allaying assumptions using equation examples. You say, “The underlying mortgages were more highly correlated than people thought, both because they often came from the same types of developments in the same regions (California, Nevada, Florida), but also because everything in the economy became highly correlated.” Please, we’d love to learn more about this ‘correlation’ THANG.;-)
What you see happening in America is a scary thing that happens in all mass media rich cultures. If you have sources of information that broadly distribute the same ideas to millions of people, the unintended consequence is that a large section of those people all engage in the same action. This creates a large scale group think that can have consequences that range from the amusing to the disastrous. So imagine 300 million people, most of whom have identical educations, have watched all the same TV, read all the same things, purchase all the same things, and have on the whole identical experiences. No individual is a robot, of course, but on the whole they tend to behave in aggregate. Get all these people, living in the wealthiest nation in the world, and put them on a buying spree on seemingly unlimited credit. Imagining the consequences is scary enough, but now we get to experience them in actuality.
So, now you have the same group mind contemplating the consequences. In addition, you have the same mass media feeding the group mind. Also, the media is developing its message by surveying the mass mind, looking at what it responds to, etc. What’s going to happen next? You can see the themes unfold in the media bit by bit. I see a strong “holding people accountable” theme which suggests no one will be happy until someone gets punished. There is also a lot of helplessness running round. Then there is the president, who strikes me more and more as the parent of last resort that the mass mind is clinging to and resenting at the same time.
Marching on Washington on a grand scale? Armed revolt? Even as some politicians and pundits call for it, it will never happen. The mass mind is not activist or radical.
The good news, if you like, is people are checking out of the mass mind more and more. Blogs and other diverse information sources, whether they are accurate or not, will create an increasing diversity of behavior. Certainly all forms of mass media are experience declines in audiences while the internet is increasing. This blog is not really the dance band on the titanic, its a little canoe with an AM radio. Hopefully it will float along with hundreds of thousands of other little canoes, some of which might crash into icebergs, but the most just keep skimming along.
I guess analysis of what went wrong within the banking/financial sectors is fruitful in developing more responsible behaviour and hopefully will get rid of “weapons of mass(financial) destruction” (Warren Buffet). However, I feel it it ESSENTIAL to remould peoples’ mindset on the ground globally and specifically in the USA. Past excessive consumption will NOT be rectified by preaching government consumer stimulus programmes and encouraging more borrowing. Such an approach merely pushes the problem down the road and IS “The Road to Hell”, to quote the current President of the European Union. The responsible message from the upcoming G20 gathering should be “People, we have to learn to live within our means!!! Governments will create jobs by funding infrastructure spending, provide an extensive social security safety net and look after the sick. We will arrange terms to help you pay off your debts, but YOU remain responsible to do so.” To my mind this is the only (slow) way to global recovery. In the words of the Rolling Stones “You can’t always get what you want, but if you try sometime, you get what you NEED”.
Am I a realist, or am I a dreamer?
It might be worthwhile to say something about *why* the correlations were so badly underestimated. Do you, as blogger/journalist Felix Salmon did, blame “Gaussian Copula” and related technology for this?
I’m interested in hearing James’ take too. In the menatime, I would partition the question into a bunch of sub-questions.
1. Can CDOs be modeled accurately at all?
I say no, which if right, renders all other questions moot. But, for the sake of argument, lets say we can try to model them.
2. If you try to model a pool, is a copula model ok?
I think you can do a lot worse. The answer depends a lot on what you want to do with the model. For trading and relative value analysis, I think it is ok. For risk analytics and ratings, I’m not so sure.
3. Should we blame the Gaussian copula model?
Let me ask you, should we trust any model in the first place? If you put blind faith in any model, it doesn’t matter whether it is Gaussian copula or what, you are asking for trouble. If you blame anyone, blame the people who put blind faith in the models. Do NOT blame the people who built the models, because if you asked them, they would have told you the weaknesses of the model themselves. Everyone ignored the caveats.
Now aside from Gaussian copulas and bad assumptions about correlations, there are many other parameter inputs into the ratings model. For example, average FICO score, average loan-to-value “LTV”, average maturity, etc etc.
Of all the model inputs, the one I think was the most egregious (even moreso than correlation) was the assumption about home price appreciation (HPA). Most model even up until 2007 assumed HPA would always be positive. In fact, most models, including those of the rating agencies, did not even have the ability to input a negative HPA into them. I joined a research group at a large structurer (that no longer exists for some reason) in August 2007 and was shocked that a big effort was underway to modify the models to incorporate negative HPA. No one had ever considered the possibility that home prices could go down, because they hadn’t in recent history.
It takes a lot of skill to PROPERLY dump a big portfolio. It’s a chess game. You can’t sell everything at once (people would rip you off bigtime) and so you have to sell it piece by piece. You have to choose your pieces carefully (these are illiquid markets, and you have to try to get good prices) and consider your hedges (if you sell one part that was offsetting another part, you might be open to very large risks). So it requires a very tight mind to do this economically.
The correlations were badly estimated because people used what’s called ‘implied correlation’. What that means is that people took the market prices in these securities and used these prices as the ‘answer’ the model gave and then figured out which model parameters were consistent with these prices.
When the market prices for these instruments changed, the ‘correlations’ suddenly changed. That’s as nobody was ‘assuming’ low correlations in the first place, nobody ‘assumed’ high correlations later. The models yielded the correlation numbers in the first and second case.
I don’t think the fundamental problem is with the model. If you use most probability based models, and if you assume correlations that look good ‘a priori’, you will end up having a very safe-looking super-senior tranche. if not, you can just increase the size of your pool until you do. That’s a fundamental aspect to almost all multivariate probability model.
There are two problems with using ANY model.
One is that the models are very sensitive to disruptions in the market. How’s that work? Well, the model parameters (ie correlation) that you use for the model are implied by market prices. So when people say ‘correlation’ is much higher now than it was, what they mean is that market prices for the various tranches IMPLY that correlation is much higher than it was. this is just math and nobody has or had a ‘view’ on this that changed and made it so. the market prices changed — partially for market illiquidity reasons — and that made ‘correlation’ change.
Two is that the way these models were used, especially models for the supersenior (‘AAA’) tranches, was to justify very high leverage. These tranches paid very small coupons and so they were then packaged into leveraged structures with all sorts of doodads that defined allowable leverage, deleveraging triggers, etc.
There are two problems with that. First, (and let me shout it) BECAUSE YOU ARE ALLOWING MODEL RISK MEASURES TO DEFINE ALLOWABLE LEVERAGE YOU NEED VERY ROBUST AND PRECISE MODELS. And these models are neither robust nor precise. It’s more or less as if you were doing an fraction (allowable debt/equity) with a very tiny number in the denominator and using that to determine allowable leverage — it has to be perfect, and it isn’t in the nature of things for it to be perfect. And second, most of these doodads in the leveraged structures that were laid on top of these instruments were never ever modeled.
One thing that’s puzzled me is that it looked to me at the time (though I only blogged about it later) that at least one likely cause of highly correlated defaults on at least subprime loans was pretty obvious: a rise in interest rates off historically low levels.
The argument, in brief: a lot of subprime loans were adjustable-rate mortgages issued to people who could barely make the payments, and certainly wouldn’t be able to keep writing the checks if the rates adjusted up. But if the rates went up, they couldn’t sell to a similarly situated buyer either (because that buyer couldn’t make the payments to begin with). Which, if that’s all valid, leaves default as a much more likely event for a lot of these people, all at once…
you’re right, that should have been caught. but i bet they were assuming that house prices would go up even faster than that or that people would just refinance out of these loans.
in that light kind of interesting that the trigger for this crisis was a severe devaluation of the dollar in terms of commodity prices (ie high gas prices).
The one part of this that I do not understand is how a default on a house results in the CDO holder receiving 0$.
No house will go down to 0$, they all get sold for something, even if it is a 80% cut in price. Who gets the money from that sale? Would not that money go to paying off the senior debt holders first? I clearly am missing something because all of this analysis seems to assume that when every mortgage in the pool busts everyone is left with 0$.
Very impressed with the recent articles and content put out by Simon and James. Would love to see an article that talks about some simple steps that the typical american family can take to help protect themselves in the coming recession. Things like:
– Should I refinance my home if able?
– Sell securities, or not?
– Take advantage of new Tax changes?
– Buy gold as a hedge?
– Add more debt since hyper-inflation is coming?
– Buy guns and ammunition?
– Write our congressman?
– etc.
Thanks, JBK
White, turn on the news, the BBC would perhaps be the best. The Europeans are protesting heavily at the scene of the forthcoming G20 meetings. “Protesting” may be a bit of an understatement, it’s on the verge of becoming a riot. Your second to last paragraph indicates that the Americans will never march on Washington. I agree. They don’t have the courage or backbone for such things. Their forefathers did, but this generation of Americans are soft, afraid and weak. Their founding fathers would probably be embarrassed by this generation of lightweights that have infected the country they founded. Besides, marching en masse on Washington doesn’t get them a new flat screen HDTV or a new Visa card, so why bother….right? Watch what happens these next few days at the G20m meeting over here in EUROPE, where we actually “get it”. I predict Obama will come out of this with his tail between his legs, and Merkel will be viewed as having won this round….we’ll see.
Except that rising interest rates would also seriously impair the ability of a strapped borrower to refinance.
So, we’re left with the idea that the borrowers would always be able to re-sell the house at a profit. But ex hypothesi, they’d be trying to sell at the same time lots of other borrowers are also trying to sell, for the exact same reason. And at the same time (as I’ve already mentioned) as the number of buyers who could afford the former prices is actually dropping, because the same nominal price, with a mortgage at new, higher rates, leaves them stuck with a dramatically higher payment.
In short: the consequences I would have expected from a rise in interest rates a priori would have included lots more sellers, and lots fewer buyers — and it’s hard to see how prices could keep going up in that environment either.
Which is easy to say now, I suppose, when it’s all already happened. But I continue to be perplexed by the failure of specialists to forsee it ahead of time…
Our-government-is-even-more-corrupt-than-yours is a pretty depressing argument to have… nevertheless, the UK is meking a strong bid for lead position with the latest revelation that a Cabinet Minister’s husband claimed for his pay-per-view porn on her government expenses.
Even the level-headed Libby Purves saw this as a likley tipping point in an article arguing that throwing good money after bad no longer impresses anyone: http://www.timesonline.co.uk/tol/comment/columnists/libby_purves/article5998537.ece.
Our government’s response was, if possible, even more astounding… merely to give Members of Parliament the chance to edit their expense claims: http://www.timesonline.co.uk/tol/news/politics/article6004038.ece
I’ll shut up after writing this, but I wanted to point to another very good posting on a similar topic. This deals with super-senior CDOs and AIG. It’s very technical, but also very readable.
http://www.acredittrader.com/?p=65
People knew.
(i) Tanta’s posts over at Calculated Risk have been addressing this since 2005. (Both CDOs — see ubernerd on Dog Tails — and destructive subprime and Alt-A loans.)
(ii) A lot of the people on Wall Street who figured out what was going on simply treated it as a good opportunity to make money via credit default swaps — see Michael Lewis on Eisman http://www.portfolio.com/news-markets/national-news/portfolio/2008/11/11/The-End-of-Wall-Streets-Boom
nice blog :)
From: A practical guide to the 2003 ISDA credit derivatives definitions:
“The arbitrage market has evolved a stage further to the extent that single tranches for a particular portfolio can be structured without the need to create the remaining tranches to “complete the capital structure”
In other words, because the securities are synthetic (read- “a bet about whether or not someone is going to pay”, and not real- read: “when someone makes a payment, you are entitled to a piece of their money”), then one could sell only “Senior” tranches.
I had really been wondering how so many of these products could have been sold because while it make ssense that many people would have been willing to buy the “safe” senior tranches, people should have been much more cautious when buying junior tranches.
Of course, if one converts junior tranches into a senior tranches, then it makes sense that so many could be sold because everyone thought it was “money good”.
From: A practical guide to the 2003 ISDA credit derivatives definitions:
“The arbitrage market has evolved a stage further to the extent that single tranches for a particular portfolio can be structured without the need to create the remaining tranches to “complete the capital structure”
In other words, because the securities are synthetic (read- “a bet about whether or not someone is going to pay”, and not real- read: “when someone makes a payment, you are entitled to a piece of their money”), then one could sell only “Senior” tranches.
I had really been wondering how so many of these products could have been sold because while it make sense that many people would have been willing to buy the “safe” senior tranches, people should have been much more cautious when buying junior tranches.
Of course, if one converts junior tranches into a senior tranches, then it makes sense that so many could be sold because everyone thought it was “money good”.
Quick question,
It appears that junior level tranches must have an inherently greater level of risk of default than the mean of the pool. Was there an active market for these in 2005-2007? Do lower level tranches make up a significant fraction of bank holdings today?
Any and all info appreciated.
The bottom line is, don’t do business with the Americans going forward…why would you do that when other markets/investments will probably pay in the long run? I’m done with them. I have kid’s I have to worry about. They have proven themselves to be untrustworthy, why do this again?
Econo, I just went to your blog and read your piece. Very well done — thank you for putting your thoughts down on “paper”.
Further to this topic: http://nymag.com/news/business/55687/
Would an additional block-level scoring system (at the APN level) have made a difference? See: http://www.homevaluepredictor.com
The paper had another interesting and counter-intuitive claim.
The expected payoff of Senior tranches is relatively *insensitive* to changes in parameters, even though parameter variation can create a large change in the rating of the Senior tranche. This holds true even for CDO^2. This bodes very well for the Geithner plan, because it indicates that at least some of these CDOs really are seriously undervalued, and that they really aren’t just worthless paper.
Hi Patrick,
Unfortunately, what you say is what people believed going into the crisis and is partly responsible for the mess we’re in.
There are a couple different aspects. One that the paper does not address (it is a good paper, but clearly coming from the “Ivory Tower”) is the fact that many investors lost their shirts due to liquidation clauses. That is, once the principal in the pool drops to a certain level, the entire structure is forced to dissolve.
I haven’t seen the detailed mathematical formulation James et al used, but I’m pretty sure (based on the use of the word “correlation”) they assumed normal distributions. So a second aspect to the crisis is RISK MANAGEMENT. Trading desks at the big banks are allocated capital according to some risk model that calculates value at risk (VaR). VaR defines a “boundary” of losses whereby any loss greater than “VaR” is considered a “tail event”. VaR does not tell you anything about how much you may lose if a tail event happens, it merely gives a number specifying the boundary where beyond “there be dragons”. So it would be MAGICAL if traders could somehow find a security where all the risk was in the tails because risk models based on VaR would be blind to such a security. Enter CDOs. CDOs represent a blind spot to risk management systems. As such, CDO desks were allocated capital way beyond anything that was sensible because the risk models were insensitive to the tail events that would cause a CDO to lose money.
I haven’t spent enough time de-constructing these CDO models because I saw how ridiculous they were and moved on to other things. Maybe it is time I did. Any CDO model that does not properly model the tails, e.g. anything that assumes normal distributions, is pretty much a bogus model that should be disregarded. A CDO is a “tail event” security and if the model does not properly handle tail events, it is not a proper model of a CDO. How do you model tail events? Some would say you can’t, in which case, you cannot model CDOs. Hence CDOs should stay out of any responsible asset manager’s portfolio.
Thanks for the reference to the paper. It finally makes clear something I’d been trying to understand. I’m not an economist, but I read your blog daily and am impressed by your explanations, candor, and willingness to hear our opinions.
no matter what you do, there is always the possibility of a tail events. there is always something that can happen that will blow up the financial sector.
what if, say, AIDS had killed the same proportion of the young to middle aged, educated, professional class in the US as it did in, say, Botswana? every insurance company would have gone bankrupt without a bailout. there’s really no way you can tell me that this is a fanciful story and it couldn’t have happened that way. it didn’t, but it certainly could have.
my point is that tail events happen, and you can’t prepare for every contingency.
your logic is that if tail events can blow up cdos then you shouldn’t use cdos. my point is that you can’t avoid tail events; you are stuck with tail events. you can do a much better job of estimating their probability and the damage they will cause. you should try to limit the scale. but you shouldn’t just throw out structured finance because of this very bad cycle.
You know what the Wizards Of Finance in the US needed to incorporate into their financial models? The “Integrity” factor. Back in the day, before the current group of complete idiots took over, lending decisions were very simple. 1. Ability to pay (debt to income) 2. Intent to pay (credit rating) 3. Equity (down payment). 4. Stability of the collateral (appreciation / depreciation) These were the major criteria that went into the decision to lend, or not. All of you Titans Of Finance over there in America, may want to go back to the old way of doing things, you know, like ONLY lend money to people who can actually repay it, and don’t bribe the ratings agencies. But…..you can’t do that, can you? That requires being honest…..so, that’s out of the question. YOUR DONE OVER THERE, WE”RE ONTO YOUR BS
I’m not suggesting that the reason CDOs should be shunned is that they are susceptible to tail risk. As you note, every security is susceptible to tail risk. I’m suggesting that the design of CDOs AMPLIFIES tail risk. In fact, all of the risk of a CDO IS tail risk. Most of the time they make money but during a tail event they can lose everything. This aspect makes them blind to risk systems and I would argue was one of the drivers of their popularity in the first place (which happened to coincide with the proliferation of VaR systems).
NOT FOR PUBLICATION!
Dear James,
Sorry to bother you with this.
Unfortunately, I have seen your “Where did all the money go?” (Financial crisis for beginners) only few weeks ago. I started then to write (off and on) this comment, but now found the subject closed.
Anyway, I hope you will read this comment. And… many thanks for your most instructive exposition.
Apart from the fact (pointed out most aptly by Justin, November 13, 2008) that you have physically destroyed, by a stroke of a pen, the $200,000 bills paid by Danny to some landowner (call him Rusty), I am afraid that you detrimentally mixed money with financial assets. To my mind, money is a claim on what you call ‘stuff’ (e.g., bread or a… house), whereas financial asset is a claim on money. In more than one way, money and financial assets are worlds apart.
Thus, if we don’t forget the 200,000 paid by Developer Danny to Landowner Rusty, the total amount of money ‘in the world’ after the initial transaction is the same $600,000 (200,000 with Rusty and 400,000 with Danny).
In fact, this answers, literally, also the question “where did all the money go?”: money that was Danny’s (200,000) went to Rusty; money that was Homebuyer Harry’s (40,000) went to Danny and money that was Banker Bonnie’s (360,000) went to Harry and then to Danny. Total: 600,000 dollars – in time one as well as time two.
[By the way, you do not inform us what, if anything, Rusty and Danny did with their money, but I shall come to that later.]
However, consequent upon the house transaction, a newly created financial asset came into being, with face value of $360,000, owned by Banker Bonnie and being, in fact, a promise-to-pay made by Harry on the basis of his future income, backed by the market value of the house [cash transferred on the basis of future-income is the heart of the economic action, but out of context here].
Now, according to your perception, the total amount of money in time one would be $1,000,000 (or your $800,000 plus Rusty’s 200,000), i.e., a total purchasing power of one million dollars. But this is totally wrong.
Assume, for example, that immediately after Harry’s house deal, a new player, John, asked Bonnie for a loan. Given John’s excellent record, steady income and solid collateral, this would have been the ideal business for Bonnie who has, according to you, $360,000. But, alas, Bonnie cannot make the deal – not because of regulations, but, simply, because she does not have a cent (at least at the beginning of time one). What she does have is a ‘promise-to-pay’, plus an access to Harry’s house.
Therefore, the financial holdings in your time one is as follows:
Rusty: $200,000 cash (i.e., money);
Danny: $ 400,000 cash (i.e., money);
Bonnie: promise-to-pay $360,000 (i.e., financial asset);
(Harry’s ownership of 10% of the house is neither money nor financial asset).
However, Bonnie could still make the desired deal with John, if she issued a bond backed by Harry’s promise, but even then, she would have to find someone with money to purchase her bond if she wanted to accommodate John.
In short, if purchasing power is the main attribute and ultimate measure of money, financial asset is the opposite of money. You can convert a ‘promise-to-pay’ into money, provided that someone would be willing to part with cash [in fact, you hold the ‘promise’ because, in the first place, money that was yours was transferred, by yourself, to others].
Comes the crash.
The only possible ‘endogenous’ cause of the crash in your $600,000 economy (or $800,000 if you insist), is that Danny and Rusty (the only people with money at the beginning of time one!) reduced, for whatever reason, their spending level, so that part of their $600,000 was hoarded. [If we were allowed to assume ‘other banks’, then part of the boys’ cash would have ended up as idle balances in those other banks, because the guys with good record would not want to increase their obligations, and, on the other hand, the banks won’t lend, anymore, to sub-prime guys).
Now, following your narrative, Harry stopped paying, and Bonnie found out that the value of the house dropped to $300,000. How did she find out about the value? Most probably, someone with cash made her an offer.
However, looking back at the financial account we made for time one, what transpired in time two was that Bonnie’s financial asset of $360,000 was obliterated and, instead, she (plus her own creditors) owned now a house for which people with money offered 300,000.
In other words, nothing ‘bad’ happened to the money (600,000) in those periods (if anything, its value in term of ‘stuff’ went up!). What has changed was the volume of the claims on money. In your economy, all such claims were annihilated, because Harry’s income could not fulfill his obligation. The vital mechanism of transferring cash against promises, without which no economic action can take place, collapsed.
Felix knows this stuff a lot better than I do. My first-order response is not that sophisticated: I think that the models (whatever the math was behind them) were based on data from one state of the world – a state in which housing prices did not fall across the board, and when they did it was only in specific areas, so if you were geographically diversified you were OK. These models were then applied to a new state of the world, in which housing prices behaved differently.
My second thought is that people’s choice of models was affected by their incentives – either to rate the securities or to sell them. I don’t know how many end investors – the people with the incentive to be skeptical – were really re-analyzing all of the underlying mortgages.
Shivz,
In your analogy, I noticed that no one ever works for a living. Everyone just seems to have this “manufactured” money in their pocket. Where is this place where money just “happens”? It sure seemed like that house just SHOT up in value in your story. How can that happen….I dont get it. I’m just a novice in all this fancy-schmancy finance stuff…..but I betcha that type of economy never has a happy ending….ya know? Seems like a huge, no-holes-barred Ponzi scheme to me, no?
adios amigos,
You should read the original, written by James, wherein the story begins with the production of ‘stuff’, namely, the construction of a house by Danny. The financial asset (360,000) was created on the basis of the buyer’s future income and the market value of the produced ‘stuff’.
Now, the financial asset vanished, not becasue of Ponzi, but because Rusty and Danny reduced their spending by hoarding money. One of the collateral damages to the economy was that the poor Harry lost his job.
Hi James: Brilliant piece. We need follow up. If you drill down to the references in the Harvard working paper you soon discover Ingo Fender at BIS who has been warning about the use of ratings on CDO’s (let alone CDO squared). Apparently no one paid any attention. We should open this up to bright sunlight if we want to prevent another catastrophe in the future. Otherwise, the really smart bankers will figure a way to game the system and beat the regulators in the future. We taxpayers are going to bail out the system but no one will explain to the taxpayers in words they can understand what happened. This is really about much more than a bunch of bad mortgages.
To help shine a bright light, I think it would be helpful to write about 2 items.
1) How much of the toxic assets on bank balance sheets are CDO’s? CDO squared? How many had to be written down because of rating downgrades?
2) How much did the banks rely of CDS to protect them from rating failures?
While you’re at it, try and explain to John Q public why he should allow banks to buy this stuff? You could also talk about credit ratings. I am hoping that somewhere along the line we will find a Richard Feynman (Challenger O ring disaster explained in simple terms by Nobel prize winning physicist who was also renowned as a teacher) who will explain “how” to regulate these markets. Probably by outlawing many of the instruments (naked CDS?) currently being used.