Yesterday I had the pleasure of attending a very good talk by Yale’s Gary Gorton on the origins of the financial crisis.
Gorton’s story is that this was a bank run, not substantially different from the bank runs that have always plagued capitalist economies. In this case, the run took place in the repo market, which is an unregulated (and largely unmonitored) industry roughly equal in size to the standard banking industry. The repo market serves large institutions (e.g. Fidelity Investments or state governments) with a lot of cash on hand that they want to stash in an interest-bearing account for a day or two. So Fidelity deposits, say, a half-billion dollars at, say, Bear Stearns, just as you might deposit five hundred dollars at your local bank. One difference, though, is that your account at your local bank is insured, whereas Fidelity’s account at Bear Stearns is not — so Fidelity, unlike you, demands collateral for its deposit. Bear Stearns complies by handing over a half-billion dollars worth of bonds, of which Fidelity takes physical possession. The next morning, Fidelity withdraws its money and returns the bonds.
The problem comes in when rumors begin to spread that some bonds might be riskier than they appear, and Fidelity starts to worry that maybe Bear Stearns is picking particularly risky bonds to hand over. Therefore Fidelity demands more than a half-billion in bonds to guarantee its half-billion dollar deposit. If there’s, say, a 10% discrepancy between the deposit and the collateral, we say that Bear Stearns has taken a half-billion dollar haircut.
Because Bear Stearns has a fixed quantity of bonds on hand, and because all of its depositors are demanding haircuts, Bear Stearns can now accept fewer deposits than before. This means that Bear Stearns has less cash on hand. This makes depositors even more worried about the security of their deposits, which means they demand larger haircuts. The effects snowball until Bear Stearns collapses. Like so:
The last time I blogged on this subject, our astute commenter Bennett Haselton raised a very good question: Bear Stearns is presumably not just sitting on Fidelity’s money; they’re investing it somewhere. Why can’t that investment serve as Fidelity’s collateral? The answer, if I understand Gorton correctly, is that the repo market is a very short-term market, typically 24 hours. For Fidelity to verify the quality of Bear Stearn’s investment project would take a week or so, by which time it’s too late for the information to be of any use. Fidelity’s ongoing concern is that Bear Stearns is pawning off its shakiest investments; to allay that concern requires due diligence; due diligence takes time; the repo market is all about getting things done NOW.
So what should we do about all this? Gorton, along with his colleague Andrew Metrick, argues that the repo market, like any banking market, is inherently susceptible to runs and therefore ought to be regulated. In this case, the regulations should focus on insuring the availability of sufficient high-quality collateral to keep depositors calm. Gorton observes that the existing policy responses to the crisis (e.g. the Dodd-Frank bill) do pretty much nothing to address this fundamental need. The Gordon/Metrick paper contains some specific proposals, which unfortunately Gorton never got to in yesterday’s talk.
For further reading, here are the same references I gave the last time I blogged on this: There is much more here and in Gorton’s book.
Since you wrote, “The problem comes in when rumors begin to spread that some bonds might be riskier than they appear, and Fidelity starts to worry that maybe Bear Stearns is picking particularly risky bonds to hand over” — could the whole problem be avoided by simply having Fidelity, rather than Bear Sterns, pick *which* of Bear Sterns’s bonds to hand over?
Or as an extension of that, you mentioned that the reason Fidelity can’t simply use Bear Sterns’s investment as collateral is because then Bear Sterns might make an unwise risky investment and pawn it off as collateral on Fidelity. But then what if Fidelity had the right to choose which Bear Sterns investment was the collateral — not necessarily the one that was made with their money?
Bear Sterns would have to continue making investments (and holding bonds) that they believe to be good ones, on average, to make a profit. Then if Fidelity were free to choose which Bear Sterns investments or bonds to use as collateral, they would have more reason to believe the collateral was worth that much.
Market participants are drawn to fragile financing because the profit manifests consistently but the risk only manifests intermittently. It’s a mix of self-deception and fraud. Contrary to Gorton, I don’t see how this is “essential to job creation”. It’s not essential to anything. It’s a mistake.
“The problem comes in when rumors begin to spread that some bonds might be riskier than they appear”
Don’t the problems actually begin when the bonds are priced as less risky than they actually are? The revelation that they were mispriced to begin with may be the immediate visible unraveling but this unravelling would not happen without the original artificial boom. Nothing to do with rumors.
This whole “it’s all about confidence not fundamentals” tract is not only wrong but it also leads to mass support of govt policies that focus on boosting short term confidence at the expense of long term fundamentals.
I’m not sure I follow the snowballing logic. I get that depositers will demand a larger haircut on risky collateral, but why would they demand an even larger haircut if Bear Stearns itself looks shaky? Are we assuming the collateral bonds are Bear Stearns bonds or are otherwise tied to Bear Stearns’ fate?
1. Doesn’t the logic actually suggest that rumours should be regulated?
2. Regulation will presumably involve things like minimum fractions of AAA debt. Basel II classifed sovereign debt as inherently AAA I believe. Successful?
Steve,
I believe the explanation for why Bear Sterns couldn’t use the investments as collateral is wrong. I believe they were using the investments as collateral. The chart above is for haircuts on “Structured Debt”. Most people in finance use the term “Structured” to mean securitizations. Securitizations (CDOs, CMOs, etc.) are exactly what caused most of the problems. The value of these securitites became suspect. Nobody knew the value since they were filled with complicated garbage. People lost faith in these securities and also just plain didn’t want to hold the Bear Sterns name anymore.
Bear Sterns used the proceeds of repos, commercial paper, etc to finance whatever it held on its balance sheet. Some of what was on balance sheet was Structured Debt, the securitizations. The repos and commercial paper were continuosly rolled, meaning the debt came due and they borrowed again. During the crisis, repo and commercial paper markets locked up, meaning the investors withdrew. Thus, Bear Sterns had long term investments that couldn’t be liquidated fast enough or at a high enough price to pay the maturing debt.
Matt
I agree with Matthew. Bear Stearns owned the structured products, for better or for worse. The money they borrowed was to finance what they already had, not to fund new projects (its like borrowing money to pay for a car you own – you can’t use that money to go buy something else, as you’ve already paid the car dealer with it!). BS could have sold the structured products, but they probably didn’t want to be dumping large quantities of those into the market in that environment.
Bennett:
Then if Fidelity were free to choose which Bear Sterns investments or bonds to use as collateral, they would have more reason to believe the collateral was worth that much.
Choosing freely is better than letting Bear Stearns choose, but it’s not much better than choosing randomly, since there’s no time to do any due diligence.
Ken B: Correct, the AAA rating proved to be completely unreliable in the crisis. Any regulation cannot be based on these assessments unless they are made much more rigorous. The fact that the ratings agencies are paid by the people they are rating introduces a fundamental bias.
Dave: also correct – it was not rumours, it was fact – the stuff was riskier than assumed.
I am sure the repo market had an amplifying effect, but it was surely not the cause of the problem. The basic cause was the amplification of risk through “slicing and dicing” of securitisation. I think a big problem arose when the lowest risk was assumed to be no risk. I believe JP Morgan insured this “low risk” tranche with AIG, and thereby did away with most of their profits. Other banks did not do this, so took higher profits in the short term, but suffered more long term.
On an even more basic level, the whole system seemed to forget the principle of incentives. It assumed the risk of default was a constant, or nearly so, and used this in the “stress tests” for the banks. They forgot that once you separated profit for the lender (through CDO’s etc.) from the need to ensure repayment, you were building in a massive increase in risk of default. Blindingly obvious in hindsight. The complexity made it impossible to know where all this risk had gone, so confidence collapsed.
@Steve & Bennett:
Aside form Steve’s point, here is where leverage comes in. If Bear has most of its bonds are serving as collateral then even with due diligence you have the same contraction.
Okay, maybe I do not understand this, but if Bear Stearns finds that it owns risky bonds, this means its assets are worth less than they were before, and it should be scaling back operations commensurately, right? Gorton’s story says that it tried to keep running business as usual, borrowing the same amount of money every day, which would obviously have led to problems. The eventual result sounds totally appropriate. Isn’t this kind of like what happens if your income suddenly drops and you try to keep charging the same amount of money to your credit cards every month?
Paul G,
Not really. It’s more like you financed your car with your credit card. For years, when the credit card came due at the end of the month, you transfered the balance to another credit card lender and kept things going. Now, all the credit card lenders realized they want out of the business. Thus, you now have nowhere to refinance and the credit card bill is due. Also, the value of the car has declined. Even if you sold your car, you can’t cover the credit card bill.
Bear Stearns couldn’t just scale back (let the balance sheet run off). There was no time. Yesterday’s debt was due now. You either borrow the same amt (do a refi) and pay yesterday’s debt or raise equity and pay off a fraction while borrowing the rest if you can.
Matt
‘Bear Stearns complies by handing over a half-billion dollars worth of bonds, of which Fidelity takes physical possession.’
I find that very hard to believe. Are you sure don’t mean legal possession. Otherwise you have a logistical nightmare with bonds being transported back and forth on the streets of NYC all the time.
Also;
‘The problem comes in when rumors begin to spread that some bonds might be riskier than they appear, and Fidelity starts to worry that maybe Bear Stearns is picking particularly risky bonds to hand over.’
Well, so…? If Fidelity can return the bonds and get their cash back what does it matter how risky the bonds are?
Patrick R Sullivan:
find that very hard to believe. Are you sure don’t mean legal possession. Otherwise you have a logistical nightmare with bonds being transported back and forth on the streets of NYC all the time.
The bonds, as I understand it, are physically moved from a vault belonging to Bear Stearns to a vault belonging to Fidelity, both located at the same central clearinghouse.
Well, so…? If Fidelity can return the bonds and get their cash back what does it matter how risky the bonds are?
The concern is that Bear Stearns will vanish overnight.
WHy isn;t Fidelity big enough to take care for themselves and write a contract that says when it lends money to Bear Stearns, Bear Stearns will not accept money that is not backed by some specified quality of asset? This contract can be in force with no end until one of the parties wants out of it. Why would Government regulators be expected to do a better job?
@Harold: My point was that the Basel II regulatory rules DEFINED sovereign debt as the equivalent of AAA for regulatory purposes. So the fault is not with Moody’s. Moody’s did better than the regulators here.
The custodian or DTC makes book-entry changes to ownership of the securities. This is particularly true for large institutional trades like the one described in the blog post above. Physical securities are not moving, not even within the same building.
ADDENDUM:
IMO the problem is not that Bear Stearns fails but that the monetary system has feedback problems so that the failure of Bear Stearns is contractionary with no countervailing expansionary force. Their failure in the current system leads to weakness in more financial institutions.
I think that if we had free banking with competitive currencies backed in nothing but the issuers assets we would not have a problem. If a currency starts to fall below par people try to spend it or trade it fr a stronger currency. All currencies cannot fall of par with one another at the same time. The issuers with the stronger currencies are encouraged by profit motive to issue more as other currencies fall.
We don’t have free banking, but my bank offers free checking. That’s pretty good, I think.
@Steve –
“Choosing freely is better than letting Bear Stearns choose, but it’s not much better than choosing randomly, since there’s no time to do any due diligence.”
Right, but presumably Bear Sterns would do their own due diligence on their investments, since they don’t know which ones they’ll have to rely on to make money, and which ones they’ll be giving out to others as collateral. So even if you’re choosing randomly, you can expect the investments or bonds to have more value, than if you’re forced to accept the investments or bonds that Bear Sterns gives you (since in that case they’ll presumably give you the ones that they think are overvalued).
Of course Ken’s right that if *all* of Bear Sterns’s bonds are being given out as collateral, then it doesn’t matter, since the crummy ones end up being given out to someone.
Interesting read. (I didn’t realize the reason there were panics in the “free banking” era was that people questioned the collateral value of STATE bonds.) Certainly the real cause of the problem WITHIN the housing/subprime/structured products markets was that home prices were inflated, and that pain was bound to be absorbed by whoever held those assets. We also don’t want to let the ratings agencies off too easily, they played a major role. But Gorton’s point is why this affected other seemingly unrelated markets, and he does present some compelling data. One quibble with his historical background though, just something to add to the annals of unintended consequences: He says “at root this change has to do with the traditional banking system becoming unprofitable in the 1980s [when] traditional banks lost market share to money market mutual funds”. He mentions the lifting of Reg Q as adding to the problem (in an inflationary environment), but my understanding is Reg Q was largely responsible for why MMMFs were created and flourished in the first place (presumably a clumsy, and unsuccessful, attempt to quell inflation).
Harold – “a big problem arose when the lowest risk was assumed to be no risk” – of course this is right.
However re: “the basic cause was the amplification of risk through “slicing and dicing” of securitization” – slicing up the risks can’t amplify the overall risk, just re-distribute it in a concentrated form to some and a diluted form to others.
@ Bennett, SL etc.: I believe repo is (supposed to be) done on specific collateral, so I’m not seeing the whole aspect of ‘who chooses’ as germane – certainly one can argue that no one knew the true value of underlying housing bonds, but I don’t think it’s like BS had a free cheapest-to-deliver option or could substitute some other investment.
Iceman: Whilst you may be technically correct, the slicing and dicing can only redistribute risk, this only applies if your assumptions are exact. Here is an example http://marginalrevolution.com/marginalrevolution/2010/05/the-dark-magic-of-structured-finance.html
This shows how creating a CDO from mortgages with a 0.05 probability of default can create tranches of AAA rated CDO’s with a default probability of 0.0005. However, if you had got your initial estimate of default risk off a bit, and it was actually 0.06, this means a 20% greater default risk – not disasterous perhaps. But in our sliced and diced securities, the CDO now has a default risk of 0.247 – or an increase of 45,000%. If that isn’t an amplification of risk, I don’t know what is.
Ken B: Is your point that the Basel II rules created a higher demand for AAA rated securities and therefore drove the creation of these CDO’s?
@Harold: No. I cited an example of a regulation that was faintly crazy but served the purposes of some, and seemed quite sane to most. A not uncommon thing. When people call for more regulation it sometimes is worth looking to see how well regulation has worked in the past. People often talk as if the appropriate regulation were obvious and easy; it is rarely either and can create perverse incentives. As happened in this case.
Harold – thanks for the link, but if I’m following it correctly (it’s kinda sloppily written and the link to the calcs doesn’t work), Tabarrok is still talking about the disproprtionate impact on 1/10th of 1/10th of the original mortgages (i.e. tranche 10 of the CDO, although at times he seems to refer to this as “the CDO”). I think this is what Gorton means by “not large enough to cause a systemic financial crisis by itself”.
I’m confused. Bear Stearns takes Fidelitys deposit for what purpose? If they can’t do anything with the cash, why would they issue bonds at a haircut for the cash? That funding would need to be worth more than the value of the bonds issued. No?
Steve,
> If there’s, say, a 10% discrepancy between the deposit and the collateral, we say that Bear Stearns has taken a half-billion dollar haircut.
I don’t understand. Do you mean ‘has taken a 50 million dollar haircut’?
Iceman: Yes, I think it is impossible to say “This was the cause”. There are many chickens and eggs.
You reflect a fundamental misunderstanding of what Dodd-Frank is supposed to do. The idea is to force financial institutions to re-separate low-return/low-risk business from high-return/high-risk business. If I take out a mortgage, or put money into savings, I want that interaction to be low-risk. Without Dodd-Frank, institutions can take my savings or mortgage money and put it into higher risk instruments. They make higher returns (on average) but do so by gambling with my money.
If a high risk institution wanted to offer consumer instruments and people chose to engage in them knowing they were high risk, and the people who made that choice had a chance to share in the (potential) rewards then I (and as I read it, Dodd-Frank) would have no objection.
The present situation is one in which the consumer isn’t given a choice, is exposed to high risk will he or no, and has no opportunity to participate in the rewards of that risk-taking. Oh, and the bank gets the Fed as backstop to ensure it won’t suffer if its bets go bad, either.
The situation is wholly crocked, and Dodd-Frank is not going to solve it, but it should re-establish the separation between low-risk and high-risk financial transactions.
Interestingly enough, in old Scotland (beginning of 19th century) the unregulated banks actually weren’t susceptible to bank runs (because the banking business as is is not ‘ecologically rational’ (to borrow a phrase from Vernon Smith), they made some provisions to do it so… And if I was reading history correctly, vast majority of bank runs (let’s say except those in the 30’s) were perfectly reasonable (i.e. the bank was really insolvent). And if I read history of the USA correctly, the bank runs were NOT pervasive; there were some crises, and even during these the bank runs were actually not like “everywhere”…
I mean describing this as “being plagued by bank runs” seems to me a little strech.
Harold – to be sure…but I think you still want a theory as to how the slicing in particular turns a ‘non-disastrous’ rise in the overall default rate into a systemic crisis. Not saying there isn’t a good and interesting one. For example creating huge amounts of synthetic exposures relative to actual underlying loans could magnify the tranche 10 effects. (Still Gorton is trying to explain how this transmitted to seemingly unrelated markets.)
Richard – yeah I saw that and think you’re right the ‘haircut’ is $50 million in the example.