June 11, 2009
How to lose on a sure-fire bet
There was a wonderful story in today's WSJ about how some big banks managed to lose some of their hard-earned TARP money.
Let me begin with a little background. A credit default swap is sometimes described as an insurance contract written against the possibility of default of a particular underlying asset. If I buy a CDS and the specified asset defaults, I get to collect money from whoever sold me the contract. If I also have a long position in the asset in question, I might consider buying a CDS written against that asset as an insurance or hedge against the possibility that the asset loses its value.
But I don't actually have to own the asset in question in order to buy a CDS from somebody else. I might want to buy a CDS as a partial hedge against some other asset I hold with which the specified security could be correlated. Or maybe I just feel like making a bet with somebody I think is dumber than I am.
The fun and games begin when multiple contracts get written on a single credit event and the notional value of outstanding contracts on that event-- the total amount of money that is promised to be paid to the buyers of those CDS in the event of a default on the underlying asset-- becomes larger than the par value of the underlying asset itself. Then it would clearly pay the party who sold those contracts to buy the underlying asset itself at par, relieve the original debtors of their burdensome obligations, and be out only $X (the underlying event) rather than some multiple of $X (all the contracts written on the event).
And so the WSJ recounts the tale of a security based on $29 million (par) worth of subprime loans in California, half of which were already delinquent or in default. Betting that the loans weren't worth $29 million sounds like easy money, and the smart guys were willing to pay 80 to 90 cents for each dollar of CDS insurance.
It appears from the WSJ account as if little Amherst Holdings of Austin, Texas was happy to sell the big guys like J.P. Morgan Chase, Royal Bank of Scotland, and Bank of America something like $130 million notional CDS on a $27 million credit event, used the proceeds to buy off and make good the underlying subprime loans, and pocketed $70 million or so for their troubles. The big guys, on the other hand, paid perhaps a hundred million and got back zip.
Said big guys, naturally, are screaming bloody murder, trying to bring in the lawyers to show that Amherst wasn't playing by the rules of the game.
For my money, the first rule we need would be a law, not a rule, that notional not exceed actual.
Barring that, here's another rule I trust: a fool and his money are soon parted.
Posted by James Hamilton at June 11, 2009 04:08 PMdigg this | reddit
Horse race bookies are a lot smarter. The way parimutuel odds are calculated they never lose money. More bets on one horse lower the payout, so the equivalent of betting on every horse wouldn't work.
There have been cases with the multiple state mega lotteries where the payout exceeds the cost of buying every possible number combination. The catch is you can't buy that many tickets and it also assumes that there is only one winner.
Posted by: robertdfeinman at June 11, 2009 04:30 PM
Hard earned TARP. Nice. I wasn't aware that Barclays was a recipient. JPM took it at gunpoint and is repaying. BAC, per the article, lost very little. But why not make this about TARP because we're also stupid enough to think legislating private contracts is a good idea.
On the other hand, cases like this might actually send a signal to banks that their sure things might not be so sure and they should think twice. But you miss that entirely and advocate a law that protects their sure things.
Posted by: Kevin at June 11, 2009 06:06 PM
its tough to legislate stupid
Posted by: JS at June 11, 2009 06:52 PM
Professor, the only thing bad I see about this is that Amherst did not take Golden Schmucks, also, to the cleaners on this.
Amherst -- next time, take money from Government Sachs, too, please.
And, good job, gents.
Posted by: jg at June 11, 2009 07:18 PM
This story ought to be framed. It exemplifies all that is wrong about the entire derivative, CDS, off balance sheet investing which occurred.
On another note nice to see Amherst beat out those 'smart' big guys.
Posted by: Colin Henderson at June 11, 2009 07:55 PM
1) You don't spend $9 to gain $1 on a put. That is the same as buying $1000 strike puts on AAPL as a way to short it.
2) You especially don't do that with easily manipulable subprime securities, ones which are overinsured.
3) Is this why i-bankers need to be paid in excess of $500k/yr, otherwise they will find employment in places where they will be appreciated? This is the 'brain drain' we are worried about ??
Hmmm.. in general, a brilliant plan: Oversell insurance on terms mentioned in #1, then make the insurance worthless at a lower cost than the payout. Same thing as me selling puts on a company I am about to buy. Better yet, maybe I can make more on the premium I earn than I have to pay for the underlying company. What a plan!! Free money!!! The guys at the banks thought there was free money in shorting subprime to zero, so they took advantage of 'free money'.
Posted by: Michael at June 11, 2009 08:11 PM
My high school wrestling coach had a piece of advice for what to do when your opponent makes a monstrously stupid mistake--NEVER give a sucker an even break. All the more satisfying when you're a 98lb weakling and your opponent is an 800lb gorilla.
My current boxing coach likes to say that, "A negative consequence is the best teacher." Hopefully the big banks will learn their lesson.
Posted by: Kevin Dick at June 11, 2009 08:13 PM
Perhaps Treasury is trying to keep the market high so as to avoid massive pay-outs from AIG?
Posted by: Mattyoung at June 11, 2009 08:14 PM
Amhearse Holdings cremated 'em.
Posted by: Paul Jerome at June 11, 2009 08:22 PM
I'm afraid you have your facts wrong. Amherst didn't "buy off and make good the underlying subprime loans." It didn't have the right to do that. The servicer, Aurora Loan Services, was the one that had the right -- called a "cleanup call" -- to purchase the outstanding mortgage loans. Amherst somehow got Aurora to exercise its cleanup calls, almost certainly at a loss, and that's what rendered the banks' CDS worthless. How did Amherst get Aurora to exercise its cleanup calls? We don't know -- Amherst refused to tell the WSJ. But the banks are definitely suspicious (as they should be).
Posted by: Economics of Contempt at June 11, 2009 09:47 PM
OPM (Other people's money) - Sigh. these guys are quite smart. They go off and lose your money and my money and our kids money... Did I hear a giant flushing sound?
Posted by: Chris J at June 11, 2009 11:46 PM
My god poster Feinman is an idiot. JPM took TARP at gunpoint? BULLSHIT
Posted by: 1 currency at June 12, 2009 12:23 AM
Interesting story. It's related, I think, to some of the discussion about short sales on stocks. Eg, with naked shorts, you could easily have the same thing, where the number of bets exceeds to total outstanding shares. A market design issue -- still unresolved as far as I know -- is whether this is a good or bad thing. My colleagues Brenner and Subrahmanyam have a view of this, but it's not likely to be the last word. See
Posted by: Dave Backus at June 12, 2009 12:35 AM
These problems only arise when the CDS is physically settled (i.e. physical exchange of the underlying security for the insured amount).
However, the CDS market is moving increasingly towards cash-settlement, where, rather than physical exchange of the securities, where the post-credit-event price is used to determine the value of the insurance payout.
So this avoids the problem of notional exceeding actual and addresses your 'first rule'. That said, I think most of the CDS-mortgage market is still physically settled, so the fun and games might continue for a while longer.
Posted by: Alan at June 12, 2009 12:52 AM
Bwahaaaaaaaaah! Time to get the money out of Wall street into other hands. Thanks! Best story I heard so far this year.
Posted by: balasr at June 12, 2009 03:32 AM
"For my money, the first rule we need would be a law, not a rule, that notional not exceed actual."
It would be easier to just pass a law stating you need an insurable interest to purchase a CDS. That would automatically limit notional to actual. Then pass a law requiring the seller of the CDS maintain the same sort of actuarial determined reserve that an insurance company would be required to maintain. The whole problem goes away at that point.
Posted by: RueTheDay at June 12, 2009 05:01 AM
THis story makes me happy.
Posted by: KevinM at June 12, 2009 06:00 AM
For Kevin and his defense of Barclays, he is ignoring the money that is laundered through other companies, for example the money that went to AIG and ended up at Barclays.
And no, there was no gun at the head of JPM, you show me any story where Paulson or anyone else pulled a gun. At best JPM and the rest might have been bribed, but there was no force. If you don’t know the difference between force and bribery you don’t know much.
As to “stupid enough to think legislating private contracts is a good idea..”, I don’t think it is a good idea, especially the paying out of trillions in taxpayer money to keep incompetent and crooked bankers in business since they failed to mind their own business properly. Without the trillions in loans, guarantees, buyback of overpriced assets JPM, Barclays, and the rest would be out of business and good riddance to them. And if you bought into them then good riddance to you too.
Posted by: DJF at June 12, 2009 06:06 AM
Occasionally this same tactic is attempted in the Treasury Repo market and the Treasury Futures market. You need not limit the amount of total contracts sold. The Future's exchanges limit the size any one counterparty can sell relative to the amount outstanding.
Posted by: Mike Rulle at June 12, 2009 06:16 AM
Keep the bonuses coming for the Big Banks' "Brightest and Best" ? Wouldn't want to lose all that talent to the competition now would we?
Posted by: David at June 12, 2009 07:02 AM
-- It would be easier to just pass a law stating you need an insurable interest to purchase a CDS. That would automatically limit notional to actual. Then pass a law requiring the seller of the CDS maintain the same sort of actuarial determined reserve that an insurance company would be required to maintain. The whole problem goes away at that point.
This sounds about right. But is there a simpler solution, such as forcing all CDS contracts trade on a central exchange? (See Mike Rulle's post).
-- These problems only arise when the CDS is physically settled (i.e. physical exchange of the underlying security for the insured amount).
I don't understand. Can you elaborate?
Posted by: pat at June 12, 2009 07:03 AM
As I note on my blog, including a mention of the underpants gnomes, the problem here is allowing people without a material interest in the item to buy insurance in it.
This was known to be potentially catastrophic, which is why why, 263 years ago, parliament passed the Marine Insurance Act of 1746, which required that anyone wanting an insurance payout demonstrate an interest in the continued existence of the property.
Posted by: Matthew G. Saroff at June 12, 2009 08:34 AM
-- These problems only arise when the CDS is physically settled (i.e. physical exchange of the underlying security for the insured amount).
It had nothing to do with settlement. The bond was made good. Period. End of story. It could be a company bond that got called. It just happened to be subprime mortgage pool.
Posted by: HZ at June 12, 2009 09:33 AM
Actually, requiring physical settlement may be a very good idea.
Forcing firms to prove an interest in order to purchase CDS could be costly; requiring them to produce that interest in the form of the underlying asset only when there is actually a default even is significantly less costly.
Theoretically, this might still allow investors to make bets on assets they don't own and then simply buy those assets quickly in the event of a default, but that would still only work so long as there is enough of the underlying asset to justify all the outstanding bets. It might even provide a sort of free partial insurance for the original owners of the assets who might be able to squeeze out a better price from a desperate buyer than they could have gotten through the normal recovery process.
Posted by: Victor Galis at June 12, 2009 09:56 AM
An old saw applies to this event:
You can never make anything fool-proof because fools are too ingenious.
It appears that these traders have been working hard to find new ways to lose money over the past year. It sounds like another bonus is in order.
Posted by: rd at June 12, 2009 02:36 PM
"Said big guys, naturally, are screaming bloody murder, trying to bring in the lawyers to show that Amherst wasn't playing by the rules of the game."
I thought these same big guys went to considerable effort some years back making sure that there weren't any rules to the game; that is, that CDS were unregulated. Am I mis-remembering?
Posted by: Ken at June 12, 2009 03:15 PM
"It had nothing to do with settlement. The bond was made good. Period. End of story. It could be a company bond that got called. It just happened to be subprime mortgage pool."
My point, HZ, was that Professor Hamilton's rule about notional not exceeding actual is not entirely relevant when cash settlement is the sole way of settling CDS transactions. I wasn't saying that the banks were caught with their pants down BECAUSE these CDS were not cash settled.
So I think we basically agree.
Victor Galis says:
"Actually, requiring physical settlement may be a very good idea.
Victor Galis says:
"Forcing firms to prove an interest in order to purchase CDS could be costly; requiring them to produce that interest in the form of the underlying asset only when there is actually a default even is significantly less costly."
Note that these issues are one of the main reasons why cash settlement arose in the first place. People (i.e. traders) wanted to buy into naked puts, without having to rush out and buy the bond in the event of a credit event (I think the argument was the potential market disruption these short-covering purchases would cause).
Now, it remains an open question about whether cash settlement is really such a good thing, especially if it distorts incentives by allowing traders of such CDS to not have to hold (or even consider holding) the underlying.
Posted by: Alan at June 12, 2009 05:30 PM
The bank guys are not that stupid as you think, Professor.
Please read between the line. The reason why they paid CDS premium of 80 or 90cents for 1$ is because they bought the underlying bonds around 20% or 10% of the par value.
If you buy the underlying at 10% of face value and pay the 90% CDS premium, then you make nothing but also lose nothing. You lose your investment of 10% and CDS premium of 90% but you get compensated 100% from CDS seller. Of course you must hold the bond.
Now read the article again carefully.
"At one point, at least $130 million of bets had been made on the performance of around $27 million in securities, according to a person familiar with the matter."
It is saying bets were made "at one point" I think it is the transaction volume not the outstanding amount.
"Since the mortgage securities were valued at just $3 million or so in the market, well below the $27 million they were redeemed for, traders believe Amherst entered into an uneconomic transaction to profit from its swap positions."
You see? bonds were trading around 11% of par value. I don't think banks bet 9$ to make 1$. They just tried to make small arbitrage.
"J.P. Morgan lost millions, while RBS and BofA suffered minimal losses, said people familiar with the matter."
One last point.
Now they are working on new scheme to move all the plain vanilla CDS to central exchange like CME. Your idea of limiting notional not exceeding actual will kill whole this idea and kill all derivative trading.
Posted by: Alphahunter at June 12, 2009 11:55 PM
"Your idea of limiting notional not exceeding actual will kill whole this idea and kill all derivative trading."
Excellent. What kind of payoffs would it take to put this plan into place?
Posted by: Crapitalist at June 13, 2009 06:20 PM
Did I get this straight... the actual owners of the underlying MBS's got paid off in full on the bonds and the banks are crying about how they didn't make money on a bet that the owners would get hit with defaults? What a bunch of babies!
Posted by: anonymos at June 14, 2009 10:16 AM
The outrage over this whole mess is clear. What isn't clear is who has profited from it. This makes it a little more clear. Why have there been less arrests then there are fingers on my hand? Let's clear our jails of mino drug offenders and put the real criminals in there: Gitmo "terrorists" (really just war criminals) and Wall Street Terrorists.
Posted by: Ryan at June 15, 2009 02:02 PM
Bank guys maybe not AS stupid as portrayed, AlphaHunter, but still plenty stupid. As there were only 27mm in bonds outstanding there was a pretty large A/L gap at the point where there is 130mm of notional CDS written. Owning the underlying is a petty hedge (not technically arbitrage) against owning the CDS due to its tiny size: it seems obvious the big guys thought they were the ones betting on a sure thing. Also, for all anyone knows the notional CDS could have been growing. Perhaps Amherst should have waited to have the bonds called at par until they had written 1 billion in CDS!
Also, not sure what tools these banks have to check outstanding notional value of CDS, but if they new the notional exceeded the underlying on such a small reference security they really should have considered the possibility that the CDS writer was effecting a bit of horse trading.
Posted by: Reg LeCrisp at June 16, 2009 06:00 AM
From Bloomberg on June 17th:
U.K. unemployment rose less than economists forecast in May and the Bank of England said the risk of a deeper slump has receded, the latest evidence to indicate the economy is past the worst.
Claims for jobless benefits rose 39,300 to 1.54 million in May, the smallest increase since July last year, the Office for National Statistics said today. Economists predicted a gain of 60,000, the median of 28 forecasts in a Bloomberg News survey shows. Separately, the Bank of England said in London “the risk of a continued sharp contraction” has “receded somewhat.”
Signs are mounting that the U.K.’s economy is recovering from the worst recession since 1979, with manufacturing activity rising for a second month in April and services activity expanded for the first time in a year last month
Posted by: Ivars at June 17, 2009 03:20 AM
Scratch that, these guys are as stupid as he thinks. There were no bonds to buy as hedges because Amherst owned them all!
Posted by: Reg LeCrisp at June 17, 2009 05:30 AM