Upon Closer Inspection
On Monday November 17, 2008, I posted a link to an article by Arnold Kling entitled, “Why Credit Default Swaps,” which forms the impetus for the hackneyed title of this article. At the time, I dubbed it a “good article,” but expressed my reservations about its conclusions. For whatever reason, I read the article for a second time earlier today. Upon closer inspection, with all due respect to Mr. Kling, my reservations have grown into outright rejection of both his facts and his conclusions.
You Forgot To Add The Basel
Kling begins his analysis with the conclusory statement that credit default swaps have no “natural seller.” That sounds brilliant. But what does it mean? Well, my personal opinion is that it means absolutely nothing. Kling seems to think that it means that credit default swaps exist only to evade regulation. You can read more conclusory remarks about the absence of natural sellers in his other article here. Let’s accept for the moment that he is correct. That is, suspend disbelief, and accept that the CDS market exists for the sole purpose of evading regulation. A natural question to ask is, which regulations are these CDS folks trying to evade? Well, Kling boldly states that banks and other unspecified entities “could hold AAA-rated or AA-rated bonds but … are precluded from holding B-rated bonds.” Unfortunately for Mr. Kling, this statement is completely false. But don’t believe me, believe the Bank For International Settlements. After all, they did write the regulations. The full regulatory text, Basel II, is available here. The Basel II risk weighting regulations are here, and I recommend you turn to page 19 of the risk weighting regulations, where they explain exactly how a few classes of B rated bonds are treated under the system.
As is evident, the fulcrum of Kling’s argument, namely the existence of some regulation that credit default swaps were engineered to evade, is a nullity. At this point, his argument falls apart. But we continue onward and analyze the broken pieces to expose more troubling aspects of his argument, namely his economic analysis of how the purported regulatory evasion worked.
Conclusione Confusione
Assume for the moment that banks are precluded from holding B rated paper, despite the fact that reality disagrees. Kling claims that what banks want is the higher returns that a well diversified portfolio of B rated paper offers. And because B rated paper is forbidden fruit in his world, banks use credit default swaps to indirectly obtain B rated paper levels of returns. But how? This is done, he claims, by purchasing protection on B rated bonds, which the regulations allow under his analysis. The protection sellers, he claims, engage in all of the diversification, and although he doesn’t mention it, I presume that the rate of protection these sellers charge is somehow discounted by virtue of this diversification process. Now, as we all know, protection buyers are just that, buyers. That is, protection costs money. In the case of a CDS, the protection buyer pays an amount similar to the risk premium paid on the underlying bond. Thus, the protection buyer gives up some of its returns in exchange for safety. Therefore, buying protection on a bond decreases the returns on a bond. And fully protecting a bond should bring the returns close to the risk free rate, else there would be opportunities for rather obvious arbitrage (though there should be some room for counterparty and liquidity risk). Kling acknowledges this, by saying that the protection seller gets to keep most of the additional returns.
While it is possible that there is some wiggle room to profit from protecting a B rated bond to the point that it’s treated as an A rated bond, there shouldn’t be much, since it’s a rather obvious scheme and arbitrageurs will eventually pick up on it and cause prices to stabilize to arbitrage-free levels. Despite this, Kling places the weight of the entire credit default swap empire on the tiny spread between the price of protection and risk premiums on bonds, which as we discussed, should not be a long term phenomenon.
Finally, as mentioned above, in Kling’s regulatory world, banks are allowed to purchase A rated paper. So why would such an elaborate ruse be necessary when banks could simply invest in an A rated tranche of a CDO comprised of a diversified portfolio of B rated bonds? It would not be. So even in Kling’s hypothetical world, his argument completely fails to explain the existence of credit default swaps.
So, Why Credit Default Swaps?
I will write an article on this topic, soon. It is my personal opinion that the CDS market has transformed credit risk into a product that can be traded like a commodity future. But the short answer is liquidity. That is, corporate bond markets are not very liquid markets. And the market for loans is even less liquid. The CDS market by contrast is a very liquid market with highly standardized agreements that allow credit risk to be sliced up and allocated in ways that are prohibitively expensive if not impossible with ordinary debt instruments.
First of all, I think one of the things Kling left out of his argument was the effect of Too Big to Fail government guarantees on the pricing of CDS contracts sold by many big banks.
But more importantly I need to take issue with this claim: “The CDS market by contrast is a very liquid market”
Isn’t it more accurate to state that the CDS market is a less illiquid market than the bond market? Markets that are traditionally considered very liquid are the Treasury bond market and the stock market. In these markets bid ask spreads are often negligible and volumes are measured in the millions.
I don’t understand how a market with as few trades as the CDS market can be considered very liquid: According to the DTCC the total number of single name contracts outstanding is 2 million and the average number of contracts outstanding per name is about 2000. A brief browse through the data found just four names with more than 10,000 contracts outstanding. Even for the indexes only 2 CDX and 2 ITRAXX categories have more than 10,000 contracts outstanding. There is just no way markets this thin could ever be consider liquid by normal financial market metrics.
The corporate bond and loan markets are even less liquid than the CDS market. Loans don’t even get “traded.” They’re syndicated out, which is a complex process fraught with transaction costs. The Treasury markets are an exception, and note I was talking about the corporate bond market, not the sovereign debt market.
Are you looking at gross or net figures from the DTCC? Please provide a link.
I got the information here: http://www.dtcc.com/products/derivserv/data/index.php. Click on “This week’s select data on all live positions in the Trade Information Warehouse”, then click “I agree” and use the pull down menu to get Table 6 for single name CDS and Table 7 for index CDS. Since I was looking at the number of contracts outstanding, I didn’t think that there was any need to look at the gross vs. net number of contracts. If you have reason to believe that the DTCC is netting some contracts before providing us with the number outstanding, please let me know/give me a link.
Nobody is likely to disagree with the point that the corporate CDS market is less illiquid than the corporate bond market. My point is simply that it is faulty logic to use the fact that the market is less illiquid than a notoriously illiquid market to support the claim that the CDS market “very liquid”. It’s a bit like saying that because we know that 2 is greater than 1, we can also conclude that 2 is equivalent to 10.
ACC,
Bid/Ask spreads in the CDS market used to be pretty tight, before all of this regulatory uncertainty. Check out the charts in:
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=891263
My understanding of the Basel arbitrage is a little different than what you’ve explained (though this is all recollection, so feel free to correct me). Under Basel, banks have capital charges against risk assets. A B bond would require capital held against it. By purchasing protection on the purchased bond, the bond is no longer deemed a risk-asset (assuming a certain rating of the protection seller). If the bond spread is greater than the spread paid for protection, under Basel the bank has achieved “risk-less” earnings that requires no capital offset.
And actually, I’d heard that some European banks were booking the PV of the entire earnings stream upfront on this arb, with no offsetting capital charges. Can’t verify that though.
I don’t mean to beat a dead horse, but the paper (thank you!) shows that bid ask spreads never fell below 5 bps and that at the time of the 2005 GM/Ford downgrades, they rose as high as 20 bps. In a very liquid market like the Treasury market bid ask spreads are typically less than 1/32 of one bp (see here: http://www.newyorkfed.org/research/epr/03v09n3/0309flem/0309flem.html).
In 2005 when the bid ask spread rose to 20 bps, it was larger than 30% of the average of the bid and offer prices. Even when the bid ask spread was 5 bps, it almost never fell below 15% of the average of the bid and offer prices. I just don’t understand how a market with a bid ask spread that is typically more than 15% of the price, can be considered a liquid market.
On page 7 the authors of your article acknowledge that the CDS market is illiquid compared to other financial markets: “Compared to other established markets, the CDS market is relatively illiquid. The bid-ask spread is high (at 23% on average) with a sizable fixed component.”
No one is saying that the CDS market is as liquid as the Treasuries or the equities markets. It doesn’t need to be. The point is that the bid/ask levels are a reasonable indication of price.
Page 19 of Basel II isn’t the point. Under Basel II, the capital charge is a function of credit quality. Under basic or advanced (IRB), the higher the rating, the lower the capital charge (the lower the risk-weighted assets). A CDS is a technically credit mitigation (CRM) which allows the bank to lower the charge as a function of the protection. It’s not about “evading regulation,” rather it’s sometimes about regulatory arbitrage or, more harmlessly, simply about using a CDS to hedge the credit risk and, thusly and appropriately, reduce the capital charge. In short, CDS can have a regulatory motive consistent with their basic job to hedge pure credit risk.
As to “why credit default swaps?,” yes you are correct that allow for trading credit risk. We don’t need to reinvent theory. The risk premium on a single-issue corporate bond includes a cost for credit, funding, interest rate, and perhaps currency, risk (four risks above the riskless rate). An asset swap is nearer to credit risk, but also includes a cost for funding (a type of liquidity if you like). A CDS is pure, tradable credit risk.
David Harper, bionicturtle.com
David,
As to page 19, yes it is on point. Kling claimed that banks are forbidden from holding B rated notes. Page 19 says otherwise.
As for regulatory arbitrage and CDSs, so what? I never said that doesn’t happen. I said that it could happen, but it’s too obvious of an arbitrage play and therefore can’t possibly explain the existence of a trillion dollar market.
As for your last point, a CDS is not “pure tradeable credit risk.” CDS prices reflect liquidity risk currency risk and a host of other factors. My point was that it’s probably the best we can do and certainly a better way to trade credit risk than plain old bonds.
Somewhat unrelated question: why is it that the corporate bond market is still so illiquid? Why is there better price discovery in the single-name corporate CDS market than in the corporate bond market? Did the introduction of TRACE have any impact on the single-name CDS market? Should it have?
There are a number of questions about CDSs being asked:
What are they supposed to be good for?
Mimicking Bonds: Insurance On Mortgages And Bonds: Hedging Certain Investments
These seem to me to be perfectly possible.
Do they work?
They do seem to work most of the time, but better numbers would be nice.
Are they very complicated?
This was my main area of interest, because I believe that they are in fact understandable, at least as to their risk, and that a lot of our problems stem from fraud, negligence, fiduciary mismanagement, and collusion, by the people marketing them.
It seems clear that their popularity had to do with investors wanting to get around capital requirements, by selling insurance that was not subject to the stricter capital requirements of other types of insurance. This seems clearer with CDOSs, where the capital requirements were also an important consideration.
But I simply don’t like blaming the investments because people misused them. I fear that leads to people being exonerated by pleading stupidity when they are guilty of something worse.These investors would have found other means if they didn’t use these.
But, finally, there is a strain of criticism that believes that CDSs and CDOs are like a Ponzi Scheme, incapable of working out at all. Or, that they are parasitic investments used only as a more acceptable form of gambling. I don’t accept these criticisms in general, but I do feel that these investments are definitely not for everyone. Certainly too risky for me. So there’s an overlap of criticism that stems from different concerns.
It seems as if, at least in the English speaking world, Charles and Felix Salmon have been given the task of explaining and defending these investments. I’ve learned so much from both of them, especially Charles, that I’ve followed a topic that never used to interest me at all.
But just as some people want CDSs banned because they suspect they are inherently fraudulent, I want them clearly explained and understood so that we can determine the causes of this crisis that have to do with human beings mismanaging other people’s money.
From my perspective, blaming the instruments is wide of the mark, but that could simply be my bias.
I hope people will indulge my attempt at understanding these issues myself in public. Perhaps other less sophisticated people like myself will benefit from the back and forth. Take care all, Don
erdosfan,
Kling is incorrect about that, you are right, of course Basel doesn’t preclude banks from holding speculative (e.g., B rated) obligors. But, p. 19 is a quibble, what he *meant* is valid, or at least arguable: Basel encouraged banks to use CDS. I agree with you, that doesn’t imply regulation explains CDS market–that goes too far–but he’s certainly correct that Basel plays a role. (Also, regarding A versus B, he probably means: by asset allocation POLICY, funds/banks are internally have constraints on holding junk credits. I know, he didn’t say that. But the point is, from both policy and regulatory [Basel] standpoint, CDS gives advantages. All I mean is, it is false to say: regulation has no role).
The “regulatory arbitrage” is different from a market arbitrage (which over time cannot persist too greatly); it’s not subject to your arbitrage argument, necessarily. The regulatory arbitrage refers to banks’ very real motives to reduce improve their return on regulatory capital. This is an “arbitrage” that can persist indefinitely and expand, unless/until regulatory “close the loopehole.” Regulatory arbitrage is when a bank seeks to maximize its RAROC where capital = economic capital.
Re CDS, yes of course a the price of a CDS reflects technical (e.g., liquidity) in addition to fundamental factors. All financial instruments reflect technical factors, of which liquidity may be the most noticeable (just as the CDS basis will reflect a host of factors) But, the point is, a CDS is *fundamentally* credit protection (i.e., protection on default risk and maybe/maybe not credit deterioration) and the CDS premium is *fundamentally* a premium for the transfer of pure credit risk (unlike a corporate bond, not funding, not interest rate risk).
David
The why of CDS over corporate bonds is very obvious — greed ans stupidity.
CDS were presumed to be nothing more than deep out-of-the-money put options whose premium could be booked as income. Selling trillions in notional, generated tens of billion in income for bonuses, and only have minimal worries about suffering any losses.
CDS also allowed a level of leverage unavailable from corporate bonds. Corporate bonds have interest rate risk that must be hedged with U. S. Treasury securities. CDS have no need for such elaborate hedges.
David,
You said “This [Regulatory Arbitrage] is an “arbitrage” that can persist indefinitely and expand, unless/until regulatory “close the loopehole.””
Nonsense. All transactions have a price. Those banks engaging in regulatory arbitrage transactions will increase the demand for those transactions and cause the prices of those transactions to stabilize at a point where the transactions no longer offer arbitrage profits.
Where ever there are obvious risk free profits, competition forces them out in efficient markets.
“Those banks engaging in regulatory arbitrage transactions will increase the demand for those transactions and cause the prices of those transactions to stabilize at a point where the transactions no longer offer arbitrage profits.”
Standard economic theory assumes scarcity. But with financial innovation it is not at all clear that there is any scarcity of products that facilitate regulatory arbitrage. Instead it seems that the number of available products to achieve this end is increasing over time.
Thus in terms of finance I think the correct model is one of a tiny group of people farming a vast expanse of land. When the population grows, you just cultivate more land — the economy is far from having to worry about its growth being limited.
Dynamic financial markets with constant innovation are very different from the simple commodity markets studied by traditional economic theory.
ACC,
Scarcity still applies. There is no limitless landscape. The scarcity in the case of a derivative is the party willing to accept risk. That is, there are only so many parties out there willing to accept risk. This pushes prices up.
But isn’t one of the goals of financial innovation an increase in the types of investors willing to accept risks. That is, 30 years ago banks carried 90% of all credit risk. Financial innovation was used to diversify the risk-carriers away from banks. Arguable the number of risk-carriers was increasing right up until the middle of 2007. Maybe the crisis is comparable to a culture built on free land that suddenly finds that it has cultivated all of the available land — and that it’s now time for a complete change in culture.
erdosfan,
I agree with Acc. You are applying a fundamental no-arbitrage truth to “regulatory arbitrage” where the “arbitrage,” having a different dynamic, is not subject to quite the same rules. Yes, of course, ceteris peribus, an increase in demand leads to price increase. But regulatory arbitrage, like ratings arbitrage, does not occupy quite the same assumptional vaccum as a classic no-arbitrage.
The regulatory arbitrage, generally, is: the bank achieves a lower regulatory cost of capital (less risk weighted assets) than is economically accurate. For example, if the true cost of a senior tranche is 50 bps, but by virtue of regulatory classification, the bank can lower the cost (arbitrarily!) to 30 bps. Now, initially, it may do that with a “cheap” CDS. Maybe the CDS costs only 10 bps. Demand may push up the cost of the CDS to 15 bsp. To your point, the instrument price increases, but the regulatory arbitrage remains quite profitable. Yes, I understand, that under a strict assumption of only the one instrument and infinite time, there must *eventually* be convergence; but it’s academic and irrelevant: for all practical purposes, Acc’s point is the relevant one, the instruments and methodologies that availed to banks, to achieve regulatory aribtrage (under Basel I, more obviously) were essentially in infinite supply.
Similarly, in regards to ratings “arbitrage,” theoretical convergence never came. As Tavakoli has well explained, the key flaw in the synthetic (arbitrage) CDOs was that the true cost (risk-adjusted) of the senior (and super senior) tranches very much exceeded the small CDS premiums. If there was arbitrage convergence, owing to increased demand, well it didn’t show up much, or it was overshadowed by the gross mis-pricing of risk.
So, my basic point about your post is that is uses Kling as a straw man. We can agree on two ideas:
1. It is true the regulations have at least some impact (are at least a determinant of) CDS.
2. It is true that regulations cannot explain all of CDS growth
So, the relevant question is, how much of the CDS growth can be explained by regulation. This question is very much complicated by the difference between Basel I (which impacts much of the recent situation) and Basel II.
David
It’s me again, the occasional novice commenter who is auditing this class. My fellow novice friends are having a spirited argument about pending impact from derivatives and hedge funds, primarily related to CDO swaps. Our collective ignorance works to sustain the argument. Thank goodness we are friends. There are two camps. One is the you-ain’t-seen nuthin-yet camp who are smugly awaiting the financial Armageddon. The other is the all-the-major hits-from-tanking derivatives-and-hedge-funds-have-already camp.
I am squarely in the second camp, using the most simlistic of logic. First, Bear Sterns tanked in March. So we must be pretty far along in the process of lifting the skirts of troubled financial institutions. Second, if there really are very big and very sick funds and or institutions just waiting to go tits-up, where and how would you hide them?
Is there any consensus about a pending Doomsday scenario among the informed and educated?
Here is my question, maybe argument.
David,
1) “Acc’s point is the relevant one, the instruments and methodologies that availed to banks, to achieve regulatory aribtrage (under Basel I, more obviously) were essentially in infinite supply.”
That is simply ridiculous. THERE ARE TWO SIDES TO EVERY TRADE. There is no infinite unlimited supplier of credit protection.
2) Kling is straw man? Are you joking me? He testified about this is exact subject before Congress. You don’t think it’s important that someone corrects the errors in his statements?
3)Finally, that regulatory arbitrage exists is nothing new. I discuss it in the article, briefly, because it is irrelevant. The point of the article, which appears to be escaping everyone, is that CDSs serve a bona fide economic purpose.
Cat Fish,
Rumors abound. I don’t buy the hype. I’d prefer to wait and see. My own prediction is that the fireworks will come from the “real economy.” That is, labor and corporate earnings, particularly in discretionary spending sectors.
erdosfan,
1) Yes, I agree “there is no infinite unlimited supply.” That’s the academic point. The relevant point is: regulatory arbitrage is not technically arbitrage (riskless profit) and the interesting point is: how/why do near-arbitrages defy convergence and persist. Acc wrote this: “Standard economic theory assumes scarcity. But with financial innovation it is not at all clear that there is any scarcity of products that facilitate regulatory arbitrage.” Now, if you want you can dissect this with a (boring) academic hammer, okay; but, during our long lifetimes before the crunch, in a practical sense, his statement is true enough. This, btw, is what I personally find so interesting about finance: none of the important theoretical ideas (e.g., no arbitrage, CAPM) ever survives the rigorous assumptions in practice. Yet they are valuable!? Yes, two sides to every trade, but that doesn’t help us understand how this happens. Don’t you agree with me that what’s more interesting is that (misplaced?) regulations might create entire persistent markets that arbitrage can’t overcome? Put another way, sometimes what is relevant is, when assumptions are violated and theory doesn’t work.
2) Yes, I do. Agree with your correction. I just meant: you over-rotate to totally dismiss the regulatory motive. That’s all I am saying: Kling, even if he mispoke (e.g., I think he maybe meant policy restrictions rather the regulary restrictions on investment versus junk/speculative credits) has an argument and his argument isn’t dismissed by showing the Basel does allow for speculative credits. The regulatory impact is real but hard to parse. My contribution, ha ha, is to suggest that it doesn’t help to say that regs have 0% role or that they have 100% role. It seems to me in must be somewhere in between.
3) I *totally* agree with you CDS serve an economic purpose. But regulatory arbitrage isn’t irrelevant. Banks have held, and will hold, economic capital in relation to (and, in some cases, as a function of) regulatory capital. Basel I shaped *a lot* of credit derivative activity; unless/until we see evidence otherwise, I think we can expect Basel II will shape this market too.
Thanks, David
David,
As to 1), Final Round. This is not a point of debate. It is not academic. It is a fact. There is a finite supply of credit protection. When the demand for protection increases, WHETHER IT IS DUE TO REGULATION OR SPACE INVADERS, the price increases.
Obvious arbitrage does not “persist” in reasonably efficient markets. Take the Treasury Strips market. Could be opportunities for arbitrage there between the Strips and the Treasuries, but there aren’t because it’s obvious.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=251499
erdosfan,
What I am trying to explain to you is that “arbitrage” is often misused, it is often a misnomer. Yes, true abitrage cannot persist in an efficient market. But very few trades are actually true arbitrage.
Regulatory arbitrage, ratings arbitrage, and i will throw this in to provide another example, merger arbitrage (risk arb) are all NOT really arbitrage in the academically pure sense you are using (i.e., riskless profits ought not persist in efficient markets. These are not this case). Technically, each is a misnomer. None of these are riskless trades that imply convergence in the efficient market.
David
Mkamdar,
Not sure I would say better price discovery in the CDS market but arguably more current. But that also makes the CDS market subject to gyrations, like the equities markets. So better is determined by what you prefer.
Google corporate bonds and liquidity. You’ll find tons of stuff.
Erdosfan “…1) “Acc’s point is the relevant one, the instruments and methodologies that availed to banks, to achieve regulatory aribtrage (under Basel I, more obviously) were essentially in infinite supply.
That is simply ridiculous. THERE ARE TWO SIDES TO EVERY TRADE. There is no infinite unlimited supplier of credit protection.”
Dumb guy here – isn’t the problem now that there aren’t two sides (uh, no buyers?)
By the way, just discovered this sight a few days ago. Thank you for writing and educating people about this complex subject.
Just to jump in with no desire to engage in the argument, I think you have both make good points and that you’re generally in agreement, with one emphasizing the more solid actualities of CDS markets and the other how they are also used. Reading the posts, I think it’s a matter of emphasis which divides you.
I think this is on topic, but I’ve become less interested in the CDS part of the story and more in the nature of the underlying bets. To explain, with a focus on the subprime loan issue, we see in the auction-rate securities how it wasn’t sustainable for a borrower’s long-term instrument to be short -term for the lender and that is the same thing with the subprime loans. The easiest examples would be 2/28 or 3/27 loans intended to be refinanced, so the usual lender’s long bet on price was actually a shortish term bet on appreciation – not merely stability – a bet that needed to be renewed by that or another lender in 2 or 3 years. But these loans were sold as actual mortgage loans, not as 2 or 3 year bets. To go just a tiny bit further, in many of these loans, there was no meaningful income coverage or borrower money at risk, so that means the lender was making a fairly pure bet not on the borrower (income) or even the collateral (money at risk) but on the market. If that had been thought through, if people had said, “we need to insure this stuff as 2 or 3 year market bets,” then we would have built the correct credit edifice on top. Maybe you would take a pure bet and say, “here’s cash equal to the price you’re paying and I don’t care what you make,” but you’d hedge that bet differently than if you treated this kind of loan as a mortgage. We could then have had puts and calls on the real estate markets, etc. and the risk of the bubble would have directly been priced into the system.
[...] George Soros has loused up explanations of how credit default swaps work. I’ve called out economists in the past for their mumblings on credit default swaps and the like, and so has Megan McArdle. [...]