Eighteenth-Century Lessons for Credit Default Swaps

hogarthcoffeehouse.jpgOf late I’ve been reading Niall Ferguson’s The Ascent of Money: A Financial History of the World, and I’ve been struck again at how similar the rise of the modern CDS market is with the rise of maritime insurance at Lloyd’s Coffee House in the early 18th century.

Originally, of course, there weren’t insurance companies. Rather, individual merchants and speculators would meet informally at Lloyd’s Coffee House in London. Suppose that there was a merchant with a ship that was about to make a voyage to America that he wished to insure against loss due to the perils of the sea. He would arrive at Lloyd’s with a written contract promising to pay him so much in the event that his ship went down. He would also arrive with money — or at least the willingness to pay it. He would then circulate through the coffee house, looking for those who were willing to sign the contract. Suppose that the contract promised to pay out 5,000 pounds in the event that the ship went down. Some people would sign for one pound of liability, some for 100 pounds, some for 1,000 pounds. Each underwriter would, of course, be paid by the merchant for their signature in proportion to the amount of liability that the underwriter took on. In the end, the merchant would have an insurance contract on his ship signed by a pool of investors who in aggregate were paid less than the insured value of the ship. Of course, while he was at Lloyd’s the merchant might pick up a bit of money by signing on as an underwriter on another merchant’s insurance contract. Before too long there were people who made their living (or at least tried to) entirely with in Lloyd’s Coffee House.

Thus maritime insurance started out much like the modern CDS market. It was an entirely over the counter trade. (Literally!) There were no reserve requirements for underwriters, there was a lot of speculation by those who didn’t have a clear idea of the risks they were taking on, as well as underwriting by experienced market participants who probably had a very good idea about the perils of the sea. And of course, there was an endemic problem of counter party risk. If an underwriter turned out to be insolvent in the event that a ship did sink off Cape Hatteras, there wasn’t much that the hapless owner could do. There was, I suppose, the consolation of debtor’s prison and the knowledge that you could drive welshing underwriters and their children into a Little-Dorrit-esque existence.

There was a two fold response to the wild world of Lloyd’s. On one side, the common law courts reacted with skepticism if not outright hostility to the contracts written in the smokey coffee rooms. There was a persistent suspicion that these contracts were little more than gambling devices, something disreputable that ought to be suppressed or limited. On the other side, the regulars at Lloyd’s realized that they needed to clean up the insurance market. The result was a system of trading within an association to which one could not gain access unless certain basic proofs of reliability were met. In the end, pay-as-you-go was replaced with insurance based on an insurance pool and investment proceeds, which gradually reduced the problem of counter party risk in insurance. (Although it far from eliminated it.)

I actually think that there are some important lessons to be learned from Lloyd’s. First, over the long term I think that it is pretty clear that the insurance-as-gambling meme turned out to be a legal dead end. Judicial hostility toward insurance contracts injected legal uncertainty into the mix of other risks, but did little or nothing to deal with the underlying problems of counter-party risk and proper risk assessment. These problems, it turned out, were solved despite legal hostility rather than because of it. Second, a completely decentralized market of spot contracts in pure risk allocation doesn’t work horribly well. Regulating the capital requirements of players in the game (or at least insuring their transparency) is absolutely necessary. Better yet is a clearing house system that eliminates counter party risk through asset pooling and the ruthless exclusion of unreliable or undercapitalized players. Not surprisingly, the commercially sophisticated parties at Lloyd’s did both of these things long before the courts and parliament managed to sort out a sensible attitude toward insurance. I suspect that the same will be true of the CDS market.

You may also like...

15 Responses

  1. Matt says:

    When I took a course on international financial regulation a few years ago, and we were covering CDS’s, and I was trying to understand them, I asked, “So, these are basically like a form of insurance, right?” The answer was yes, they are like a form of insurance, but all of those involved are very careful to say they are not insurance and in fact not like insurance at all, the reason being the insurance is regulated and controlled, while CDS’s were not, allowing for all sorts of fun and innovation. It does not seem that this has turned out well at all now, and that perhaps some breaks on this sort of innovation might be in order.

  2. Nate Oman says:

    Of course, there is an element of insurance in every contract. One of the things that strict liability does is allocated certain risks to one party rather than the other. On the other hand, CDSs are clearly insurance in more than this sense.

  3. Mike Zimmer says:

    There are two types of credit default swaps. The first are “insurance-like” in that they are connected to securitized assets and “insure” the failure of the underlying mortgages/credit card debt, etc., to be paid. I have heard that is something $7 trillion of the $53 trillion dollars in CDSs. The second type are not even “insurance-like” because they are not linked to any securitized assets but are simply bets on the payback of the mortgages, etc., making up the securitzed package. These are pure gambles and, as such, are so toxic.

    I don’t own the Chicago Cubs, but I did bet they would make it to the World Series. I lost. If I could call my loss part of the CDS mess, I might be able to get my money back. Perhaps, these noncollateralized CDSs should just be knocked out as unenforceable gambling contracts: If you “won,” by betting a pool would not pay back and that happened, you don’t get your money. But, those who “lost” that bet, and are on the other side of the CDS would not have to pay. All is square. Such a CDS is not an asset but it is not a liability either. Does that solve the toxicity issue?

  4. Nate Oman says:

    I wonder, however, if there might be value in these naked CDSs as it were in terms of hedging. For example, if a relatively thick market develops in a particular kind of bet so that the market aggregates information about the probability of an event well, then that market price conveys an important bit of information that could be used to structure transactions. In a sense this is what futures contracts in commodities where neither side takes delivery does. I am not sure why it makes sense to do this with commodities but not other kinds of assets. Likewise, it seems to me that the naked bets could be used as hedges, and therefore could serve a legitimate risk management function. Initially the law of contracts tried to knock out a lot of insurance contracts with the doctrine of an insurable interest, but the results have not been entirely edifying. It seems to me that the big problem created by CDSs was the systemic uncertainty caused by counter party risk. This, it seems to me, could be dealt with through some sort of a clearing house or exchange mechanism. I am less convinced that we can or should draw legal distinctions between “good” insurance CDSs and “bad” speculative CDSs.

  5. I like your analogy but would like to take it one step further.

    I spent the first 20 years of my career in marine reinsurance (in simplest terms, insurance of insurance companies). In the mid 1980’s the reinsurance market nearly sank Lloyd’s for what seems to be one of the problems plaguing the CDS market. Everybody reinsured everyone else creating what was referred to as the LMX spiral (London Market Excess). It became nearly impossible for reinsurers to quantify their exposures as large losses would wind their way through the market. Claims would come in from all directions and then be ceded back out through retrocession covers which to make matters even more confusing is reinsurance of reinsurance. The big loss that was the ultimate catalyst for change was the large North Sea oil rig, Piper Alpha. Soon after Lloyd’s transformed itself from a market of names (individual risk bearers) to corporations which was backed by significant capital and more liquidity. Other controls were put into place to eliminate the retrocessional spirals and give underwriters half a chance to quantify the risk on their books. It ain’t perfect but it’s better than it was.

    I’m no Wall Street quant, but this CDS mess smells a lot like the reinsurance near debacle of the ’80’s.

  6. A.J. Sutter says:

    Apropos of “there might be value in these naked CDSs as it were in terms of hedging”: Le monde diplomatique published an interesting bar chart in the November 2008 issue (@20), compiled mainly from UN sources. It showed the following average daily values of different kinds of trade in 2007 (my readings are approximate, since it’s a graph, not a table):

    global commerce (“real” economy): ~$40BN

    global GNP: ~$130BN

    Stock markets: ~$130BN

    Forex markets: ~$1.6 Trillion

    Derivatives: ~$3.9 Trillion

    Clearly the risks being hedged, if that’s what’s being done, don’t reflect risks in the real economy (which, one is told, is what the original function of insurance was supposed to be). They would be risks in the financial economy, without any relationship to the real economy. The distinction between such risks and gambling eludes me. To say that they represent “value” is to distort the teachings of classical economics.

    I’m agnositc as to whether there might be a reasonable argument that one shouldn’t draw legal distinctions between “good” and “bad” CDSs. But if you’re basing your legal reasoning on economics, then denying a distinction would suggest that all are “bad”. Otherwise, the implications of your notions of how “value” is created (namely, that “value” created without labor is 40x-100x times that of all value created by humanity) are quite scary, as a matter of social policy.

  7. JP says:


    Nowhere does Nate, or anyone else, suggest that the total amount of all derivatives traded is a measure of “value.” Nor does it make any sense whatsoever to compare that amount to GNP or most other total trade values.

    The point is that a large enough, well-regulated speculative market can create real value by spreading risk and aggregating information.

  8. A.J. Sutter says:

    JP, thanks for your reply. My comment was apropos of Nate’s comment at 2008/12/09 6:44 PM, in which he speculates, “I wonder, however, if there might be value in these naked CDSs as it were in terms of hedging.” You yourself are less tentative, and say that under certain conditions such a large market “can create real value” (emphasis added). But neither you nor Nate suggest how that is possible or what sort of value that may be.

    Under classical theories of economics (including both Smith and Marx), labor is a necessary component of value. Global GNP includes, among other things, the value of the wages of everyone in the world. So the paradox arises when the aggregate (numerical) value of derivatives is of the order of 10x-100x GNP, and all these derivatives are asserted to “create (real) value”. I understand how from an accounting point of view such a *numerical* value could arise from such trade; the issue is whether this constitutes “value” in the classical sense. Because the classical sense, and not the numerical one, is the sense of value that is invoked by the rhetorical trope “creates value”, which is often used by proponents of liberal economic policy to justify and exalt those policies, notwithstanding the social inequalities and other results (vide the news today or any day recently) thereof.

  9. Nate Oman says:

    AJ, the short answer is that I reject the notion that real economic value is created by labor. Rather, it seems to me that economic value arises when we provide people with either goods or services that they are willing to pay for. It is the satisfaction of expressed preferences rather than labor that creates value.

    A thick CDS market even in “naked wagers” would create value by (1) providing a hedging device; and, (2) by providing information. To give a concrete analogy, a futures market consists from one point of view of a set of nake wagers on future flucuations in commodity prices. However, a farmer might still use a future to manage risk (as opposed say to crop insurance) and future markets, by aggregating information about future events, assist in planning my business to the extent that future commodity prices are an important input. The value comes from the fact that the market allows the reallocation of risk to those who are able to bear it at a lower cost and through more efficient allocation of resources due to better information. This is not a defense of the current system of over the counter CDSs, but it is, I think, a reasonable set of arguments in favor of thick market in credit dervitives, even when those derivitives are naked bets on future financial events.

  10. JP says:


    Like Nate, I don’t accept the labor theory of value. (Arguably, neither did Smith, though he certainly isn’t clear on the issue.) Both Nate and I, I think, have suggested how naked speculative trades can create value and what sort of value that is: basically, risk allocation and information aggregation.

    I want to emphasize again, though, that this value is not accurately measured by the total amount of all trades. Accordingly, it doesn’t tell us much to say that “the aggregate (numerical) value of derivatives is of the order of 10x-100x GNP.”

    To the extent that this particular quantification is used to support policy arguments (e.g., ‘Wall Street salaries are justified by the aggregate value of all trades’), I think we are in agreement that this is inappropriate.

  11. A.J. Sutter says:

    Nate, thanks for your comment. As for your definition of value, fair enough. This doesn’t resolve the issue raised by JP of whether the amount of risk is based on the entire total (notional) trade value, or something else. Obviously, there isn’t any limitation on the aggregate value of naked wagers. To pick up on Mike Zimmer’s baseball example, the total amount wagered on the World Series by everyone in the world is much bigger that the aggregate amount than will be paid to the team who wins the game. The more people who bet, and the more money they bet, is there more “value” created? I take it you would say yes. How much value — the face value of the bets, or some fraction of that? If a fraction, how do you compute it? Are bets on opposite sides of a contingency netted against each other, or their absolute values added?

    The way that the credit derivatives industry estimates risk is based on “value at risk” (VaR) models, which compute an expectation value that is much lower than the notional value. (I leave aside the question of whether the usual industry methodology for computing VaR is reasonable; see, e.g., Janet Tavakoli’s Structured Finance and Collateralized Debt Obligations (2nd ed. 2008) @92ff, and the work of Benoit Mandelbrot on distributions with infinite variance. Moreover, the typical industry distributions are computed on the basis of historical data, which raises another reasonableness issue (recall what your mutual fund prospectus says about history).) Expectation values are not reality, though. When people underestimate the risk, as has happened within historical memory, all losing bets are called, and notional value ceases to be merely notional. Indeed, Tavakoli @ 52-53 mentions some examples of common deals in which the protection seller must pay the full notional amount, or something close. Certainly it’s the protection buyer’s preference to get the full notional amount or a big chunk thereof. In that situation, what’s the principle on which to distinguish “real” value (in your sense of satisfied preferences) from notional value?

    If we can’t distinguish notional value from “satisfied preferences” value, some interesting social conclusions follow. The number of people who participate in the credit derivatives market is quite small, compared to the population of the world. Let’s suppose there’s a million of them. Also, to be conservative, let’s use annualized figures. The estimated notional amount of credit derivatives at the end of 2007 was roughly $60 Trillion. (Since many contracts are confidential, the outstanding notional value is probably underestimated; also, I’m not including other types of derivatives, such as commodity belly futures, forex swaps, etc.). According to the World Bank, world GDP was about $54 Trillion in 2007; and the world population is about 6.7 billion. The satisfaction of those 1 million peoples’ preferences would be more important than satisfying the rest of the world’s (as represented by global GNP), i.e., each participant in the market would be more important than about 6,700 other people. (Of course, if fewer people are directly involved in the CDS industry, they are proportionally more “important”.) So the benign-sounding rhetoric of “satisfaction of expressed preference[s]” masks a rather stratified view of society.

    If we consider the usual practice of macroeconomists, though, we may want to re-think including the full notional value as “value”. You mention satisfying peoples’ preferences by “provid[ing] people with either goods or services.” These are the very things that are included in GDP and GNP. But we know that the notional value of the derivatives markets is greater than global GDP. If this notional value is indeed the value of a service, as you suggest, then macroeconomists have been understating GDP by at least 100% — even more, if we add in the notional value of derivatives other than credit derivatives.

    So equating “value” to the satisfaction of preferences seems not to be as straightforward as it sounds, in this situation. As for “aggregating information about future events,” this is about as neat a euphemism for gambling as they come. But to the extent you’re relying on an aggregation of individual preferences, it’s also a notion based on fallacies of composition, as well as being plagued by some significant analytical difficulties in achieving this aggregation, as pointed out by Gérard Debreu among others. (He is the Nobelist who, along with K. Arrow, proved the first and second welfare theorems, which are the closest things there are to proofs of the efficiency of markets.) This critique, which is internal to neoclassical economics and not some rant inspired by Adam Smith (from whose views on value you stray), is discussed in the theorem and references mentioned in my comment to this recent post of Dave Hoffman’s.

  12. Nate Oman says:

    A.J.: I have no idea how to quantify the real value added of derivatives. I pretty sure that it is not equal to the face value of the transactions, but that point is neither here nor there as to the argument I was making. I really don’t care if you call derivatives a form of “gambling” so long as we are clear about the potential functional virtues of gambling. I think that inTrade markets provide valuable information. As an analytic matter, I don’t think it matters all that much if you call it gambling or something else, so long as one understands how such markets aggregate information. Frankly, I fail to see how credit derivatives are any more suspect than ordinary commidity future contracts. The problem is not that people are making bets on future events. It is that they are making bets on future events in such a way as to create massive systemic risk. We can have the bets without the systemic risk. Commodity markets, for example, are not regularlly brought to the brink of armageddon by the Chicago Mercantile Exchange.

  13. A.J. Sutter says:

    JP, it looks like my immediately previous comment crossed with yours. As that comment explains, there are circumstances in which the notional value of a trade becomes the amount of the protection seller’s obligation. While VaR valuation is usually far lower than the notional value, recent events illustrate that VaR can be a kind of convenient and optimistic fiction.

    As for information aggregation, the credit derivatives market is notoriously non-transparent. (See, again, Tavakoli.) Moreover, I think there may be some confusion betwen the value of naked wagers and the value of information about those wagers. This distinction seems to become somewhat blurred as one reads from the beginning to the end of Nate’s 12/10 19:52 comment, for example. In the derivatives industry, most people who place wagers won’t have complete information about the bets other people are placing. So the manner in which naked wagers per se create value remains somewhat mysterious. (As does the manner in which markets can “aggregate information abut future events,” as distinguished from information about peoples’ beliefs and professed intentions about future events, consistently with the principles of special relativity.)

    No one has stepped up to the plate to offer any principle (other than VaR or notional or “trade” value) that could be used for determining the value of credit derivatives, yet the assertions of value persist. Suppose the value is greater than zero, but less than the value of a 2-liter Coke — mightn’t the policy conclusions we draw from that be different from if the value were some significant fraction (or multiple) of GDP? (If not, why not?) If Nate, JP or anyone else wants to protest that the value is clearly more than that, then please explain how you are justifying even your approximate quantification.

    If you can’t explain your principles of measuring value, especially when relying on neoclassical economics theory to define value (viz., as satisfaction of preferences), it seems quite irresponsible to look at law or anything else through the prism of that theory. Ditto for making reassuring assertions of “value” in CDSs and other naked bets, notwithstanding the social havoc they cause. Disguised by the scientistic apparatus of NCE, such unfounded assertions actually advance, in the best case, (which I assume applies to JP and Nate) a faith-based agenda, and in worse cases, an ideologically-motivated or self-serving one.

  14. JP says:


    I missed your comment until now. I guess I don’t really understand what you’re arguing.

    There are circumstances in which the value of my homeowners insurance policy becomes equal to something close to the value of my home (e.g., my house burns down). That doesn’t mean that the value of my policy is equal to the value of my home, much less that the value of the property insurance market is equivalent to the value of all property insured.

    In addition to allowing me to hedge the risk of disaster, a robust market in homeowners’ insurance also aggregates information about the risk of things like fires, and in turn improves the ability and incentives to ameliorate those risks. (The value of information aggregation is value of information from naked wagers, not just about them.)

    That the current credit derivatives market is “notoriously non-transparent” is in no way inconsistent with Nate’s post. Early insurance markets were non-transparent too. [As I type that, I realize this comment basically just restates Nate’s original analogy, so I guess we’ve come full circle].

    To the extent you suggest that our inability to quantify the value of a CDS market means we should assume it has little or no value, I simply disagree. I think commodities markets are valuable, though not equal to the total of all commodities traded (or the value of labor invested in creating and maintaining it). I don’t know how to quantify the value of a public criminal justice system, but I think we should have one.

  15. A.J. Sutter says:

    JP, thanks for your comment. My primary point is that it’s either naive or disingenuous to argue for something having value in an economic sense when you can’t quantify what that value is, or even describe the algorithm for its quantification.

    I agree with you that we should have a criminal justice system, but I don’t rely on an economic argument for that. The most important “value” of the criminal justice system is a social one, IMHO.

    One of the downsides to (or programs of) economic analysis of the law is that social and economic values are allowed to become conflated. The assertion of something having “value” in one sense is allowed to stand in for its having value in the other.

    In the present case, my issue was that Nate used the vocabulary of economics to assert value in the CDS market. Yet neither you, Nate nor anyone else has proposed how to determine that value (other than me, following Tavakoli, which you reject). Instead, the value of the market is said to be the value arising from the very unclear notion of “aggregation of information”. The implication seems to be that this aggregation has some social value (as also suggested by your criminal justice analogy). So some rhetorical slight of hand is afoot.

    Of course, if I’m mistaken, and you’re asserting that the aggregation has an economic value, please describe your principle for calculating that value.

    A secondary point concerns aggregation per se. Though often invoked, this aggregation is never clarified beyond a meaningless abstraction. It’s similar to the argument that there is a market-aggregated “demand curve” that is aggregated from individual “demand curves”. In fact, though, if one tries to do the math, it isn’t at all straightforward how to perform this aggregation. The inability to draw macroscale conclusions from the aggregation of microscale preferences is the subject of the Sonnenschein-Mantel-Debreu theorem, which is discussed in the cited references cross-referenced at the end of my 2008/12/11 post, as are Gérard Debreu’s reservations about the summation process.

    My overarching point is that if one isn’t capable of doing the economic analysis beyond vague assertions, it’s a dangerous and irresponsible obfuscation to drag those assertions into an analysis of law. Wovon man nicht sprechen kann, darüber muss man schweigen.