How the Prisoner’s Dilemma and Unintended Consequences are Accelerating the Credit Crisis

Two readers wrote to me concerning phenomena we’ve mentioned upon occasion in the expanding credit crunch, and it seemed a good opportunity to discuss them longer form.

There are two separate, but related threads: we are now seeing a lot of “every man for himself” behavior (liquidity hoarding is one of many examples) that seem rational (or at least defensible) on an individual basis, but are destructive to the financial system as a whole. The second is that, per Richard Bookstaber, our financial system is “tightly coupled” and in tightly coupled systems, risk reduction measures (which too often look at risks in isolation) will typically have the perverse effect of increasing risks. As he explained:

Tight coupling is a term I have borrowed from systems engineering. A tightly coupled process progresses from one stage to the next with no opportunity to intervene. If things are moving out of control, you can’t pull an emergency lever and stop the process while a committee convenes to analyze the situation. Examples of tightly coupled processes include a space shuttle launch, a nuclear power plant moving toward criticality and even something as prosaic as bread baking.

In financial markets tight coupling comes from the feedback between mechanistic trading, price changes and subsequent trading based on the price changes. The mechanistic trading can result from a computer-based program or contractual requirements to reduce leverage when things turn bad.

Eugene Linden, who has written extensively on animal behavior as well as markets, gave this observation:

The problem facing the credit markets right now is yet another iteration of the “prisoner’s dilemma” from game theory, at least in the sense that participants know that if everybody takes the stance of “every man for himself” the markets will crater, but they also know that if they rush for the exits there’s a chance that they will get out the door relatively unscathed. Studies of the problem suggest that the more anonymous the context, the more likely that players will adopt “every man for himself,” and, of course there’s nothing more anonymous than markets. Nature has a long time to work out solutions for problems, and it turns out that a number of animals have converged on the same optimal solution that game theorists have worked out. It’s called “tit for tat,” and it simply means that if someone extends trust to you reciprocate that trust, and if not, not. The best example comes from vampire bats. When a bat is short on blood it will call on a copain for a sip, and if its bat buddy does the right thing, then the thirsty bat will reciprocate at some point in the future when the tables are turned.

It is wonderfully perverse that vampire bats are more community-minded than Wall Street.

The problem now is, save perhaps within the dealer community itself, many players deal with each other on an anonymous, one-off, or transactional basis. So the opportunity to discipline bad behavior is diminished considerably (but ironically, one of the big factors behind Bears’ demise was anger in the community that it had behaved badly both in the LTCM crisis by being the only firm called by the Fed who refused to participate, and its reluctance to shore up its failed hedge funds last June).

Now consider how this conspires with the second element, the perverse outcomes that result from trying to reduce risk in a tightly coupled system. We had written about these examples of efforts to fix the housing/credit crunch backfiring. I’ll start with the first, which is that aggressive cuts at the short end of the yield curve initially did nothing to lower long-term rates, which are the basis for pricing most mortgages; the later cuts have steepened the curve, making matters worse.

Reader Lune came to similar observations independently and put them together well, so we’ll continue with her list:

We’ve already seen the law of unintended consequences so far:

1) Congress raises conforming limits on Fannie/Freddie to help unfreeze the mortgage market. Result: agency spreads skyrocket, bringing down Bear and a host of hedge funds. Mortgage markets still remain frozen.

2) Fed opens TSLF to unfreeze mortgage market. Result: Carlyle goes bankrupt as people rapidly arbitrage the difference between holding MBS in firms that can and can’t access the new credit facility. Mortgage markets remain frozen.

Now we have 3) Fed opens TSLF to broker-dealers. Given the track record of our esteemed Fed so far, I shudder to think what the unintended consequences of this one will be, and I’m disturbed that it’s very likely that no one has thought about that while running around in a panic shooting from the hip at any shadow that comes up. Anyway, here’s my speculation…

The Fed is already close to tapping its full balance sheet. The trigger for the collapse of the past few weeks has been the rise of agency spreads, which is the cause not the effect of all the implosions we’ve seen so far. So to stop the panic, the Fed would have to intervene in the agency market. But it’s remaining reserves of ~$400bil is tiny compared to the amount of debt out there. Furthermore, even a full faith govt. guarantee is unlikely to stop the rise in premiums (witness Ginnie Mae debt, where spreads are increasing even with a govt. guarantee). This is partially because of panic, and partially because agency debt will have fundamentally different behavior when it includes all the extra debt Congress is talking about stuffing it with. So with that uncertainty and unpredictability, it’s no wonder spreads are increasing.

As the spreads continue to claim more casualties, more firms will line up for funding (when do hedge funds get to drink directly from the punch bowl? At this rate, probably in a week or two), and the Fed, unable to say no, will have to start issuing treasuries to expand its balance sheet. Within a matter of a month or two, the Fed will find itself with a trillion or so dollars of impaired debt in a “repo” that can’t ever be recalled (some because the counterparty’s balance sheet is still too weak, others because the counterparty has gone BK). The ultimate casualty? The Fed itself, unable to lower interest rates below 0%, facing default on collateral on its hands, and counterparties (central banks) unwilling to trust the Fed to manage the dollar any longer.

Oh yeah, and mortgage markets will still be frozen.

And she continued later with another possible unintended result:

I’m wondering: if the demise of Carlyle and BSC was hastened because they were firms that couldn’t access Fed money and thus were foreclosed by firms that could, what will happen Monday? I’m thinking hedge funds, unable to access the Fed directly, will be eaten alive by the IBs.

Why? Because I’m figuring they’ll find it safer to shut down hedge funds, take their collateral and convert it into Treasuries, even at the usual Fed haircut, rather than deal with the prolonged uncertainty and volatility of working with their hedge fund clients for an orderly unwind of their positions.

When there was no choice but to choose option #2, plenty of IBs bent over backward to try to keep the hedgies afloat, lest the market collapse. But now, better to shut them down, stuff the Fed with the remaining crap, and sleep better at night knowing your collateral is now in Treasuries rather than illiquid and opaque hedge fund positions. Which IB out there wouldn’t be willing to convert their whole CDS position into treasuries even at a 50% discount (especially since with a repo, if the CDSes don’t default, you’ll get them back at par when the storm has settled)?

Altogether plausible. Let’s hope this is not what come to pass.

Print Friendly, PDF & Email


  1. Anonymous

    That was the finest summary I have read.

    The only thing I can add is the cardinal element. The opacity of the MBS securities has been elevated to the black out stage. No one will buy the stuff because they can’t be valued.

    The rating agency rudder has been snapped off by fraud. The Fed will have to develop a mechanism for people to value the MBS securities and interest rates have to increase.

  2. tracy

    The idea of converting hedge funds to treasuries should stir anyone thinking this is just a minor glitch into outright fear. First, if the idea comes to one holder of hedge funds, it will come to others. Second, a race to the window is likely because there is no way in hell’s green earth there are enough treasuries available to convert even 50% of the funds out there – IMO. I could be wrong, but unless the amount of treasuries available to trade is somewhere in the multiple trillions, the Fed will usurp the Treasury in printing funny money.

  3. fmo

    interesting thought…

    if you believe that the new term facility will be unlimited and stay open forever then the fed has granted all the ibs a compound put

    they can put the crap to the facility whenever they want and as a nonrecourse repo the result will be a put out of the money by whatever the haircut is

    therefore if the haircut is measured relative to face, there is provably no incentive to early exercise (ie putting the stuff to the facility)


    if you worry that the facility will choke or shut or something, then you think, “i shouldnt panic, but if i do, i should panic first,” or if its the case that the lendable amount even of level iii assets is somehow subject to degrade based on market conditions vs. face amount…

    and boom

    thar she blows

    how many people can get thru the revolving door at once?

  4. Lune

    Prisoner’s dilemma is a great game. I learned C programming genetic algorithms to develop optimal Prisoner’s dilemma strategies. Fun times.

    Actually, there is one 2-bit (meaning you only look at the previous 2 decisions) strategy that beats tit-for-tat. In tit-for-tat, if you rat the other guy out, but he still trusts you, then you trust him on the next play, hoping that both of you now start to trust each other.

    In the new strategy, you keep ratting him out until he finally no longer trusts you.

    To put a sociological spin on it: tit-for-tat will reciprocate with trust if you trust it. But if you assume some people are suckers who will continue to trust you even after you violate that trust, and take advantage of that, then you can come out ahead. Which strategy does Wall St. seem to be playing…?

  5. Max

    The post is a nice follow-up to the problem of trust. Politicians, and apparently Fed chairmen, have an amazingly vulgar understanding of trust, e.g. they think they can make something trustworthy simply by decree, like extending the mortgage limits for GSEs.

  6. fuguez

    I have to agree with the first poster: the finest article that I have read as well.

    Game theory and behavioural ecology strike again.

  7. Anonymous

    So we create a housing floor by requiring that all home loans be refinanced based on current market prices (rules – must lived in home five years and other things that would support you NOT being the home short term investment business, but a home to live in for awile).

    The banks would lose, but seeing security now in the lender, might well be willing to loan. (Where necessary, the Feds would provide a soft second down payment at 20% loan amt – need to go back to old/wise rules).

    The big losers would be the BS and alike. Ok by me. Much rather the Feds help the little guy, than BS.

    Would appreciate a reply to this idea, thanks

  8. Yves Smith

    Anon of 2:19 PM,

    Your proposal is not dissimilar to what would have been achieved by the proposed changes in bankruptcy laws, which were not particularly well described in the popular press (which, captured by the opponents, generally simply said that “judges could rewrite mortgages’).

    The changes would have put the treatment of residential mortgages on the same footing as those for investment properties and commercial mortgages in bankruptcy. The judge would write down the value of the mortgage to current market value (it’s a secured credit, so you are merely making sure the mortgage is no higher than the value of the security); the amount in excess would be added to the borrower’s other unsecured debts and paid whatever percentage the other unsecured creditors got. The judge would then set a payment schedule on the secured debt.

    This has major advantages to the banks: they don’t need to worry about being sued by investors in securitized deals over loan mods; their servicers get paid their fees first (which is a better deal than they get now). People I know who work in affordable housing are mystified that the banks are fighting it.

    And yes, you are right that the prospect of governmental intervention ought to make the banks more willing to do mods, which is the best of the bad outcomes open to them.

    The one issue I have with your formulation is the five year requirement. Most of the mortgages that are in trouble are 2005 and later vintages. People who bought 5 years ago could have done a refi.

    From a practical standpoint, it is hard to tell speculators from homeowners (a lot of the speculators lied and said their purchase was for a primary residence, so you’d need to spend hard dollars to do due diligence to determine otherwise).

  9. Observer

    Don’t accept simplistic views of game theory and especially the variants of Prisoner’s Dilemma. There are more satisficing solutions than mentioned above, and 2-player games are dangerously close to misrepresentation for most real world situations.

    That said, Satyajit Das (again) has excellent game-theoretical analysis of markets and risk in his book.

    And going back to a theme from last summer, remember that there is a real difference between risk and uncertainty (as first espoused by Frank Knight in 1921). Not to say there aren’t a lot of twists and turns in all this. Here’s a useful summary:

  10. Anonymous

    You can play an amusing game of what if…

    1. Bear Stearns died, in part, because hedge funds stampeded out the door, withdrawing their prime brokerage business.

    2. Bear Stearns almost certainly would have survived if the TSLF had kicked in immediately instead of March 27, or if the emergency measure of allowing broker dealers to access the discount window had been introduced before its demise instead of in the immediate wake of its demise.

    3. In some parallel universe where Bear Stearns survived instead of dying, Bear executives are right now plotting to kill off and plunder for collateral some of the very hedge funds that, in our universe, killed them off first.

    In the words of Vladimir Lenin: кто кого? (kto kovo?). “Who to whom?”. Ie, who dominates/destroys whom? Who can do what to whom?

    And by the way, aren’t hedge funds big writers of CDS protection? So after a bunch of hedgies get killed off, maybe Mr. Market will give the unintended consequences merry-go-round another hard whirl…

  11. Hank

    > if you assume some people are
    > suckers who will continue to
    > trust you even after you
    > violate that trust, and take
    > advantage of that, then you
    > can come out ahead. Which
    > strategy does Wall St. seem
    > to be playing…?

    Since the New Deal. Well documented, academically, too.

    “… securities legislation can best be understood as an effort to reestablish the viability of what has been labeled the “American dream”…. as a response to a moral crisis of capitalism, generated by the “immoral behavior” of the capitalist elite…. to establish the moral legitimacy of capitalism by restoring trust in the existing system. … it would merely be symbolic and used as propaganda to maintain the status quo.
    … examining the private correspondence and the actions of the regulators during the early years of the SEC act. We believe our analysis shows that the early SEC commissioners had a commitment to the private property rights paradigm, and were unwilling to confront the monied interests.”

    That study is just one example.

    Google Scholar will find many more studies.

    What puzzles me more than anything, as a mere bystander reading everything available, is why the illusion is so strong.

    Has anyone opened a betting pool on when everything will be back to normal, the markets will again have always had the answers to everything, and the government will once again have always been just a drag on the proper functioning of the markets?

    I’d give it eight or ten months.

  12. Lune

    I guess I should have been a little less dramatic about hedge funds getting “eaten alive” by their banker/brokers. Right now, there are certainly a lot of hedge funds that probably can’t meet their margin requirements if they were to be called immediately. So IBs could force them to liquidate. But if one IB (say GS, for example) starts to do this, the news that GS is trying to kill off its hedge funds would spread quite quickly, and other hedgie customers of GS would likely flee to a broker who could provide some sort of guarantee that they get time and money to help them get through the mess. This would start a stampede out the door that would get noticed quickly by the short speculators, triggering a run like that on BSC.

    So in other words, IBs and hedgies are in equally precarious positions. A loss of confidence of one by the other could be deadly, but since it works both ways, perhaps they’ll find some uneasy balance.

  13. nick fertilemind capital

    tightly coupled systems respond to shocks poorly. loosely coupled hetrogenous systems are much better. Unfortunately Basel ii will make the system more homogenous and tightly coupled. Responses to shocks will be unified and amplified by “risk” management. 08 will be nothing compared with 11.

  14. Thai McGreivy

    A professor by the name of Kristian Lindgren of Goteborg University in Sweden has developed some interesting models for this in a field known as complex adaptive systems.

    One model in particular, designed as a parable for society and the economy, combines two classic games: one being The Prisoner’s dilemma and the other being a game known as The Game of Life.

    Anyway the models have some very interesting conclusions: namely permanent social and economic stability are fundamentally impossible.

    All systems eventually implode– but at least new systems always eventually emerge from this implosion.

    The notion that stability can be permanently achieved thru ANY approach (free market Liertarian, Command economy highly regulated, hybrid economy “the third way”, etc…) is fundamenatlly wrong. All are the same in the end.

    There is a very well written summary of this (an easy read) on page 226 of Eric Beinhocker’s The Origin of Wealth.

    So while tightly coupled systems are more efficient (the prisoner’s dilemma analogy would be ‘always cooperates’ (as opposed to TIT FOR TAT), the benefit of that increased efficiency from ‘always cooperate’ (e.g. a tightly couupled system) IS ALWAYS 100% offset by the increased risk of destruction from a non-cooperator (known as ‘always defect’ in the language of prisoner’s dilemma competition).

    Energy in the universe is ALWAYS conserved

  15. Thai McGreivy

    PS– if you want to see The Game of Life in action, click the link I left in my comment about and then click ‘basic simulations’ in the left upper corner. From there you can click the ‘Run Simulation’ link.


Comments are closed.