Wednesday, October 24, 2012

Algorithms, Arbitrage, and Overreaction on Intrade

There were some startling price movements in the presidential contracts on Intrade yesterday. Here's the price and volume chart for the Romney contract, which pays out $10 if he is elected and nothing otherwise:



At 7:52am the price of this contract stood at 40 (this is a percentage of the contract face value, so represents $4.00). Over the next two hours the price edged up to 42, with about 4500 contracts traded. That's where things stood at 9:58. Over the next three minutes the price rose sharply to 48.5, with a further 1700 contracts traded. This was followed by sharp oscillating movements between the peak and 42, which can be seen as the red blur at around 10am in the chart. By 10:15 the price had fallen to 43 and the oscillations had mostly ceased. An hour later the price was back down to 41, with about 14,000 contracts and $63,000 having changed hands over three hours.

What caused this unusual price behavior? There's been some talk of attempted price manipulation, but I have my doubts because the trader who was buying aggressively over this period was extremely naive. (I am fairly certain that this was a single trader). Throughout the buying frenzy, the Obama contract never fell below 57 and there was a substantial block of bids at or above this price. The trader who was buying Romney at 48 could have made the same bet for 43 by simply selling the Obama contract at 57. In fact, he would have obtained a slightly superior contract, which would pay off if any person other than Obama were to win, including but not limited to Romney.

This fact also explains the oscillations in the Romney price, and the decline to 57 under selling pressure of the Obama price. Any individual who had posted ask prices in the 43-48 range in the Romney market had these orders met by the crazed buyer, and could then sell Obama above 57 for an immediate arbitrage profit. As it happens, there are algorithms active on Intrade that do precisely this. They post ask prices that, together with with highest bid in the complementary contract, add up to slightly more than 100. As soon as these orders trade, they sell the complementary contract at once, booking a riskless profit. These algorithms posted prices in the 42-43 range over the period in question, and the buyer repeatedly traded through them to reach the higher Romney asks. Hence the red blur in the chart. Only when the buyer gave up or ran out of funds did the price settle down.

If this was not an instance of attempted manipulation, then what was it? I suspect that it was an overzealous response to reports of a major announcement concerning the presidential race, promised by Donald Trump. Further frenzied activity in the presidential and state level markets took place in the evening, as speculation about the nature of the announcement started to spread.

This whole bizarre episode tells us very little about the presidential race, but does shed some light on how these markets work. Changes in one market spill over instantaneously to changes in linked markets via arbitrage, some of it executed algorithmically. And algorithms that work very effectively when rare can end up with disastrous results if copied. For instance, if two algorithms were to follow the strategy outlined above, it is possible that only one of them may be able to complete the second sale since the first mover would have snapped up the existing bids. This kind of game is currently being played in the world of high frequency trading, but with much higher stakes and considerably more serious economic consequences.  

Sunday, October 14, 2012

Of Bulls and Bair

Sheila Bair's new book, Bull by the Horns, is both a crisis narrative and a thoughtful reflection on economic institutions and policy. The crisis narrative, with its revealing first-hand accounts of high-level meetings, high-stakes negotiations, behind-the-scenes jockeying, and clashing personalities will attract the most immediate attention. But it's the economic analysis that will constitute the more enduring contribution.

Among the many highlights are the following: a discussion of the linkages between securitization, credit derivatives and loan modifications, an exploration of the trade-off between regulatory capture and regulatory arbitrage, an intriguing question about the optimal timing of auctions for failing banks, a proposal for ending too big to fail that relies on simplification and asset segregation rather than balance sheet contraction, a full-throated defense of sensible financial regulation, and a passionate critique of bailouts for the powerful and politically connected even when such transactions appear to generate an accounting profit.

Let's start with securitization, derivatives and loan modifications. Under the traditional model of mortgage lending, there are strong incentives for creditors to modify delinquent loans if the costs of doing so are lower than the very substantial deadweight losses that result from foreclosure. But pooling and tranching of mortgage loans creates a conflict of interest within the group of investors. As long as foreclosures are not widespread enough to affect holders of the overcollateralized senior tranches, all losses are inflicted upon those with junior claims. In contrast, loan modifications lower payments to all tranches, and will thus be resisted by holders of senior claims unless they truly see disaster looming. One consequence of this "tranche warfare" is that servicers, fearing lawsuits, will be inclined to favor foreclosure over modification.

But this is not the end of the story. Bair points out that the interests of those using credit derivatives to bet on declines in home values are aligned with those of holders of senior tranches, as long as the latter continue to believe that foreclosures will not become widespread enough to eat into their protected positions. This is interesting because these two groups are taking quite different price views: one is long and the other short credit risk. Bair notes that some of the "early resistance" to FDIC loan modification initiatives came from fund managers who "had purchased CDS protection against losses on mortgage-backed securities they did not own." The irony is that they were joined in this resistance by holders of senior tranches who were relying (overoptimistically, as it turned out) on the protective buffer provided by the holders of junior claims.

Another interesting discussion in the book concerns the trade-off between regulatory arbitrage and regulatory capture. A fragmented regulatory structure with a variety of norms and standards encourages financial institutions to shop for the weakest regulator. In the lead up to the crisis, such regulatory shopping occurred between banks and nonbanks, with mortgage brokers and securities firms operating outside the stronger regulations imposed on insured banks. But Bair also notes that the "three biggest problem institutions among insured banks - Citigroup, Wachovia, and WaMu - had not shopped for charters; they had been with the same regulator for decades. The problem was that their regulators did not have independence from them."

This is the problem of regulatory capture. Bair argues that while a single monolithic regulator would put an end to regulatory arbitrage, it could worsen the problem of regulatory capture: "a diversity of views and the ability of one agency to look over the shoulder of another is a good check against regulators becoming too close to the entities they regulate." It's a point that she has made before, and clearly believes (with considerable justification) that the FDIC has provided such checks and balances in the past. It was able to do so in part through its power under the law and in part through the power to persuade; yet another reminder of the continued relevance of Albert Hirschman's notion of voice.

A very different kind of trade-off concerns the timing of auctions for failing banks. One of the policies that Bair favored at the FDIC was the quick sale of failing banks prior to closure in order to avoid a period of government stewardship. She recognizes, however, that there are some costs to this. Bids from prospective buyers who have not had time to closely examine the asset pool of the failing institution will tend to be lower than the expected value of these assets, given the need to maintain adequate margins of safety in the face of risk. Waiting until a more precise estimate of the value of the bank's assets can be obtained can therefore result in higher bids on average. But there are also costs to waiting: a "deterioration of franchise value" occurs as large depositors and business customers look elsewhere, and this can offset any gains from a more precise valuation of the asset pool. Bair seems to have concluded that sale before closure was always the best course of action, but I suspect that this need not be the case, especially when the asset pool is characterized by high expected value but great uncertainty, and the bulk of deposits are insured. In any case, it's a question deserving of systematic study.

Bair describes herself as a lifelong Republican and McCain voter; she is contemplating a write-in vote for Jon Huntsman this November. Yet she seems quite immune to partisan loyalties and pressures. Her description of Barney Frank is positively affectionate. She has kind words for some Democrats (such as Elizabeth Warren and Mark Warner), but offers blistering criticism of others (Robert Rubin, Tim Geithner and Larry Summers in particular). Among Republicans too, she is discerning: Bob Corker's efforts on financial reform are lauded, but the "deregulatory dogma" overseen by Alan Greenspan comes under forceful attack. She laments the "disdain" for government and its "regulatory function" and describes as a "delusion" the idea that markets are self-regulating. These are not the views of a political partisan.

On policy, Bair is opposed to the use of derivatives for two-sided speculation (when neither party is hedging) and would require an insurable interest for the purchase of protection against default. She describes synthetic CDOs and naked CDSs as "a game of fantasy football" with no limit to the size of wagers that can be placed. She wants a "lifetime ban on regulators working for financial institutions they have regulated." And she argues, as did James Tobin a generation ago, that the attraction of the financial sector for some of the best and brightest of our youth is detrimental to long term economic growth and prosperity.

No review of this book would be complete without mention of the bailouts, which troubled Bair from the outset, and which she now feels were excessive:
To this day, I wonder if we overreacted... Yes, action had to be taken, but the generosity of the response still troubles me... Granted, we were dealing with an emergency and had to act quickly. And the actions did stave off a broader financial crisis. But the unfairness of it and the lack of hard analysis showing the necessity of it trouble me to this day. The mere fact that a bunch of large financial institutions is going to lose money does not a systemic event make... Throughout the crisis and its aftermath, the smaller banks - which didn't benefit at all from government largesse - did a much better job of lending than the big institutions did. 
What bothers her most of all is the claim that the bailouts were justified because they made an accounting profit:
The thing I hate hearing most when people talk about the crisis is that the bailouts "saved the system" or ended up "making money." Participating in bailout measures was the most distasteful thing I have ever had to do, and those ex post facto rationalizations make my skin crawl... The bailouts, while stabilizing the financial system in the short term, have created a long-term drag on our economy. Because we propped up the mismanaged institutions, our financial sector remains bloated... We did not force financial institutions to shed their bad assets and recognize their losses... Economic growth is sluggish, unemployment remains high. The housing market still struggles. I hope that our economy continues to improve. But it will do so despite the bailouts, not because of them.
The ideal policy, according to Bair, would have been to put insolvent institutions into the "bankruptcy-like resolution process" used routinely by the FDIC, but she recognizes that the legal basis for doing so was not available at the time. She therefore signed on to measures that were instinctively repugnant to her, and tried to corral and contain them to the extent possible.

The argument that the bailouts "made money" is specious for two reasons. First, the funds provided were given well below market value, and the cost to taxpayers should be computed relative to the value of the service provided. If insurance is provided at a fraction of the actuarially fair price, and no claim is made over the period of insurance (so the insurer makes money), this does not mean that there was no subsidy in the first place. Steve Waldman has made this point very effectively in the past. Furthermore, the cost to taxpayers should take into account any loss of revenues from more sluggish growth. If Bair is right to argue that the bailouts were excessively generous, to the point that growth prospects were damaged for an extended period, the loss of tax revenue must be included in any assessment of the cost of the bailout.

As noted above, there are many accounts in the book of meetings and decisions, and a number of speculative inferences about the actions and intentions of others. Some of these will be hotly disputed. But focusing on these details is to miss the larger point. The crisis offers us an opportunity to think about the flaws in our economic and political system and how some of these might be fixed. It also suggests interesting directions in which economic theorizing could be advanced. The book helps with both efforts, and it would be a pity if these substantive contributions were drowned out in a debate over conversations and personalities.