Sunday, August 21, 2011

The Theory of Information-Insensitive Debt Prompts Some Head-Scratching Questions

Here's Part 3 on the magic pigs of high finance, information-insensitive debt (everyone still awake?). Last time we looked at the concept using a common-sense definition of the term. Now let's try to figure out where the theory is unsatisfying on a more granular level, by using Gorton’s own words.

First, to show us what qualifies as information-insensitive debt, he offers examples: high-grade corporate debt, government bonds (presumably U.S. Treasuries, and not Greek 10-year bonds), and AAA rated asset-backed securities.

In different places, he characterizes such debt as (the bold is mine):
[Debt that] "is very liquid because its value rarely changes and so it can be traded without fear that some people have secret information about the value of the debt. If speculators can learn information that is private (only they know it), then they can take advantage of the less informed in trade. This is not a problem if the value of the security is not sensitive to such information."
Also (page 7 of the same document, Slapped In the Face by the Invisible Hand):
"Bank debt is designed to be informationally-insensitive, that is, these bonds are not subject to adverse selection when traded because it is not profitable to produce private information to speculate in these bonds."
These definitions sound impressive, in that arid academic way, but what do they mean when applied to real debt in the wild? For example, the first one doesn't really make sense. He's saying that speculators can't take advantage of the less informed while trading information-insensitive debt, even after learning private information, because "the value of the security is not sensitive to such information."


What debt could Gorton possibly be thinking of here? Does he really believe that even Treasuries are immune to being profited upon, by someone who possesses private information? If I wiretap the Federal Reserve Board meeting, and learn the Fed is about to announce an operation to purchase $800 billion of Treasuries, he doesn't think that gives me an advantage trading in this market? (Note: the second definition does add "it is not profitable" to produce private information, but this opens a new can of worms, which we’ll soon see.)

Also, U.S. Treasuries are very liquid, but it’s not true that their value rarely changes -- their value changes constantly. Now, do the bonds trade without fear that some people possess secret information about that value? For the most part, yes -- but there may also be certain junk bonds that trade without fear that some people have secret information about them. If so, could these junk bonds qualify?

And what if the junk bonds are "information insensitive" for six years, then the company reveals itself to be tottering near bankruptcy and their price becomes volatile? Do they suddenly become information sensitive, or were they always information sensitive but only seemed information insensitive?

In other words, is this quality of being “information insensitive” only ascertained after empirical evidence of how the security actually behaves? Or is a security considered information insensitive only if we can’t imagine a situation in which someone could profitably produce private information to speculate in the debt? But, honestly, no security exists for which that’s an inherent quality, as the thought experiment for Treasuries shows.


And "it is not profitable to produce private information" raises many fresh questions. What's profitable for a trader to do at any given moment depends on many variables that seem as though they should have little to do with information sensitivity.

A trade may be profitable simply because I can lay my hands on large-enough blocks of securities to make chasing a minuscule gain on each one worthwhile. Or, a narrower trading spread may allow me to turn a profit more easily. (Of course more-liquid securities tend to trade with narrower spreads, which leads to a Gortonian paradox, as being liquid is supposed to be a sign of information-insensitivity.) Profitability also hinges on what I pay my workforce -- so does a security become information sensitive simply because I’ve got six traders in Bangladesh who work at one-sixth the salary of their U.S. counterparts? Also how does "private information" factor in? What if it's profitable to ransack Company X's Dumpster for trading information. Does that make its debt information sensitive until the Dumpster is relocated to a more secure place?

Some of the above may sound a bit picayune. But here are the takeaway points: (1) If so many questions can be posed, doesn’t information-insensitive debt sound like a theory that presents a false dichotomy at best? (2) There are so many trades, and so many price movements on securities (especially liquid ones), how do you sort out evidence that proves a security is information insensitive, instead of the opposite?

Next: The troublesome analogy that Gary Gorton’s theory leads to.


  1. This is a couple of years after you wrote this, but I came across your post while doing some research. I think you're conflating the difference between interest rate arbitrage and credit risk arbitrage. Under your example, knowing that the Fed is going to engage in a big QE would allow me to make money in trading, but this is based on interest rate fluctuations. No one is predicting a credit default. So you see mild fluctuations in the value of your "informationally insensitive debt" based on inflation/rate volatility, but you don't see huge volatility based on credit risk.

    I agree that Gorton's explanation of his thesis is somewhat difficult to follow, but I think the best way to understand it is to simply (as Morgan Ricks, a law professor formerly from Treasury, does) replace the phrase "informationally insensitive debt" with "money". Treasuries are a form of money, deposits are a form of money, and in recent years, high grade corporate securities, AAA private-label MBS, and other similarly highly graded liabilities all served the function of money.

    Money fluctuates in value, sure, but this is inflation-based fluctuation. It still trades at par. $1 is $1 is $1, even if that $1 may be worth slightly more or less than it was worth last week. What Gorton is describing in his theory of banking panics is a situation in which people start questioning whether their $1 in money (bank deposits in the old days, now AAA-rated collateral) is actually worth $1. Combine that with the inherent leverage and illiquidity of banking and you have the recipe for a panic, which is what he claims occurred in Fall 2008.

  2. First, thanks for the comment. Now, on to your points.

    Regarding Treasuries and credit risk:
    (1) Of course I’m conflating interest rate risk and credit risk. So is Gorton. Read the excerpt above. He uses “value” three times, and he makes no attempt to qualify this as “value only as it relates to credit risk.” Credit risk and rate risk both can affect the value of a security a lot (“mild fluctuations” based on rate volatility isn’t always true; if credit risk is small, the effect of rate risk can be much larger and thus more significant for an investor, and what about high inflation environments?). Still, if you don’t like my “rate risk” example, what if I construct a different hypothetical, involving say my obtaining secret information about coming changes that will affect liquidity premiums on TIPS, which allows me to turn a nice profit? (2) Here’s the nub of the problem: You and I start the day with $1,000 cash. I find out, through a source at the Fed, about an operation to be announced later that day that will drive down rates and make a $1,000 10-year Treasury (which at that moment is trading at par) worth $1,030. You keep your cash. I buy the Treasury. At the end of the day I cash out for $1,030. So my “money” is worth more than your “money”? No. (For one, Treasuries are “near money,” not “money.”) What I’ve done is earn a trading profit using information “that is private” (only I know it, or at least, only I am free to trade on it, unlike my source) that has allowed me to “take advantage of the less informed.” (See Gorton above) (3) Treasuries do have some credit risk. Maybe you remember this recent Bloomberg story from Oct. 15: “Rates on Treasury bills due in October soared and the U.S. attracted the least demand at weekly bill auctions since 2009 as Senate attempts to end the fiscal impasse were put on hold, increasing speculation of a default.” So if I had “secret knowledge” that the Congress would not lift the debt ceiling before Oct. 15, why would that not allow me to profit in the market for Treasuries?

    Look, you may be thinking: Yes, but generally it’s going to be very hard to make money trading on secret information about Treasuries. True, but I think that just argues that, if you do want to use an “information sensitive” paradigm for understanding debt, the proper framing shouldn’t be “sensitive” versus “insensitive”, but rather a messier continuum. All traded debt is information sensitive to some degree, except for actual currency or FDIC-insured deposits.

    What I think is very dangerous about Gorton is that he’s constructed a theory that helps justify the opacity of AAA securitization tranches that are used like money. I think this is bad, very bad -- in terms of implications for a healthy market, for the propagation of risky regulatory arbitrage, for the growth of risky complexity, and so on. If you aren’t familiar with games played with securitizations, you might want to look here for starters:

  3. profitable for "who"? really? it's for whom!