The Li(e)for investigation is metastasizing at an incredible place. Both Bloomberg and FT have run long pieces on the problems with the rate-setting mechanism, and how to fix them. Investigations (10, according to FT) are proceeding around the world, in the US, UK, Europe, Japan, Canada, and Singapore. There is a criminal investigation in the US. Numerous banks are scurrying to cooperate in the hopes of receiving leniency-the prisoner’s dilemma in action.
Some of the reporting is confusing. There are at least two different motives for the alleged misstatements, one likely being more legally fraught for the banks. The first motive is that during the throes of the crisis, banks had an incentive to misreport the rates at which they could borrow unsecured in order to avoid the perception that they were on the brink. The other motive is that traders wanted to influence LIBOR in order to advantage derivatives positions tied to LIBOR-the latter likely presents the banks and individual traders with far more acute legal problems.
Then there remains the issue of what to do about it. The most obvious remedy is to base LIBOR on actually borrowing transactions, in the same way as is done with EONIA and SONIA overnight (hence the “ON”) borrowings. This is a reprise of the data hub idea I advocated for energy trades back in 2003(!).
There is pushback against this idea, however. One argument is that particularly during times of stress only a few banks are actually able to borrow unsecured for 3, 6, or 12 months, so the rate would be based on transactions from a small set of institutions.
I don’t get this objection, frankly. So it’s better to base the most important reference rates in the world on what banks that cannot borrow unsecured say they could borrow at, than to base it on the rates that banks that actually do borrow unsecured pay? Really? The first alternative is an oxymoron. The second might result in rates from only a few banks being used, but far, far better to base an index that is supposed to reflect the actual cost of unsecured borrowing (by those able to do it) as represented by real transactions then to use the oxymoronic fantasy rates of those who can’t do it.
And if nobody can borrow unsecured in any volume-that’s important information in and of itself. That raises the question of the advisability of basing so many derivatives contracts on something that has an appreciable probability of not existing from time to time. Better to recognize reality than MSU and pretend it’s real.
Here’s a thought: BBA could continue to report a LIBOR based on the traditional methodology, but just modify its name to LIBORGIGO. Truth in advertising, dontcha know.
This raises a broader point. Cash settlement is often raised as a panacea for contract design. Delivery-settled contracts have problems-so go to cash settlement! But as I’ve argued since 1990, in many markets cash settlement is problematic due to the opacity of cash markets, and/or the absence of cash market transactions. Cash settlement works where there is a liquid and vibrant and transparent cash market (e.g., S&P500 stock index futures) or centralized collection of cash transactions prices (e.g., live hogs). Everywhere else, problematic indeed.
LIBOR is the most pronounced example of that today. Natural gas was so in the late-90s and early-00s. Right now, the Platts’ methods for oil price assessments are under considerable scrutiny. Market participants have always been skeptical (though they often mute their criticism given Platts’s heft), but IOSCO has issued a very critical and questioning appraisal, and raises the question of whether further regulation is necessary.
It’s deja vu all over again, folks. Been there, done that. Went through all these issues with the energy data hub in ’03. I can show you the scars. I can tell you some stories (e.g., the trade group guy who asked me what it would take to get me to back off with my proposal.)
The fundamental problems are that cash settlement is hard if there aren’t, well, a lot of cash transactions going on. Even if there are a decent number of cash trades, that doesn’t help if people don’t report. Given the public good nature of more informative price indices, there are under-incentives to report. Indeed, it can be worse than that: reporting can be a private bad.
This raises a justification for mandatory reporting, and that is actually one of the defensible parts of Frankendodd. That can’t fix the problem of a lack of transactions, but it can make the most of the transactions that are there.
But this also points out the value of delivery-settled derivatives. The ability to make delivery gives market participants the opportunity to put their money where their mouths are in a way that is impossible with cash settlement, especially cash settlement based on self-reported prices and assessments. Even if delivery doesn’t occur all that much, the ability to make delivery keeps prices honest (noting, of course, that the delivery mechanism can be abused, but further noting that’s a problem that can be addressed by appropriate anti-corner measures.)