A golden rule of market practice and regulation should surely be never to trust prices – and certainly never to encourage actual transactions using such prices – when they are created not by supply and demand in competitive markets but according to what self-interested parties would like others to believe them to be.
It's hard to imagine a more clear-cut and widespread violation of this rule than the rise of Libor as a global interest rate benchmark.
That big banks have an incentive to manipulate Libor under certain conditions should be obvious – in times of market stress, a bank's reported Libor rate can affect not only its borrowing costs but its solvency.
And the evidence surely bears this out. Consider September 2008, as Lehman and AIG hurtled towards disaster.
The chart below compares Libor to an alternative benchmark established by ICAP: NYFR, or the New York Funding Rate. The operative difference between Libor and NYFR is that the latter is based on anonymous reports from major banks. In normal times, the two rates are highly correlated. September 2008 was not a normal time.
During that month, a bank revealing publicly that it could only borrow at elevated rates naturally put itself at risk of suffering a bank run or lending halt. It should not be surprising, therefore, that banks would be less honest about their rate reports when their names were attached to them – as they were with Libor.
Libor therefore drops well below NYFR, suggesting clear and deliberate underreporting. This is consistent with the admission of a Barclays official to the New York Fed a few months earlier "So, we know that we're not posting an honest Libor…we are doing it, because if we didn't do it, it draws unwanted attention on ourselves."
Perhaps paradoxically, Libor as a benchmark can be damaging even when the bank rate reports that determine it are wholly accurate.
Central bankers necessarily spend a great deal of time studying market data that they believe to be forward-looking indicators of the economy's health, and one such is the Libor-OIS spread. OIS, the Overnight Indexed Swap rate, differs from Libor in a significant way.
Libor, being the rate at which major banks can (supposedly) borrow from each other unsecured by collateral for three months, contains a default risk component which is absent in OIS (which merely reflects market expectations of the overnight unsecured rate over a three-month period). Libor is therefore generally higher than OIS.
European Central bank board member Benoît Cœuré, echoing thoughts expressed by Alan Greenspan and others in the past, recently referred to the gap between Libor and OIS as "a standard measure of tensions in unsecured markets". It goes up when such tensions go up, and down when such tensions go down.
The Libor-OIS spread can be low, however, even when banks are in appalling financial health. How is this possible?
The problem starts when official bodies start tracking such a measure to determine whether they need to do something. The reason is that when statistical measures are targeted for policy purposes they tend to lose the information content that recommended them for that role in the first place.
This common pitfall in economic policymaking has been termed "Goodhart's Law," having been first articulated by British economist and former Bank of England Monetary Policy Committee member Charles Goodhart in 1975.
Could Goodhart's Law be at work with the Libor-OIS spread? It would appear so.
In March this year, after the ECB ended its long-term refinancing operations, which provided banks with over €1 trillion in three-year 1%-interest loans, the Libor-OIS spread continued the downward trend it started on after the programme was launched last December – as shown in the chart below.
By Cœuré's logic, this indicated that LTRO had succeeded in addressing earlier worries about the ability of major European banks to attract vital funding.
But look at what happens to the price of five-year credit default swaps on the members of the Libor bank panel after LTRO ends – it soars. A huge, and highly unusual, gap opens up between the CDS price and the Libor-OIS spread.
(The insert to the chart shows that CDS prices and the Libor-OIS spread were highly correlated over the two years to the start of LTRO, but that this relationship collapses thereafter.)
This indicates that whereas banks are happy to lend to each other for three months, given that they're now awash with ECB cash, the end of LTRO combined with a renewed deterioration of Spanish and Italian sovereign bond prices led to rapidly revived fears of bank defaults within five years.
In other words, the LTRO policy intervention significantly reduced the information content of the Libor-OIS spread. Bank CDS prices are now a more valid indicator of the health of the eurozone banking system – which is poor and deteriorating.
The big picture on Libor is one of flawed polling data used inappropriately to create real and potentially damaging economic distortions. It's time to bid Libor good riddance.
Benn Steil is director of international economics at the Council on Foreign Relations and co-winner of the 2010 Hayek Book Prize. Dinah Walker is an analyst at the Council.
This article appears in full on CFR.org by permission of its original publisher. It was originally available here (Subscription required).