The Power of Empirical Screens to Trigger Investigations: Evidence from the Libor-Alleged Conspiracy and Manipulation and Other Successes

By Rosa M. Abrantes-Metz on July 25, 2012

Since March 2011, worldwide investigations on the alleged conspiracy to manipulate the London Interbank Offered Rate (Libor) have been making news. Regulatory agencies across Europe, the United States, and Canada are investigating whether the major participating banks manipulated the Libor at an artificially low level after the eruption of the financial crisis, and also whether collusion among banks occurred prior to that. What assisted in triggering these investigations?

The Libor has been called the world’s most important number. It is a primary benchmark for global short-term interest rates and is used as the basis for settlement of interest rate contracts on many of the world’s major futures and options exchanges, as well as most over-the-counter and lending transactions. All told, hundreds of trillions of U.S. dollars in contracts, instruments, and transactions reference the Libor.[1]

If the banks did indeed manipulate the Libor (either individually or through coordinated behavior) it would have profound ramifications. With hundreds of trillions of dollars of securities and derivatives written referencing the Libor, even a small manipulation could potentially distort capital allocations all over the world.

But when and how did such suspicions get started, at least publicly? Arguably, the first major markers were articles by the Wall Street Journal (WSJ) in April and May of 2008 putting forward the possibility of the downward manipulation since January 2008. The WSJ’s analysis was then supplemented by my work with co-authors Albert Metz, Michael Kraten, and Gim Seow in an August 2008 paper in which we first put forward the possibility of collusion between banks starting significantly before the financial crisis. How did the WSJ and my work flag such possibilities? Both analyses used what are commonly known as screens.

Screens use data such as prices, costs, market shares, bids, transaction quotes, spreads, and volumes to identify patterns that are anomalous or highly improbable under ordinary competitive conditions. Screens can signal the possibility of cheating in a market or industry, mark when such cheating began, and identify who may be engaging in it. Screens are used by regulatory agencies, plaintiffs, and defendants worldwide. While they assist in the detection of alleged wrongdoing, they are equally effective in rebutting such allegations.

In the case of the Libor, in combination the WSJ’s screens and my own screens flagged as early as 2008 several items: (i) that the Libor was essentially constant day-in and day-out for a long period of months prior to the financial crisis; (ii) the identity of the underlying quotes by most of the participating banks; (iii) the unresponsiveness of the Libor to varying relevant benchmarks at least since 2006 and throughout August 8, 2007; (iv) the Libor’s unresponsiveness to changing market conditions particularly in late Spring and early Summer of 2007; and (v) the possibility of too low Libor after the beginning of the financial crisis starting in August 2007 when compared to other benchmarks for the cost of borrowing.

Though this has by far been the largest potential illegal behavior ever flagged by screens, (with the potential for over $30 billion in damages as reported this week by Crain’s magazine) , I should note that these methods have previously successfully detected illegal behavior in financial markets in the United States. including both an alleged conspiracy among dealers in the NASDAQ throughout the mid-1990’s, and stock options backdating and springloading cases from the mid-2000’s. I explained these two successful applications of screens in detection in my November 2010 article titled The Power of Screens to Trigger Investigations in which I also elaborated on the abnormal patterns I observed on the Libor and why they deserved a closer look, prior to public knowledge of investigations in this area. I commented on these successful uses of screens in a recent interview with Competition Policy International published last month.

Other successful applications of screens in detection have occurred in various markets and countries as well. Recent examples include the detection of bid-rigging in the Mexican pharmaceutical market  and of price-fixing in Brazilian gasoline markets.

Clearly, the power of these tools to flag illegal behavior has been further enhanced with the Libor case, and I predict that their use in detection as well as on the defense side will be expanding even faster in the near future (on the defensive use of screens, see “Conspiracy Screens: Practical Defense Perspectives).” I have certainly used screens as often to assist in establishing the unlikeliness of illegal behavior as I have used them to detect the possibility of such behavior. But can screens also play a role in preventing illegal behavior? I fully expect so and, as I noted in a previous Romy Beat, the Commodity Futures Trading Commission (CFTC) has required, in their settlement agreement, that these methods be used internally by Barclays.

I expect these events represent the beginning of a very significant expansion of the adoption of screens worldwide. For those interested in learning more about the multiple uses of screens, these are explained in detail in my 2010 article with Professor Patrick Bajari titled Screens for Conspiracies and Their Multiple Applications

[1] The Libor is supposed to measure the rate at which large banks can borrow unsecured funds from other banks at various short-term maturities in various currencies. The U.S. dollar-denominated Libor, for example, is set as follows: On a daily basis the 16 participating banks are surveyed by the British Bankers Association and submit sealed quotes which answer [a]t what rate could you borrow funds, were you to do so by asking for and then accepting inter-bank offers in a reasonable market size just prior to 11:00 a.m. London time? The Libor is then computed by averaging over the middle eight quotes, discarding the four highest and the four lowest. After the Libor is computed, all quotes become public.