When the Chicago Mercantile Exchange launched its Eurodollar futures contract in 1981, it devised a nearly foolproof way to calculate the London interbank offered rate.
The CME would select 20 banks at random from a much larger pool and survey those banks for their perception of interest rates. It would discard a subset of the highest and lowest and average the responses that remained. At a randomly selected time within the next 90 minutes, the CME would conduct a second survey with a fresh random set of banks. Finally, it would publish the average of the results of the two surveys without revealing the identities of the banks that participated.
This is in sharp contrast to the now-infamous procedure of the British Bankers' Association which uses a smaller sample of the same banks every day and displays their rates for the world to see.
In September 1996, the CME applied to the U.S. Commodity Futures Trading Commission for permission to switch from its own, sensible Libor calculation to the BBA's. Presumably, the CME worried that a rival exchange would steal market share by starting a contract tied to BBA Libor, which was and is the standard floating rate index for swaps, loans and notes.
On Nov. 18, 1996, I filed a public comment letter with the CFTC objecting to the CME's plan. Here are some excerpts from that letter:
"In a severe funding crisis … banks might respond to the BBA survey with a rate substantially below their true lending rate. Since the BBA survey results appear on Telerate [a Bloomberg-like information service widely used at the time], the banks may want to hide the extent of their troubles…The CME randomly draws a limited number of reference banks from a larger pool on the futures settlement date. This ensures that banks will not know ahead of time whether they will be able to participate in the survey. The BBA, on the other hand, surveys the same sixteen banks every day. To see how a bank could exploit this feature of the BBA survey to manipulate the index, suppose that the four high quotes are expected to be the three Japanese banks (Bank of Tokyo/Mitsubishi, Fuji, and Sumitomo Trust) and the Bank of China. Then, any of the remaining banks can raise their quote, and the effect will flow directly to the final index. A British bank, for example, might raise its quote by 12.5 basis points. Since eight banks make up the average, the average will rise by 12.5/8 = 1.5626 basis points. If two banks worked together, they could raise the average by 3 basis points.
“The CME claims the BBA survey will self-correct if markets become more volatile. They argue that ‘[t]he outstanding notional value of instruments tied to the BBA fixings is enormous…’ But enormous markets create enormous temptations. The CME argument works only to the extent that we rely on BBA members to look beyond their self-interests. Again, without impugning any of the BBA banks, we do not consider this to be a sound basis for predicting human behavior.”
That was more than 15 years ago. No one should be surprised that banks would suppress their posted rates in a funding crisis or that they might manipulate the survey for gain. It was easy to see this coming.
The CFTC ignored my letter, sent on the stationery of the derivatives subsidiary of Japan's largest bank, as well as a similar letter from Salomon Brothers. (They are available today on request from the CFTC.) The CME converted the gigantic Eurodollar futures market to BBA Libor starting in January 1997 and crystalized the dominance of the BBA. So while the CFTC is dishing out nine-figure fines to banks caught up in the scandal, it might want to consider another possible culprit: itself.