A Mechanism for LIBOR

For years, traders colluded to rig LIBOR, the rate at which banks lend to each other.

The system for setting LIBOR relied on banks to tell the truth but encouraged them to lie. Is there a better way?

The London Interbank Offered Rate (LIBOR) is supposed to represent the average interest rate that banks use to lend to and borrow from one another. The rate is then used as a benchmark for pricing a number of financial products.

The ongoing LIBOR scandal has revealed that the rate is vulnerable to manipulation by the major banks so that they can profit from their exposure to the benchmark. Manipulating one of the rates by even a fraction of a basis point can bring substantial gains: the market for derivative and loan products that use LIBOR rates has been estimated at more than $300 trillion.

While banks have already been fined over $9 billion for inappropriate submissions and repeated attempts to manipulate LIBOR, this is unlikely to stop them. The underlying process that determines LIBOR needs to be fixed.

In this paper, Joel Shapiro and his co-authors propose a new mechanism for LIBOR that reduces the possibility of manipulation and produces an unbiased estimate of the true rate.

In their model, LIBOR is set daily at 11am, as it is now. But instead of the current method by which the administrator asks representatives of a panel of banks to estimate the rate at which they could borrow from each other, the rate is based on actual transactions conducted by the panel within the previous 23 hours and 59 minutes. A transaction consists of the bank borrowing an amount at a fixed rate from another bank in the interbank market; this does not need to be a LIBOR panel bank. The transaction rates are submitted automatically, and LIBOR is calculated by the administrator as a simple average.

This system is still open to manipulation, however, as banks can set whatever rate they like in the transactions. In order to minimise this, the model developed by Shapiro et al allows the administrator to set fines for transactions that appear to be manipulated.

Manipulation is detected using a comparison rate that consists of the set of transaction rates, along with the results of a ‘revealed preference algorithm’ (RPA) run immediately after the LIBOR calculation is made. The RPA requests from the banks the rate at which they would lend to a given bank. As the LIBOR calculation does not include these RPA rates, banks cannot influence LIBOR by their actions in the algorithm. In addition, the RPA includes a mechanism to ensure truthful reporting.

The proposed mechanism has two additional benefits. The first is that the revealed preference algorithm allows the LIBOR mechanism to continue operating even if there are no transactions – something that can happen in illiquid markets. Secondly, it can break the vicious cycle that started with the decline of panel-sizes for benchmarks – caused partly by banks becoming more aware of reputational risks and partly by the drop in the number of transactions in less liquid markets, which reduces incentives to participate on benchmark panels. A more accurate benchmark will improve liquidity and incentivise greater participation in the benchmark panel.


A mechanism for LIBOR is co-authored by Brian Coulter, Joel Shapiro and Peter Zimmerman, and is published in Review of Finance.