Do two Libor reforms reduce the effect of incentives on submitted rates?

Abstract

Libor is a set of survey-based reference interest rates for an estimated $350 trillion in financial instruments. In 2013, administrators adopted several reforms designed to improve Libor. Two key reforms were (1) eliminate Libor collection for unpopular currencies and maturities, and (2) defer the public release of individual bank Libor submissions for 90 days. The intent of the first reform is to reduce bias and subjectivity by increasing the probability that an actual transaction is underpinning a Libor submission. The intent of the second reform is to reduce public signaling bias in the Libor submission; however, the period of anonymity granted in this reform also has the potential to increase bias from other directional incentives. Using an experiment, I examine whether estimates are less influenced by directional incentives when a closely-matching transaction is in the underlying dataset and whether a period of anonymity allows more influence of directional incentives. I collect data for two dependent variables but find statistically significant results only using the second. I find that the influence of directional incentives decreases when a closely-matching transaction is in the underlying dataset but that it increases when participants have a period of anonymity. These findings lend support to the first key reform but show Libor administrators an unintended consequence of the second key reform.