Are interest rate fixings fixed? an analysis of libor and euribor

Market reference interest rates such as Libor and Euribor play an important role in many financial contracts around the world. The integrity of these instruments, and of the markets, themselves, depends crucial on the confidence that market participants place in the reliability and veracity of these rates. Unfortunately, developments in London and in other financial centers have shaken this confidence, due to widespread allegations of manipulation. While prosecutors are engaged in taking action against the purported manipulators, regulators, including the Bank of England and the European Central Bank, are grappling with the issue of how to reform the rate–setting process, without creating too much confusion about the nature of the contracts, or inducing potential litigation among contracting parties that use these rates as benchmarks in their contracts. We believe that our analysis provides useful additional findings for this reform.

In this paper, we quantify and explore the incentives and potential effects of manipulations for Libor and Euribor, in detail. The focus of our study is on the analysis of the individual submissions of the panel banks for the calculations of the respective benchmark rates for the time period January 2005 to June 2012. We present results for the AUD Libor, USD Libor and Euribor. In our analysis, we explicitly take into account the possibility of collusion between several market participants. Furthermore, our setup allows us to quantify potential manipulation effects for the actual rate-setting process in place at present, and compare it to several alternative rate-fixing procedures. Moreover, we can directly analyze the effect of the panel size and the underlying methodology for eliciting rate submissions on manipulation outcomes.

Our results show that the cross-sectional volatility of individual submissions is high, i.e., the screening for manipulation is hindered by noise in the data. In line with the related literature, we find that, in this case, the detection of concrete manipulations by particular banks based solely on their submissions is almost impossible to prove. Thus, in our main research question, we focus on the underlying incentive for manipulation. We quantify the potential effects of manipulations on the final fixing for different benchmark rates and rate-setting procedures by considering simultaneous manipulation attempts by up to three banks. Overall, we find that the panel size and the calculation method significantly influences such incentives, i.e., a large panel size and the use of the median of the submissions (instead of trimmed mean approaches) substainally reduce the effects of individual banks on the final rate. Although a change in the calculation methodology could be implemented fairly easily, the increase in the panel size for the Libor rates in its current setup could be more difficult. Given that banks are explicitly asked about their own funding rate for Libor, enlarging the sample might introduce even more heterogeneity, in terms of credit and liquidity, across panel banks. Thus, increasing the sample size might only be feasible when asking about the money market funding costs of a (hypothetical) prime bank as in the case of the Euribor.