Statistical evidence about LIBOR manipulation: A "Sherlock Holmes" investigation

This paper contributes to the crucial problem of LIBOR malfunctioning due to its manipulation by banks, a phenomenon described clearly in the FSA Inquiry Report published in September 2012. After applying classical tests of non-stationarity to a series of participating banks' LIBOR quotes, we detected some significant breaks in the data that correspond to significant economic events, namely Lehman Brothers' bankruptcy or central banks' decisions. Our conclusion is that a possible manipulation, confirmed ex post in the FSA report, can be deduced throughout the crisis period as a result of the behavior of the coefficients in a linear three-regime Threshold Regression model. Finally, we applied an original procedure to detect those banks most likely to build cartels – or at least homogeneous groups – during the important period of turmoil under review, namely August 2007 through November 2012. This method, called hierarchical clustering analysis, helped us to precise the conclusions of the abovementioned FSA inquiry.