CMStatistics 2017: Start Registration
View Submission - CFE
Title: Residual based detection of market dislocations Authors:  Leopold Soegner - Institute for Advanced Studies (Austria) [presenting]
Martin Wagner - University of Klagenfurt (Austria)
Abstract: Multivariate monitoring procedures are developed to detect market dislocations, e.g., in covered or uncovered interest parities. The simultaneous consideration of multiple cointegration relationships, in the above example corresponding to different maturities, is intended to increase the monitoring performance compared to univariate approaches. The procedures are based on monitoring variants of well-known non-parametric cointegration rank statistics. In case the parameters of the considered relationship have to be estimated rather than assumed to be known from theoretical arguments, modified least squares estimation techniques need to be used for asymptotically nuisance parameter free limiting distributions of the monitoring procedures. The different monitoring procedures are compared by means of simulations and applied to interest rate parity data for eight countries.