Most of the available monthly interest data series consist of monthlyaverages of daily observations. It is well-known that this averaging introduces spurious autocorrelation effectsin the first differences of the series. It isexactly this differenced series we are interested in when estimatinginterest rate risk exposures e.g. This paperpresents a method to filter this autocorrelation component from theaveraged series. In addition we investigate thepotential effect of averaging on duration analysis, viz. whenestimating the relationship between interest rates andfinancial market variables like equity or bond prices. In contrast tointerest rates the latter price series are readilyavailable in ultimo month form. We find that combining monthlyreturns on market variables with changes inaveraged interest rates leads to serious biases in estimatedcorrelations (R2s), regression coefficients (durations)and their significance (t-statistics). Our theoretical findings areconfirmed by an empirical investigation of USinterest rates and their relationship with US equities (S&P 500Index).
interest rates duration averaging time series properties spurious autocorrelation