Investigating Long Memory in Yield Spreads

Abstract

This article investigates the presence of long memory in bond yield spreads, looking at the spread between Moody’s Baa and Aaa corporate bond yields as well as the spread between each of these series and the 10-year Treasury bond yield. Using the aggregated series and wavelet OLS estimator methods, the authors test the hypothesis that yield spreads follow a fractionally integrated process. We find significant evidence of the presence of long memory in these spreads, which suggests that a time element exists and should be accounted for when analyzing yield spreads, making investment allocation decisions, or using yield spreads as forecasting tools.

TOPICS: Fixed-income portfolio management, big data/machine learning, analysis of individual factors/risk premia