A Simple Timing Strategy for Canadian Fixed Income Portfolios

Abstract

Abstract Fixed income return enhancement can be obtained in the Canadian marketplace when one conditions on past stock market movements. A simple timing rule is demonstrated to increase holding period return significantly. It is likely that this strategy is capitalizing on increases in ex ante return due to higher levels of risk aversion induced by lower societal wealth. Nevertheless, although these opportunities were available to all, it is shown that Canadian fixed income portfolio managers failed to seize them. Resume On peu ameliorer le rendement des instruments a taux fixe sur le marche canadien en se basant sur les mouvements passes du marche boursier II est prouve que le simple choix du bon moment peut augmenter de fa,con significative le rendement enregistre au cours de la periode d'investissement. Cette strategie tire probablement parti des augmentations du rendement ex ante imputable i des niveaux eleves d'aversion pour le risque induits par une richesse sociale amoindrie. Toutefois, bien que tous aient eu acces rt ces possibilites, l'etude revele que les gestionnaires canadiens de portefeuille d'instruments a taux fixe n'en ont pas profite. Portfolio managers are forever seeking ways to enhance return. For active fixed income managers this amounts in large part to searching for tools that allow one to occasionally time interest rate movements and make anticipatory duration-switching adjustments. Nevertheless, conventional wisdom and the empirical evidence tell us that interest rate forecasting is a very difficult undertaking.1 Nevertheless, the unpredictability of interest rate changes does not necessarily close the door on return enhancement through holding-period-return prediction. To see this most simply, suppose we have only 1-year and 2-year zero-coupon bonds. Further assume that the best prediction for the future 1-year interest rate is always the current 1-year interest rate. If managers wish to maximize expected return on their portfolio over a 1year holding period, the correct strategy will always be to buy two-period bonds when the yield curve is upwardsloping, but to otherwise stay short. Of course, in this hypothetical world, the long-short spread is entirely accounted for by what is called the term premiunm. Its two principal definitions, which coalesce in this twoperiod example, are a) the gap between the implicit forward interest rate, as calculated from the current yield curve and the corresponding expected future spot rate, and b) the expected holding period return of a bond net of the rate whose length corresponds to the measurement horizon.2 In reality, the slope of the term structure is also partly the result of anticipated changes in yields. Still, if one could estimate term premia (however defined), scope would be indeed opened up for duration-switching techniques. A technique, based on what researchers have recently learned about the nature of term premia and what factors have an impact on their magnitude, is tested here. This is at the heart of current term structure research. The expectations model of the term structure of interest rates, which argues that the term premia embedded in the yield curve are time-invariant, has now been discredited by abundant evidence. It is apparent that premia are positively related to the long-short spread, various sources of uncertainty (whether in the bond market or throughout the economy), and the current interest rate level (though the record is far from perfect on the latter point).3 Most striking has been the recent finding that holding period bond returns net of the short-term interest rate (hereafter excess returns) are related to lagged movements in the aggregate stock market.4 Ilmanen (1995) calls the ratio of the past level of the stock market to today's level, which is just the inverse of the most recent gross aggregate stock market return (ignoring dividends), inverse relative wealth. …