ModelsofTimeVaryingExpectedPriceChangeand.docVIP

Models of Conditional Variance for Bond Prices by Naren Pappu and S. Lakshmivarahan George Lynn Cross Research Professor School of Computer Science University of Oklahoma Norman, Oklahoma 73019 and Duane R. Stock* Price Investments Professor 205A Adams Hall Price College of Business University of Oklahoma Norman, Oklahoma 73019 Fax: 405.325.7688 Email: dstock@ March 1, 2005 The latter author wishes to thank the Price College of Business and Bank of Oklahoma for support. Also, Shengguang Qian provided valuable research assistance. *Please send all correspondence to the last listed author.  Abstract Evidence of predictability of financial time series along with recent advances in modeling time dependent conditional variance clearly call for incorporating such theory into expressions for expected bond price changes and variance of price changes. Thus, we derive new first and second order expressions for conditional variance of price change based on these advances. Among other things, we express conditional (expected) variance of price changes as a combination of innovations in simultaneously estimated expected change in yield and conditional variance of yield change. Also, we examine how conditional variance is affected by duration, yield volatility, expected change in yield, and convexity. Portfolio managers can use the results to enhance expected performance analysis. For example, in contrast to many theoretical papers, we find that variance of price changes depends on bond convexity. More specifically, convexity may either increase or decrease conditional variance of price change dependent upon time specific conditions. Bond portfolio managers should thus reassess the role convexity plays in their portfolio choices. That is, they may wish to be compensated for positive convexity in terms of greater expected return or required yield. Introduction The potential distribution of price changes from any investment is extremely important to an