INTEREST-RATE RISK AND THE TERM STRUCTURE OF INTEREST RATES.

WHILE there have been frequent attempts to explain yield-maturity relationships on loans differing only in length of time to maturity, known as the term structure of interest rates, considerable difficulty has been encountered in formulating generally acceptable empirical tests in order to reconcile workable theory with observed behavior in the bond market. Until the work of David Meiselman, comprehensive empirical testing of interest-rate expectations was lacking.[2] Meiselman developed an error-learning model and, testing it with David Durand's basic corporate-yield data for the 1900-54 period, concluded, that interest-rate expectations are revised whenever previously held expectations prove to be in error. Moreover, the evidence was said to support the unbiased expectational theory.[3] In this paper the term structure of interest rates is analyzed empirically, using monthly Treasury yield-curve data that appeared in the Treasury Bulletin for the 1954-63 period. Using Meiselman's error-learning model, I found that a significant portion of the movements in forward rates of interest for the Treasury data under review can be explained by errors in forecasting the one-year actual rate of interest. In contrast to Meiselman's results, however, the constant-term values in the regression studies are consistently positive and significantly different from zero, an occurrence suggesting that the explanation of forward rates can be improved by introduction of a risk variable. Section II of this paper develops a multiple-regression model in which the deviation of actual forward-rate level from accustomed level is employed as an additional independent variable. Results of multiple-regression studies are seen to be consistent with risk varying inversely with the level of interest rates relative to an accustomed interest-rate range. [ABSTRACT FROM AUTHOR]