Modelling the Yields on Australian Coupon Paying Bonds

Before coupon bond data can be used to make term structure inferences it must be adjusted to account for the coupon effect. This paper compares the performance of two alternative adjustment methods, namely a duration based adjustment of term and a zero coupon method that adjusts yields. The comparison of the two adjustment methods was carried out on unique data set on Australian traded bonds. The duration adjustment method models coupon bond yields as a function of duration and convexity. The estimation results indicated that despite the apparent attractiveness of high convexity to bond holders that convexity was not a significant determinant of yields. In making the zero coupon adjustment we employed a simple polynomial functional form for the zero coupon yield curve. This method, not only provided a good fit to the data, but also produced a plausible implied zero coupon forward rate curve and a zero coupon volatility structure. It was concluded that while the zero coupon method of adjusting coupon rates may be slightly more complex to implement, it was preferred because, it provided a slightly better estimate bond prices and yields, and as expected, it gave a very much better estimate of spot zero coupon rates that in tum could be used to provide zero coupon forward rate and volatility estimates.