in the above sections, we apply a maximum-likelihood estimation procedure for a combined
autoregressive model and a GARCH (1,1) model.
***Insert Table 6 here***
Panel A of Table 6 reports that yield spreads are negatively related to both the inflation
rate and the real government yield for all callable indices. It is interesting to note that the
coefficients of both explanatory variables are very similar in magnitude. We interpret this as
evidence that the influence of riskless term structure on yield spreads is due to mainly
nominal riskless rates. Based on a Wald test we fail to reject the hypothesis that β1 = β2. Thus,
we can rewrite regression (5) as:
∆S =β0 +β1(∆π+∆YLT,R)+ε (7)
where (∆π + ∆YLT,R) is the monthly change in nominal government yield. Thus, these results
are in line with the fact that the moneyness of the call provision is based on the level of
nominal rates rather than real rates, which will induce a negative relationship between the
yield premium attributed to callability and the nominal riskless rate. Not surprisingly, for the
economically noncallable BBB-rated bonds during the 01:1995-07:2001 period the
relationship is insignificant with both the inflation rate and the real rate (Panel B of Table 6).
Note that for the same period, for the economically callable AA- and A-rated bonds the
relationship is significantly negative with both variables.
5.6. Collin-Dufresne et al.’s (2001) Model with Real Rates
To check that the above analysis with the real interest rate is not affected by model
specification, we test the most comprehensive regression model - Collin-Dufresne et al.’s
(2001) - with real government rates. Table 7 outlines the results of this estimation. In Panel A,
which reports the results for the entire sample, the only significant factors are the term
structure slope changes and the stock index return. The interest rate factor is not significant
24