when real rates are used. This is in line with the expectation of a negative relationship
between the yield premium attributed to callability and the nominal riskless rate rather than
the real rate. Panel B of Table 7 shows similar results for the 01:1995-07:2001 period.
***Insert Table 7 here***
We use the difference between the nominal interest rate and inflation rate as our measure
for real yield in the above regressions16. We find that once the real interest rate is used the
relationship between the bond yield spread and riskless interest rate becomes insignificant
even for AA and A indices, which are mainly composed of callable bonds. This result shows
that once the relationship between callability and the nominal interest rate is factored out, no
relationship remains between the interest rate and credit spread.
To further test the robustness of our results, we use the return of Real Return Bonds
(RRBs) issued by Government of Canada since 1991. This is a better proxy for the real
yield.17 We run the most comprehensive regression model - Collin-Dufresne et al.’s (2001) -
with rate of return on RRBs. We report both OLS and AR-GARCH results in Table 8. Once
again, the results are very similar to those of Table 7.
***Insert Table 8 here***
To sum up, irrespective of the model tested, namely the Longstaff and Schwartz (1995)
two-factor model, Duffee’s (1998) model, or Collin-Dufresne et al.’s (2001) comprehensive
model, we find evidence the negative relationship between the yield spread and the
government yield is due mainly to the effects of the call provision. Once the impact of the
call option is removed, i.e. when considering BBB-rated economically noncallable bonds,
the yield spread - government yield relation is not significant. When we use real government
yields, we find that the yield spread - riskless rate relationship is insignificant for both
25