including a broad range of macroeconomic variables as well as the debt composition indicators
at our disposal. Therefore we follow a sequential approach in Tables 4 to 6 and show results
from regressing Pi first on macro variables, then separately on a set of variables capturing broad
characteristics of the debt portfolio and then finally on detailed statistics on the debt composition11.
Table 5 shows the results of regressing our performance indicators on the average maturity
of debt, the average debt and deficit to GDP ratio, real GDP growth and inflation. The point
estimates suggest that issuing longer maturity debt helps improve fiscal insurance although the
effect is nowhere significant. Similarly large values of debt tend to worsen fiscal insurance (although
again not at any conventional significant levels) whereas surprisingly a large primary deficit leads
to better performance and this latter effect sometimes is of borderline significance. The results
regarding inflation are mixed in terms of the sign of the impact but always statistically insignificant
which is the same as for GDP growth although here the point estimates suggest faster growth helps
improve fiscal insurance.
Table 6 shows results of regressing our performance indicators on average maturity and the
proportion of fixed rate, foreign currency denominated and indexed debt. Once again the regressions
reveal few statistically significant effects. The point estimates suggest that foreign currency debt
may contribute to fiscal insurance but the only statistically significant finding is that higher levels of
indexed debt improve fiscal insurance. Table 7 adds the proportion of short and long run debt and
continues to find few results of significance, there is some very weak evidence that longer term debt
and indexed debt helps achieve better fiscal insurance but the explanatory power of the equations
are very poor. Similar results were found irrespective of the combination of variables used in the
regression
Summarizing these results it seems there is very little relationship between observed cross
country differences in debt structure and the degree of fiscal insurance achieved. Only occasionally
are any variables significant and even then not in a consistent manner and the R2 from all regressions
is invariably very poor. Whether the government issues short, medium or long; fixed, variable or
11Some governments have made increasing use of derivatives in manging the debt. The consequence is that the
reported debt statistics may not capture the effective maturity structure of debt. This may contribute to the weak
statistical evidence we find although in the absence of reliable data it is impossible to comment on the importance of
this bias.
21