performance of France and particularly Australia improves while that of Netherlands and Ireland
deteriorates. To provide greater insight into the differences in debt management across countries
Table 4 shows summary statistics. Norway tends to score well across our performance indicators
but in terms of the proportion of fixed rate debt issued and how much was short term and how
much longer term its portfolio structure is almost exactly the average across all countries. The two
areas where Norway is distinctive is the amount of foreign currency debt it issues and the average
maturity of its debt, where only the UK issues longer average maturity. In addition to these features
of its debt structure Norway also stands out because throughout the period it tends to run fiscal
surpluses and reduces it level of debt. Belgium and the Netherlands tend to be assessed poorly by
our indicators and from Table 4 their debt structure stands out in having above average issuance
of fixed rate debt and more long term debt but despite this the overall debt structure has below
average maturity - suggesting larger issuance of medium term debt.
While Table 4 offers some insights into differences in debt structure and how this may account
for the differences in Table 3 we now consider more formally whether variations in achieved fiscal in-
surance is linked to cross country differences in debt composition and macroeconomic performance.
We do this by estimating a relationship of the form :
Pi = a,Xi + βfZi + Ui (6)
where Pi denotes one of our performance indicators for debt management, Xi denotes a vector
of macroeconomic variables (e.g. the average level of debt or deficit in a country, inflation or
GDP growth) and Zi a vector of variables describing the portfolio composition of country i (e.g.
proportion of indexed debt, proportion of variable rate debt etc.). Finding significant variables in
Zi is critical if we are to make recommendations for debt management. In estimating (6) we face
severe data limitations. For reliable inference our performance measures need to be identified over
our whole sample and so we cannot use a time series approach in estimating (6). This problem is
reinforced by the fact that many of our key variables.such as maturity of debt, do not change much
over time. As a consequence we are reliant on cross sectional variation to identify the determinants
of debt management. We therefore estimate the equation in cross sectional form. Further, we
only have data for 12 OECD countries so we cannot estimate this equation whilst simultaneously
20