that the levels of the two different debt measures are strongly correlated and that they share the
same order of integration (see the Unit root tests). The two series are also very strongly correlated
in differences which suggests that fluctuations in bond prices offer limited fiscal insurance. Both
debt series show a strong negative correlation in first differences between debt and GDP e.g. debt
changes behave counter cyclically. The correlation between changes in the debt∕GDP ratio and
the primary deficit are invariably positive. That the outstanding value of debt should increase
with the primary deficit is to be expected but the fact that the market value does as well suggests
limited fiscal insurance. However, in the majority of cases the market value of debt shows a weaker
correlation with the primary deficit than the outstanding value of debt suggesting at least some
fiscal insurance has been achieved.
6 Determinants of OECD Debt Management Performance
Table 3 shows estimates of our preferred measures10 of fiscal insurance. While there is clearly
variation across countries the general message of Table 3 confirms the findings of Marcet and
Scott (2004) based on US data - the behaviour of OECD debt and deficits is not consistent with
the complete market outcome. Examining the relative persistence of debt compared to deficits
and the impact effect of deficit shocks shows no evidence at all in favour of complete market
outcomes - debt displays substantially more persistence than deficits. However, the main point of
Table 3 is not to adjudicate between complete and incomplete markets but to assess the relative
performance of debt management across countries. Focusing on the relative persistence measures
Φifc and Φ2fe we find that at all horizons Norway performs well whereas Belgium and Netherlands
tend to be the worst performers. At horizons of a year we also find evidence of relatively good
performance in Germany, Ireland, Australia and Austria although performance in these countries
rapidly deteriorates. From the impact measure there is once more evidence of good performance in
Norway and Australia and also interestingly Netherlands. Belgium once more shows relative poor
performance as do Austria and UK. The relative dynamic measure Φ* again confirms the good
performance of Norway and the bad performance of Belgium and suggests that at longer lags the
10Our results are unaffected if we also include all 7 performance measures.
19