indicator essentially combines information in Ψ∙2∕.∙ and Im and so not surprisingly performs well.
Notice that in the case of persistent shocks under incomplete markets Φj is initially extremely
negative and then increases slowly but for all lags and also in the case of i.i.d shocks the statistic
discriminates clearly between complete and incomplete markets, as can be seen by looking at the
quartile values.
Summarizing across these measures we have that ɑʌ^v * ∕σω* is a poor indicator of Hscal in-
surance; pω*jβ* and σ^*∕σω* perform better in the sense that for a given simulation they can
discriminate between complete and incomplete markets but comparing across all three simulations
suggests they too are unreliable measures. By contrast Φ2fc5 Im and Φj prove themselves to be
reliable indicators in an absolute sense of the difference between complete and incomplete markets.
As a consequence it is these measures we will focus on in our empirical analysis.
5 Data
Key to our concept of fiscal insurance are fluctuations in the market value of debt. This creates
an empirical problem as official published data on government debt record the outstanding value
of debt to be repaid. To construct market value data we use the approximation suggested by
Butkiewicz (1983), namely :
ИЦ ≈ Bt ' '
t 11 + ntVt
where Bt is the outstanding value of government debt, n is the average maturity of debt, c the
average coupon rate and y the average redemption yield on government debt. A data appendix
lists how we use this approximation across OECD countries but the majority of data is taken from
the OECD’s Central Government Debt Statistical Yearbook and Missale (1999) with the yield data
taken from Global Financial Data. For Australia, Germany, Italy, Netherlands, UK and US we are
able to construct a series for 1970-2000, for Canada and Belgium from 1976, for Ireland from 1977,
Austria from 1981 and Norway from 1984. All variables are deflated using the GDP deflator.
Table 2 reports a number of statistics to help gauge the differences between our estimates of the
market value of debt and the official outstanding value series and gives some crude indication of
the extent of fiscal insurance achieved. Given that ɪ^ɑ* should be stationary it is not surprising
18