central and sub-central tiers of government should not be ignored when analysing fiscal
consolidations.
The structure of the remainder of the paper is as follows. In Section 2 we discuss the data and
briefly summarise the importance of sub-central tiers of government in the countries in our
sample. Section 3 presents a discussion of the methodology employed to identify fiscal
consolidation attempts and to determine the success or failure of each attempt. Section 4
presents an analysis of episodes of fiscal consolidation, focussing on their size, the extent to
which the attempts are co-ordinated across tiers of government, and the composition of the
adjustments. In Section 5 we switch the focus from general government consolidations to sub-
central consolidations and examine their composition. Section 6 concludes.
2: Data and the nature of sub-central fiscal autonomy
2.1 Data
The basic data used throughout our analysis are annual, and are taken primarily from IMF
Government Financial Statistics (GFS), 2002 edition but supplemented with general
government debt data from the OECD Statistical Compendium. GFS is the key source of
internationally comparable data relating to fiscal variables disaggregated by tier of
government. A major advantage of GFS data over that provided in OECD revenue statistics is
in the richness of the decomposition of expenditure and revenue available3. We are able to
focus on fifteen OECD countries: our unbalanced panel dataset covers as much of the period
1970-1999 as feasible for each country, and totals 336 observations4.
One weakness in GFS relates to the composition of tax revenues available to sub-central tiers
of government. Essentially, no distinction is made between tax revenues from tax sharing,
which involves little real autonomy of the sub-central tier, and ‘own taxes’ where the sub-
central authorities exercise some degree of independent control over the tax rate, the tax base
or both. We have been able to supplement the GFS data with more detailed information
relating to 1995 using OECD (1999) for the majority of countries and with data collected by
Jonathan Rodden at MIT in the cases of the USA and Canada. Unfortunately we have been
unable to obtain this information in the cases of Australia and France.
3 Some well known caveats apply to the use of these data, see for example the discussions World Bank [2001] and
Ebel and Yilmaz [2002] and Darby et al. [2003]. In particular, to the extent that central government exert influence
through directives, GFS data will overstate the true extent of sub-central expenditure autonomy. Nonetheless the
GFS data remain the best available.
4 Sample periods are as follows: Australia 1980 - 98, Austria 1970 - 94, Belgium 1978 - 98, Canada 1979 - 99,
Denmark 1977 - 99, Finland 1972 - 93, France 1978 - 97, Germany 1974 - 98, Netherlands 1975 - 97, Norway
1980 - 98, Spain 1980 - 1997, Sweden 1978 - 98, UK 1974 - 98, and USA 1980 - 99.