Public debt around the world: a new data set of central government debt

Abstract

Commonly used data sets on the level of public debt provide incomplete country and period coverage. This article presents a new data set that includes complete series of central government debt for 89 countries over the period 1991 to 2005 and for 7 other countries for the period 1993 to 2005.

Acknowledgments

We would like to thank Eduardo Borensztein and Eduardo Levy-Yeyati for helpful comments. This article was written when both authors were with the Research Department of Inter-American Development Bank. The views expressed in this article are those of the authors and do not necessarily represent those of the institutions with which they are or have been affiliated. The data used in this article should not be considered part of the official statistics of any institution with which the author are or have been affiliated. While the authors made an effort to assemble the best possible data set, they cannot guarantee the accuracy of the data reported here and accept no responsibility whatsoever for the consequences of their use.

Notes

¹ We realized that obtaining data on public debt was a serious problem when we started working on a paper aimed at measuring the determinants of debt growth (Campos et al., 2006). Data availability is particularly limited in regard to domestically issued public debt. Data on external public debt for developing countries are generally available from the World Bank's Global Development Finance (GDF) data set; even in this case, however, the data present some problems, since GDF separates the public and private components only for long-term debt, and it does not report data for industrial countries.

² The International Monetary Fund (2003) assembled a dataset comprising up to 54 countries (34 developing and 20 industrial) but does not make the data available.

³ The five largest countries not included in groups 1 and 2 are: Ukraine (ranked 51st, our data start in 1999); Kazakhstan (ranked 53rd, our data start in 1997 and end in 2003); Croatia (ranked 57th, our data start in 1995); Libya (ranked 63rd, our data end in 2002); and Zimbabwe (ranked 65th, our data end in 1997).

⁴ IFS sends forms with precise definitions that the countries must fill and send back, and in theory the statistical department of the IMF does conduct a quality control. However, the large discrepancies between IFS and data and Article IV data (which are the ones used by IMF economists) suggest that the quality of IFS data is not very high (for evidence on discrepancies between these two sources of data see Pellecchio and Cady (2005)). Note that we did perform some quality check on the data. In particular, we graphed the evolution of the debt over GDP for each country and whenever we observed either a big jump or a level of debt that was not consistent with our priors, we investigated the cause of this problem and, if appropriate, corrected the data.

⁵ Note that IFS data often differ from the information reported in Article IV documents (Pellechio and Cady, 2005). Whenever possible we tried to reconcile information from different sources, starting from the levels reported by those that we deemed most reliable and completing the series based on growth rates from alternative sources.