The Public Financial Management (PFM) Performance Measurement Framework, an indicator-based assessment tool developed by the Public Expenditure and Financial Accountability (PEFA) initiative, was launched in 2005 and has been applied so far in over 60 countries. PEFA reports provide detailed accounts of the performance of PFM systems along various dimensions. This paper is based on the results of the 57 PEFA assessments completed as of August 2007. It looks at comparative cross-country PFM performance, overall and across the different budget dimensions defined by the PEFA methodology (out-turns, cross-cutting features, budget cycle), and analyses differences linked to certain country characteristics which might have an influence over PFM system performance, using both bivariate and multivariate analysis. It is based on a numerical conversion of the letter-scores used in the assessments, a methodology which can be considered controversial but which nevertheless yields some interesting results.
Two overall findings become immediately evident. First of all, there is a large variation in overall average scores, ranging from Norway’s maximum score of 3.44 (roughly equivalent to a B+) to a minimum score of 1.46 (roughly equivalent to a D+). There are 14 countries that fall below the 2.00 mark (i.e. whose average score is below C), including countries from a range of regions and with different levels of income. The second immediate interesting finding is that average scores tend to deteriorate the further one moves down the various phases of the budget cycle. Given the large number of countries and dimensions involved, overall comparisons are not very useful in terms of detecting specific issues and trends. Instead, the paper compares PEFA scores looking at different country characteristics, including region, population size, and level of income, degree of dependency on foreign aid or natural resources, and strength of democratic institutions.
Taken one by one, these dimensions show some interesting trends, with countries in certain categories showing a better performance than others. However, such binary associations are not necessarily significant from a statistical point of view, and could therefore be potentially misleading. Multivariate regression analysis results highlight that the main factors which are correlated to variations in the overall PEFA score in a statistically significant way are the level of income, country size as measured by the log of the total population, and the degree of aid dependency. Regarding income levels, as expected an increase in per capita income is associated with an increase in overall average PEFA scores. Aid dependency levels are also significant in all models shown, with a positive coefficient but associated with very small changes in PEFA scores, meaning that higher aid dependency levels are associated with marginal improvements in PEFA scores. Finally, as far as population size is concerned, results seem to indicate that larger country size is generally associated with better PFM system performance.
Despite the limitations of the existing data, the analysis highlights some of the interesting comparisons that the existence of such data allows. PEFA assessments are a unique source of information which sheds light on an aspect of governance which until very recently had been mostly overlooked. As more and more assessments are carried out, and repeated in various countries, the availability of more data will allow for a more significant and robust analysis of the determinants and consequences of improvements in PFM system performance, with the potential to generate firmer and more nuanced conclusions. Such analysis could be complemented with a structured comparison of country case studies, in order to allow for a deeper investigation of the large number of factors that are likely to affect the quality of PFM systems and its evolution over time.