Discussion

The fiscal convergence criteria of the Maastricht Treaty and the Stability and Growth Pact have stimulated a large strand of economic literature. Most of this literature is on how to interpret the available data, how to perform cyclical adjustments, how to identify the underlying budgetary position and the discretionary behaviour of government, etc. The infrastructure of indicators (how the deficit is defined; how it is compiled; how reliable and accurate it is; should it be compiled on a cash or on an accrual basis; should it cover a narrowly defined government or a wider public sector) has received much less attention. Balassone, Franco and Zotteri are among the rare authors to have devoted attention to these issues. And they have done it with quite a provocative spirit, even daring to consider that the indicators that are relevant for fiscal surveillance in the EU have been a ‘misleading compass’, as they put it in previous and related papers (Balassone, Franco and Zotteri 2006). The effectiveness of fiscal rules depends crucially on the quality of the fiscal indicators.

The quality of statistical indicators can be generically defined as ‘fitness for use’. It is, in fact, a complex multidimensional concept; it covers relevance, timeliness, coherence, completeness, comparability across time and space, accuracy, reliability, transparency, etc. The topic of the chapter is on reliability; in technical terms, reliability refers to the closeness of the initial values to the subsequently revised figures. The authors note that the EU fiscal indicators do not score high in terms of reliability. This is worrying for two reasons: First, in the implementation of the EU fiscal framework, the most relevant decisions – such as deciding whether an excessive deficit exists or has been corrected, if a given country is complying with Council recommendations, or even if a country has fulfilled the criteria to enter the euro area, or ultimately whether the Council will have to impose sanctions – are taken on the basis of the first outcomes. The subsequently revised data – though of better quality – appear too late to have any decisive impact on the policy decisions. So if the first outcomes are subject to large revisions afterwards, important decisions will be taken on the basis of wrong data. A second reason to be worried is that, as the authors point out, given that the fiscal rule is based on a rather simple indicator, one would expect it to be reliable. Presumably, other more complex fiscal indicators were rejected (implicitly if not explicitly) because the government deficit would be simple, easy to compile and reliable. Balassone et al. establish a link between lack of reliability and opportunistic manipulation,

i.e. political pressure to get data that are rosier than the reality, at least the first outcomes even if these are revised afterwards. If the lack of reliability is directly connected with opportunistic manipulation, then the solutions the authors propose seem to be quite appropriate: (i) increase the scrutiny of the government accounts and (ii) strengthen the status of the data compilers, by giving more independence and accountability to the

national statistical institutes. In the last two or three years, the Commission has taken several steps in this direction, including all the suggestions by the authors. Obviously, it is very early to say how fruitful such steps are. But are political pressures and opportunistic behaviour the main cause of the lack of

reliability of fiscal data in the EU? Or are there other causes? If the lack of reliability of government accounts was directly and decisively connected with political pressures, one would expect the countries with less reliable data to be those that have had more problems with the respect of the EU fiscal reference values, and vice versa. The three countries that the authors use in their case studies (Greece, Italy and Portugal) would confirm the conjecture. Yet we could also expect similar problems with the French and German figures. Now, if we consider data reported by Member States since 1994, France and Germany have the most reliable deficit data in the EU.2 So the question is: what do the French and Germans have that prevents them having to resort to the same gimmickries as Greece, Italy and Portugal? (Or, do they resort to other tricks?) If I now apply the model proposed in the chapter, I am led to conclude that France and Germany bear larger costs (Cx) for entering in data manipulations with the purpose of hiding their deficits. Why? Is this because of institutional reasons? Are the French and German statistical institutes more independent than in other Member States? It would be interesting to research what can be done to increase the reputation costs for producing unreliable data in Greece, Italy and Portugal. If we try identifying which are the EU Member States with the less reliable accounts we

get some surprising results. Among the countries that have reported the less reliable data we find not only Greece, Portugal and Italy, but also Sweden, Luxembourg and Denmark. Now, these are among the countries that have had less trouble in complying with the EU fiscal framework. Of course, the data revisions in Sweden and Denmark did not catch public opinion, as did the ones in Italy and Portugal, not to say Greece, but the magnitudes are not dissimilar. My point here is that the opportunistic manipulation of data certainly has an implication of the reliability of data, but there are other factors at play. It may be the reputation of the data compilers, irrespective of their status of independence, or the resources, competence or expertise available to each statistical institute, or even the more or less complex institutional arrangements in each country, or because some countries enter into more complex transactions than others. Or perhaps the reliability difficulties would simply reflect the fact that our accounting rules are excessively complex and unreliability is somewhat inevitable, or very costly to avoid. We do not know. We also need more research on this. The authors identify two fragilities of the deficit indicator: (i) the fact that it is measured

in an accrual basis and (ii) the fact that it is measured net of financial transactions. Do the authors believe that these two fragilities are such that the deficit definition should be changed? Would a cash deficit including some financial transactions be a better, and more reliable, definition? My understanding is that the authors believe that the current deficit definition is still preferable to any other, and they simply want to pinpoint the areas that need to be closely scrutinized by Eurostat.