Summary for non-specialists  

This paper provides evidence on the size, composition and cyclicality of discretionary tax measures (DTM), using a new database developed by the Output Gap Working Group. While their average magnitude is fairly limited over a long period with discretionary tax cuts being offset by discretionary tax hikes, they can be non-negligible at any given point in time. The cyclical pattern of DTM appears irregular and depends on the policy regime. While small pro-cyclical discretionary tax cuts were seen during the pre-crisis period, larger counter-cyclical tax breaks were adopted at the start of the crisis period, followed by pro-cyclical tax hikes in a context of substantial public finance consolidation. The paper also examines the impact of DTM on tax elasticities in the EU for broad tax categories over the period 2001-12:  DTM do not seem to explain the bulk of the large short-term fluctuation in gross elasticities of tax receipts to GDP. The availability of DTM also allows for an analytical illustrative exercise, computing variants of the cyclically adjusted balance (CAB) based on time-varying elasticities (net of discretionary measures) instead of on constant elasticities. However, the indicators turn out to be extremely erratic and plagued by statistical 'noise', which makes them difficult to interpret in practice. The fact that elasticities change sign frequently and that their strong movements offset each other over a number of years also suggests that the short-term variations may largely be driven by time lags between revenue collection and revenue bases. Therefore, the CAB variants cannot be seen as an adequate solution for addressing the issues faced by the CAB.