Introduction

The concept of potential output has played an important role in economic analysis and policy debate at least since the 1960s. Usually defined as the productive capacity that would be feasible under full utilisation of all factors of production, it is a central reference variable in economic theories and provides the starting point for analysing the current status of the economy. Potential output is also a key reference variable in empirical economic research and is regularly employed by national and international economic research institutes and economic advisory boards to analyse the business cycle. Potential output is of importance in economic policy for disentangling structural problems and business cycle phenomena, in medium-term budget planning and especially for providing an orientation variable for monetary policy. The potential growth concept also plays a pivotal role at the European level in the context of the Lisbon Agenda and, at the German level, in the definition of necessary reforms and reform priorities. In spite of the widespread use of the concept of potential output in economic theory and empirical applications as well as in economic policy debates, the historical background and the assumptions inherent to the concept of potential output regarding economic relationships as well as concerning epistemological perspectives are rarely made transparent, let alone critically questioned. Apparently welldefined and unequivocal potential figures are quoted and used in empirical practice and economic policy debates as a matter of course and with a confidence that cannot be taken for granted given the assumptions and limitations of conventional methods. It is doubtful, for example, whether some of the fundamentally retrospective empirical procedures often used to determine potential output, such as univariate filter-based methods for example, can be appropriately applied in the context of a genuinely forward-looking concept (i.e. the future growth perspectives of an economy measured in terms of potential output). It is also doubtful whether labour input in production function-based methods can be reliably projected into the future given that such volumes are influenced in their turn by changes in labour market structures, technological trends or macroeconomic developments.