Introduction

The main aim of this chapter is to provide estimates of trend rates of growth in output per person based on modern methods of time series analysis. This is of interest for several reasons:

1. (i) we can obtain more reliable estimates of changes in trend growth and their statistical significance than hitherto;

2. (ii) this will allow us to put both the Golden Age of European growth in the earlier post-war period and the more recent slowdown into a long-run context;

3. (iii) in investigating whether the hypothesis of a unit root in output per person can be rejected, the results obtained inform not only statistical debate but also throw light on controversies in growth theory.

Our analysis is based almost entirely on the standard set of long-run data provided by Maddison (1991). To this we add only the recent estimates for Spain constructed by Prados (1993). The Maddison data have been used by a number of econometricians interested in unit root and segmented trend models and keen to avail themselves of long runs of data: see, for example, Ben-David and Papell (1993), Duck (1992) and Raj (1992).

These authors engage the data from the standpoint of the econometrics literature and, in particular, the debate triggered off by Perron (1989), who argued that the failure to allow for adverse exogenous shocks, such as the Great Crash of 1929 and the 1973 oil crisis, had led researchers erroneously to fail to reject the unit root hypothesis for USA GDP.