Following the boom in 2000, economic growth in Austria slowed down steadily and even turned negative in the third and fourth quarter of 2001. A fall in the seasonally adjusted real GDP over two successive quarters is the conventional definition of a recession in the economic debate. More accurately, however, the business cycle is defined by the variations in the utilisation of productive capacities or by the deviation from potential output (output gap). For practical purposes, this is often done by subtracting trend growth from the actual growth figure. The growth rates obtained in this way, nevertheless, exhibit a certain lead vis-à-vis the business cycle. The fact that the phenomenon of the cycle is reflected differently in the various economic time series is giving rise to different approaches in business cycle research. Disagreement among academics also extends to the issue of adequate adjustment for the trend component. Generally recognised and largely uncontroversial, on the other hand, is the importance of a high-quality adjustment procedure for seasonal and calendar effects, which has an important influence on the results of different cyclical measurement procedures. When applying the rule of two successive quarters of negative growth to Austrian GDP data corrected for seasonal effects and variations in the number of working days, only 5 recessions have occurred since 1954. In this case, no correction has been made for a trend growth component. Since trend growth was significantly higher between 1954 and the mid-1970s than in the following years, growth in that period remained positive even in periods of marked slowdown that may well be regarded as recessions in a cyclical sense. Thus, a procedure focussing exclusively on growth rates will give only an unprecise picture of the business cycle pattern. A further source of differences in dating turning points are the several possibilities to treat the irregular residual component resulting from the statistical analysis. In this way, still applying the two-negative-quarters rule, the number of recessions in Austria since 1954 is reduced to three, when adjustment is made only for seasonal and calendar effects, leaving the irregular component in the time series. A method of dating business cycles similar to the one developed by the NBER in the USA delivers no clear-cut results when applied to the Austrian data. The monthly series for employment, industrial production and wholesale turnover adjusted, for that purpose, for seasonal and calendar effects exhibited considerable variations in the high-frequency area, even after the irregular component had been eliminated. No firm conclusions on the dating of turning points could be drawn either, after the time series had been considerably smoothed using moving averages.
