The Measurement of Potential Output for Austria
Most empirical approaches to measuring potential output (PO) are based on variants of the method of trend extraction. This is also true for estimating PO on the basis of production functions. Structural approaches rely on trend extraction methods, in connection with determining potential employment or the long-term path of productivity for example. Thus, trend adjustment methods are of central importance for the measurement of PO, regardless of whether PO is estimated through a structural or astructural approach. These considerations have led some economists to attempt to combine astructural methods (e.g., mechanistic trend extrapolation) with structural approaches in such a way that the disadvantages of both methods are diminished. One such method of combining structural and astructural elements computes aggregate PO on the basis of a trend adjustment method (e.g., the HP filter) and on the basis of specific "structural information" deduced from price-wage equations. This article has refined this technique. The extension concerns above all the method of optimal extraction of cyclical components from GDP and the unemployment rate through the use of bivariate structural time series models. Thus, PO and the "natural unemployment rate" UNAT are determined without the aid of direct structural information. PO and UNAT are defined as "unobservable variables" and follow a "smooth stochastic trend". Through the use of price and wage equations for the "optimal" extraction of cyclical components of GDP and of the unemployment rate, information on the system and structure of the economy enters in an indirect way into the computation of PO and UNAT. This link is established by explicitly taking into account the close connection between the cyclical components of GDP (BC) and the unemployment rate (UC), known as Okun's Law, and the influence of these two cyclical variables on wage and price inflation. This ensures the neutrality of PO with regard to inflation in this approach. Potential output exhibited the highest annual growth rates (over 5 percent) in the early seventies (1970-1973) and in 1989. The lowest growth rate for PO (less than 1 percent) was computed for 1983 and 1984. In the period after 1989, a year of record growth, PO growth shows a marked downwards tendency. PO growth during the last three quarters of the sample period (1995:(1) to 1995:(3)) is estimated to be just slightly above 1 percent, marginally higher than during the troughs of the sample period. Since the beginning of the 1990s, actual GDP was almost always below PO. The PO estimates imply that the average length of the cycle in the variables BC and UC is 28 quarters. The cyclical component of the unemployment rate lags behind that of GDP by close to 3 quarters on average.