Permanent Components in Swiss Macroeconomic Variables

In this paper the estimation of stochastic trend components of Swiss time series, especially of real per capita GDP, is performed within a multivariate framework. The model includes real per capita consumption, real per capita investment, real per capita balances, and also a short-term interest rate and a measure for inflation. All series but the interest rate are adjusted for only deterministic seasonality. The cointegration analysis gives evidence for 3 cointegrating relationships and hence for 3 common trends. These are identified as being induced by a real productivity shock, an inflation shock and a real interest rate shock. To isolate the structural shocks the method proposed by KING et al. (1991) is used. The impulse response functions depict that the main part of fluctuations in real variables is due to the productivity shock. The variance decomposition attributes the main fraction of the forecast error variance to the productivity shock, in the long as well as in the short run. Even in the short run, the inflation and the real interest rate shock have only a minor influence on fluctuations in the time series. This could be interpreted as evidence for the assertion of real business cycles theory, that real permanent shocks are the most important source of business cycles fluctuations. However, nominal shocks seem to have a positive short-run effect on GDP and can explain part of the forecast error variance in real variables when applying the method to a set of Census X-ll adjusted data. Finally, the comparison of trend components of GDP estimated by alternative methods shows that the paths of multivariate estimates are similar and reach their peaks before contraction periods sooner than an univariate estimate. While the former are already decreasing or at least stagnating, the latter trend component is still increasing.