seasonal adjustment and measuring persistence in output

abstract: this paper presents evidence on the following question: by how much does an unexpected change in real gdp of 1 percent change this series in the long-run? to shed light on the robustness of the various methods suggested in the literature to answer this question, we filter quarterly austrian data on real gdp by three seasonal adjustment methods. applying a variety of time series techniques to the resulting series allows us to report two main findings. first, unexpected shocks affect gdp in the long-run. this finding contradicts conventional wisdom about the generating mechanism of business cycles and confirms results by campbell and mankiw (1987) for u.s. data. second, quantitative measures of persistence are not robust with respect to different seasonal adjustment methods. in addition, we present monte carlo evidence showing that the census x-11 method for seasonal adjustment can be a source of spuriously high measures of persistence if seasonal differencing is the appropriate adjustment method.