On testing for symmetry in business cycles

Abstract

Business cycle forecasting has become an important part of short and medium term economic planning. Such forecasting, however, is often very intricate, as business cycles are not at all periodic, just recurrent. Furthermore, they often include irregular timing and varying amplitudes. When patterns and relationships are very irregular there are no simple reliable business cycle forecasting procedures. In practice there is, somewhere, a limit for business cycle predictability, and it is often worthwhile to examine empirically the various theoretical regularity assumptions. One important regularity issue concerns the business cycle symmetry assumption. The present paper empirically tests the hypothesis of symmetry around business cycle turning points in some economic time series. Two test procedures are applied. One is based on the analysis of transition probabilities between expansion and recession regimes. The second procedure tests symmetry versus asymmetry through skewness statistics. The analysis is based on detrending through the use of linear deterministic trends as well as by Beveridge-Nelson decompositions.