Extracting an average business cycle and its stylized facts in the U.S.

Andreas Groth

Ecole Normale Supérieure, Laboratoire de Météorologie Dynamique, Paris, France

Much effort has focused on the identification of business cycle and its stylized facts from macroeconomic data. In this work we apply different approaches on U.S. data to answer the questions of leads and lags between economic variables and variations around an average business cycle behavior. In a first step we apply the concept of phase synchronization to quantify the behavior of expansion and recession, and we demonstrate a clear phase locking between several economic variables. With this information we try to cope in a next step with the problem of extracting an average dynamical behavior from short noise time series. For this reason we apply the concept of multivariate singular spectral analysis and extract information about the variability and sensitivity along an average business cycle. General properties are in consistency with the literature, where it exhibits both deepness and steepness in production and employment rate, and a three-phase shape in which the high-growth period is explained by changes in inventories and investment.