Agnostische Schätzung und Zerlegung von Produktionslücken

This paper uses a Bayesian dynamic index model to extract common trends and cycles from large datasets in a novel agnostic way. We show how to derive an agnostic output gap measure from these estimates, applying the procedure to a data set of quarterly time series compiled for the U.S. and Switzerland. We show that our derived output gap measure is a good reflection of the U.S. and Swiss business cycle and has a high correlation with alternative estimates of the output gap. We also show that two common cycles estimated for U.S. – one on the frequency of business cycles and the other at a lower frequency – dominate in explaining fluctuations in the output gap. Both cycles are strongly related to production, investment, labour market and money market variables, with the former being dominated by production and investment variables and the latter by labour market variables.