Title: Semiparametric selection and optimal weighting of leading economic indicators
Authors: Henri Nyberg - University of Turku (Finland) [presenting]
Heikki Kauppi - University of Turku (Finland)
Abstract: A semiparametric econometric procedure is developed for getting maximal signalling predictive power out from a potentially large set of leading indicators to the state of the business cycle. The procedure is transparent, largely automated and meets demands from users working at different forecast horizons by selecting and weighting to obtain an optimal linear combination of leading indicators. The application with the U.S. data demonstrates the superiority of our procedure, providing a valuable complement to the existing methods and the benchmark composite index of leading indicators.