Abstract
In this paper we conduct an assessment of the forecasting performance of a particular state space specication in which predictors are explicitly modelled in the measurement equation and the forecasting objective is included in the state vector. The set of models that arises belongs to the class of dynamic factor models. This specication allows for supervision of the estimated factors for the specic variable to forecast. We apply this method to forecast selected macroeconomic variables using information from the yield curve. In particular we consider a model in which factor loadings pertaining to the yield curve have the Nelson and Siegel (1987) structure and one variable at a time is forecast. We compare the forecasting performance of this specication to the one of benchmark models such as principal components regression, partial least squares, AR(p) and MA(q) processes. We present results using yield data from the Gurkaynak, Sack and Wright dataset and macro data from FRED.

CREATES arranges lunch seminars on a regular basis, normally on Tuesdays. The speakers are usually CREATES members, mostly PhD-students and postdocs. The speakers can also be visiting PhD-students or professors.

Each lunch seminar is followed by a discussion of the paper presented.

Organizers: Anders Bredahl Kock and Cristina Amado.

Place: CREATES meeting room, M303, building 2628 (M)

www.creates.au.dk