Trimbur, Thomas M. - In: Journal of Forecasting 25 (2006) 4, pp. 247-273
This article develops a new method for detrending time series. It is shown how, in a Bayesian framework, a generalized version of the Hodrick-Prescott filter is obtained by specifying prior densities on the signal-to-noise ratio (q) in the underlying unobserved components model. This helps...