Relatore:
Estela Bee Dagum (University of Bologna)
Abstract:
For detecting in real time the short term trend which includes the impact of an incomplete business cycle if present, official statistical agencies generally produce estimates derived from asymmetric moving average techniques that were developed by Musgrave in 1964. However, the use of the latter introduces revisions as new observations are added to the series as well as delays in detecting true turning points. In this paper, we use a reproducing kernel methodology to derive asymmetric filters that converge fast and monotonically to the corresponding symmetric one. We consider three specific criteria of bandwidth selection based on the minimization of: (1) the distance between the transfer functions of asymmetric and symmetric filters, (2) the distance between the gain functions of asymmetric and symmetric filters, and (3) the phase shift function over the domain of the signal. We show theoretically that any of the three criteria produces real time trend-cycle filters to be preferred with respect to the Musgrave ones from the viewpoint of revisions and time delay to detect true turning points but the bandwidth selection based on (2) is the best. The asymmetric filters can be applied in many fields, such as, macroeconomic, finance, health, hydrology, meteorology, criminology, physics, labor markets, utilities, and so on. In fact, in any time series where the impact of trend jointly with cyclical variations is of relevance. We have chosen a set of leading, coincident, and lagging indicators of the U.S. economy to illustrate our theoretical conclusions.
Organizzatori
Proff. Simone Giannerini e Alessandra Luati