Preliminary estimation of transfer function weights : a two-step regression approach

Sammanfattning: In economic time series modelling, dynamic relationships frequently have to be modelled where the explanatory variables influence the dependent variable over more than one period. Such dynamic relationships are found in business cycle forecasting with leading indicators, in marketing models describing the relationship between advertising and sales, and in many traditional econometric models. In this dissertation the transfer function model proposed by Box and Jenkins is used to describe the dynamic structure. There are several approaches that could be used to specify the model. A two-step regression approach is proposed by the author and tested by three simulation studies. Finally, the regression approach and two other approaches are used to identify transfer function models for the Swedish Index of Industrial Production using financial variables as leading indicators.