Time-varying Parameter Regression Research Articles

Interfiber Competition in Textile Mills Over Time

AbstractCotton and synthetic fiber competition in textile mills between 1961-1990 was examined using a time-varying parameter regression model. Results indicate that the structure of demand for cotton is not stable and cotton's share responses to changes in the prices of cotton and synthetic fiber vary over time. Cotton and synthetic fiber competition in textile mill use is essentially between cotton and noncellulosic fiber. Cellulosic fiber is not a cotton competitor.

A note on flexible least squares

In a recent paper Kalaba and Tesfatsion (1988b, app. B) mention that the recurrence relations characterizing their Flexible Least Squares (FLS) approach when applied to linear systems [e.g., Kalaba and Tesfatsion (1986, 1988a)] are analogous to those of the more familiar Kalman Filter (KF) when the appropriate definitions are used. However, they fail to point out that, in that same case, the statistical model for which this KF is optimal is just a standard time-varying parameter regression model with certain restrictions imposed. These restrictions are derived in the present note.

A Framework for Time Varying Parameter Regression Modeling

A framework for time varying parameter regression models is developed and employed in modeling and forecasting price expectations, using the Livingston data. Alternative model formulations, which include various choices for both the stochastic processes generating the varying parameters and the sets of explanatory variables, are examined and compared by using this framework. These models, some of which have appeared elsewhere and some of which are new, are estimated and used to assess the expectations formation process.