(1989), "Two Exact, Non-Arbitrary and General Methods of Decomposing Temporal Change", Economics Letters, (30), pp. 151-156.

A common goal in economics is to attribute changes in a variable Y between two points in time to various causes, when the only information available is an equation expressing Y as a matrix product of several variables, and the values of all variables at the two dates. Past methods of decomposing temporal changes into various components are found to suffer from one or more of the following problems: they can be arbitrary due to the uncertainty concerning whether to expand about the initial or final time period, inexact due to a residual term, and specific to cases in which the variable is a product of two matrices. This paper presents two decomposition methods which overcome all three shortcomings of earlier methods.