Computing Transitional Dynamics in Recursive Growth Models: The Method of Progressive Paths.


The Method of Progressive Paths is a numerical algorithm for computing the transitional dynamics of nonlinear recursive growth models, especially when the transitions depend on multiple state variables. I restrict my attention to saddle point problems, which are encountered quite frequently in dynamic modeling.

Progressive Paths has four important advantages. First, the building blocks of the method, time paths of particular economies, are "natural" concepts, not artificially introduced for computational purposes. Second, the method exploits the user's knowledge of linear systems of ordinary differential equations. The analysis relies only on systems of ODE's - no PDE's or functional equations. Third, Progressive Paths can be used to study decentralized dynamic equilibria because the focus is on a system of differential equations - not on a Bellman equation or value function. Finally, the programming required is quite minimal; actual application of the method needs only for the user to write loops that iterate difference equations! Therefore, the prerequisites are relatively modest, especially from the computer science and numerical mathematics departments.

The paper has a self-contained discussion of implicit finite difference methods as applied to systems of ODE's as well as two MATLAB programs that implement some of the methods.

Click here to send email to me requesting a PostScript copy of the paper. The request should include your email address and institutional affiliation.

Two MATLAB programs are included as Appendices of this paper. Click to download ascii versions of the programs.