Finding numerical results to large scale economic models using path-following algorithms: a vintage capital example

This paper describes the numerical optimization methods used in Berger (2001) to find the complete time paths of key economic variables in neoclassical vintage capital models. An interior and a non-interior point method are discussed. Both of the methods are part of the general class of "path-following" algorithms. These algorithms can be efficiently applied to convex programming problems; and due to the standard shape of production and utility functions, many economic problems can be written as convex programming problems. Vintage capital models add scale and complexity to standard growth models because one must now handle the dynamics of multiple capital stocks. This increase in complexity will often prevent the discovery (or existence) of closed form solutions, making numerical solutions of the type found in Berger (2001) necessary.