Structural breaks and regime switching models: theoretical extensions and applications

Abstract

In 3 essays, regime switching and structural break models are explored and used in the fields of International Economics, Health Economics, and Macroeconomics. (1) Characterizing the Real Exchange Rate in a Switching AR(1) and Unit Root Model best represents its behavior in our long horizon data for 16 countries, which raises questions regarding the common practices of utilizing single process long horizon regressions, panel analysis, and structural breaks. (2) Using the SEER-Medicare database, the burden of illness of colorectal cancer patients from 1991-2002 is shown to have a break in the average first-year cost of treatment coinciding with the FDA approval date of irinotecan, a chemotherapy agent. (3) Simulations exploring the finite sample properties of endogenous structural breakpoint tests show that the performances of the nonlinear and linear forms are identical, but bootstrapped critical values should be used in small samples. Using the NAIRU as an example, the finite sample dangers of these tests are apparent.