Saturday, June 28, 2008

More on Change Points

A few months ago I posted on an entry on using estimates of "turning points" in financial time series to study the economic costs of conflict. I've recently learned of a nice little "change point" estimation suite that is part of the MCMCpack for R that allows one to use Bayesian estimation to locate change points and, wonderfully, to compare the fit of various change point models. The suite allows estimation on time series outcomes associated with various distributions. At the EITM seminars, Robert Walker demonstrated how easy it is to use by showing some examples for studying structural breaks in patterns of international exchange rate policies. Very cool. Jong Hee Park also lists a working paper demonstrating applications on his website, although the paper is not linked there. Thus, go forth and discover "structural breaks" in time series data!