PARAMETRIC CHANGE POINT ESTIMATION, TESTING AND CONFIDENCE INTERVAL APPLIED IN BUSINESS

Abstract

In many applications like finance, industry and medicine, it is important to consider that the model parameters may undergo changes at unknown moment in time. This paper deals with estimation, testing and confidence interval of a change point for a univariate variable which is assumed to be normally distributed. To detect a possible change point, we use a Schwarz Information Criterion (SIC) statistic whose asymptotic distribution under the null hypothesis is determined. The percentile bootstrap method is used to construct the confidence interval of the estimated change point. The developed tools and methods are applied to the 1987 – 1988 US trade deficit data. Our results show that a significant change in US trade deficit occurred in November 1987. Further, it is shown that the percentile bootstrap confidence intervals are not always symmetrical.