Essays on theoretical and empirical aspects of structural break models.

摘要

This dissertation analyzes theoretical and empirical aspects of structural break models. The first chapter proposes a test for the slope of a trend function when it is a priori unknown whether the series is trend-stationary or contains an autoregressive unit root. The procedure is based on a Feasible Quasi Generalized Least Squares method from a first-order autoregressive specification. The autoregressive estimate is obtained by the Ordinary Least Squares method applied to detrended data and is truncated to take value 1 whenever the estimate is in some specified neighborhood of 1. This implies that inference on the slope parameter can be performed using the standard Normal distribution in both cases.; The second chapter extends the analysis to the case of testing for changes in level or slope of the trend function. When the break dates are known, our test statistic has a chi-square limit distribution in both the stationary and unit root cases. When the break dates are unknown, a version of our test has nearly identical limit distributions in the two cases so that a testing procedure with nearly the same size can be obtained. We show our procedure to be substantially more powerful than currently available alternatives.; The third chapter pertains to an empirical analysis involving a structural break. We analyze the reaction function of the Japanese monetary authorities in deciding when to intervene in the foreign exchange markets using daily Japanese data from April 1, 1991 to December 31, 2002. We document a regime change in June 21, 1995 when Dr. Sakakibara became in charge of intervention policy in Japan.; The fourth chapter investigates real exchange rates dynamics. A standard model with transportation costs implies that real exchange rates should follow a band threshold model where the process is a random walk within the bands and mean-reverting outside them. Because of technological improvements, these bands should narrow over time. We examine whether this is the case in Japan. The evidence indicates that such a feature is not supported by the data, casting doubts on the relevance of such models to describe the behavior of exchange rates.