Identification of currency jumps in the Rand to US Dollar exchange rate

Abstract

Modeling most financial assets, including exchange rates, proves to be a rather difficult task. One cause of this difficulty is the occasional unpredictable and large price movements that occur, known as jumps. Thus, including a jump component into a model may improve the accuracy of a model for financial assets. In order to do this jumps need to be identified. The question is whether this is possible or not? Different tests have been developed to identify jumps, many of which are based upon the theory of bipower variation. In this dissertation we will be studying the concepts of quadratic variation and bipower variation in order to study various tests based on bipower variation. We will additionally be applying one of the tests to identify the jumps in the Rand to United States Dollar exchange rate. Having found the dates of jumps we assess potential causes and find that most jumps in the exchange rate are caused by political instability, most of which occurs in South Africa.