Modeling dependency: application to currency

Abstract

The primary purpose of this dissertation is to investigate the behavior of the elements of the foreign exchange market, the largest financial market in the world. Whilst the market itself is not new, the concept of currency as an alternative asset, is. Apart from growing awareness of the attractiveness of foreign exchange as an asset class, the recent huge growth in foreign investing combined with record high levels of currency volatility raised the importance and immediacy of foreign exchange risk. This dissertation applies several copulas to model the dependency between a chosen currency pair, and employs a superior goodness-of-fit test recently proposed in the copula literature. Applying the selected copula (from the goodness-of-fit test) for the calculation of Value at Risk, it is shown that the selected copula offers superior protection to the organization with only one-third the failure rate compared to the classical correlation-based Value at Risk. In addition, regression techniques are applied on interbank foreign exchange intraday trades to investigate the factors behind the massive volume of foreign exchange trades. Volatility and investments in foreign equity are found to be the key reasons driving foreign exchange trades in 1998; in 2008, the ‘carry’ trade became the most prominent factor.