Event-related Exchange Rate Forecasting

Specialeforsvar ved Frederik Vibeager Hummel

Event-related Exchange Rate Forecasting

 

Abstract: Political events may have substantial effects on exchange rates, one example being the EU membership referendum in the United Kingdom. The large price movements caused by these events differ from what is usually considered in jump models, since the time point is known in advance. This thesis presents a dynamical model of the exchange rate that incorporates this, modelling the event as a jump occurring at a deterministic time point known in advance. The jump size is modelled as a mixture distribution, using risk neutral event probabilities derived from betting quotes as mixture parameters. Under this model, we derive a closed form solution for the price of a European call option, and use this to calibrate the model to observed market prices on currency options. Doing this, we obtain densities of the exchange rate on the day following the event. The model works well when applied to GDPUSD options leading up to the EU membership referendum, where we are able to obtain densities that leads to accurate forecast of the exchange rate when conditioning on the event. When we apply the model to USDMXN options leading up to the US Presidential election of 2016, the model yields less optimistic results. In addition, we discuss different approaches to hedging an option within this model, and finally, we present an extension of Mertons Jump Diffusion Model, showing how the event can be incorporated in to more sophisticated models 

 

Vejleder:  Rolf Poulsen
Censor:    Bjarne Astrup Jensen, CBS