Exchange Rate Economics

According to Bloomberg, a daily volume in the worth of 7.5 trillion dollars in currencies is exchanged every day (Bloomberg, 2022). Another survey conducted by Credit Suisse revealed that 78% of the asked firms in Switzerland declared that they had to deal in the currency of EUR (Luzerner Zeitung, 2022). For these reasons, the evolution of the exchange rate is a relevant risk factor for Swiss businesses. The Thesis therefore aims to derive a comprehensive model that allows SME companies to predict the future exchange rates based on historical data.

To be able to construct the model, the Thesis first identifies the most important exchange rate drivers based on a literature research which are the oil price, the interest rate, the inflation rate, the growth rate and the terms of trade.

Secondly, because in the literature the statistically significant results are found for different time horizons, the influence of the lag is analysed. The linear regression of different lags reveals that the time horizon considered has an impact on the explanatory power of the exchange rate drivers.

Finally, two different models based on multiple linear regression are formulated with the aim to explain the fluctuation in the exchange rate of USD/CHF. The first model is a MLR model based on the exchange rate drivers identified in the literature research. The second model is a Taylor rule based MLR model that was developed by Chang and Matsuki (2022) and in this paper is applied to the USD/CHF exchange rate. For both models, a stationary and a non-stationary version were created. Furthermore, for the MLR model based on the exchange rate drivers, an optimised non-stationary version was created to improve the condition of no multicollinearity among explanatory variables.

For both models, the versions with the highest predictive power were the stationary ones. The optimised non-stationary MLR model is able to explain 88% of the fluctuation in the USD/CHF exchange rate whereas the nominal Taylor rule based model is only able to explain 70% of the fluctuation. However, none of the models fulfils all criteria of good fit.

Nonetheless, the Thesis succeeds in developing a comprehensive model using historical data to predict the future exchange rate development.