Forecasting Combination: An Application For Exchange Rates

This paper tries to forecast exchange rates by comparing forecasting methods that take into account cointegration and methods that do not. The first finding is that taking into account cointegration provides better forecasting results.

Furthermore, the factor model with cointegration provides the smallest forecasting errors, but when compared with penalized maximum likelihood, the differences are not always significant. In addition, we show that a forecast combination of all the methods used provides better exchange rates forecast accuracy.

Download the full whitepaper now at the link beneath Related Files/Links

Supporting documents

Click link to download and view these files