dc.description.abstract | This thesis investigates the predictive ability of fundamental economic and financial indicators on the EUR/NOK exchange rate. In doing so, we explore the emerging field of density forecasting, in addition to the standard point forecasting literature. Using a set of well-established empirical models, we construct short-term pseudo out-of-sample forecasts for the exchange rate. The results are benchmarked against a naïve random walk model, using a range of evaluation statistics grounded in the literature. The empirical analysis reveals that no models significantly outperform the random walk model using neither a point nor density forecast approach. However, we find evidence that fundamental models outperform in terms of forecasting appreciation tail risk at the one-month horizon. Furthermore, we find that a simple normal distribution is a better fit compared to an empirically backed skewed t-distribution derived from quantile regression. Our findings add to the growing strand of literature investigating the Meese & Rogoff puzzle from a density forecast perspective. | en_US |