This dissertation studies the cause of the forward premium anomaly. The anomaly refers to the violation of uncovered interest parity (UIP) in the form of a negative estimate of the slope coefficient in the regression of the future percent change in the spot exchange rate measured in units of domestic currency per foreign currency on the interest rate differential between the home and foreign countries. The result is anomalous since UIP predicts the coefficient to be one. The deviation from UIP is an excess return, which a portion of the literature has viewed as a risk premium. I first reexamine conventional risk factors as explanations of the anomaly that were previously evaluated in the literature. In line with prior findings, I do find evidence that these factors are explanations of the anomaly. I next proceed to a new factor, rare disasters, recently proposed as a solution to the anomaly, but not empirically tested. Rare disasters are found to be systematically related to the excess currency return. Countries that on average have economic disasters more often have a higher excess currency return. I next present evidence that the risk premium is positively correlated with the lagged domestic-foreign inflation differential between countries, a new finding. To further understand the economic relations as to why either rare disaster or inflation could give rise to the forward premium anomaly, economic models are developed separately for each risk factor. In regards to inflation I study a two country dynamic stochastic general equilibrium model (DSGE) with nominal rigidities. I find that the model is capable of providing a resolution to the forward premium anomaly. The rare disasters model has less success in explaining the anomaly.