| Summary: | In this study, volatility of the TRL/USD, TRL/EUR and TRL/GBP series in the Turkish exchange rate market are modeled by employing moving average, AR, ARMA models and ARCH process and the performances of the models are compared according to their out-of-sample forecasts. The forecasting performance of the Value-at-Risk measurement based on different volatility forecasting models are investigated by adopting the Basle Committee back testing criteria. The effect of the latest global financial crisis on the risk measurement techniques is investigated. The results showed that, according to RMSE criteria GARCH family models and according to MAE criteria AR models are superior to other models in estimating the exchange rate volatility. It is observed that the financial crisis does not too much effect on the order of the volatility forecasting models; however, the performances of the models converge to the worst performing model during the financial crisis period. When the Value-at-Risk performances of the underlying models are compared EWMA and GARCH family models are found to be more accurate than other models. It is seen that the performances of the models are worsen with the financial crisis.
|
|---|