Simple Measures of Exchange Rate Volatility: Applying a Combined Artificial Neural Network and Stochastic Volatility Model
Exchange rate fluctuations have a great impact on the Japanese economy. This paper tries to present a simple and effective approach for predicting volatility of US/JPY. The approach showed that the ANN-SV model can enhance the volatility forecasting ability of the traditional GARCH model and SV model. The results indicate that the proposed hybrid GA-ANN-SV model could not provide better volatility forecasts than ANN-SV in US/JPY exchange rate. We also have observed the same effect for US/EUR, US/GBP and US/CHF. Because of the simplicity and effectiveness of the approach, it is promising for US/JPY currency.
Japan is a country with high dependence on trade, 99% of crude oil is dependent on imports, but also exports a lot of capital and equipment goods (Nishimura and Hirayama, 2013; Thorbecke, 2015). In contrast, a large proportion of Japanese manufacturers’ trade surpluses are exposed to exchange rate risk. Therefore, the yen exchange rate fluctuations in particular by the manufacturers of special attention. Forecasting exchange rate volatility is an essential step in many financial decision makings. GARCH family of models has been extensively used in finance and economics, particularly for estimating volatility (Bentes, 2015; Abounoori et al, 2016). However, GARCH family processes are parametric models that assume a linear correlation structure in the data. They are restricted to stationary and normality distribution of variables and errors. These assumptions are not true in real life situations. Therefore, GARCH family models may not capture nonlinear patterns in the data (Hajizadeh et al, 2012).
Currently, by proposing a hybrid model to overcome the limitations of the GARCH-based mode ...