Testing long memory in exchange rate return and volatility: A research on the 2008 global financial crisis

Information has a significant impact on financial asset value for investors and market players trading in financial markets. With the information entering the market, investors take positions and decision makers take some steps in line with the information received. With the developing technology, information has been becoming accessible to everyone. The Efficient Market Hypothesis puts the concept of knowledge at the forefront in terms of price movements of financial assets and the effectiveness of financial markets. If the information and news reaching the market can affect the financial asset price quickly and effectively, the market is stated as effective. If the future price of the financial asset can be estimated by looking at the past market price, mean if long memory exists, it is indicated that there is no effectiveness of market. In these markets, investors are able to earn a profit over the average market return. Since all transactions between countries are carried out through foreign exchange, foreign exchange markets have become the market that is always considered. The volatility that may occur in foreign exchange markets has the capacity to affect all countries that have economic activity with that country, even indirectly. Excessive volatility arising in the foreign exchange market can spread to all financial markets and affect them negatively. Most investors will not want to invest easily in a high volatile country market. This study is performed in order to test the presence of long memory in returns and volatility series by using daily closing prices of the three exchange rates which are the US Dollar / Turkish Lira, Euro / Turkish Lira and the Pound / Turkish Lira traded in Turkey's foreign exchange market, the market of the emerging country, in the dates between 01/02/2002 and 12/31/2019. Taking the 2008 Global Financial Crisis into account, the data set is divided into 3 different periods. ARFIMA was used to test the existence of long memory in return; FIAPARCH, which is an asymmetric model, was chosen to test the presence of long memory in volatility; and existence of dual long memory was tested through ARFIMA-FIAPARCH model. The analysis concluded that long memory in return, Euro return series exists only in the pre-crisis period. For the whole data set in all exchange rates, long memory is determined in volatility. It is determined that there is a dual long memory in return and volatility for the dollar return series in the crisis period and post-crisis period, and for the Euro return series in the pre-crisis period. In the FIAPARCH model, asymmetry parameter γ is found significant for the whole data set and in negative value. This means that positive information shocks are more dominant on volatility than negative information shocks, that is, cause more volatility. Besides, ARFIMA-FIAPARCH model was concluded as the most suitable VaR model for all exchange rates as a result of Value at Risk analysis.