Intraday and inter-market volatility of foreign exchange rates of Philippines, Indonesia and Thailand using ARCH/GARCH model, Johansen cointergration test and Granger causality

• Main Authors: Lim, Sarah Phoebe K., Recto, Ericka Francesca M., Reyes, Tanya T., Won, Hyun Jae
• Format: text
• Language: English
• Published: Animo Repository 2016
• Subjects: Foreign exchange rates > Philippines; Foreign exchange rates > Indonesia; Foreign exchange rates > Thailand
• Online Access: https://animorepository.dlsu.edu.ph/etd_bachelors/7761
• Institution: De La Salle University

This research paper mainly determines the intraday volatility of the Philippine peso, Indonesian rupiah, and Thai baht vis-a-vis the U.S. dollar. In line with his, this paper also prove the reasons and causes behind the intraday volatility by showing the level of currency volatility, trends in currency volatility, and the factors affecting currency volatility. Hence, this paper uses three statistical test such as ARCH/GARCH model to determine the level of currency volatility, the Johansen cointegration Ttst to see the cointegration in the currency volatility, and lastly the Granger causality to determine the economic factors that affects the currency volatility. Using a total sample of 50,620 observations of hourly open prices, five economic factors, and four non economic factors which consisted of three ASEAN countries, over a period of five years (2011-2015), the ARCH/GARCH test recognized GARCH (1,1) as the best fit model for PHP while ARCH (1) was the best fit model for IDR and THB, there were trends in the currency volatility, only two economic factors in Thailand validated, all the non-economic factors somehow had an effect to currency volatility, and the Philippine peso was seen to be volatile while the Indonesia rupiah and Thai baht were both seen to be more stationary. The researchers validated their study by using ARCH/GARCH to forecast for January-June 2016. The best fit models for each currency were used and results showed that the forecast was almost accurate to the actual prices of January-June 2016. Therefore, the best fit models can be used to predict future outcomes.