Intraday Patterns In Foreign Exchange Returns And Realized Volatility PDF Download

This thesis examines patterns in the FX returns and implied volatilities using daily return and implied volatility data for four major exchange rates for a period of January 1994 to December 2003. The existence of the patterns could indicate that the FX market is not efficient and could provide a basis for the construction of the trading strategies. Volatility tends to rise prior to the announcement of both scheduled and unscheduled news and fall on the announcement day. The "sign effect", indicated by the bad news having stronger impact on the volatility than good news, tends to weaken in post euro period. We find a strong evidence of the day of the week effect in the FX returns and implied volatilities, indicated by (i) positive Thursday and negative Friday returns, (ii) positive implied volatility changes on Monday and Tuesday and (iii) negative implied volatility changes on Thursday and Friday. The intraweek patterns have become more significant after the introduction of euro. We confirm the holiday and January effect that tends to strengthen in the "bad" years characterized by low GDP growth rate, and tends to weaken in the "good" years characterized by high GDP growth rate. We find a strong relation between implied volatility and contemporaneous returns, which is strongly affected by the news announcements, stronger for small returns and whose significance declines following the introduction of euro. There is also some evidence of the extreme levels of the implied volatility predicting following day returns, which is found to be particularly significant for negative (as opposed to positive) returns and for extremely large increases (as opposed to decreases)in the level of the implied volatility. The evidence presented in this thesis contributes to the existing research on FX anomalies, with the main contribution centring around a significant impact of euro.