This study analyzes the realized volatility and discrete jump volatility of Korean won–US dollar exchange rate returns using  high-frequency five-minute returns from June 1, 2010, to April 30, 2021. This study considers several periodicity filters, such as MAD, ShortH, and WSD, to obtain increasingly efficient and robust jump estimators. This study estimate the volatility and jumps using the maximum outlyingness and local robust variance with truncated power variation. Although most studies adopted the standard normal Z-type jump statistics of the standard normal distribution, this study adopts the Gumbel distribution to determine whether a significant daily and intraday jump occurs. The empirical tests utilize 951, 336 five-minute exchange rate returns and the empirical results indicate that many large and discontinuous exchange rate jumps occurred in the 2010s. However, when this study considers the periodicity filters such as MAD, ShortH, and WSD, jump probabilities seem to be lower using Gumbel statistics in the 2010s. Moreover, in the Lee and Mykland (LM), Lee-Hannig (LH), Laurent-Shi (LS), and combined LH and LS jump statistics, if this study includes periodicity filters, the empirical returns always have significantly lower jump probabilities using local robust variance with average truncated power variation. Therefore, overall, when we utilize MAD, ShortH, and WSD, the five-minute returns of Korean won–US dollar exchange rates have considerably lower intraday jump probabilities. If the periodicity filters of volatility are not considered, the jump probabilities may be overestimated. 

* Reference

- Author: Chae-Deug YI(Department of International Trade, Pusan National University)

- Title of original paper: Exchange rate volatility and intraday jump probability with periodicity filters using a local robust variance

- https://doi.org/10.1016/j.frl.2023.103821   

- Journal: Finance Research Letters