وابستگی نوسانات حافظه بلند مدت، تجمع زمانی و بحران ارز کره ای: نقش نرخ دلارus برنده شده کره ای با فرکانس بالا

Using a new type of 5 min high frequency dataset consisting of real time Korean won (KRW)–US dollar ($) exchange rates, this paper characterizes the volatility process of high frequency returns. The semi-parametric local Whittle estimation is applied to estimate the long memory dependency in the volatility process of the 5 min KRW–$ returns and the temporally aggregated returns data. The estimation results present that the underlying long memory dependency in the volatility process appears to be generally consistent across various temporally aggregated returns and that the exogenous shocks and the multiple breaks associated with the crisis in the market seem to induce greater long memory dependency during the crisis.

Financial data such as foreign exchange rates, stock prices and commodity prices have been the subject of many empirical studies, most of which have analyzed daily or lower frequency data. Since the early 1980s, however, many empirical studies have dealt with higher frequency intraday prices because gathering financial data has become easier due to the rapid development in computer technology. In particular, foreign exchange markets, which have no geographical location and no business hour limitations, provide a set of complete intraday time-series data covering a worldwide 24 h market. The exchange rates used for most empirical studies are the quotes from large data suppliers such as Reuters. As the actual transaction prices and trading volume are not known to the public, the quotes are intended to be used by market participants as a general indication of the markets, and these indicative prices appear to closely match the true prices in the markets. This new type of high frequency intraday price is important for the empirical analysis of foreign exchange markets. The large number of observations enhances the significance of the statistical study, and it can increase the ability to analyze finer details of the behavior of different market participants. For more details about this new type of high frequency data, see Müller et al. (1990), Baillie and Bollerslev (1991), Dacorogna et al. (1993), Bollerslev and Domowitz (1993), Andersen and Bollerslev, 1997 and Andersen and Bollerslev, 1998, Baillie et al. (2000), Baillie and Han (2002) and Han (2002b). This paper is concerned with some of the intriguing features of the high frequency 5 min Korean won (KRW)–US dollar ($) foreign exchange rate. In particular, it explores both intraday periodicity and long memory dependency in the volatility process of high frequency spot returns. Even though these features have been widely documented by previous empirical studies that analyzed high frequency foreign exchange rates such as those of Andersen and Bollerslev, 1997 and Andersen and Bollerslev, 1998, Baillie et al. (2000) and Baillie and Han (2002), only a small proportion of the literature has studied the cases of high frequency exchange rates for Asian countries including the KRW–$ exchange rate. Hence, this paper demonstrates the dynamics of the intraday return volatility process in the high frequency KRW–$ returns over various frequencies and investigates the effects of the Korean currency crisis on the long memory dependency in the volatility process. The rest of this paper is organized as follows. Section 2 shows the construction of the 5 min high frequency Korean won–US dollar returns data and discusses the basic properties of the volatility process of the high frequency returns. The long memory property and the intraday periodicity are presented in the volatility process of the 5 min KRW–$ returns, which are widely documented by previous empirical studies for high frequency foreign exchange rates. Section 3 describes the local Whittle (LW) estimation and presents the estimation results for the long memory dependency in the volatility process of the KRW–$ high frequency returns over various frequencies. The empirical estimation results indicate that the long memory dependency seems to be generally consistent over various temporally aggregated returns data implying that the underlying dependency appears to be invariant to the sampling frequency. Section 4 investigates the effects of the currency crisis in Korea on the long memory dependency in the volatility process of the high frequency KRW–$ returns. The long memory dependency is found to be greatest during the crisis, which seems to be due to the shocks and the breaks associated with the crisis. In addition, the analogous estimations for each individual month during the crisis provide supporting evidence. Section 5 provides a brief conclusion.

This paper has considered the volatility process of 5 min high frequency Korean won–US dollar returns data. In particular, in order to investigate the long memory dependency and temporal aggregations in the volatility process, the 5 min returns are estimated by the semi-parametric local Whittle estimation method which explicitly ignores the effects of the intraday periodicity. Interestingly, the estimates of the long memory volatility dependency of the high frequency KRW–$ returns are very consistent across time aggregations, which suggest that underlying long memory volatility is invariant to the temporal aggregations. This paper has also analyzed the effects of the currency crisis in Korea on the long memory dependency in the volatility process of the 5 min KRW–$ returns. The estimation results show that long memory dependency in the volatility process is most significant during the crisis, in which the Korean foreign exchange market experienced significant shocks and breaks associated with the crisis. The analogous estimation of long memory dependency for each individual month provides additional supporting evidence. Thus, the long memory dependency in the volatility process of the KRW–$ high frequency returns appears to be the underlying feature of the process which is invariant to temporal aggregation, and the exogenous shocks and breaks associated with the crisis seem to affect the long memory property significantly.