한국 원화의 대엔 환율의 시계열적 특성 (Time Series Analysis of Won/Yen Exchange Rate)

서지정보

· 발행기관 : 한국무역학회
· 수록지 정보 : 무역학회지 / 36권 / 3호 / 1 ~ 22페이지
· 저자명 : 이우리, 이영광

초록

This study analyzes the time series characteristics of daily Won/Yen exchange rate, using the time series data from April 1994 to February 2011, whether it has nonlinear dependency or not. It separates time period into the overall period and the normal period which excludes the two crisis periods, that is, 1997 currency crisis and 2008 global financial crisis. Daily time series exchange rate data was transformed into normal series by logarithmic transformation and differencing. First,according to the test result, auto-correlation analysis for the transformed series data shows that it has linear or non-linear dependence. Second, the analysis about the time series data transformed(AR residuals) from which linearity was removed shows that it has non-linear dependence, especially stronger in the whole period. Third,the test result about the series(AR-GARCH residuals) from which linearity and non-linear stochastic property were removed shows that it has non-linear dependence, especially in the whole period. Fourth, BDS test value of AR residuals is lowered than that of AR residuals in the whole period, but made no difference with the normal period. It implies that the non-linear series includes the non-linear relation, not stochastic relation.
However, we cannot conclude that it has the deterministic relation.
Therefore, the time series data of Won/Yen rate is estimated to have, to some extent, linear and nonlinear dependence, although the data and testing tool have some problems. However, we cannot say that such a non-linearity does not have non-linear determinacy, that is, chaotic characteristic.

영어초록

This study analyzes the time series characteristics of daily Won/Yen exchange rate, using the time series data from April 1994 to February 2011, whether it has nonlinear dependency or not. It separates time period into the overall period and the normal period which excludes the two crisis periods, that is, 1997 currency crisis and 2008 global financial crisis. Daily time series exchange rate data was transformed into normal series by logarithmic transformation and differencing. First,according to the test result, auto-correlation analysis for the transformed series data shows that it has linear or non-linear dependence. Second, the analysis about the time series data transformed(AR residuals) from which linearity was removed shows that it has non-linear dependence, especially stronger in the whole period. Third,the test result about the series(AR-GARCH residuals) from which linearity and non-linear stochastic property were removed shows that it has non-linear dependence, especially in the whole period. Fourth, BDS test value of AR residuals is lowered than that of AR residuals in the whole period, but made no difference with the normal period. It implies that the non-linear series includes the non-linear relation, not stochastic relation.
However, we cannot conclude that it has the deterministic relation.
Therefore, the time series data of Won/Yen rate is estimated to have, to some extent, linear and nonlinear dependence, although the data and testing tool have some problems. However, we cannot say that such a non-linearity does not have non-linear determinacy, that is, chaotic characteristic.

참고자료

· 없음