초록 close

금융시계열에서 카오스의 존재가능성은 단기예측이 가능하다는 점과 확률과정과 유사하게 보이는 복잡한 동태적 특징을 가진다는 점에서 초기부터 많은 관심을 불러 일으켰다. 금융시계열에 대한 국내외 실증연구들은 주가수익률에서 비선형성(nonlinearity)이 존재하고 있음을 제기하고 있다.


Much academic interest had been devoted to chaos theory since 1980s in that the existence of chaotic structure in financial time series implied short-run predictability and chaotic dynamics looked like random seemingly though it was generated by deterministic process.The results of many empirical studies showed that the existence of non-linearity in financial time series(including Korean stock market data) was considerably universal feature. These results are confirmed by our analysis focusing on KOSPI 200 data series. The distinguishable aspects of our empirical study is that most statistical techniques of detecting non-linearity (or chaos) in time series are described and performed comparatively. So far, in main financial theories asset prices have been considered to follow linear stochastic process. However, as stated above, Data Generating Process(DGP) of KOSPI did not exhibit linear structure, especially log return series of KOSPI showed chaotic dynamics in a view of largest LE estimates and BDS test statistics. The existence of non-linearity in stock prices implies at least that standard asset pricing model as such CAPM is not valid. It also means that well-known tradeoff relationship between risk and returns may not be effective.From the descriptive statistics of KOSPI(level and return series), stock index do not follow normal distribution, the approach to measure financial risk as standard deviation leads to false estimation, specifically standard deviation would be underestimation(overestimation) true volatility. In conclusion, our study reveals that the DGP of KOSPI(level) seemed to be nonliear structure in variance term, the DGP of KOSPI(returns) be chaotic dynamics. Such results have key implications regarding the interpretation of stock market behaviors. For example, if returns have chaotic property(i.e. SDIC), it means that investors may use trend-following strategies for excess returns.