LOG IN창 닫기

  • 회원님의 아이디와 패스워드를 입력해 주세요.
  • 회원이 아니시면 아래 [회원가입]을 눌러 회원가입을 해주시기 바랍니다.

아이디 저장

   

아이디 중복검사창 닫기

HONGGIDONG
사용 가능한 회원 아이디 입니다.

E-mail 중복확인창 닫기

honggildong@naver.com
사용 가능한 E-mail 주소 입니다.

우편번호 검색창 닫기

검색

SEARCH창 닫기

비밀번호 찾기

아이디

성명

E-mail

학술자료 검색

선물 시장 거래 활동과 주식 시장 변동성의 상호 작용 : 구조형 벡터 자기회귀 모형

  • 고광수 부산대학교 경영대학 부교수
  • 김태우 부산대학교 대학원 경영학과 박사과정
  • 백미연 부산대학교 대학원 경영학과 박사과정
  • 옥기율 부산대학교 경영대학 교수
본 연구는 구조형 벡터 자기회귀 모형(SVAR: Structural Vector Auto-Regression)을 이용하여, 주식 시장의 변동성 및 선물 시장의 수익률과 선물 거래 활동의 상호 작용을 검증하였다. 기존 연구와 합리적 추론을 바탕으로 식별 문제를 해결하였고, 과도 식별을 이용하여 동시적인 효과, 시차 효과, 전체적인 효과의 유의성을 검증하였다. 분석 결과는 다음과 같다. 첫째, 주식 시장의 변동성 충격은 선물 거래량을 동시적으로 증가시켜, 선물 거래자들은 비대칭적 정보에 의거하여 포지션을 결정한다. 둘째, 각 충격에 대한 미결제 약정의 반응 분석은 비정보 거래자인 헤저의 수요 특성을 잘 보여주었다. 셋째, 선물 시장의 거래량 충격은 주식 시장의 변동성을 감소시키는 긍정적인 역할을 한다. 넷째, 가설 검증을 통해, 동시적인 효과의 중요성을 확인하고, 전체적인 효과에서 다양한 경로를 통한 시차 효과도 중요한 작용을 한다. 마지막으로, 분산 분해와 충격 반응 분석은 가설 검증 결과를 지지하였다. 본 연구는 SVAR 모형을 통해 변수 간 동시적인 관계가 매우 중요함을 보여주었고, 나아가 선물 시장 정책 및 규제의 근거가 될 수 있는 선물 시장과 현물 시장의 유기적 관계를 분석하였다는 데에 의의가 있다.
시장 변동성; 선물 수익률; 선물 거래량; 미결제 약정; 구조형 벡터 자기회귀 모형; Market Volatility; Futures Return; Futures Trading; Open Interest; Structural VAR

The Dynamics of Futures Trading and Stock Market Volatility Using Structural Vector Auto-Regression

  • Kwangsoo Ko
  • Taewoo Kim
  • Miyoun Paek
  • Ki Yool Ohk
To understand the dynamics of futures trading and stock market volatility, this study examines the relationships among the salient endogenous variables: stock market volatility, futures return, trading volume, and open interest. We use a structural vector auto-regression (SVAR) model, which is identified based on the empirical results of previous studies and reasonable inferences about capital markets. To just-identify the SVAR model, first, we assume that futures return shock does not have a contemporaneous effect on stock market volatility. This does not mean that it has no effect at all on stock market volatility. In fact, we conjecture that it must have some effect on the stock market volatility through lagged relationships among the variables. Second, futures return and trading volume shocks are assumed to have temporary effects on each variable. For these restrictions, we follow the method of Blanchard and Quah (1989). Finally, we believe open interest shock does not have contemporaneous effects on stock market volatility, futures return and trading volume because open interest shock is generally viewed as a natural aftermath of trading. Given the above just-identification of our SVAR model, we investigate the contemporaneous, lagged, and overall effects by over-identifying restrictions. The overall effect is classified into two components: contemporaneous and lagged effects. As known, a contemporaneous effect is defined as a concurrent impact of each shock on each variable. Lagged effects are estimated from the reduced-form VAR. Observing these two types of effects, we study the nature of the dynamic relations that exist among endogenous variables. A graphical representation of dynamic relations could be seen from impulse-response function analysis. Variance decomposition is also done to evaluate relative importance of each shock. Our major empirical findings are as follows. First, the volatility shock of stock market increases futures trading volume concurrently. Futures traders are very sensitive to spot volatility shock. In the sense of Black (1986) and Hong (2000), informed and uninformed traders quickly adjust their futures positions based on asymmetric information; consequently, such behavior increases futures trading volume. Second, the responses of open interest to other shocks reflect demand characteristics of hedgers who are not informed. If open interest is determined by hedgers’ demand as mentioned by Bessembinder and Seguin (1993), the contemporaneous effects of various shocks on open interest show the demand characteristics of hedgers, i.e., uninformed traders. The positive effect of volatility shock on open interest implies that volatility shock increases hedgers’ open interest. This is interpreted as hedgers’ behavior to manage spot market volatility. Statistically insignificant is the effect of futures return shock on open interest, which is concurrent irrelevance of futures return with hedgers’ demand behavior. On the other hand, volume shock decreases open interest concurrently. We interpret this phenomenon as hedgers’ liquidation of open interest, which results in increases in the trading volume. Third, futures trading plays a positive role in decreasing stock market volatility. The negative effect of volume shock on stock market volatility is consistent with the fact that the stock market uncertainty may be decreased by noise traders’ behavior. This finding supports the results of Cox (1976), Danthine (1978), Kwon and Park (1997), and Ohk (2005). Fourth, the impulse response function analysis also supports the results of hypothesis tests. Responses of volatility to various shocks fade away much more slowly than those of other variables. Stock market volatility does not respond to open interest at all. Negative response of stock market volatility to volume shock disappears more than twenty days later. On the other hand, responses of futures return, trading volume, and open interest to various shocks quickly disappear within 2 days at most. This behavior is consistent with the results of reducedand structural-form VAR estimation. Fifth, variance decomposition results show relative importance of each shock. Stock market volatility is explained by its own shock (69.48%) and volume shock (24.08%), while most of both futures return and open interest are explained by their own shock (94.02%) in the long-run. Similar to the case of stock market volatility, trading volume is also explained by volatility shock (66.56%) and its own shock (32.09%). These results confirm the close relationship between stock market volatility and trading volume. Finally, test results of hypotheses re-confirm the importance of concurrent relationships and indirect lag effects among overall effects. The validity of reduce-form VAR, however, is not evinced by the fact that indirect lag effects are important, because the contemporaneous effects are too substantial to overlook. Hence, we conclude that the SVAR estimation, accompanied by impulse response function and variance decomposition analyses, is an appropriate method for studying the relationships among the stock and futures market variables. While many other studies have previously investigated the same issue, our study stands out from the rest in research methodology. Most of previous studies estimate the dynamic relationships among the same variables as in this paper, but they used reduced-form VAR. This means that they are not able to consider contemporaneous effects of each variable. Unlike them, this study investigates an important issue of the dynamic relationships among stock and futures market variables using structural VAR. In this way, this paper makes an important contribution to extant research by suggesting that contemporaneous effects should be considered.