The Stock Index Period Relationship Before And After The 2015 Stock Market Crash

During the 2015 Chinese stock market crash,stock index futures were traded very actively,and the ratio of their daily trading volume to open interest was extremely high,accompanied by large absolute futures-cash basis in some days.As a result,many people blamed stock index futures for the market crash and strict restrictions were put on the trade of futures contracts.Since then much research has been done to investigate the role played by stock index futures in the market crash,producing few conclusive results.This paper addresses the topic from the perspective of futures-cash price relation,particularly the lead-lag relationship between the intraday return of stock index futures and the underlying stock index.Applying two simple but effective methods?i.e.vector autoregression and multivariate regression—to minute-by-minute high frequency transaction data,the lead-lag relationship between stock index futures return and cash market return in pre-crash,mid-crash,and post-crash period are analyzed and compared.Relationships under different market conditions are also compared.The result is that the lead from futures return to spot return in the mid-crash period was the same or somewhat weaker,not stronger,compared to pre-crash and post-crash period.Besides,during the period of market crash,the lead from futures return to spot return was significantly stronger when the stock index was rising or falling sharply than when the stock index was changing mildly.However,contradicting conventional wisdom,the lead was not stronger when the stock index was falling sharply than when it was rising sharply,implying that at that time stock index futures did not give cash market an asymmetric downward push as many people suggest.Those results undermine to some extent the argument that stock index futures caused the market crash.In this paper,some innovations and improvements are made to the prototype of the two methods.When estimating the reciprocal leading relationship between futures and spot return in each trading day using vector autoregression model,this paper tries not only to answer the question whether there is a lead in each direction,but also to compute the lower and upper bounds for the span of the leading relation through a series of joint significance tests.As for the multivariate regression method utilized in this paper,in addition to the original regression equation,in which spot return is set as dependent variable and the lead,contemporaneous and lag terms of futures return are set as independent variables to estimate the lead-lag relation from spot return to futures return during a period a time,with heteroscedasticity and autocorrelation of error terms accounted for applying Hansen's correction,dummy variables are added to the original equation to assess whether this relationship will change significantly under different market conditions.Another innovation of this paper is that this paper tackles the problem from various angles-different trading days,different period of time,different market conditions in the same period of time.Two methods combined with different analysis angles yield consistent results,confirming the robustness of this paper's conclusion.