自2013年以來,中國已先后建立了七個區域性碳市場。研究中國碳市場之間的波動溢出效應,有助于發現在碳市場運行初期起價格引導作用的區域市場,為其它地區碳市場的發展和全國碳市場的建立提供建議。本文選擇廣東、湖北和深圳三個交易量最大的區域碳市場為樣本,利用多元GARCH(1,1)-BEKK模型,檢驗其波動溢出效應。為了消除履約期的影響,本文根據履約期將樣本劃分為階段一(2014年7月1日—2015年6月30日)和階段二(2015年7月1日—2016年7月19日),分階段進行了檢驗,并從市場有效性的視角解釋了檢驗結果。波動溢出效應檢驗結果顯示:在階段一內,廣東碳市場對湖北碳市場、湖北碳市場對深圳碳市場、深圳碳市場對廣東碳市場分別存在單向的波動溢出效應;在階段二內,只存在深圳碳市場對廣東碳市場的單向波動溢出效應;在整個樣本期內,只存在深圳碳市場對廣東碳市場的單向波動溢出效應。基于方差比的市場有效性檢驗結果顯示:無論是階段一、階段二或整個樣本期,三個碳市場均沒有達到弱式有效。本文進一步在多重分形檢驗的基礎上,利用多重分形譜寬度,比較了三個碳市場的非有效性程度,結果顯示:在階段一、階段二和整個樣本期,三個碳市場有效性大小關系分別為:湖北>廣東>深圳、深圳>湖北>廣東、廣東>深圳>湖北。可見,三個碳市場早期的波動溢出效應不完全符合市場有效性大小關系,而較晚期的波動溢出效應符合市場有效性大小關系,這證明了中國碳市場在階段二相比階段一更為成熟。本文最后對中國碳排放權交易市場的發展提出了政策建議。

Since 2013,seven regional carbon trading markets have been set up in China. The study on the volatility spillover effects among the markets will help to identify the leading market in price discovery during the prior stage and provide suggestions for the further development of other regional markets and the establishment of the national carbon market. This paper selects three carbon markets with the largest trading volumes,which are Guangdong,Shenzhen and Hubei respectively,employs the multivariate GARCH( 1,1)-BEKK model to analyze the volatility spillovers on Chinese carbon markets and explains the results from the perspective of market efficiency. To eliminate the effect of compliance,the samples are divided into two phases according to the compliance date:Phase I covers the period from July 1st,2014 to June 30 th,2015 and Phase II covers the period from July 1st,2015 to July 19 th,2016. The empirical results from the volatility spillover tests show that,during Phase I,there are significant spillover effects from Guangdong to Hubei,Hubei to Shenzhen and Shenzhen to Guangdong,respectively. During Phase II as well as the whole sampling period,the spillover effect is only significant from Shenzhen to Guangdong. Empirical results from variance ratio tests indicate that,either in Phase I,Phase II or in the whole sampling period,the three carbon markets have not reached weak-efficiency. Furthermore,based on multi-fractal tests,the spectrum width is calculated to compare the non-efficiency of the three markets and the results show that,during Phase I,II and the sampling periods,the market efficiency degrees are respectively: Hubei > Guangdong > Shenzhen,Shenzhen > Hubei > Guangdong,Guangdong > Shenzhen > Hubei. As it can be seen,the volatility spillover effects completely comply with the efficiency degree during the later stage but not during the earlier stage,which indicates the Chinese carbon markets are more mature in Phase II compared to Phase I. Finally,the paper puts forward policy suggestions for the future development of Chinese carbon markets.