過去文獻對於傳統線性之共整合分析，無法描述股價指數期貨與現貨兩者間不對稱之非線性調整過程，因此，本研究希望藉由Enders and Granger(1998)的門檻共整合模型與Tsay(1998)的多元門檻模型，來探討臺指現貨、臺指期貨與摩臺指期貨的非線性價格關聯性。實證結果發現，非線性的門檻模型比傳統的線性誤差修正模型，較能顯示出其長期均衡及短期動態關係而且有比較高的解釋能力。任意兩變數間皆存在雙向回饋關係，亦即三變數的變動會相互影響。在價格發現能力以摩臺指期貨最佳，最差的為臺指現貨。投資者可將摩臺指期貨當成領先指標，藉由觀察其價格的變動，作為投資操作或避險套利的資訊。 Previous analyses using linear cointegrative approach failed to describe the asymmetry of nonlinear relationship between stock index and stock index futures. This study uses threshold cointegration and multiple threshold models to investigate the asymmetry of long-run equilibrium and nonlinear relationships among Taiwan stock index(TS), Taiwan stock index futures(TX), and MSCI Taiwan stock index futures(MX), and to explore the determined process of price discovery. The empirical researches indicate a threshold cointegration and nonlinear relationship exist among TS, TX, and MX. MX is the best indicator in price discovery process, while TS the worst. However, there are reciprocal feedback relationships between any two of them which imply any variation of these three indicators cross-influence among them in two regimes. These data suggest MX is the leading indicator of price fluctuation and informative in investing or hedging.
