本研究主要利用非線性模型來探討價差(基差之負值)與台灣期貨和現貨股價指數的關聯性,並分析期貨市場中交易者的投資行爲與策略。實証方法以Hansen and Seo(2002)的門檻共整合模型爲基礎,延伸出三區塊門檻共整合模型,估計出台股期貨與現貨之價差的兩個門檻值。實證結果發現,價差調整回均衡值的速度,隨價差偏離零值的程度越大而越快,且此均數復歸(mean reversion)現象存在非線性結構,又大正價差的復歸速度較大逆價差高。再者利用實務的投資操作模擬,分析投資人如何因應不同的價差變化,作出較適當的交易策略。實證模擬結果中隱含:(1)將價差分成三區塊來分析,較傳統上僅分正/負兩區塊價差爲佳;(2)投資者在開盤時似有過度反應的現象;(3)投資人在期貨市場有獲得超額報酬的可能性;(4)反向操作策略(逆向回饋交易)在台灣期貨市場是可獲利的。
The purpose of this research is to examine empirically the relationship between futures and spot prices in Taiwan stock markets employing a nonlinear model. We attempt to understand how the spread (or negative of basis) behavior affects the market price change and therefore the trader's strategy and behavior. This study extends the Hansen and Seo (2002) model to three-regime threshold cointegration model and estimate two threshold parameters of the spreads. Furthermore, our empirical results show that the speed of adjustment of the spread towards its equilibrium is found to be an increasing function of the size of the spread deviation from zero and the spreads exhibit nonlinearly mean-reverting behavior. The bigger the spread is, the greater the speed of mean reversion is. And large positive spreads show greater mean reversion than large negative spreads. A simulation of trading strategies is conducted to determine the optimal trading strategy under different spreads. In particular, there are four findings from simulation: (1) we find that three-regime model generally outperforms two-regime model. It is better to use three-regime model in analyzing the stock market. (2) Market could over-react to the spread in the beginning of the open section; (3) the participants of the futures market may have excess return; (4) the contrarian strategy (negative feedback trading) works in the future market of Taiwan.