台灣毛豬市場批發價格與交易量之非線性關係探討－－雙變量門檻迴歸模型之應用

On the Non-Linear Relationship between the Volume of Transaction and Price in the Taiwan Pork Market-An Application of Bivariate Threshold Regression Model

本研究使用1981年1月至2009年12月台灣地區毛豬的價格與交易量資料，應用Tsay（1998）的多變量門檻迴歸模型來探討當d期前的交易量大於或低於某一門檻水準時，價格與交易量之間的互動關係是否會有所不同。實證結果發現：當11期前的交易量為一種極端小交易量時，本期若產生一個正向價格衝擊時，交易量後續將不會有顯著的反應關係，但若在本期產生一個正向（負向）的交易量衝擊時，價格將在第6期出現一種負（正）向的反應關係。另一方面，當11期前的交易量不是極端小的交易量時，本期若產生一個正（負）向價格衝擊時，交易量將在第2-4期及第14期出現一種負（正）向的顯著反應關係；但若在本期產生一個正（負）向交易量的衝擊時，後續的價格走勢將不會有顯著的反應關係。 Applying Tsay's multi-variate threshold regression model to volume of transaction and price of pork of Taiwan (January 1981 through December 2009), this paper explores the different interaction patterns between price and volume of tranaction. The results are encouraging. First, when the transaction volume of pork 11 periods before is very small, we observe the following: (i) to a positive shock on the current price, the responses of future volumes are insignificant, and (ii) to a positive (negative) shock on the current volume, the price six periods later is expected to decreases (increases). Second, when the volume 11 periods before is not very small, we observe that (i) the volumes 2~4 and 14 periods later respond negatively (positively) to a positive (negative) shock to current price, and (ii) future price changes are insignificant in response to a positive (negative) shock to current volume of transaction in the Taiwan pork market.

Applying Tsay's multi-variate threshold regression model to volume of transaction and price of pork of Taiwan (January 1981 through December 2009), this paper explores the different interaction patterns between price and volume of tranaction. The results are encouraging. First, when the transaction volume of pork 11 periods before is very small, we observe the following: (i) to a positive shock on the current price, the responses of future volumes are insignificant, and (ii) to a positive (negative) shock on the current volume, the price six periods later is expected to decreases (increases). Second, when the volume 11 periods before is not very small, we observe that (i) the volumes 2~4 and 14 periods later respond negatively (positively) to a positive (negative) shock to current price, and (ii) future price changes are insignificant in response to a positive (negative) shock to current volume of transaction in the Taiwan pork market.