台灣毛豬市場批發價格的非線性模型分析

The Non-linear Model Analysis of Wholesaler Price in Taiwanese Hog Markets

台灣毛豬產業無論就肉類消費、農業生產與對外貿易均佔農業部門相當重要的地位，也是農業經濟領域重要的研究議題。本文以平滑轉換自我迴歸模型（STAR），檢定台灣地區第一等級8個地區毛豬市場批發價格是否呈非線性走勢，並描述其動態調整行為及進行樣本外預測；此外討論口蹄疫事件及市場重整對豬價的影響，進而比較消費地與產地批發市場之價格分析的差異。實證結果顯示，1997年口蹄疫事件及2000年市場重整為造成豬價呈非線性走勢的重要因素，其中台北縣、桃園縣、台中縣大安區、台中市、彰化縣、雲林縣等6地區呈現對稱調整的ESTAR模型走勢，表示豬價上升及下降具有相同的動態結構，而中間區域則不同於外部區域的動態過程。不同地，屏東縣、高雄縣鳳山區則呈現不對稱調整的LSTAR模型走勢，表示豬價在門檻值的上、下具有不同走勢。此外，8個地區的毛豬價格在調整速度、調整型態、門檻值內、外區間的動態走勢、落遲現象以及價格預測的影響效果都有不同的表現；而消費地與產地批發市場也呈現不同的非線性動態走勢。

This paper applies the smooth transition autoregressive (STAR) model to investigate whether the hog wholesaler price with first grade show a non-linear path or not in 8 areas of Taiwan, and to describe the dynamic adjustment behaviors along with the out-of-sample forecasting. In addition, we will discuss the effects of epidemic FMD' s (Foot and Mouth Disease) shocks and market reorganizations on hog' s price, as well as the analysis of price comparison in consumer and wholesaler markets. Our empirical evidences support that the FMD in 1997 and the market reorganization in 2000 both are substantial factors which have caused the hog price to follow a non-linear path. Next, in 6 areas (Taipei county, Taoyuan county, Daan area, Changhua county, Yunlin county), the hog prices have a tendency of symmetrical adjusted EST AR model, which therefore indicates the hog price has the same dynamic structures while the price shows motion within upper and lower regimes, but the dynamic process in the middle areas differs from the external areas. However, the difference sourced from Pingtung and Fongshan, showing the tendency of asymmetrical adjustment under LSTAR model, in these areas, we suggest that all series display the asymmetric path above and below the threshold value. Furthermore, the hog price in 8 areas has an entirely distinct manifestation in adjustment speed, type, threshold value, external dynamic path, lag phenomenon and price forecasting results etc. Finally, hog price also presents inconsistency of non-linear dynamic path in consumer and wholesaler markets.