為了解氣候變遷對台灣農業之影響,本研究針對台灣最大宗農產品─稻米─建立一無法觀察效果縱橫資料模型,以2001至2010年台灣各縣市的縱橫資料進行實證。為控制社經解釋變數的內生性,採用工具變數估計法。研究結果顯示,生長季之有效積溫、累積降雨量、累積日照時數與稻米單位面積產量之間存在顯著的非線性關係,有效積溫、累積降雨量每增加1%,稻米產量分別約減少0.76%、0.05%,累積日照時數每增加1%,稻米產量約增加0.28%。累積日照時數影響稻米單位面積產量的顯著性,意味著過去僅納入氣溫和降雨量的計量實證模型可能存在內生性問題,進而導致統計偏誤。欲確保估計品質,設計氣候如何解釋農產量之計量實證模型時,日射量或日照時數應為候選之關鍵氣候因子。最後,利用氣候解釋變數的估計係數,推估2001至2010年間,氣候變遷對台灣稻米生產活動造成的經濟損失約1,104百萬新台幣,此估計值讓我們更具體的理解氣候變遷對台灣農業之影響。
To move Taiwan’s climate change policy forward, improved analyses of climate impacts on agricultural sectors using rigorous methodology and high quality data are called for. This study establishes an unobserved effects panel data model to examine the impacts of weather on rice yields in Taiwan based upon panel data for 21 prefectures over 2001-2010. In order to deal with the endogeneity of socioeconomic variables, we estimate the model by instrumental variables. We find that there are significant nonlinear relationships between rice yields and effective temperature summation, accumulated precipitation as well as accumulated sunshine hours during the growing season. When effective temperature summation and accumulated precipitation increase by 1%, rice yields will decrease by 0.76% and 0.05%, respectively. On the other hand, when accumulated sunshine hours increases by 1%, rice yields will increase by 0.28%. Lastly, using estimated weather coefficients, we show that climate change caused a net economic loss of 1,104 million NTD in Taiwan’s rice sector during 2001 to 2010.