本文利用Hayami和Ruttan（1985）之兩習段CES函數法，測定臺灣農業在戰前時期之要素使用偏向，實證結果支持「誘發性技術變動或誘發性創新假設」，亦即要素相對價格之變動會誘發要素之使用偏向。而技術變動偏向之測度結果亦顯示，戰前臺灣農業屬勞動要素與變動投入（肥料）要素使用偏向，而固定投入要素節省之方式，耕地之要素變動偏向則屬節省但呈波動變化。同時亦發現戰前臺灣在1920年代因蓬萊米稻種之大量推廣而有所謂「生物性創新」，導致耕地生產力與勞動生產力之提高。美國、日本和臺灣在同一戰前時期之跨國比較結果亦顯示:誘發性技術變動均存在，但因各國之農業資源的豐幅各殊，個別資源間的比例互有差異，因此要素使用偏向亦有不同。

　　　　　This paper adopted the Hayami and Ruttan's (1985) two-stage CES production function approach to measure technical bias of inputs in prewar Taiwan agriculture. Our findings have supported the existence of the induced innovation hypothesis in that period. Empirical result of technical bias measurement indicated that the technical change in prewar Taiwan agriculture was biased in the labor-using, which is opposite to previous research by Shih (1980). While directions of technical change for other inputs were fertilizer-using and fix input-saving, the land input tended to be toward input saving but fluctuated over time. It is also noticed that substantial increase of land and labor productivities of Taiwan agriculture in 1920s can be regarded as a biological innovation of the yield-increasing type resulted form the successful diffusion of a new Japonica rice. Finally, results of the cross-country comparison have indicated that the induced technical change hypothesis were all evidenced for prewar Taiwan, Japan and the U.S. agriculture.