Are farmers learning by doing? Experience in Taiwan

There has been an increasing concern for agricultural development problems faced by developing countries. Identifying the major differences and similarities in agricultural growth patterns between developed and developing countries is important. This study proposes an empirically implementable method that: (1) accounts for the gain in efficiency accompanying learning; and (2) permits an empirical test of the learning-by-doing hypothesis in Taiwan agriculture. Empirical testing of the learning-by-doing hypothesis proceeds in two stages. The first stage examines the dynamic structure of production agriculture in Taiwan. The results indicate long lags in the adjustment of both agricultural capital and labor, and also suggest that asset fixity is an important characteristic of Taiwan production agriculture. The second stage tests the learning-by-doing hypothesis with dynamic adjustment as the maintained hypothesis. The results suggest rejecting the null hypothesis of the no learning effect in favor of the presence of learning. Except for variable input demand, most of the empirical estimates have signs consistent with those predicted by the comparative statics analyses. These results also suggest that farmers in Taiwan are learning through experience, and that the dynamic learning model is a valid specification showing their intertemporal optimization behavior. An earlier study indicated that technical change is consistently biased against labor but towards capital and intermediate inputs for U.S. production agriculture. This study demonstrates that material-using, labor-using, and capital saving were characteristics of technical change in Taiwan production agriculture during the past two decades. This pattern of factor bias illustrates the possible influence of resource endowments on technological change.