The Measurement and Sources of Technological Change Biases, with an Application to Postwar Japanese Agriculture

Abstract

Technological change bias in agriculture has important effects on and is affected by other changes in an economy. Labour-saving technological change, for example, enables more farm labour to migrate to the non-farm sector (Kako, 1978). A land-using bias, on the other hand, may stimulate efficiency differentiation among farm groups and lead to a rise in farm land prices (Lee 1980a), while a machinery-using bias accelerates the rate of investment in agricultural machinery. All of these affect the income distribution as well as the economic growth path of the economy. In spite of its importance, there are relatively few empirical studies of biases in technological change. The purpose of this paper is to present, first, an empirical method of measuring technological change biases in many-factor production with an application to postwar Japanese agriculture, and then to investigate the factors that guided the evolution of biases in agriculture. Sources of biases have been one of the critical concerns of growth economics. Relative factor prices have been asserted to be the prime motivator of biases. However, land size per farm, output price and innovation lags are also important factors affecting technological change biases. The fundamental method used here to test these hypotheses is to measure the biases in four regions among which economic variables have been lifferent or have moved at different rates, and then to examine the relationship between the measured biases and the economic variables. Since the Japanese agriculture sector has undergone substantial technological changes in the rapid economic growth since the Second World War, it provides an ideal case study which is of considerable relevance to similar but less developed countries in South-east Asia. In the first section, we present a theory of measuring biases in many-factor production, using a production function approach. In Section II, the homogeneous translog production function is estimated with micro-farm data on rice production in postwar Japan. Section III analyses the characteristic structure of technological change in postwar Japanese agriculture, and then investigates the sources of the biases following the method presented. In the final section, we draw conclusions about the sources and implications of the measured biases.