Disentangling the directions of technical change: a new growth accounting method and application to green technical change

This paper develops a parsimonious framework to compute the bias of technical change and disentangles factor-saving innovation from factor substitution. It does not require estimating complex elasticities of substitution between factors. I apply this framework to industry-level data and answer key questions: What has been the bias of technical change across US and EU industries? Does an increase in the price of one factor spur specific factor-saving innovation? Can we accurately forecast the evolution of factor shares? Is green technical change skill-biased? I represent technical change as an increase in the technology menu available, it results in a shift in the local production function around a specific capital to labour ratio. I first validate the model with only capital and labour, then extend it to energy, high and low skills labour, material and services. I find that most industries are energy-saving and capital-biased, with a growing trend of labour-saving technical change along with capital substituting labour. I find significant evidence of labour-saving technical change induced by the rising cost of labour in a number of industries. This framework provides better out-of-sample forecasts of factor shares than CES, Cobb-Douglas and Leontief functions.