Firm-Heterogeneous Biased Technological Change: A nonparametric approach under endogeneity

•We consider technological change that is biased and firm-heterogeneous.•We propose a nonparametric framework to estimate factor biases under endogeneity.•We confirm in a Monte Carlo simulation that we adequately deal with endogeneity.•We apply our framework on a rich dataset of Belgian manufacturing firms.•We show how firm characteristics relate to firm-specific biased technological change. We propose a fully nonparametric framework to test to what extent technological change is factor-biased and heterogeneous. We show in a Monte Carlo simulation that our framework resolves the endogeneity issue between productivity and input choice and provides accurate estimates of firm-specific biases. For all Belgian manufacturing industries analyzed, we reject the predominant assumption of Hicks-neutral technological change over the period 1996–2015. We find that technological change is skill-biased, capital saving and domestic materials using. Moreover, we find significant heterogeneity in the pattern of technological change between and within industries. Relying on a rich dataset of firm characteristics, we provide robust indications that firm-level technological change can be attributed to specific firm strategies and technological characteristics.