A constrained nonparametric regression analysis of factor-biased technical change and TFP growth at the firm level

Using firm-level data for Belgium, we study the validity of Hicks neutrality in several sectors that cover the spectrum of knowledge intensity. We find that Hicks neutrality is clearly not supported by the data in different sectors. The results are not sensitive to altering the specification of the technology by including firm age and R&D into the analysis. We also reject Hicks neutrality for a balanced sample, pointing to `within-firm' factor-biased technical change and we also find factor-biased technical change in the pre-crisis era, indicating that unobserved heterogeneity in demand does not drive the results. Overall, our results point towards low-skilled laboursaving and materials-using technical change. So far, this has received little attention and may be linked to ofshoring and global value chain networks. Finally, we show that nonparametric estimates of TFP change that allow for factor biases support the evidence of the recent slowdown in TFP growth in many manufacturing sectors in Belgium. Estimations of TFP and technical change are shown to be sensitive to the estimation method and the specification of the factor bias of technical change.