Reconsidering Non-neutral Technical Change (with Ming Li, Maastricht University)
It is not a coincidence that the current debates on inequality coincide with a renewed interest in non-neutral (particularly skill-biased) technical change. In this paper, we revisit the measurement of non-neutral technical change, keeping in mind the inequality discussion and with an emphasis on skill-biased technical change. We start with a simple 2-firm model to present an intuition of where possible measurement problems of non-neutral technical change might come from, and how important accounting for inefficiency is in mitigating these problems. Subsequently, we generalize this model and formally derive the measurement bias that can exist in many non-neutral technical change measurements, as well as the conditions under which this bias can exist. Next, we use simulations based on the latest version of the Penn World Data to provide a first assessment of the magnitude of this bias. Finally, we use the World Input-Output Database to estimate non-neutral technical change for industry data covering 40 countries over the period 1995-2009. We indeed find biases in the estimates of skill-biased technical change, and discuss the importance of these findings for the inequality debate.