The fact that labor productivity growth is different across sectors is well known. Average annual labor productivity growth between 1960 and 2017 in the US, for instance, was 2.49% in the goods sector, much higher than the 1.53% in low-skilled and the 0.72% in high-skilled services. However, there is no consensus on the origins of these differences.
In this paper we study the drivers of sectoral labor productivity growth in a productionside framework. We consider various types of occupational labor as distinct factors in the sectoral production faction, allow for technological change to be sector-and-factor specific, and infer the evolution of the sector-and-factor specific technologies over time directly from the data drawing on the model’s optimality conditions.
Our results show that technological change has been very far from neutral. That we do not impose that sectoral technological change is factor-neutral, nor that factor-specific technological change is uniform across sectors, turns out to be crucial. Technologies have evolved at very differential rates, both across factors within each sector and across sectors for a given occupation or type of capital. In particular, amongst the labor-augmenting technologies those augmenting routine occupations have been growing the fastest in all sectors, but at very different rates: at 5.59% per year in goods, at 2.92% in low-skilled services and at 1.32% in high-skilled services.
Through a series of counterfactual simulations, we study the role of inputs and of technological change in labor productivity growth. We find that the single most important driver of sectoral labor productivity growth differences are the sector-specific growth rates of routine labor augmenting technologies. Without sector-specific routine labor augmenting technological change, labor productivity growth would have been almost equalized across sectors. This type of technological change explains at least 59 percent of labor productivity growth in low-skilled services, 74 percent in goods and 21 percent in high-skilled services. Moreover, in terms of labor productivity growth in the aggregate, we show that the contribution of routine labor augmenting technological change is large and increasing over time. In its absence aggregate growth would have been lower by about a third between 1960-1990, and there would have been hardly any growth over 1990-2017.
These counterfactuals also allow us to evaluate the role of various other channels proposed in the literature for sectoral productivity growth differences, such as differential capital intensities and capital accumulation, or differential sectoral intensities in occupational employment and technological change specific to occupations. We show that while capital accumulation contributes to labor productivity growth (without it growth would have been 39 percent lower on average), it does not generate the sectoral differences observed in the data. Similarly we find that in terms of occupational employment structure, differences across sectors as well as changes over time within sectors hardly matter for sectoral labor productivity differences.
You can download the complete paper here .