Technology Adoption, Learning by Doing, and Productivity: A Study of Steel Refining Furnaces

Models of vintage-capital learning by doing predict an initial fall in productivity after the introduction of new technology. This paper examines the impact of new technology on plant-level productivity in the Japanese steel industry in the 1950s and 1960s. The introduction of the basic oxygen furnace was the greatest breakthrough in the steel refining process in the last century. We estimate production function, taking account of the differences in technology between the refining furnaces owned by a plant. Estimation results indicate that a more productive plant was likely to adopt the new technology, and that the adoption would be timed to occur right after the peak of the productivity level achieved with the old technology. We have found that the adoption of the new technology primarily accounted not only for the industry's productivity slowdown in the early 1960s, but also for the industry's remarkable growth in the post-war period. These results are robust to endogeneity in the choice of input and technology.