Reallocation and Technology

Roughly half of the productivity gain in the steel industry came from efficient minimills displacing older and less productive vertically integrated steel plants.

Between 1963 and 2002, the productivity of the U.S. steel industry soared as producers shuttered outdated mills and laid workers off. In Reallocation and Technology: Evidence from the U.S. Steel Industry (NBER Working Paper No. 18739), Allan Collard-Wexler and Jan De Loecker find that a key driver of that productivity was the arrival of a new technology: the minimill. They estimate that roughly half of the productivity gain in the steel industry came from efficient minimills displacing older and less productive vertically integrated steel plants. Furthermore, the increased competition from minimills caused the remaining vertically integrated producers to become more productive as well.

"[T]he main reason for the rapid productivity growth and the associated decline in employment is neither a steady drop in steel consumption, nor a consequence of globalization. Nor is it a displacement of production away from the Midwest," they write. Although it is well-established that an industry's reallocation of resources can substantially boost productivity, this is one of the first studies to provide a mechanism -- in this case, the introduction of a new technology -- for how it works.

The effects can last over decades. Between 1963 and 2005, the steel industry lost some 75 percent of its employees, but its product shipments by the end of the period nearly matched levels at the beginning. Worker productivity soared five-fold and total factor productivity rose by 38 percent.

By looking at plant-level data, the authors are able to reconstruct how the steel industry became more productive. The first minimill was built in the late 1950s. In the 1960s, minimills began to make inroads into the lower end of the steel market, primarily steel bar, because on average they were 11 percent more productive than the traditional big, vertically integrated plants. About one third of the industry's overall productivity surge can be attributed to production of bar and other low-end products moving to minimills. But the vertically integrated sector didn't stand still. It began to close its bar-producing mills. The authors calculate that these closures automatically made the vertically integrated producers more productive, because those mills making steel sheet were already more productive than those making bars. This reallocation of resources -- rather than improvements made at the surviving big mills -- also explains a large portion of the industry's growth in productivity. By 2002, the average vertically integrated mill was as productive as a minimill, although the two made different products. Minimills themselves also have made productivity improvements, allowing them to begin to compete in higher-margin products. "Our results suggest that the total effect of minimill entry on industry-wide productivity growth was 87 percent," the authors conclude -- some 50 percent directly from the entry of minimills and another 37 percent from the gains among vertically integrated mills.

--Laurent Belsie