Combinatorial growth with physical constraints: evidence from electronic miniaturization

In the past sixty years, transistor sizes and weights have decreased by 50 percent every eighteen months, following Moore's Law. Smaller and lighter electronics have increased productivity in virtually every industry and spurred the creation of entirely new sectors of the economy. However, while the effect of the increasing quality of computers and electronics on GDP has been widely studied, the question of how electronic miniaturization affects economic growth has been unexplored. To quantify the effect of electronic miniaturization on GDP, this paper builds an economic growth model that incorporates physical constraints on firms' production sets. This model allows for new types of productivity spillovers that are driven by products' physical characteristics. Not only are there spillovers from changes in industry productivity, but also, there can be "size spillovers," where the miniaturization of one industry's product leads to miniaturization of products that are downstream in the supply chain, reflecting how transistor miniaturization has led to the decrease in size of a large variety of electronic products. Using a new data set of product weights and sizes, we test the predictions of the model and show that Moore's Law accounts for approximately 3.5 percent of all productivity growth in the 1982-2007 period, and for 37.5 percent of the productivity growth in heavy manufacturing industries. The results are robust under multiple specifications, and increase in strength during the 1997-2007 subperiod.