Measuring Moore's Law: Evidence from Price, Cost, and Quality Indexes
"Moore's Law" in the semiconductor manufacturing industry is the predictable evolution of a manufacturing technology platform that has continuously reduced the costs of fabricating electronic circuits since the mid-1960s. This chapter reviews how this progression delivered 20 to 30 percent average annual transistor manufacturing cost declines while it lasted. We review how other characteristics associated with smaller feature sizes created additional economic value, and assess their historical trends. These benefits pose special challenges for price and innovation measurement, and motivate quality adjustment methods when measuring semiconductor product prices. Empirical analysis reveals a slowdown in Moore's Law as measured by quality-adjusted prices for the highest volume products: memory chips, custom chip designs outsourced to foundries, and Intel microprocessors. We assess whether Intel microprocessor price trends might diverge from those of other semiconductor chips. A model of chip characteristics' effects on microprocessor performance is specified and tested in a structural econometric model of processor performance. A small set of characteristics explains 99% of variance across processor models' performance on common benchmarks. The diverse effects of changing characteristics on manufacturing costs create a rationale for their inclusion in hedonic price models separate from their demand-side effects on computer performance.