Does Hedonic Price Indexing Change Our Interpretation of Economic History? Evidence from Swedish Electrification

Rapid price decreases for ICT-products in the 1990s have been largely attributed to the introduction of hedonic price indexes. Would hedonic price indexing also have large effects on measured price and productivity during other technological breakthroughs? This paper investigates the impact of hedonic and matched model methods on historical data for electric motors in Sweden 1900-35. The results show that during the productivity boom of the 1920s, the constant prices for electric motors decreased by 9.7 and 8.1 percent per year depending on whether hedonic or matched model price indexes were used. This indicates high productivity growth in the industry producing electric motors 1920-29. In contrast to Sweden, the US annual total factor productivity was only, according to current best estimates, 3.5 percent in Electric machinery compared to 5.3 percent in manufacturing 1920-29. However, hedonic price indexes were not used to calculate US productivity. Moreover, in comparison to the matched model, the hedonic price index on average overestimates price decreases when prices are decreasing and overestimates price increases when prices are increasing. However, the total effect of the two different price indexes remains approximately the same in 1900-35. Finally, it is shown that the price decreases for electric motors in the 1920s are not in par with the price decreases for ICT-equipment in the 1990s, even if hedonic indexing is used.