On using relative prices to measure capital-specific technological progress

Recently, Greenwood, Hercowitz and Krusell (GHK) have identified the relative price of (new) capital with capital-specific technological progress. In a two-sector growth model, however, the relative price of capital equals the ratio of the productivity processes in the two sectors. Restrictions from this model are used with data on wages and prices to construct measures of productivity growth and test the GHK identification, which is easily rejected by the data. This raises questions about various measures of the contribution that capital-specific technological progress might make to the economy. This identification also induces a negative correlation between the resulting measures of capital-specific and economy-wide technological change, which potentially explains why papers employing this identification find that capital-specific technological change accelerated in the mid-1970s. We impose structure on the productivity measures based on their long-run behavior and find evidence of a slowdown in productivity in the 1970s that is common to both sectors and an acceleration in the mid-1990s that is exclusive to the capital sector.