Is embodied technology the result of upstream R&D? industry-level evidence

In this paper, I develop an industry-level index of capital-embodied R&D by capturing the extent of research and development directed at the capital goods in which a given industry invests. Compiling and adjusting data from the National Science Foundation and Commerce Department, I construct industry-level, time-series measures of this index and investigate its properties. The data allow me to identify the R&D directed at the development of specific types of capital rather than incorrectly assuming industry R&D spending is equivalent to R&D directed at the industry's product, an assumption typically made in the R&D literature. It is first shown that R&D directed at a type of capital is a good measure of its technological change. The constructed index of an industry's capital-embodied R&D is then compared to rates of embodied technological change estimated using plant-level manufacturing data. The index of embodied R&D is found to be positively and significantly related to the estimated rates of embodied technological change. Likewise, embodied R&D is shown to have a positive and significant effect on conventionally-measured total factor productivity growth (i.e. the Solow Residual). This has two implications. First, the capital component of the Solow Residual is generally mismeasured as it does not adequately capture technological change. Second, the constructed index of embodied R&D is proportional to true embodied technological change. Rates of embodied technological change are thus imputed for non-manufacturing industries using the estimated relationship between embodied R&D and embodied technological change found in the manufacturing data.