Measuring Knowledge with Patent Data: an Application to Low Carbon Energy Technologies

We estimate a latent factor model (LFM) to compute an index that measures the quality of an extensive data set of inventions related to Low Carbon Energy Technologies (LCETs) and patented by seven countries during 1980-2010. We use the quality index to compute the stock of knowledge accumulated in the fifteen analyzed LCETs. We investigate the composition of the stock of knowledge and find that important substitutions between technologies have taken place: technologies such as solar thermal and nuclear have been progressively replaced by wind power, solar photovoltaic and to a less extent by few other technologies. This substitution effect can be decomposed into quantity (the number of inventions) and quality (the quality of inventions). Investigating the latter, the quality of nuclear-related inventions has decreased whereas it has increased for solar photovoltaic (PV), wind power and energy storage inventions. Few newer technologies, i.e. hydrogen and sea energy, also show signs of an increase of their average quality of inventions over the last years of the data set. We go further and investigate the inventions distribution in terms of quality and conclude that the potential for signifcant inventions related to nuclear technology has decreased over time whereas higher levels of quality have been reached in newer technological areas. A cross-country comparison is conducted to assess the innovation performance of the seven countries covered by our study. We conclude that technology policies are less efficient when demand-pull and supply-push approaches are not coupled.