This paper introduces a new data-set containing estimates of the “quality” of each patent granted in England in the period 1617-1850. Following the approach of Lanjouw and Schankerman (2004), the quality indicator is constructed as a composite index integrating different pieces of information on the characteristics of the patent or of the patentee.

The first piece of information we use is the Reference Index of Patents of Invention, 1617-1852 edited by Bennet Woodcroft and published in 1862. For each patent, the book provides a list of references (either to technical and engineering literature or to legal proceedings) where the specification is mentioned. Our assumption is that the "visibility" of each patent in Woodcroft's Reference Index will provide a reasonable proxy for its technical and economic significance. This approach is analogous to the use of patent citations for assessing the value of patents in the contemporary literature. The Reference Index has been already used by Nuvolari and Tartari (2011) in a first exploratory attempt of constructing a patent quality indicator for English patents.

The second piece of information we use is whether the patent was involved in legal suits or on whether the patentee requested an extension of the patent to Parliament. We assume that only patents of non-trivial economic value were likely to be involved in legal suits or extension procedures.

The third piece of information we consider is the inclusion of the patentee in biographical dictionaries such as the Oxford Dictionary of National Biography and other more specialized biographical dictionaries. The underlying assumption is that patents taken by the inventors that were included in these reference works were inventions of particular economic or technological significance. This approach has been pioneered by Khan and Sokoloff (1993)

The fourth piece of information we use is the inclusion of the patent in a number of reference texts of history of technology. Again, we assume that patents covering inventions mentioned in reference works of history of technology were of relatively high quality.

We aggregate these different pieces of information in a quality score using a latent factor model similar to that adopted by Lanjouw and Schankerman (2004) for contemporary patents. The reliability of our quality indicator is corroborated by the fact that what we regard, with the benefit of hindsight, as the critical breakthroughs of the industrial revolution (eg, Watt's steam engine, Arkwright's water-frame, Tennant's chlorine bleaching,
etc.) are all characterized by very high quality scores.

The final step of our exercise is to re-asses the patterns of innovation characteristic of the first industrial revolution using our new indicator patent quality.
Concerning the timing of the industrial revolution we find a clustering of high-quality patents in the period 1760-1790. Concerning the scope of the industrial revolution, our findings indicate that, although total patents were relatively widespread, high-quality patents (i.e., those covering technological blockbusters) were much more concentrated across industries, being localized especially in three industries: textiles, engines, and iron production . If we regard productivity growth as an outcome of the sustained improvement and extension of these macroinventions, our results indicate that the patent records evidence may indeed be consistent with a view of the Industrial Revolution as a process driven by a few revolutionary innovations localized in a relatively circumscribed segment of the economy.
 

VIEW PAPER