Almost by definition, radical innovations create a need to revise existing classification systems. As a result, the evolution of technological classification systems reflects technological evolution. We propose three sets of findings regarding classification volatility in the U.S. Patent Classification System.
First, we study the evolution of the number of distinct classes. Reconstructed time series based on the current classification scheme are very different from historical data. This suggests that using the current classification to analyze the past produces a distorted view of the evolution of the system.
Second, we study growth rates and relative size of classes. The size distribution is exponential but older classes are not larger. To explain this pattern we propose a simple stochastic growth model in which classes are regularly split at random.
Third, we study reclassification. The share of patents that are in a different class now than they were at birth can be quite high - for instance it reaches for 40% for 1976. Reclassification mostly occurs across classes belonging to the same 1-digit NBER category, but not always. We also document that reclassified patents tend to be more cited than non-reclassified ones, even after controlling for grant year and class of origin.
More generally we argue that classification changes and patent reclassification are quite common instances, reveal interesting information about technological evolution, and must be taken into account when using classification as a basis for forecasting.
This is a joint work with Daniel Kim, and a preliminary draft is available upon request.


Research Programmes