Kodak's Rochester share contradicts itself
The introduction states Kodak employed "more than one-half" of all Rochester inventors at its 1996 peak, but Section II specifies 49% — mathematically less than half.
Document
Feedback
The Effect of High-Tech Clusters on the Productivity of Top Inventors
Enrico Moretti · UC Berkeley · American Economic Review 111(10)
I. Introduction
A stark illustration of how a dominant employer shapes a local ecosystem is the trajectory of Kodak in Rochester, NY. At its peak in 1996, Kodak employed more than one-half of all the inventors in Rochester. The subsequent collapse of the company offers a natural experiment.
High-tech innovation is concentrated in a small number of dense urban clusters that produce a disproportionate share of US patents.
I.A. Empirical approach
The baseline specification relates the log number of patents to the log size of the local cluster, excluding the inventor's own contribution.
For the Rochester event study I complement the baseline with a propensity-score reweighting strategy. The weight for inventors who are not in Rochester is 1/(1−p), where p is estimated using a logit. Data and estimation details are reported in Moretti (2021a).
II. Data and variables
A patent's generality captures the breadth of fields that cite it, following Hall, Jaffe, and Trajtenberg (2001).
More specifically, generality is defined as 1 − Σk pik where pik is the share of citations to patent class k. The sum is the Herfindahl concentration index.
I augment the patent panel with city-level demographic and economic controls merged at the MSA-year level.
Wrong propensity weight for the ATT estimator
Table 2 weights control inventors by 1/(1−p), but the ATT estimand requires p/(1−p). With low base rates the current weight barely upweights the closest cities.
Missing square in the Herfindahl formula
Generality is written as 1 − Σ pik, but the Herfindahl index uses squared shares: 1 − Σ pik2. Without the exponent every score collapses to zero.