Not addressed

Kodak's Rochester share contradicts itself

The introduction still says "more than one-half" of Rochester inventors while Section II keeps the 49% figure. The contradiction flagged last time is unchanged.

Fully addressed

Wrong propensity weight for the ATT estimator

Reweighting now uses p/(1−p) for the ATT estimand, and the closest-city effects in Table 2 move as expected. Resolved in this revision.

Partially addressed

Missing square in the Herfindahl formula

The revision restores the squared shares in generality 1 − Σ p_ik², but Table 3 still prints the old uncorrected scores — text and table disagree.