Note(s): The dependent variable is lncei (natural log of carbon emission intensity). Model 1 uses lnst (EV stock) as the EV penetration indicator; Model 2 uses lnsa (EV sales) for robustness. L.lncei denotes the one-period lag of lncei. Variable definitions follow Table 1. The System GMM estimator uses the second-to-fourth lags of lncei as GMM-style instruments with the collapse option to limit instrument count; remaining explanatory variables enter as IV-style instruments (Blundell and Bond, 1998; Roodman, 2009). AR(1) and AR(2) are Arellano-Bond tests for first-order and second-order serial correlation in the differenced residuals, with the null of no serial correlation; AR(1) is expected to reject and AR(2) is expected not to reject Hansen's J test of overidentifying restrictions, with the null that the instruments are jointly valid. Instruments reports the total number of instruments used, and Groups reports the number of countries in the panel. Robust standard errors in parentheses. ***p < 0.01, **p < 0.05 and *p < 0.1