Table 7—

Expansions

	(1)	(2)	(3)	(4)
Post	−0.0044 (0.0039)	−0.0154 (0.0091)	−0.0002 (0.0031)	−0.0046 (0.0102)
Post × x			−0.0076 (0.0072)	−0.0134 (0.0141)
Satellite	Y	Y	Y	Y
Cont. × year	Y	Y	Y	Y
City × cal. mo.	Y	Y	Y	Y
Climate × cont.	Y	Y	Y	Y
Mean AOD	0.44	0.63	0.44	0.63
R 2	0.78	0.73	0.78	0.73
Number of events	143	42	143	42
Number of cities	70	20	70	20
Observations	52,268	15,595	52,268	15,595

Notes: Dependent variable is mean AOD in a 10km disk with centroid in the city center. Column 1 includes all expansions anywhere in the world (includes old subway cities outside of our main sample). Column 2 includes expansions restricted to the set of new subway cities used in Table 2. Column 3 is the same as 1 but includes the interaction of treatment with indicator for AOD in 2000 above 0.36 (median AOD in 2000 in the sample of new subway cities used in Table 2). Column 4 is the same as 2 but includes the interaction of treatment with indicator for AOD in 2000 above.36 (median AOD in 2000 in the sample of new subway cities used in Table 2). All specifications control for event fixed effects, event-specific pre-window indicators, event-specific post-window indicators, and event-specific period-0 indicators. Climate controls are pixel count and linear and quadratic terms in temperature, precipitation, cloud cover, vapor pressure and frost days. Standard errors clustered at the event level are in parentheses.