Regression results using matched samples.

This table reports the results on the effects of flooding on firm performance using the matched sample. To construct the matched sample, we first define treatment firms as firms exposed to flooding in a year during our sample period. We then match each treatment firm with a control firm that is not exposed to flooding in that year and is located in the same province, operating in the same industry (2-digit industry classification by the China Securities Regulatory Commission), and having the closest total assets as the treatment firm. We also require the matched pair of treatment and control firms to be located very close to each other, with the geographic distance being less than 100 kilometres or 50 kilometres, so that they are likely to face very similar local economic conditions and crisis management abilities. In Panel A, we require the geographic distance between the matched treatment and control firms to be less than 100 kilometres. In Panel B, we further restrict the geographic distance to be less than 50 kilometres. Control variables used in Table 2 are included in columns 2 and 4. Variable definitions are in Table A1 in the S1 Appendix. Robust standard errors are clustered at the city-year level and reported in parentheses. *, ** and *** indicate the significance levels at 10%, 5% and 1%, respectively.