Table 3.
The effects of pollution on deaths and cases.
| (1) |
(2) |
(3) |
(4) |
(5) |
(6) |
(7) |
(8) |
|
|---|---|---|---|---|---|---|---|---|
| Log COVID-19 Deaths | Log COVID-19 Deaths | Log COVID-19 Deaths | Log COVID-19 Deaths | Log Confirmed COVID-19 Cases | Log Confirmed COVID-19 Cases | Log Confirmed COVID-19 Cases | Log Confirmed COVID-19 Cases | |
| Treated Counties After the Rollback | 0.1331∗∗∗ (0.0160) |
0.1533∗∗∗ (0.0168) |
0.1055∗∗∗ (0.0272) |
0.1410∗∗∗ (0.0348) |
0.6856∗∗∗ (0.0450) |
0.7078∗∗∗ (0.0479) |
0.5296∗∗∗ (0.0962) |
0.2308∗∗ (0.0989) |
| With State Fixed Effects and controls | X | X | ||||||
| With County Fixed Effects and daily controls | X | X | X | X | X | X | ||
| With County-Specific Linear Time Trends | X | X | ||||||
| Limited to Counties with Population Density >250 in the Control Group | X | X | X | X | ||||
| Limited to Populations between 10K and 1.64 million | X | X | X | X | X | X | X | X |
| Mean of the Dependent Variable | 0.194 | 0.194 | 0.194 | 0.194 | 1.148 | 1.148 | 1.148 | 1.148 |
| County-Day Observations | 137716 | 137815 | 84126 | 84126 | 137716 | 137815 | 84126 | 84126 |
Notes: Columns 1–4 present the results for different regression specifications with the log of COVID-19 deaths as the outcome. Columns 5–8 present the same specifications with the log of confirmed COVID-19 cases as the outcome. Columns 1 and 5 show the results for the sample of counties with 1 or more TRI sites, no deaths in the period before the rollback, and limited to populations between 10,000 and 1.64 million using state fixed effects, controlling for total population, population density, percent white, percent Black, percent Hispanic, poverty rate, the unemployment rate, median income, and the percent of workers who are likely to be essential. Columns 2 and 6 show the results on the same sample with county fixed effects. Columns 3 and 7 show our preferred specification that further limits the control group to counties with a population density of more than 250 persons/mi. Columns 4 and 8 show the results from Columns 3 and 7 with county-specific linear time trends. All models also control for an indicator for being after the EPA’s rollback, social distancing, stay at home orders, re-openings, mask mandates, days since the first COVID death, total tests, weather, and day of the week, county and month fixed effects. Columns 1-4 additionally control for daily confirmed COVID-19 cases. Standard errors are clustered at the county level and are in parenthesis. Coefficients labeled as ∗∗∗, ∗∗, and ∗ are statistically significant at the 1, 5, and 10 percent levels, respectively.