Appendix Table 5.
Effects of Hurricane Within 30km on Birth Outcomes - Hazard Models
| Low Birth Weight (<2500g) | Gestation (weeks) | Any Abnormal Conditions of Newborn (1996–2004) | Any Complications of Labor/Delivery (1996–2004) | |
|---|---|---|---|---|
| A. IV with Mother Fixed Effects | ||||
|
| ||||
| Hurricane 1st Trimester | −0.0029 (0.0091) | −0.0360 (0.0629) | 0.0348** (0.0136) | 0.0194 (0.0194) |
| Hurricane 2nd Trimester | −0.0042 (0.0085) | −0.0188 (0.0602) | 0.0113 (0.0141) | 0.0205 (0.0192) |
| Hurricane 3rd Trimester | 0.0050 (0.0056) | 0.0400 (0.0361) | 0.0227** (0.0087) | 0.0294** (0.0118) |
|
| ||||
| B. IV with Mother Fixed Effects | ||||
|
| ||||
| Hurricane Any Time During Pregnancy | −0.0011 (0.0050) | 0.0229 (0.0338) | 0.0266*** (0.0080) | 0.0288** (0.0110) |
|
| ||||
| C. IV with Mother Fixed Effects - Placebo Models | ||||
|
| ||||
| Hurricane 6 mo. After Birth | −0.0076 (0.0082) | 0.0181 (0.0543) | −0.0240 (0.0131) | 0.0045 (0.0195) |
|
| ||||
| N | 1,440,809 | 1,440,996 | 896,390 | 896,390 |
Notes: Each column in each panel is a separate regression. See notes under Table 1 for information on the sample. For each outcome, births by mothers who have at most one child with non-missing data for that outcome are omitted.
In these models, each child has a unique observation for each month of the 3rd trimester until the time of birth (i.e., children born in the 7th month of pregnancy have one observation, children born in the 8th month of pregnancy have 2 observations, and children born in the 9th month of pregnancy have 3 observations). Exposure to the hurricane is calculated by considering trimesters from the time of conception (estimated using the child’s birth date and gestation length). Exposure in the 3rd trimester is measured on a monthly level. The regressions include mother fixed effects and the key coefficients of interest are instrumented by indicators for a storm/hurricane at the time of the current pregnancy within a 30km radius of the mother’s first pregnancy residence.
All regressions include time-varying controls for mother’s age, education, and marital status, as well as child’s birth order, conception year and conception month, and an indicator for living within 30km of an area that was ever affected by a storm or hurricane over the time period. Robust standard errors are clustered on the mother.
Significance levels:
p<0.10
p<0.05
p<0.001