刮风时：农业火灾风险、父母投资和长期结果

Appendix Hai Hong(China Academy for Rural Development, Kevin Chen(1.China Academy for Rural Development,School of Public Affairs, AppendixA:additionaltables andfigures Additionaltables Notes:This table presents the heterogeneity of our baselineresults, using the rural sample.We consider theheterogene-ouseffectsregarding three variables: mother’s education level (whether completed middle school), family income level(whether thefamily incomeis above the median value), and whether the familyisengaged ingrainproduction.The depend-ent variable is a normalized health index withagreater valuerepresentingworse healthconditions.Observations areat thecounty-cohort level, with each cohort defined byitsbirth month. Individual controls include gender, age, father’s educationand age, mother’s education and age, family income, family size, andnumber of siblings. Meteorological controls include Notes: This table presents theestimated results forthealternative clustering adjustment, using the rural sample. The dependent variablein col-umns (1) and (2) is the unhealthiness index, in columns (3) and (4), it’s the standardized word test score, in columns (5) and (6), it’s the positivemeasure of non-cognitive ability, while in columns (7) and (8), it’s the negative measure of non-cognitive ability. In odd columns, wetwo-way clus-ter the standard error at the county and province by birth year level, whereas in even columns, we cluster the standard error at the prefecture level.Observations areat the county-cohort level, with each cohort defined byitsbirth month.All regressions include both individual and weather con-trols. Individual controls include gender, age, father’s education and age, mother’s education and age, family income, family size, and number of Notes: This table presents theestimated results forcontrolling for additional pollution sources, using the rural sample.In odd columns, we in-clude controls for whether households use straws as their primary source of fuel. While in even columns, we additionally control for the fertilizerexpenses asa proxy forpotential exposure to water pollution.Thedependent variablein columns (1) and (2) is the unhealthiness index, in columns(3) and (4), it’s the standardized word test score, in columns (5) and (6), it’s the positive measure of non-cognitive ability, while in columns (7) and(8), it’s the negative measure of non-cognitive ability.Observations areat the county-cohort level, with each cohort defined byitsbirthmonth.Allregressions include both individual and weather controls. Individual controls include gender, age, father’s education and age, mother’s education Notes: This table presents theestimated results forcontrolling for additionalinfrastructureand facility,using the rural sample.In odd columns, weinclude controls forhealth and educational facilities (i.e., number ofkindergartensand primaryschools, and number ofhospitalsandpharmacies). Whilein even columns, we additionally control for theinfrastructure constructionatthe village level (i.e.,access to electricity, road, and railway).Thedependentvariablein columns (1) and (2) is the unhealthiness index, in columns (3) and (4), it’s the standardized word test score, in columns(5) and (6), it’s the pos-itive measure of non-cognitive ability, while in columns(7) and (8), it’s the negative measure of non-cognitive ability.Observations areat the county-co-hort level, with each cohort defined byitsbirth month.All regressions include both individual and weather controls. Individual controls include gender, Notes: This table presents theestimated results forcontrolling forcontemporaneousweather and pollution conditions, using the rural sample.In oddcolumns, we include controls forcontemporaneous weather conditions(i.e.,weather controls at the survey month). While in even columns, we addition-ally control forcontemporaneous pollution conditions(i.e.,PM2.5at the survey month).Thedependent variablein columns (1) and (2) is the unhealthinessindex, in columns (3) and (4), it’s the standardized word test score, in columns (5) and (6), it’s the positive measure of non-cognitive ability, while in col-umns (7) and (8), it’s the negative measure of non-cognitive ability.Observations areat the county-cohort level, with each cohort defined byitsbirthmonth.All regressions include both individual and weather controls. Individual controls include gender, age, father’s education and age, mother’s educa- Notes: This table presents theestimated results foraccounting for potential measurement errors, usingatwo-stepIV ap-proach.The dependent variable is theunhealthiness indexin column (1),thestandardized word test scorein column (2),thepositive measure of non-cognitive abilityin column (3), andthenegative measure of non-cognitive abilityin column (4).Ob-servations areat the county-cohort level, with each cohort defined byitsbirth month.All regressions include both individualand weather controls. Individual controls include gender, age, fat