Dust Storms and Birth Outcomes: Evidence from Taiwan

In this paper, we examine whether early-life exposure to ambient air pollution, such as particulate matter that is 2.5 micrometers in diameter or smaller (PM_2.5), affects newborns. Despite mounting evidence suggesting adverse health impacts of ambient air pollution, studies providing quasi-experimental evidence on the causal effects have been lacking. Our study is expected to fill in this gap in the literature by focusing on the association between dust storms and infant health in Taiwan.

A major hindrance in identifying health impacts of air pollution is the lack of variation in pollution exposure (measured at each local area) that is driven by factors unrelated to the local conditions affecting health outcomes such as the local economy and access to health care. In Taiwan we observe one unique source of exogenous variation in the local pollution, which is the presence of northeasterly winds (occurring mainly during winter months) that can transport pollutants originating from China to Taiwan. As a result, newborns living in cities and townships (of Taiwan) that are similar in observable characteristics can be affected by different levels of air pollution, because of the locations of the cities and townships relative to the pathways of the northeasterly winds. This empirical setting will allow us to find a match of two locations that are observationally similar except that one is downwind of the imported pollution but the other is not, and such matched locations will enable us to attribute an observed difference in health outcomes of the newborns between the two locations to the difference in their exposure to imported pollution caused by the northeasterly winds.

To implement our empirical study, we need information on weather, air quality, and birth outcomes. First, the weather information is derived from Taiwan’s Central Weather Bureau. We have monthly temperature and rainfall information from 68 stations, and wind direction and wind speed information from 27 stations in Taiwan. We use daily wind directions in Pengjiayu, an island to the north of Taiwan, to proxy the pathways of northeasterly winds blew from China to Taiwan. Second, we draw air quality information from Taiwan Environment Data Warehouse. The pollutions studied in this paper include PM₁₀, PM_2.5, and SO₂. In the end, the birth outcomes are derived from Taiwan birth certification data (TBCD). TBCD cover all birth in Taiwan, and contain information on birth weight, gestational age, birth county, gender, and the age and education of both parents at the time of the birth.

The summaries statistics indicate that the air pollution levels in Taiwan are higher during the months of northeasterly winds, and are extremely high during the periods of dust storms. Our results suggest a strong relationship between in utero exposure to air pollution and worse birth outcomes.