Child Health and the Impact of Maternal Fasting

National Bureau Of Economic Research, Cambridge MA

Investigators

Abstract

? DESCRIPTION (provided by applicant): Low birth weight, which is partly determined through poor in utero nutrition, has been shown to be associated with adverse health and socioeconomic outcomes, both in childhood and later in life. Prior research has used plausibly exogenous events, such as famines, to identify a causal link between in utero nourishment and birth weight. These estimates of the effect of nutrition on birth weight suffer, however, from survivorship bias. The proposed research exploits a natural experiment provided by the mild nutritional shock of maternal fasting during the Islamic holy month of Ramadan. The idiosyncratic variation in the timing of Ramadan allows us to plausibly estimate causal effects of an intent-to-treat effect for children who in utero were potentially exposed to maternal fasting. Given the significant short- term and long-term effects of low birth weight on child development and adult health, the findings of the proposed research have considerable value for policy makers weighing the costs and benefits of policy interventions related to fasting and dieting during pregnancy. The study will use population birth and death records from Germany, a country with a Muslim population of around 4 million, for the years from 1980 to 2013. We will have data on approximately 1.7 million births to Muslim mothers during this sample period. The long observation period allows us to net out all seasonality from the data, as Ramadan occurs during all twelve months of the Gregorian calendar during a 33-year period. Moreover, unlike previous studies the German data allow us to definitively identify Muslim and non-Muslim women and also provide exact information on birth weight, dates of birth, and a wide array of covariates. A unique feature of the data allows us to identify multiple births to the same mother and control for unobserved between-family heterogeneity and to exploit variation in potential fasting behavior across siblings. We will also be able to examine intergenerational effects of in utero exposure to potential fasting, as the birth records also contain exact information about the mother's birth date. Lastly, by linking the birth records to death records, we will be used to assess the effect of fasting on neonatal mortality. By using such a large data set over a long period of time, we will provide the most precise and best-identified quasi-experimental estimates of the effect of fasting on birth outcomes.