Intergenerational Transmission of Mother-to-Child Health: Evidence from Cebu, the Philippines

Using rich longitudinal data from the Philippines, we causally estimate the transmission of maternal health to child health, from birth through preadolescence. To obtain exogenous variation in maternal health we exploit an array of data on climate (temperature, precipitation and cyclone) shocks surrounding the time of the mother’s birth and during her early childhood. This large set of instruments is weak, with first stage F-statistics hovering around one. We contribute a new machine learning method of choosing/forming optimal instruments in a setting of many valid but weak instrumental variables, a method based on singular value analysis (SVA). This method out-performs other options, and results in first stage F-statistics hovering around 50. Using this optimal instrument, we show that mother’s health causally transmits to child health from birth through childhood and into adolescences. Some of this transmission is due to the effect of mother’s health on child health at birth and then the effect of birth health on later health. Yet, the transmission of maternal height to child height persists into adolescence beyond its effect on health at birth and in early life. Moreover, the transmission of height increases rather than diminishes as the child ages. This may be, in part, due to a transmission of growth velocity that causes the health advantage of having a taller mother to widen with age.