Weather Shocks and Health at Birth in Colombia
By Mabl Andalon, Joao Pedro Azvedo, Carlos Rodrigues-Castela, Vivian Sanfelice, Daniel Valderrama
This research paper investigates the impact of in utero exposure to extreme weather related shocks on child’s outcomes at birth. This is relevant because the (1) frequency of extreme weather shocks is increasing with climate change. In addition, (2) literature shows that health at birth is a huge determinant of individual wellbeing later in life, as well as intergenerational mobility since it also affects academic outcomes and earnings. As such, identifying shocks that affect child’s birth outcomes is important from a policymaking perspective.
The setting for this empirical work is rural Colombia over the period of 1999-2008 because it is one of the most prone places in the world to extreme weather events, rainfall and temperature shocks with a lot of variation on a yearly basis. Due to these features, this is the first study to consider fetal exposure to extremes of rainfall (drought or flood) and temperature (extremely cold or hot) on child’s outcomes at birth in a developing country context.
The idea behind the model, is that in utero weather related shocks affect child’s outcomes at birth in rural areas, through affecting family’s food consumption, increasing mother’s stress levels or changing the environment and the spread of disease.
The method used to answer the question is Ordinary Least Squares, considering weather shocks are random. Variable of interest (Y), is outcome of child i, born in municipality j, in month m and year y, and the independent variable is Shock j,m,y which indicates whether the child was exposed to a hot or cold weather shock in his municipality in at least one month while in utero. Model controls for birth municipality fixed effects, month of birth fixed effects, birth year fixed effects and clustered standard errors. The same model is estimated by varying the timing of exposure to the shock, at trimester. I think that this strategy is appropriate considering random weather shocks; however, I believe that more controls should be added such as mother’s health, stress level, earnings or income, consumption, and other factors that can affect children’s outcomes at birth and increase the internal validity. Possibly, difference-in difference method could have been used to explore the weather shock effects for those in utero exposed and not in utero exposed during specific weather events and control for more of time invariant variables.
Data for this research is collected from the Colombian National Registry of live births (1999-2008), a nationally representative sample of 406,063 first time births from rural areas, with information at birth (weight, length, gestational age and the Apgar score) important information the mother, method of delivery and birth location. This data is merged with time series temperature and rainfall data from University of East Anglia Climate Research Unit based on mother’s municipality of residence to assess the child’s in utero exposure to weather shocks. The main advantage of the data set is that it is nationally representative and externally valid; whereas, the main disadvantage, is that the dataset only records surviving fetuses which can pose a fetus selection concern. Shocks that might have led to fetal deaths of weak fetuses are missing therefore, the estimates might understate the true negative effect of these shocks.
Results show that at birth outcomes of Colombian newborns, are very sensitive to in utero exposure to weather shocks, and the effects are especially sensitive during the first trimester of pregnancy. At this time, a cold shock increased the proportion of full-time births by 0.6 percentage points and the share of healthy born babies by 0.4 percentage points. Exposure to heat waves at any point of pregnancy reduced the share of full-term born babies by 0.5 percentage points and the share of healthy born babies by 0.4 percentage points. No effect was found as a result of rainfall shocks. Since exposure to weather is random and unpredictable the authors claim that birth outcomes should be insensitive to inclusion of mother and child covariates; the results are robust and answer the research question.
The authors were very creative in constructing the dataset by merging birth data with climate data and extending the birth measurements by including Apgar score. However, I am not entirely convinced from the results considering the missing data on fetal deaths from weather shocks and family/mother’s characteristics that were not included in the estimation of the model. Other than that, despite the fact that climatic change is random, Colombia is considered to be very prone to extreme weather event, which can possibly affect mothers’ health immunity and preparation in comparison to regions that are not used to experiencing these events.
In Utero Ramadan Exposure and Children’s Academic Performance
Authors: Douglas Almond, Bhashkar Mazumder and Reyn van Ewijk
This research paper asks, whether being in utero exposed to Ramadan fasting, affects academic outcomes for Muslim children. Most of the research looks at the impact of malnutrition on longer term health outcomes such as diabetes and health disease, but little on academic achievement. Finding aspects in nutritional environment that can be modifiable and positively impact human capital is very important because prior to birth (prenatal) investments might be more effective, with lasting effects to adulthood, and cost less than early childhood interventions.
The setting for the empirical study is England, and the population of interest is Pakistani/Bangladeshi Muslim children during 1998 to 2007. Ramadan is a Muslim religion practice and is considered exogenous on observable characteristics because it falls on different times every year which allows us to consider it as a natural experiment. In many aspects, Ramadan fasting during pregnancy resembles other nutrition characteristics such as mothers dieting, meal skipping, morning sickness and so on so the results can be generalizable in terms of restricted nutrition when we talk about poor developing countries.
Ramadan fasting, considered as nutritional deprivation, seems to send the body into “accelerated starvation” affecting blood glucose levels of the mothers, alter their neuro-endocrine system and increase storage of fat in as fast as within 12 hours of fasting These changes can affect the development of the embryo; specifically, because the brain is so sensitive in this stage of early development, it can be irreversibly damaged while forming cognitive skills.
Authors use Difference in Difference to estimate the intent to treat of Muslim children in utero exposed to Ramadan on their academic outcomes at age 7, in comparison to Caribbean students who are not Muslim, thus not in utero exposed. The outcome of interest is the test score of student i, in census area g and the independent variable is Eig x Muslim, which is an interaction term for the Muslim child having been in utero exposed to Ramadan controlling for ethnicity, month of birth, sex, and geographic effects. To validate the strategy, ‘falsification tests’ of the effect of Ramadan on Caribbean Students compared to white British as a control group and find 0 effect. The model seems to be very well established especially because it separates seasonal effects (age cut off for school entry, affecting academic performance based on time of birth), through including dummy variable for month. The main disadvantage is the lack of control for comparable characteristics between Muslim and Caribbean children such as parental characteristics, family education, income etc. Just because these two groups share a few similar characteristics does not necessarily mean we have a valid comparison.
Research uses data from Key Stage 1 Teacher Assessment in England and use a unique student identifier to link those scores with other student level data from the Pupil Level Annual School Census. Because this data does not have information on religion of the child, the authors assign Muslim religion to Pakistani/Bangladeshi students based on Census data where 92% of this population group reports to be Muslim. I think the dataset gives a good overview of the population of interest; however, not having actual accurate data on the religion of these children and whether the mother was actually fasting compromises our conclusions for this study. Because there is no data on the differential rates of fasting for mothers during pregnancy, our effect size might be larger due to that, but unfortunately with no data we are not able to tell. I think that we should take into consideration that even if the pregnant mothers were not necessarily fasting as some Muslim scholars argue that pregnant women are not allowed to fast Ramadan, the fact that the rest of family fasts might also affect the pregnant mother by changing her nutrition. Another explanation might be that mothers were not even aware of their pregnancy in the early period, therefore fasted Ramadan during that time.
Assuming that all children have a nine-month gestation period and excluding children that were certainly not in utero exposed, the results show significant negative effects in math, reading and writing scores in the first trimester of pregnancy. Largest effects are seen in the third month of pregnancy when in utero exposed children scored lower in math scores by 0.08 standard deviations compared to those not exposed and they are robust. This creates a gap of about 20% of overall test scores between Muslim children compared to national average.
I believe the results are significant considering that Ramadan last only one month, and larger exposure to similar nutrition deficiencies can have even greater impacts, especially in the developing world. I think the paper is very clear and gives a very informational background on biomedical mechanisms that can affect the mother during nutritional deficiencies and presented a good case for prenatal intervention to affect human capital.