Ministry of health in a developing country.

Empirical Question
You have recently been elected minister of health in a developing country. Congrats! After reading news about the 2019 Noble Prize for Economics being awarded to Abhijit Banerjee, Esther Duflo and Michael Kremer, you have been inspired to adopt their evidence-based approach. You resolve to allocate your health budget toward programs that deliver the greatest health improvement β€œbang for buck”. Your announce this intention and the Twitterati likes and nods approvingly.
A group of epidemiologists sees your announcement and seeks a meeting with you. They argue that your best bet is to invest in improving the health of young children and pregnant women. You are puzzled by this and worry that they just want you to fund their research.
(a) Why might small improvements in the maternal and early-life health environment have long- run impacts?

(b) The epidemiologists say you should therefore allocate more of the health budget to an early- life nutrition program. To persuade you, they show you empirical evidence indicating a strong negative correlation between birth weight and adult earnings. Would this evidence convince you? If not, explain why and discuss the likely direction of bias.

(c) The epidemiologists say that the 2019 Nobel Prize has made it easier for them to raise money for experimental studies. So, if you don’t believe them, they will run an experiment and get back to you. How would you respond to them?

One of the epidemiologists shares with you the results from a study in several Scandinavian countries including Norway, where babies are classified as being Very Low Birth Weight if they are born below 1500 gms and given additional nutrition during the first month of life. The following graph and regression table illustrates the main empirical result. The regression specification estimated is
𝑦𝑖 = 𝛼 + 𝛽(π΅π‘–π‘Ÿπ‘‘h 𝑀𝑒𝑖𝑔h𝑑 𝑗𝑒𝑠𝑑 π‘π‘’π‘™π‘œπ‘€ 1500𝑖) + π›Ύπ·π‘’π‘šπ‘œπ‘”π‘Ÿπ‘Žπ‘h𝑖𝑐 πΆπ‘œπ‘›π‘‘π‘Ÿπ‘œπ‘™π‘ π‘– + πœ€π‘–
The key outcome variable is the 10th grade math score and the units of that variable are in standard deviations relative to the mean. So, a score of 0.1 indicates that the person scored 0.1 standard deviations above the mean.

(d) The epidemiologists think you do not know any econometrics, so they smugly ask you to interpret the graph and regression table, assuming they can impress you. How would you respond when they ask you to interpret the graph and regression table?

(e) During a cabinet meeting later in the day, the education minister – who is a career schoolteacher – presents a proposal. She is planning to double teachers’ pay and expects this to improve 10th grade scores by 0.1 standard deviations. The prime minister then asks you for your views. What would you say about the relative cost-effectiveness of the early-childhood intervention relative to this proposed policy change?
(f) Later that evening, you reflect on the Norway study and consider whether you should allocate more of the health ministry’s budget to neonatal nutrition programs. Comment on the internal validity and external validity of the Norway study. By internal validity, I mean, do you believe the results of the study and do you think of these estimates as causal? By external validity, I mean, do you expect to find similar effects if you ran a similar program in your own (poor) country? Also note that your country uses 2000 gms as the cutoff to classify children as Very Low Birth Weight.

(g) You decide to read the Norway study in more detail and learn that the neonatal nutrition intervention actually reduced infant mortality. You think, β€œthis seems great! Two birds with one stone”. But then you begin to worry how this might affect the estimated effect on 10th grade math scores. How do you think this selective mortality would bias the estimated effect and why?

(h) You are eventually convinced that neonatal nutrition is something that your government should devote considerably more resources to, and give a speech about this issue at the annual convention of your political party. However, most of your party members don’t seem interested in this as they don’t think it will be a good political strategy. Explain why there may be limited political returns for early-life health investments.

Sample Solution