I just read the paper. It’s more a literature review plus data analysis than a classic meta-analysis (i.e., a paper aggregating the results of many different observations into a single statistical pooled estimate).
I interpret “Improving health is not the best way to increase growth” as: growth usually leads to better health (sure!), and that (very plausibly) investment in economic development (on average) tends to be more cost-effective than investing in health, in the long-term.
However, for EAs, I’d first remark that what matters is not so much “what’s the causal direction in the growth-health correlation?” – which is David Weil’s point – but “what prevents the 3rd world from developing—low GDP or poor health?”. The first question would have for scope even current US and Norway. Since growth trajectories are path dependent and affected by many different things, we should distinguish analyses related to current low and high-income countries, and account for the possibility that different countries will find distinct paths to growth (at least if I understand D. Rodrik’s main point). E.x.: the question “why Senegal still has a low GDP per capita and life expectancy” may turn out to be quite unrelated to “why France has increased its HDI in XXth century”. I’d be marveled if health statistics (which, at least for latin America, includes violence) didn’t play a role in the first case; my personal anecdotal evidence is that prospects for life deeply affect one’s plans, so that, e.g., it’s really hard to design savings and insurance systems, with stable interest rates, in countries where people do not expect to reach old age.
Second, I am very wary of certain conclusions/analysis:
- “differences in life expectancy at birth tend to be far smaller than differences in life expectancy at birth” (p. 3 of the file – 626 in the book)… so what? Life expectancy is still significantly different across countries, income levels and regions, otherwise every age pyramid would be similar to all the others, except that in poor countries it’d have a larger base. Second, if you’re using life expectancy as a proxy for health in general, child mortality is relevant because it provides information about other sanitary conditions.
David Weil calculates the return to health using height as a proxy. I will suspend my judgement until further inquiry and maybe look at the raw data, but I suspect of regressional Goodhart.
I just read the paper. It’s more a literature review plus data analysis than a classic meta-analysis (i.e., a paper aggregating the results of many different observations into a single statistical pooled estimate).
I interpret “Improving health is not the best way to increase growth” as: growth usually leads to better health (sure!), and that (very plausibly) investment in economic development (on average) tends to be more cost-effective than investing in health, in the long-term.
However, for EAs, I’d first remark that what matters is not so much “what’s the causal direction in the growth-health correlation?” – which is David Weil’s point – but “what prevents the 3rd world from developing—low GDP or poor health?”. The first question would have for scope even current US and Norway. Since growth trajectories are path dependent and affected by many different things, we should distinguish analyses related to current low and high-income countries, and account for the possibility that different countries will find distinct paths to growth (at least if I understand D. Rodrik’s main point). E.x.: the question “why Senegal still has a low GDP per capita and life expectancy” may turn out to be quite unrelated to “why France has increased its HDI in XXth century”. I’d be marveled if health statistics (which, at least for latin America, includes violence) didn’t play a role in the first case; my personal anecdotal evidence is that prospects for life deeply affect one’s plans, so that, e.g., it’s really hard to design savings and insurance systems, with stable interest rates, in countries where people do not expect to reach old age.
Second, I am very wary of certain conclusions/analysis:
- “differences in life expectancy at birth tend to be far smaller than differences in life expectancy at birth” (p. 3 of the file – 626 in the book)… so what? Life expectancy is still significantly different across countries, income levels and regions, otherwise every age pyramid would be similar to all the others, except that in poor countries it’d have a larger base. Second, if you’re using life expectancy as a proxy for health in general, child mortality is relevant because it provides information about other sanitary conditions.
David Weil calculates the return to health using height as a proxy. I will suspend my judgement until further inquiry and maybe look at the raw data, but I suspect of regressional Goodhart.