Yesterday, I presented paper written by Randall Kuhn: “Routes to Low Mortality in Poor Countries Revisited”, published by Population and Development Review 36(4): 655-692 in 2010. Actually, Kuhn revised the paper written by John Caldwell, “Routes to Low Mortality in Poor Countries”, published by Population and Development Review 12(2): 171-220 in 1986. So, after a quarter century, Kuhn revised the paper, especially indicators and determinant of health outcome. Unfortunately, I cannot upload two papers due to copy right reason.
Basically, from the World Development Report 1984, from 99 third world countries, Caldwell used GNP/capita as income indicator and infant mortality rate (also child mortality rate and life expectation at birth) as health indicator. All countries were ranked between the 1st (the worst) to 99th (the best) according to GNP/capita. Then, all countries were ranked based on infant mortality rate. Finally, Cadlwell calculated relative ranking of infant mortality rate ranking to GNP/capita ranking.
For example, Sri Lanka was the 23rd (income) and the 85th (infant mortality rate). Thus, ranking of infant mortality rate relative to income was +62 (=85-23). Cadlwell divided into two groups: superior health achiever (relative ranking is equal or more than 25) and poor health achiever (relative ranking is equal or less than negative 25). We can find superior health achiever, for example, Kerala (actually Kerala is a region in India, not a country), Sri Lanka, China, Burma (now Myanmar), Jamaica, India, Zaire (now Congo), Tanzania, Kenya, Costa Rica, Gana, and Thailand. Also, for poor health achiever, for example, Oman, Saudi Arabia, Iran, Libya, Algeria, Iraq, Yemen, Morocco, Ivory Coast, Senegal, and Sierra Leone.
Then, Kuhn used the data derived from 2009 WHO World Health Statistics. Income indicator was changed, from GNP/capita to GDP/capita with purchasing power parity (PPP). Child mortality rate was used as standard indicator rather than infant mortality rate. And what was the implication? Well, using same list of country from Caldwell, there was strong convergence in relative ranking, both infant mortality and life expectancy at birth.
Caldwell also analyzed relationship between religion and health achievement in some cases. For example, Cadlwell discussed Islamic countries and poor health achiever and Buddhism in Sri Lanka, Thailand, Vietnam, and Burma. In contrast, Kuhn discussed that religious identity should be analyzed with care. In my humble opinion, be careful about definition, for example, Islam and Muslim countries, what definition exactly we use, and which indicator we can use to define. For example, Islamic Economy Index is an example how we define the adherence of Islamic countries to Islamic economic principle. Ireland, Denmark, and Luxembourg are the top countries. In contrast, Saudi Arabia is the 91th position! That’s what I say, be careful about religious identity. In econometrics, I can use the concept of interaction effect, including more than one factor as independent variable. Let’s say, religous and geographic interact, then these variables affect dependent variable (in this case is health).
According to this paper, there are four group of determinant: (1) education and gender, (2) health system spending, (3) poverty, and (4) governance and society. I am not going to discuss more detail. However, I focus to discuss from my perspective. What is the best health indicator to measure exactly the empirical fact? What is causality and connection among the indicators? Can we connect among these indicators? Let’s pick children mortality rate as a health indicator. Is it the best indicator to measure health condition? However, we also have other group of indicator: health service indicator, morbidity statistics, and demographic indicator (Hardiman, and Midgley, 1982). Health service indicator, for example, number of hospital, number of doctors, number of nurse, etc. Morbidity statistics, for example, HIV/AIDS prevalence, malaria prevalence, tuberculosis prevalence, etc. Demographic indicator, for example, infant mortality rate, children mortality rate, etc. Let’s think about other indicator. 
Actually, I don’t get logic behind the threshold of +25 and -25 as Caldwell used in his paper and +30/-30 as Kuhn used in his paper. Why did Caldwell use +25/-25? Why did Kuhn use +30/-30? What is reason behind this? Also, they compare only from cross-section data. Otherwise, it is better to use panel data, especially to compare with the initial condition of each country. Longitudinal data allows us to see the progress of health achievement between different years.
Then, talking about ranking in mathematics concept. The ranking concept is more intuitive and easy to be understood, especially for policy maker. However, the ranking system does not reflect “the gap” among the countries. For example, assume there are three countries: A, B, and C. Based on life expectancy at birth, the country has 57 yrs, 60 yrs, and 75 yrs, for A, B, and C, respectively. Then, country A can be assigned as the 1st rank, country B as the 2nd rank, and country C as the 3rd rank. From rank 1 to rank 2, the gap is 3 years. But from rank 2 to rank 3, the gap is 15 years!
What I’ve mentioned from this paper is to include other factor (eg: geographic) factor, let’s say for interaction effect. For example, type of disease in Sub-Saharan Africa is different than Central Asia. Type of disease in tropical countries is different than Arabic countries. For instance, remember that HIV/AIDS is predominant in Sub-Saharan Africa. In this case, I always prefer to use interaction effect from more than one variable.
To sum up, intuitively, this paper contributed good explanation by Caldwell and Kuhn, especially Caldwell used mixed methodology between quantitative and qualitative. Well, maybe we can improved the methodology.
