2. Geography, Child Survival, and Health Expenditure
More than 10 Million children younger than 5 years die every year. Most frequent causes of
death are neonatal disorder (33%), diarrhoea (22%), pneunomia (21%) and malaria (9%). They
are mainly observed in the tropical zones of Sub-Saharan Africa (41%) and South Asia (34 %).
About 53 % of all child deaths can be attributed to being underweight. Black et al. (2003)
- the survey from which these facts are taken - identifies undernutrition and vitamin A- and
zinc-deficiencies as the underlying cause of a substantial proportion of all child deaths.
Most child deaths could have been avoided if more money were spent on nutrition and health
care. Not surprisingly, empirical cross-country studies usually find a strong correlation of child
mortality or, more generally, life expectancy and health with income per capita and causality run-
ning from income to health (see Pritchett and Summers, 1996, and the literature cited therein).
Although income per capita is already to some extend predetermined indirectly through geog-
raphy (see Introduction) some authors (among others Bloom and Sachs, 1998, Schultz, 1999,
Sachs, 2003) argue in favor of an independent influence of geography that makes some regions
inherently less healthy than others . For example, falciporom malaria needs an ambient temper-
ature of 22o C or above during the incubation period. Winter frost eliminates the prevalence of
many pathogens and parasites (Masters and McMillan, 2001).
In this paper we will therefore analyze the choice of parents who are able to control the number
of births (n) but only partly the survival of their children. Survival probability (π) consists of
a fundamental part and a controllable part. The fundamental survival rate (π) depends on
geographic location and the state of economic development (summarized by average income
per capita). The controllable part depends on the fraction of parental income devoted to child
expenditure (h).
We assume that an additional unit child expenditure is more effective in preventing child death
when fundamental child survival probability is low. In other words, at low fundamental survival
probabilities child expenditure is mainly motivated by the desire to have surviving offsprings and
consists therefore to a large extent of nutrition and health care expenditure. When fundamental
survival probabilities are already very high, an additional unit of h is relatively ineffective
in terms of increasing child survival and is therefore mainly motivated by the utility derived
directly from child expenditure. This utility could originate either from a “warm glow of giving”
(Andreoni, 1989) or a preference for having higher quality children (Becker, 1960). It helps to