xi
also be encouraged. However, the purchase
of such technologies may depend on who
controls household income, as there is evi-
dence that men are often unwilling to pay
for them. The social constraints and nutri-
tional benefits of such technologies need to
be considered in policies affecting their
availability.
4. Nutritional status depends, of course, on
food intake, but in some cases, health con-
ditions may be more constraining than food
intakes on nutritional well-being. This was
DeWaal's (1989) conclusion, for instance,
in the case of the famine in Darfur, Sudan in
1984/85. How food consumption gains are
realized may also determine whether, and
to what extent, increased food consumption
translates into improved nutritional status.
For instance, technologies (e.g., irrigation)
which increase food consumption, via in-
creased agricultural productivity and farm
incomes, may have adverse health side ef-
fects which outweigh consumption benefits,
resulting in diminished nutritional welfare.
Another example may be distributions of
food aid that encourage migration to feed-
ing camps where there may be serious prob-
lems of infectious diseases. DeWaal (1989),
in fact, goes so far as to conclude that food
aid played no role in preventing starvation
in Darfur's 198485 famine, and that if,
instead, "Darfur had been provided with
clean water, better sanitation, and measles
vaccination, most or even all of the famine
deaths could have been prevented." While
this conclusion seems exaggerated, the point
that it is not enough only to look at provid-
ing food as a solution to malnutrition is a
good one.
Implications for Food Security
Policy Research
1. Food security researchers need to define
more carefully the variables they are pur-
porting to analyze
*
and explain how these
conceptual variables relate to the proxy
indicators used to measure them. For in-
stance, anthropometric data (measurements
of body size) should not be (as they often
are) implicitly equated with nutritional sta-
tus (the level of nutrients available to body
tissues). Also, empirical studies are fraught
with problems of data unreliability and un-
observed variables, the implications of
which are frequently overlooked.
2. Because careful descriptions of exactly how
data were generated, and the problems in-
volved, as well as access to the raw data it-
self, is missing from most of the literature,
readers are forced to engage in a lot of "blind
faith" in accepting conclusions which the
authors derive. Reducing the necessity of
blind faith acceptance of results could be en-
couraged by agencies which fund research by
requiring, for instance, that reports be attached
by summaries of the raw data used in order
that analyses may be replicated.
3. Empirical findings suggesting that low in-
come elasticities of calorie consumption at
sample (or subsample) mean income lev-
els imply that income generation is only
weakly linked with food consumption are
often very misleading. The elasticity at the
mean for any sample (or subsample), no
matter how it is disaggregated, will inevita-
bly underestimate the elasticity facing the
poorest households in the sample. Two pos-
sible alternatives are to calculate elasticities
for only those below a certain minimal food
*
Understanding Linkages among Food Availability,
Access, Consumption, and Nutrition in Africa sug-
gests some definitions.