9
in the 1970s were not caused by absolute de-
clines in available food, but rather by food price
inflation which depleted the purchasing power
of low-income households.
Alderman found in Ghana, however, that
responses to food prices differed by region, and
in some cases, price increases were correlated
with increases in household food consumption.
Presumably, these households were net food
sellers--thus, the price increases were associ-
ated with increased real incomes for these house-
holds. Another reason for differences in re-
sponses to food price increases is that food
baskets may differ across regions and house-
holds, as well as across time. For instance, food
grains might be more expensive in urban areas
but this price difference may not be so impor-
tant because nongrain consumption might be
more important in the diet relative to rural areas
(Ravallion 1990; von Braun et al. 1993).
But the link from agricultural growth to
broad-based income growth and food security
is not an automatic one. Not only does increased
production not necessarily lead to improved
food security, it may even exacerbate food in-
security. The means by which food production
gains are achieved are important. Policies, for
instance, which encourage greater production
among large-scale producers, but hurt the pur-
chasing power of low-income producers, would
exacerbate, rather than reduce, food insecurity.
For instance, Michigan State University research
conducted in Rwanda and in Zimbabwe (see
case example) have demonstrated that although
government marketing or pricing policies may
stimulate food production and rural income,
leading to modest supply increases, they may
also exacerbate food insecurity for the poorest
rural households who (contrary to the conven-
tional wisdom among many policymakers and
researchers) are often net buyers, rather than
net sellers, of food and, consequently, are hurt
more than helped by higher food prices. These
households, it has been shown, may rely on
nonfarm activities or nonfood cash crops for an
appreciable share of their incomes and on food
purchases, rather than production, for a large
share of their consumption.
Methodological and Measurement
Issues
Efforts to estimate the extent to which food
availability is linked to food access is hindered
by a number of measurement problems. For
instance, as Hay (1978) points out, "while it is
relatively easy to estimate imports and the
amount of domestic production which enters
the market through official channels, it is un-
commonly difficult to estimate informal trans-
actions, black market dealings and leaks across
the border to a neighboring state." Moreover,
such estimates of commercial food availability
do not account for the great proportion of food
in many African countries which is produced
for home consumption, and thus "does not pass
through commercial channels where it might be
(easily) monitored" (Poleman 1983). Such pro-
duction for home consumption is by no means
necessarily correlated with changes in commer-
cial (market) food availability.
Food production estimates are an alterna-
tive to estimating food availability through
market supply estimates. This approach, too,
entails serious difficulties, and these difficul-
ties may vary in different contexts, according to
the varying complexity involved in estimating
outputs. For instance, Poleman (1983) cites a
finding that estimates of irrigated rice produc-
tion (which is relatively easy to measure) in
Malaysia and Sri Lanka may have underesti-
mated calorie availabilities by 10 to 15 percent.
And he notes that such undercounting may be
far worse elsewhere. As Poleman points out,
"output that is not seen is not counted, and if
communications are poor, a great deal is not
seen." Production estimates may be particu-
larly difficult in tropical areas where "many
food crops are not grown in pure stands but
mixed-planted in fields of bewildering com-
plexity" (Poleman 1983).