51
season’ dummy explains the largest parts of child admissions while at the same time it
renders the parameter of the seasonal price component (with which it is strongly correlated)
statistically non significant. Finally, the number of feeding centres rises markedly and
significantly the number of child admissions (suggesting that public policy may help in
treating the backlog of chronically malnourished children), but does not alter the significance
and size of the parameters of the three price components being investigated. All this
obviously suggests that to improve food security and reduce child malnutrition there is a
need to intensify efforts both to increase long term maize supply, as well as to smooth the
price seasonal cycle by investing in storage and cereal banks and by providing a broader
access to credit. Thirdly, the impact of the famine price component on child admissions on
occasion of food crises is also important but is less large (and less statistically significant) than
the seasonal price component (whose parameter is about six times bigger than that of
famine price component). This result may be biased, however, by the fact that lack of data on
child malnutrition prevented to include in the analysis the famine year of 2002. The adjusted
R2 of the regressions in Table 14 are broadly acceptable, while the value of the Durbin
Watson in Model 2 possibly signals the omission of other variables or the presence of
spurious correlation.
Table 14 - Malawi and Niger. OLS log-log regression of number of child admission to feeding
centres on different price components
Malawi 2003-9
Niger 2006-10
(1)
(2)
(3)
(1)
(2)
(3)
Constant
5.914***
1.139***
5.710***
1.062**
-0.613
1.056**
Log trend price/100
component
- 0.769***
- 0.572***
-0.702***
12.05***
10.624***
11.84***
Log seasonal price
component
1.676***
1.908***
-0.189
1.926***
1.727***
1.089
Log Famine price
component
0.333
0.219*
0.331^
-2.881***
-2.012**
-2.699***
Log Residual term
0.239
0.744
0.381
-2.567
-3.341
-2.183
Log number
feeding centres
1.205***
0.427***
……
Hunger season
dummy
0.782***
0.317***
Adj R2
0-30
0.86
0.39
0.84
0.86
0.85
F statistics
9.91***
107.93***
11.73***
72.84***
73.13***
66.6***
Durbin Watson
1.30
0.85
1.44
1.04
1.27
1.21
N.obs
84
84
84
60
60
60
Note: *, **, *** indicate that the parameters are significantly different
from zero at the 10, 5 and 1 percent level of probability.
Source: authors’ calculation on official data.
The analysis on Niger (covering only the years 2006-2010) points to a somewhat different
story. All three models presented indicate that the main factor explaining the increase in
admission of malnourished children to feeding centers is the price trend component, a fact
broadly consistent with the data of Table 3 and Figure 6 which seem to suggest that the
outright stagnation in millet production per capita (linked to ineffective food production
policies and to a rapid population growth) lead to a steady increase in millet prices and to
52
rising child admissions to feeding centers over the long term. Also in this case, Models 1 and
2 suggest that the seasonal component of the price of millet appears to be a major
determinant of child malnutrition, a fact that - as in the case of Malawi – is confirmed by the
significance of the dummy ‘hunger season’ and by the fact that the seasonal price
component is no longer significant when the dummy is included, a fact that suggests that
the two variables move closely in line with each other. Finally, the famine price component is
strongly significant and negative, a fact that should not surprise as during the period covered
(2006-2010) millet prices have hardly jumped during the food crisis of 2010 (see Figures 18
and 22) despite a major drop in food production, thanks to a quick increase in imports and to
effective government responses to the crisis. Also in this case, it appears that the control
variable ‘log number of feeding centers in operation’ (a policy variable – which rose sharply in
2010 but not before) explains a non-negligible part of the log of the total number of children
admitted to feeding centers. In both Niger and Malawi, the residual term is not significantly
different from zero. Also in this case, the statistical tests seem acceptable, while the stability
of the parameters across models suggests they are robust.
8. Overall conclusions and some policy lessons
This paper has shown that in Niger and Malawi (and possibly other SSA countries with similar
structural characteristics) the drivers of domestic staple prices have to be found not only – or
not primarily – in changes of international food prices. These are obviously important, but
may not be the main culprit of persistently high and rising levels of child malnutrition in parts
of SSA. Changes in domestic food production, the persistence of strong food price
seasonality, and recurrent famines/crises have exerted an important upward pressure on
domestic prices and – through them – on the nutritional status of children even during years
of stable or falling international prices.
This paper is not meant to provide policy prescriptions. But some suggestions emerge from it
about areas on which policy research ought to be intensified. First of all, the comparison
between Malawi and Niger suggests that policies aiming at intensifying agricultural
production and raising land yields - especially among smallholders - (as done in a
controversial way in Malawi by subsidizing high yielding seeds, fertilizers and other inputs)
may help reducing child malnutrition. Several studies (Dorward et al. 2010, IFPRI 2011) on the
Farm Input Subsidy Program introduced in 2005 in Malawi for maize confirm its role in
increasing the fertilizer rate of consumption. Subsidized programs for seeds and fertilizers
should involve the private sector from the beginning and facilitate a transition towards
market-based arrangements (von Braun et al. 2008) and should not substitute for an
intensification of traditional input use, such as traditional varieties, labor, manure (Abdoulaye
et al. 2000, Reeder & Tisdall 2007). At the same time, the lack of an alternative long term
agricultural policy after the SAP approach in agriculture of the last 20 years appears to be a
key determinant of child malnutrition in Niger. As noted in section 4, areas where research
should focus further are agricultural intensification, technological innovation, control of
population growth and greater budgetary allocations than at present. Several of these
concerns were raised by the recent “Conférence Internationale sur la Sécurité Alimentaire et
Nutritionnelle au Niger” of March 2011, which noted that the structural cause of the recurring
famines and food crises was slow food production and rapid population growth. In the
absence of improvements in these areas, Niger and several other nations in the region will
continue to be affected in the future by recurrent famines and devastating food price
seasonality.