2
Figure 2 - Trends in FAO’s Food Price Index (January 2007- May 2011)
Source: FAO’s World Food Situation, 2011
For instance, in the case of the two countries analyzed in this paper – Niger and Malawi – the
evidence that international food prices are a main driver of domestic prices is weak and the
same applies to other African countries ( Blein & Longo 2009, Diallo et al 2009, Benson et al.
2008).
To assess the extent of this price transmission, we carried out an econometric test of the
impact of international food prices on domestic food prices over 2000-2010 in Malawi and
Niger following the model proposed by Minot (2011). This model makes a series of rather
stringent assumptions, i.e. homogeneous cereal products, competition among numerous
small traders, perfect information, no trade taxes or other policy barriers to trade, and no
transportation and transaction costs. In the above mentioned analysis, Minot carries out his
estimation by means of the vector error correction model which assumes that the domestic
food price is affected by the world price and vice versa. This assumption does not hold in the
case of Malawi and Niger, two small countries whose prices do not affect world prices. For
this reason equation (1) is estimated with the OLS method.
Given these assumptions, the price transmission occurs via short and long term lagged
adjustments in the domestic price as described by the following econometric relation:
(
)
t
,
i
D
1
t,
i
w
1
t
,
i
w
1
t,
i
D
1
t
,
i
D
t,
i
P
P
P
P
P
ε
∆
ρ
∆
δ
β
θ
α
∆
+
+
+
−
+
=
−
−
−
−
(1)
where
D
t,
i
P
is the natural logarithm of the domestic price of the commodity taken into
consideration (maize for Malawi, millet for Niger) in country
i
(
Niger
Malawi
i
,
=
) converted
into US dollars in month
t
,
w
t
P
is the natural logarithm of the international price of maize or
3
millet
2
in US dollars,
∆
is the time difference operator (
1
t
t
t
P
P
P
−
−
=
∆
),
δ
δ
β
θ
α
and
,
,
,
are
estimated parameters and
t
ε
the error term.
In equation (1) the first term in parenthesis represents the long term transmission of world
prices on domestic prices (scaled by the parameter ). The subsequent two terms measure
the short term impact of the lagged increments ( ) of the natural logarithm of domestic and
international prices. As
is unknown we estimate relation (1) after developing the
parenthesis. A successful test of the hypothesis of the transmission of world prices on
domestic prices requires that the estimated parameters and are positive and statistically
significant
3
. The regression results are presented hereafter in Table 1, panels A (Malawi) and B
(Niger). The results in Panel A show that the long run elasticity of the Malawian price of maize
in relation to its long run world price
is pretty low and barely significant. In turn, the
short run elasticity of the domestic price relative to the world price (
δ
) is positive but not
statistically significant. In contrast, the autoregressive term (
ρ
) is significant and shows that
approximately 55 percent of the change in the Malawian maize price will be transmitted to
the domestic price of the commodity in the subsequent year. Finally and interestingly,
α
(the
trend component) is negative and significant.
Panel B on Niger shows that the long run relationship between world cereal prices is
negative and highly significant, suggesting that an increase in the average world price of
cereals reduces the domestic price of millet (a phenomenon which is difficult to explain, and
which may point to a rapid increase in domestic output of millet on occasion of rises in
international cereal prices), while the short run elasticity of the domestic millet price relative
to the average world price of cereals (
δ
) is positive, sizeable and statistically significant. This
suggests that there are important short term but not long term effects of international cereal
prices on the domestic price of millet. However, the autoregressive term (
ρ
) is significant
and slightly bigger than the prior term. It shows that approximately 57 percent of the change
of the domestic price of millet in Niger will be carried over into next year. Finally and contrary
to what observed in the case of Malawi, is positive and significant.
Overall, with the exception of a possible short term effect on the domestic price of millet (and
this may be due to the hypotheses made to proxy the international price of millet, see
footnote 5) the results of Table 1 do not seem to support the view that, over 2000-2010, the
price of maize in Malawi and of millet in Niger were mainly influenced by their world prices.
The possible reasons of these results are the moderate dependence of these countries on
imports of these staples and their greater dependence on local crop conditions, whether due
to shocks, agricultural and food storage policies or seasonal variations (World Bank 2011a;
Asian Development Bank 2011, Ellis & White 2010).
For instance, Hauenstein Swan et al.
(2009) provide evidence that the link between global price rises and malnutrition levels is not
as clear as that one between seasonal price rises and malnutrition.
2
For Niger, we could not use the international price of millet as such commodity is not globally traded. We used
instead the simple average of the prices of wheat, rice, sorghum and maize – which assumes that in any case the
domestic price of millet cannot deviate excessively from the international price of cereals.
3
To derive (the long term elasticity of price transmission) we estimate
, then obtain as follows:
.