FOOD PRICES AND SEASONALITY - Food Security, Food Prices and Climate Variability - page 123

Table 6.1 shows the average price during poor, average and good years, as measured by
remote sensing data. Although the table shows nice, neat differences in prices during different
periods, the reality of food price data is far messier. Many things affect price levels, everything
from elections and government policies about taxes and imports, to the amount of grain in
storage from the previous year's crop, and what the rumors are about demand in different
markets. Commodity price determination is very complicated, and given the number of years
economists have spent working on developing effective models to forecast commodity price
movements, many of the elements that determine prices are only marginally predicable (Baffes
and Gardner, 2003; Deaton and Laroque, 1992; Tomek and Myers, 1993; Trostle et al .,
2011).
Figure 6.4 shows the same data used to create Table 6.1 in a histogram format for each
country. The variability in actual observed prices in each year (good, average and poor) shows
how difficult it is to demonstrate the impact of growing season variability on each country.
The next chapter will discuss how each market needs to be treated individually, since group-
ing all markets according to which country they are in assumes a relationship between markets
within a country that often works against understanding the dynamics between the environ-
ment where crops are grown and the market in which those crops are sold. Markets in Niger
that trade more with Nigeria than with other cities in Niger, farmers who have long-standing
relationships with traders in markets outside their immediate neighborhood or country, dif-
ferences in storage capacity, wealth of the consumers and demand for particular product—all
work against being able to use the country as the level of analysis.
Another factor complicating the relationship seen between NDVI anomaly and millet
price dynamics analyzed here is the assumption that the vegetation index in a 25-mile area
around each market is really the area of interest. The vegetation index used to determine the
quality of the growing season in Figure 6.4 is the area around each market, not a broader
region or based on any specific information that this area actually has grain to sell. Many agri-
cultural regions produce excess grain and seek the market with the best price and adequate
capacity to absorb the grain produced when it is available, even if it is not the nearest. The
relationship between supplier and seller are undoubtedly as complex in Africa as anywhere -
assuming that all farmers simply walk to the nearest market is an assumption that contributes
to uncertainty ( Figure 6.4 ). Thus expanding the geographic scope of the NDVI data integ-
rated into this assessment is an essential next step in understanding the connection between
poor growing conditions as measured by satellite data, reductions in yield and the impact of
these changes on food prices.

TABLE 6.1 Millet prices in CFA/kg during August and September, using NDVI anomaly during the
same period in Burkina, Mali and Niger over 1982-2007

NDVI anomaly Burkina Faso Mali Niger

Below average 118.5707 107.1924 125.2000

Average 115.6478 99.5688 117.5255

Above average 101.0583 96.8137 112.9464

Note
NDVI anomalies are defined as above average and below average vegetation greenness during August and September,
the peak growing season months, over the 25-year record.