ANNEX 1
NOTES ON THE DATA
Value of crop output: Output figures are derived from the FAO Crop Production Indices,
except for Zimbabwe, which uses smallholder crop output only, as derived from Ministry of
Agriculture (1997), to separate smallholder production trends from the large-scale (mainly
European) sector. FAO crop output data for various years was rescaled to world market prices
during the 1989-91 period. Thus, the indices are in real terms and do not reflect variations in crop
price relatives or exchange rates over time. The indices were then converted to agricultural gross
domestic product (GDP) in 1994 US$.
Land and labor: The measures of labor and land are in physical units. The land variable is area
cultivated and under permanent crops, as reported by FAO, except for Zimbabwe, Senegal and
Ethiopia where actual area cultivated is used (based on Ministry of Agriculture (1997); FAO
(1997); and CSA (1997), respectively). The labor variable is population in rural areas, as
reported by FAO. This is an imperfect proxy for agricultural labor force; however, it is likely to
be highly correlated with rural population except in the event of shifts in labor allocation between
farm and nonfarm activities over time.
Fertilizer use: This variable is represented by total fertilizer consumed in the crop year (in
thousands of tons), as reported by FAO (1996).
Rainfall: The measure of rainfall is the national average of the total annual precipitation for
numerous local rainfall stations, weighted by its long-term average. The index is biased toward
agriculture, i.e., rainfall in the wettest areas is given a relatively higher weight than dry areas. The
relevant annual period is determined in accordance with the crop cycle, and therefore differs from
country to country. The data and methodology are drawn from Gommes and Petrassi (1994).
Reform: Reform is modeled simply as a binary variable taking on a value of zero before the
initiation of significant sectoral reform and a value of one thereafter. The initiation of significant
sectoral reforms for each country is considered as follows: Burkina Faso (1985); Ethiopia (1990);
Kenya (1989); Mali (1985); Senegal (1985); Zambia (1993); and Zimbabwe (1993). Although de
jure reforms began earlier in some countries (e.g., the legislation abolishing the grain board's legal
monopoly in Mali was passed in 1981), the dates used are based on the authors' estimates of when
the reforms in fact began to be implemented.
35