202
Data on noncommercial farms are from an enumerated sample survey conducted annually from
1971 to 1976 and also from 1982 to 1986, with no data collected by the CSO in the period from 1978
to 1981. While data were apparently collected and published for the 1977 season (CSO 1989b), these
data are not available. Available data are broken down by province but not by farm area size. Using
a two-stage design, a stratified random sample of geographic areas was first selected, with strata
corresponding to ecological zone. In the second stage, two random samples were selected, one of
households reporting livestock holdings, the other of households not reporting livestock. These details
are generally explained in the introductions to the reports. While in most years farmers are asked to
provide information on the agricultural period just completed, it should be noted that for a few
provinces in 1982, data are based on recalled responses collected as part of the 1983 survey.
Additional limited data were collected on farm expenditures and revenues and are briefly
inspected here. Significant difficulties with the consistency of categories in the collection of these data
over time make it quite difficult to analyze relationships between prices, input uses, and output. A
cursory inspection of these data for commercial farms is provided below.
Price indexes for bundles of consumer goods as well as for numerous commodity groups are
available in the CSO's Monthly Digest of Statistics. Several editions of these digests have been
compiled to compute a price series for the years 1962 to 1990, adjusted to the common base year of
1985. These data are also presented below. Consumer price indexes are used to deflate expenditure
and revenue data to permit examination of real price trends independent of inflation. As will be seen,
this is particularly important for data from the late 1980s when Zambia experienced significant
hyperinflation.
Anomalies are apparent in the data sources. For example, in the 1987 commercial farms
report, the same harvested area and production for soybeans in the Copperbelt province is given for
two separate farm sizes categories. Such a coincidence seems unlikely. In Luapula province in 1990,
farms of one size are reported to have produced one kilogram of wheat but then to have sold 37. Since
reports do not distinguish sales from present harvest versus past harvests, it is impossible to determine
whether this is an error or due to stock adjustment.
Given problems in data collection and reporting, policy recommendations arising out of
analyses of these data should be regarded as tentative and subject to verification in more detailed
retrospective studies of selected areas. Longitudinal analyses that follow particular farming households
over time seem particularly warranted if an understanding of structural change in agriculture is to be
gained.
III. National production trends
The aforementioned problems with the data notwithstanding, a number of important
characteristics of production structure in Zambia can be discerned. Table 7.2 presents harvested crop
areas for all farms in the nine regions of Zambia as a percentage of national totals. Data in the table
are three-year averages drawn from CSO crop forecasts for 1991 and 1992, and from actual area
figures for 1990. Of Zambia's 942,362 hectares planted nationwide in 1990-92 (three-year average),
63.8 percent are comprised of maize, followed by groundnuts (8.5 percent), seed cotton (7.3 percent),
millet (5.4 percent), sorghum (4.4 percent), sunflower (3.4 percent), soybeans (2.9 percent), mixed
beans (2.2 percent), rice (1.4 percent), and assorted other crop enterprises (0.7 percent).'
Total crop area is based only on data for crops that were reported. Area of unreported crops (horticultural crops,
cassava, other vegetables) is not included. Note also that, in official publications, tree crops are reported as numbers of
trees rather than area of planting; estimates of tree-crop area are thus not possible.