will understate increases (or overstate decreases) in agricultural labor productivity (Kelly et al.
1995).5 Nonetheless, the measures of labor and land productivity can indicate important changes
in agricultural systems performance, especially if kept in context of the changes in the use of other
factors.
The figures highlight several apparent trends.
• Land and labor productivity have increased between the 1985-89 period and the 1990-95
period in three of the seven countries examined: Ethiopia, Mali, and Burkina Faso
(Figures 3, 5, and 7). By 1991, major food and input market reforms had been initiated in
each of these countries. The rise in land and labor productivity has been associated with
increased use of fertilizer in each of these countries (by 260% in Ethiopia, 61% in Burkina
Faso, and 17% in Mali) between 1980-89 and 1990-95. The increase in fertilizer use in
Ethiopia has occurred under the continuation of subsidized state input distribution (until
1997) and alleged restrictions on private investment. In Mali and Burkina Faso, however,
fertilizer use rose despite the elimination of fertilizer subsidies in part because of policy
reform, crop productivity improvements and increases in world prices that served to raise
the returns to rice and cotton production.6
In some of these cases, e.g., Ethiopia, the restrictions imposed on producers and
marketing agents during the control period depressed investment incentives so much that
the simple elimination of these restrictions increased agricultural growth. However, as
argued later, the emerging market-oriented systems of farm finance, input delivery and
commodity marketing also suffer from a number of unresolved constraints that impede the
potential for future productivity growth.
• Land and labor productivity both declined during the 1990-95 period in three countries:
Zimbabwe, Zambia, and Senegal (Figures 1, 4, and 6). Fertilizer sales to smallholders
declined by 25%, 17%, and 15% in these countries,7 respectively, between the 1985-89
and 1990-95 period. In each of these cases, the reforms were associated with the
withdrawal of state support to key producers in the form of input subsidies, concessional
5 A more accurate indicator of labor productivity would have been value of crop output per unit of
agricultural laborer, but time series data on agricultural labor was unavailable in most cases. Use of rural
population data will give similar trend results to those using agricultural labor data as long as the
proportion of rural population engaged in agriculture was relatively constant over the sample period.
Productivity trends will be biased upward (downward) if the share of the rural population in agriculture
increased (decreased) over the sample period.
6 In Mali, for example, the reform of the state’s role in the management of irrigated rice production and
rehabilitation of rice perimeters increased the returns to fertilizer application on irrigated rice despite the
elimination of 15% to 25% subsidies on the value of fertilizer (Cisse 1997).
7 However, if fertilizer distributed concessionally to smallholders is counted, fertilizer use actually
increased in Zimbabwe by 13% in the post-reform period.