Does conservation agriculture technology reduce farm household poverty? Evidence from rural Zambia

Abstract

This study examines the adoption and impact of conservation agriculture on farm household poverty in Zambia using both provincial and average poverty lines, given that the poverty line is a key driver to the determination of household poverty. We probe the basis of setting a singular average poverty line for all provinces. The recursive bivariate probit model results reveal that conservation agriculture significantly reduces household poverty. However, the average poverty line either over or under estimates provincial poverty status. We also find that education, access to credit and information, social networks and machinery influence adoption of conservation agriculture technology.

Keywords:

• poverty line
• impact evaluation
• recursive bivariate probit
• conservation agriculture technology
• Zambia

Notes

1 It is estimated that only about 200,000 farmers had adopted CA in Zambia by the year 2013 compared to the expected 600,000 by 2015.

2 Jesuit Centre for Theological Reflection (JCTR) is a prominent local non-state actor that has single-handedly brought cost of basic household need in Zambia to the limelight since 1991 through its monthly provincial surveys on costs of living across Zambia under the basic needs basket (BNB) framework.

3 Ministry of Agriculture & Cooperatives and The Central Statistical Office, Zambia 2010 and 2011 Crop Forecast Surveys.

4 Agricultural camp in Zambia is a management unit of agricultural camp officer comprising a catchment area of up to eight different zones of different villages.

5 The Magoye ripper is a Zambian innovation which was developed in 1996 at Zambian Ministry of Agriculture and Cooperatives research station of Magoye (Dutch funding).

6 Engel’s law states that with given tastes or preferences, as income increases, the proportion of income spent on food diminishes.

7 At the time of the survey, US1$ was ZMW5.37.

8 The full information maximum likelihood method was estimated using Stata version 12 (StataMP 12).