The poverty impacts of improved soybean technologies in Malawi

ABSTRACT

Improved soybean varieties and agronomic practices have been widely disseminated to smallholder farmers in Malawi over the last 15 years. However, there is no empirical evidence on the welfare impacts of adopting improved soybean technologies. This paper estimated the poverty impacts of adopting improved soybean technologies using data from 1,234 households in six soybean growing districts accounting for over 80% of the total soybean production in the country. The results from an endogenous switching regression model showed that 32% of the sample households adopted improved soybean varieties and agronomic practices. The adoption benefits were higher for female-headed households and increased with the household head’s education and cultivated land areas. A comparison of the observed and counterfactual incomes for adopters based on the international poverty line of US$1.90 per capita per day showed a 4.16 percentage-point reduction in poverty among the sample households, translating to over 150,000 people lifted out of poverty. The household head’s education level, household size, cultivated land area, livestock size, and asset ownership are associated with the daily per capita income. The results point to the need for scaling up of improved soybean varieties and agronomic practices for greater impacts on poverty reduction among smallholders in Malawi.

KEYWORDS:

• Soybeans
• poverty
• impacts
• endogenous switching regression
• Malawi

JEL CLASSIFICATIONS:

• I32
• O33
• Q16
• Q18

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Daily per capita income is defined as the total amount of annual food and non-food household expenditure adjusted for purchasing power parity and then divided by household size and number of days in a year.

2 Based on simple falsification test, a variable can be a valid selection instrument if it will affect adoption decision but it will not affect the outcome variables in our case daily per capita income among farm households that did not adopt (Di Falco, Veronesi, and Yesuf 2011).

3 Surveybe software is a tool that helps to design electronic computer-assisted personal interview (CAPI) questionnaires and collect and export analysis-ready data (https://surveybe.com/).

4 Malawi education system categories as illiterate those who do not read and write; Standards 1 to 8 are similar to Grades 1 to 8; Forms 1 to 4 are similar to Grades 9 to 12.

5 The national poverty level for Malawi is 70.8% which is different from the poverty level in this paper. This could be because we used data collected only from rural Malawi and the effect of crop failure a year before the survey.

6 Computed as the percentage reduction divided by the counterfactual poverty headcount ratio, that is,${\displaystyle \frac{0.8883-0.8467}{0.8883}\times 100.}$