Can legume production boost nutrition and income? Empirical evidence from Ethiopia and Malawi

Description

This dataset supports the research manuscript titled "Can legume production boost nutrition and income? Empirical evidence from Ethiopia and Malawi." It was derived from the publicly available nationally representative Integrated Household Survey Panel datasets from Ethiopia and Malawi, published at World Bank website. The panel dataset is from three survey years from 2013-2019. The data is in form of Stata file, with its do file and excel sheet containing figures derived from it. We isolated crop production variables by focusing on three legumes, namely soybeans, common bean and groundnuts. We also computed the nutrient intake like carbohydrates, proteins, vitamin A, thiamine, niacin, riboflavin, zinc and iron and income expenditure from the 7-day dietary intake, and other socio-economic variables. The outcome variables are the nutrient intake and food expenditure variables, while the rest are the independent variables.

Steps to reproduce

First of all, we downloaded the nationally representative integrated household datasets from the World Bank website. We downloaded data for Ethiopia and Malawi collected from 2013 to 2019. We then isolated two modules relevant for our study, which are household and agriculture modules. We also isolated important variables like household demographics (household size, age, gender, education level), main livelihood source, income level, household and farm assets, livestock number owned per household, food types and quantities consumed in a week, farm or land ownership size, number of people consumed, cost of food consumed and many more. For the agriculture module, we isolated crop production variables to get the legume production quantities per household. After isolating these variables, we clean them to reduce the noise. We used Stata 18 as a software package for analysis. To quantify the nutrient intakes, we used the Food Composition Tables for Malawi, Ethiopia and Africa because they contain nutrient quantities foreach type of food. After computing nutrient intake from consumed items, both purchased and produced, we calculated the nutrient gap by dividing the consumed quantities by the recommended daily allowances (RDAs) for each nutrient, adjusted for age, sex, and lactation status in women . Since the food items were computed per week, we divided them by 7 to get the recommended daily intake, and household size to get the per capita nutrient intake. Food expenditure was adjusted for inflation using the Consumer Price Index to reflect current values. We then converted all amounts to a common denomination using exchange rates from the World Bank website. We then start analyzing the data using the do file which is included in the zip folder. We used Fractional Probit regression model with mundlak effects. Because production is endogenous, we used instruments like livelihood source; gender; household size; farm implementation; distance to the district center; rainfall; household income; farm size; residential location; number of male household members between 15 and 55 years of age; and number of children between 6 and 14 years of age. These were tested and analyzed using the Two-Stage Least Squares approach (2SLS) and tested for validity.

Institutions

• MwAPATA Institute

  Central Region, Lilongwe
• Lilongwe University of Agriculture and Natural Resources

  Central Region, Lilongwe

Categories

Related Links

Licence