Accounting for Agriculture in Development: Impact of a Digital Platform for Farm Finances

We propose to study the impact of a farm financial management tool implemented by People’s Action for National Integration (PANI) -- a social development organisation working in underdeveloped regions of Uttar Pradesh. PANI developed a mobile application which enables farmers to record and review their expenses and income over the agricultural season. Downloaded on the mobile phone of farmers, the application requests farmers to record their expenses (specific to practices along the growing season including preparing land, seed and sowing, soil health, plant growth, pesticide, irrigation, and harvesting), as well as their harvest value (including both quantity, verified through a crop cutting exercise on a square meter of cultivated land, and price received if the crop is sold). This information is reviewed monthly allowing farmers to update their records on a regular basis.

PANI plans to extend their programme across Eastern Uttar Pradesh in late 2024. We propose to leverage this new role-out to investigate the impact of farm financial management for farmers using the platform. Using a randomised control trial, our experiment will randomly allocated farmers to either receive the app or not. In the first agricultural season, farmers having been selected to receive the app will be visited on a monthly basis by PANI community workers to encourage them to update their financial records on the platform. This data will be used to calculate crop-wise the total costs, value of harvest, and returns to investment -- broadly termed as financial summary. At the end of the season, PANI community workers will visit the farmers to explain their financial summary benchmarked to the top 25% of producers.

This project will employ empirical methods for robust causal inference on the behaviour of agricultural market participants to further our understanding of barriers to economic growth within the sector and consequently highlight potential paths to reducing income/wealth disparities.

2024-11-01

2026-11-01

The key outcome variables in our analysis will be:
1. Crop choice
2. Expenses on inputs
2. Value of harvest
3. Farm Profit, which will take into account imputed costs of family labour

Some of the secondary outcome variables in our analysis will be:
1. Saving
2. Credit
3. Wellbeing

Intervention Villages: Randomly select 75 Gram Panchayats (GP) to be reached by PANI as part of their programme. Among the list of programme participants identified by PANI in each GP, equally split 40 farmers to either randomly receive the financial management app or not. A comparison of the farmers having received the app with those that did not will enable us to measure the direct impact of farm financial management on production choices.

Spill over Villages: Farmers using the financial management app may discuss the information retrieved from this app and changes to their production choices with other farmers in their community. This spread of information may bias the treatment effect. We therefore propose to randomly select 20 GPs, with 20 farmers per GP, to be reached by PANI as part of their programme in which no farmer will receive the app. A comparison of the non-users of the app in the intervention villages with farmers in this spill-over villages will allow us to measure the presence of information diffusion and its impact on production choices.

Pure Control Villages: In order to obtain pure control villages, we propose to randomly sample 50 GPs, with 20 farmers per GP, identified by PANI as eligible for the programme but not included in their final sample. Firstly, a comparison of these pure control villages with the intervention villages will give us a measure of the impact of PANI’s overall programme including financial management. Secondly, a comparison of these pure control villages with the spill-over villages will give us a measure of the impact of PANI’s programme excluding financial management. Thirdly, a further comparison of the above two effects will enable us to capture the marginal impact from providing financial management over and above just information on agricultural practices.

Not available

Randomisation will be conducted in office using a computer, based on the lists of GPs and farming households identified by PANI as beneficiaries of their programme.

For the purpose of the study design, the intervention and spill-over GPs will be randomly selected from the total list of GPs identified by PANI. Within the intervention villages, the farmers will be randomly selected to either receive the financial management app or not.

No

4400 households across 145 Gram Panchayats

4400 households across 145 Gram Panchayats

Number of households and GPs in each treatment arm:
1. The intervention arm will include 3000 households across 75 GPs (40 households/GP)
(a) 1500 household will receive the app and 1500 will not
2. The spill over arm will include 400 households across 20 GPs (20 households/GP).
3. The pure control arm will include 1000 households across 50 GPs (20 households/GP)

Using data from a survey in 2022 among 3451 farming households across 40 GPs in the neighbouring district of Balrampur in Uttar Pradesh (part of a project conducted by Camille Boudot-Reddy, co-PI on this project) for a study evaluating PANIs layered vegetable farming model, we conduct power calculations to identify a suitable sample size. Our results, indicate that a sample size of 3000 households will allow us to estimate a 6% change in harvest value, 3% change in expenses and 12% change in profits with 95% confidence.