Enhancing Dairy Farmer Income and Livelihood Diversification through Capacity Building: Evidence from a Cluster-Randomized control Trial in Uttar Pradesh, Bihar, Madhya Pradesh, and Rajasthan

The intervention consists of a structured capacity-building and training program delivered to dairy cooperative member households through selected Dairy Cooperative Societies (DCSs) in rural settings of Uttar Pradesh, Madhya Pradesh, Rajasthan, and Bihar. Training modules will be delivered by technical experts from IRMA, ICAR, NDDB, State Dairy Federations, and Milk Unions and will focus on improved dairy management practices, animal nutrition, fodder production and conservation, animal health and breeding, clean milk production, record keeping, financial literacy, market linkages, and livelihood diversification opportunities. Training will be delivered through classroom sessions, demonstrations, exposure visits, and follow-up advisory support. Farmers in treatment DCSs will receive the intervention, while farmers in control DCSs will continue with business-as-usual services available through the cooperative system during the study period.

2026-08-01

2026-08-10

1. Annual dairy income of participating households (INR)
2. Livelihood Diversification Index
3. Milk productivity (average liters of milk produced per animal per day)
4. Adoption Index of improved dairy management practices

Annual dairy income (INR) of DCS members will be measured based on various kind of products they produce and sell at DCS.
The Livelihood Diversification Index will be constructed using information on the number and relative contribution of income-generating activities undertaken by the household, including dairy, crop agriculture, wage employment, livestock-related enterprises, and non-farm businesses.

The Dairy Practices Adoption Index will be constructed from a set of recommended practices promoted during training, including fodder cultivation, balanced feeding, vaccination, deworming, artificial insemination, hygienic milk handling, record keeping, and cooperative participation. The index will be standardized and aggregated using equally weighted components.

1. Livestock asset ownership
2. Number of milch animals owned
3. Women's participation in dairy decision-making
4. Cooperative participation and meeting attendance
5. Access to veterinary and extension services
6. Milk marketed through cooperative channels
7. Household resilience to income shocks

Knowledge scores will be generated from responses to objective questions related to dairy management, animal health, feeding, breeding, and marketing practices. Women's participation will be measured using indicators related to involvement in dairy production, income utilization, and household decision-making. Household resilience will be assessed through coping-capacity indicators and diversification measures.

The study uses a cluster-randomized controlled trial design. Four Indian states (Uttar Pradesh, Madhya Pradesh, Rajasthan, and Bihar) are included. In each state, two Milk Unions are selected, one with relatively high average milk procurement and one with relatively low average milk procurement. Within each Milk Union, six Dairy Cooperative Societies (DCSs) are selected. DCSs are pair-matched on key pre-randomization characteristics including milk procurement, membership size, female membership share, SC/ST membership share, and village population. Within each matched pair, one DCS is randomly assigned to treatment and one to control. Fifty dairy cooperative member households are randomly selected from each DCS. Baseline and endline surveys will be conducted to estimate intervention impacts using an intention-to-treat framework.

Not available

Matched-pair cluster randomization conducted by computer using reproducible random-number generation. Within each Milk Union, six DCSs are matched into three pairs based on pre-treatment characteristics. One DCS from each pair is randomly assigned to treatment and the other to control using computer-generated random numbers.

Dairy Cooperative Society (DCS) (cluster level). Individual dairy farmer households are sampled within randomized DCSs but treatment assignment occurs at the DCS level.

Yes

48 Dairy Cooperative Societies (DCSs):
• 24 Treatment DCSs
• 24 Control DCSs

2,400 dairy farmer households. (48 DCSs × 50 farmers per DCS)

1. Treatment: 24 DCSs, approximately 1,200 dairy farmer households
2. Control: 24 DCSs, approximately 1,200 dairy farmer households

Total: 48 DCSs and approximately 2,400 dairy farmer households.

Power calculations assume 48 clusters, 50 farmer households per cluster, significance level α = 0.05, statistical power = 0.80, and intra-cluster correlation (ICC) between 0.05 and 0.10. Under these assumptions, the study is powered to detect a minimum standardized effect size of approximately 0.25–0.30 standard deviations for the primary outcomes. For annual dairy income, this corresponds to detecting approximately a 15–20 percent change relative to the baseline mean, depending on observed baseline variance.