A multi-criteria decision analysis for the public distribution system of food grains in Indian Punjab: Towards decentralized food policies in developing countries

In 2014, the government of the state of Indian Punjab amended its existing system of public distribution of food grains by launching the “new atta-daal (wheat-pulses) scheme”. Instead of distributing subsidized food grains monthly to its underprivileged citizens, the state started distributing grains on a biannual basis (i.e., every 6 months). The state claims that the new system yields logistical and monetary benefits that favour the state treasury; however, beneficiaries have voiced concerns regarding grain handling and one-time payments. To achieve a beneficial outcome for multiple stakeholders with conflicting objectives, a formal analysis of different policy alternatives is necessary. This research paper applies multi-criteria decision analysis to evaluate alternative public grain distribution schemes for the district of Ludhiana in Indian Punjab. First, an objectives hierarchy and measures for the decision problem were developed using emergent codes from qualitative data collected via semi-structured interviews, focus group discussions, and guided conversations with 40 participants representing eight different stakeholders. Using the rank-order method, weights were then assessed to identify the relative importance of these measures. Five feasible grain distribution policy alternatives were generated, and data describing the performance of these alternatives with respect to each measure were collected via Likert scale surveys of over 300 beneficiary households across 14 different villages (rural) and localities (urban) of Ludhiana district in Punjab. Value functions were assigned to each measure to identify the best performing alternative, and sensitivity analysis was performed to establish the robustness of the final solution. Overall, beneficiaries preferred a 6-monthly distribution system with improvements, such as better communication and disciplined distribution.