Rural Development Project Impact Assessment on Poverty and Food Security in West Garo Hills Villages (Meghalaya)

Introduction Assessing rural development projects is crucial to understand how well the ex-ante expectation align with effective outcomes. In this context, it is important examine their materialization and evaluate their influence on project participation. When conducting this type of analysis in low-income countries, the goal often consists in comparing changes in poverty and food security after the implementation of the program. To this end, two commonly used indicators are the Multi-Dimensional Poverty Index (MDPI) and the Food Consumption Score (FCS). The purpose of this study is to assess the impact of Project C3S’ (Production of Appropriate Food: Sufficient, Safe, and Sustainable) ten-year activities on agricultural, nutritional, and socio-economic development of rural Garo villages around St. Alphonsa Church Parish (Meghalaya, India). In particular, we evaluated the effect of participating to the project by estimating the Average Treatment Effect (ATE) with respect to the two aforementioned measures: the MDPI and the FCS. Data and research methodology Our data collection methodology involved structured questionnaires divided into five sections: one addressing the elementary items pertaining to the MDPI, the second dealing with the basic components of the FCS, then three general section investigating agricultural activities, socio-economic characteristics and demographics. To ensure a balanced comparison between beneficiaries (B) and non-beneficiaries (NB) of the project, we fixed the proportion of control units relative to B to approximately 1.5. Then, to minimize the effect of spatial spillovers, we set a minimum distance of 3 km between data collected from treated and non-treated samples. Our final sample consists of 108 NB household and 78 B household. We analyzed the treatment effect of being a beneficiary on both the MDPI and FCS using Inverse Probability Weighting (IPW) and Doubly Robust (DR) methods for consistency check. In both cases, we estimated the propensity score via a linear logistic regression model including different variables linked to household’s characteristics and activities. The difference between B and NB households were evaluated at level of confidence of 90% and 95%. Discussion of results Both samples exhibit medium values that exceed the necessary threshold to be considered below the MultiDimensional Poverty level. Both the IPW and the DR estimator suggest that the average difference between the two groups is not significant using defined confidence levels. In the other hand, the mean value of the FCS in both groups shows that the interviewed households exhibit high dietary standards. In particular, we register higher FCS values in the B group (B mean – 65.89 ± 22.88) compared to the NB sample (NB mean – 60.20 ± 18.66). The ATEs obtained using either IPW or the DR methods also suggest a positive mean difference, with significance at 10% for both and only IPW at 5%. Main conclusions Our results show that the intervention had a positive impact on food security of the households in the project, as evidenced by an increase FCS resulting from the comparison between treated and untreated household, after adjusting for confounders. On other hand, there is no impact on poverty (as measured by the MDPI): this outcome suggests where interventions should be focused in the future.