Though studies on multidimensional poverty have been gaining research and programmatic attention, no attempt has been made to understand the association of multidimensional poverty with consumption poverty in India. Using data from the National Family Health Survey-4, 2015–16, this paper examined the association and spatial clustering of multidimensional and consumption poverty in the districts of India. Context-specific indicators were chosen to provide robust estimates of multidimensional poverty. The Alkire and Foster method was used to estimate the indices of multidimensional poverty. The spatial patterns of multidimensional and consumption poverty were examined using Moran’s I statistics, Local Indicator of Spatial Association, and cluster maps. A set of spatial regression models was used to understand the predictors of multidimensional poverty. The results suggest that 30.3% of the population in India was multidimensionally poor, with an average intensity of poverty of 44.2% and a multidimensional poverty index of 0.13. The state variations in multidimensional poverty were high. The univariate Moran’s I statistic of multidimensional poverty was 0.75, while that of consumption poverty was 0.56, suggesting that multidimensional poverty was spatially clustered. Though spatial regression model shows multidimensional poverty is positively associated to consumption poverty, the extent of association is limited. Besides, fertility level, share of rural population, health insurance, and percentage of scheduled caste population were significant predictors of multidimensional poverty. Based on the results, we suggest that multidimensional poverty measures may be integrated along with consumption poverty and that districts with high levels of multidimensional and consumption poverty should be prioritized for evidence-based planning.

尽管对多维贫困的研究已经获得了研究和计划的关注，但还没有试图了解印度多维贫困与消费贫困的关系。本文使用 2015-16 年全国家庭健康调查 4 的数据，研究了印度各地区多维贫困和消费贫困的关联和空间聚类。选择特定背景的指标来提供对多维贫困的可靠估计。Alkire 和 Foster 方法用于估计多维贫困指数。使用 Moran's I 统计量、空间关联的本地指标和聚类地图检查了多维和消费贫困的空间模式。一组空间回归模型用于理解多维贫困的预测因素。结果表明，印度 30.3% 的人口多维贫困，平均贫困强度为 44.2%，多维贫困指数为 0.13。多维贫困状态差异较大。多维贫困的单变量 Moran's I 统计量为 0.75，而消费贫困的单变量 Moran's I 统计量为 0.56，表明多维贫困在空间上是聚集的。虽然空间回归模型显示多维贫困与消费贫困呈正相关，但关联程度有限。此外，生育水平、农村人口比例、医疗保险和在册种姓人口百分比是多维贫困的重要预测因素。