An assessment of multidimensional water poverty in India: an application of Alkire–Foster dual cut-off approach

Abstract

Proper measurement of water poverty is essential for designing appropriate and effective water policies, particularly in emerging economies like India. Against this backdrop, this study proposes to estimate India’s Multidimensional Water Poverty Index (MWPI) by applying the Alkire–Foster approach (two stages identification process) to quantify water deprivation at the household level. This methodology captures several attributes to understand the complex issues related to households’ water deprivation, thus enabling decomposition of the overall MWPI into dimensions, as well as state, rural and urban classifications. This study analysed two rounds of Indian Human Development Survey data (IHDS 2004–05 and 2011–12) to find that water poverty in India is 44.9 per cent and 40.9 per cent for 2004–05 and 2011–12, respectively. By decomposing the country’s MWPI for both rounds, Orissa and Bihar emerged as the states with the highest MWPI for rounds 1 and 2, respectively. Furthermore, access to water sources and sanitation were identified as significant contributors to India’s MWPI for both rounds. The uniqueness of this study lies in it being the first to estimate the multidimensional water poverty index using the Alkire–Foster approach. This study provides insightful data for policymakers to prioritise lower- or higher-intensity water-poverty households to ensure improvements in the overall MWPI. Another significant intervention is the identification of specific indicators that majorly impact the MWPI.

Appendices

Appendix

See Table 8 and 9.

Table 8 Average Poverty Gap

Table 9 Attrition analysis of households in IHDS dataset

Appendix A

2.1 Average poverty gap

The Average poverty gap (G) is found by taking the mean of the normalised deprivation gap for water-poor households across indicators where they are deprived. This involves computing the normalised deprivation gap (g_ij) (${g}_{ij}= \frac{{z}_{j}-{x}_{ij}}{{z}_{j}}$) which represents how far below deprivation cut-off $({z}_{j})$ the household’s achievement ${(x}_{ij})$ is in each dimension. To find G, we take the mean of the weighted normalised gap for deprived households at 30 per cent cut-off. The weighted aspects account for the importance of each dimension and is determined by the weights assigned to different indicators. Therefore, the average intensity of deprivations among the identified water-poor households contributes to our multidimensional water-poverty depth. The average poverty gap estimated for round 1 and round 2 of the dataset using in this paper is shown in Table 7.

Appendix B

3.1 Attrition analysis

The model is related to a liner probability model. The binary outcome variable Output has a value of one if the household is not included in the study and zero otherwise. The household characteristics are based on values from IHDS-I and IHDS-II and match the controls used in the primary analysis. State level fixed effects are reported. Standard errors that are clustered are shown in parentheses. **p < 0.05, ***p < 0.01, *p < 0.10.