Multidimensional Poverty Index — Explained
Detailed Explanation
What does the National MPI count?
India's National Multidimensional Poverty Index measures overlapping deprivations in health, education and living standards. NITI Aayog's 2023 Progress Review uses three equally weighted dimensions represented by twelve indicators.
A household's deprivation score is the sum of the weights of indicators in which it is deprived. The poverty cut-off then identifies those whose weighted score reaches at least one third. Counting deprived indicators without considering their weights can therefore give the wrong classification.
The index is designed to capture several aspects of poverty together. It is not an income total or a ranking based on the number of government schemes received. Each indicator has a specified deprivation rule, and the same broad label can conceal important detail. For example, the drinking-water indicator concerns both source quality and access time under the report's definition. A generic statement that a household has some water does not settle its classification.
How are the dimensions and indicators weighted?
Health receives one third of the total weight. Within it, nutrition has weight one sixth, while child-adolescent mortality and maternal health each have weight one twelfth. Education also receives one third, split equally between years of schooling and school attendance, each weighted one sixth. Living standards receive the remaining third, divided equally among seven indicators, each weighted one twenty-first.
The living-standard indicators are cooking fuel, sanitation, drinking water, electricity, housing, assets and bank account. The unequal number of indicators across dimensions explains why their individual weights differ. Equal dimension weights do not imply twelve equally weighted indicators. This is a central examination trap: replacing the specified weights with one twelfth for each indicator changes the model rather than simplifying its arithmetic.
What does a worked household example show?
Consider a hypothetical household deprived in nutrition and years of schooling, with no other deprivations for this exercise. The weighted score is one sixth plus one sixth, or one third. It reaches the poverty cut-off. Now consider a household deprived in four living-standard indicators. Its score is four twenty-firsts, approximately 19.05 per cent, which does not reach one third. More deprived indicators do not necessarily imply a larger weighted score.
The example does not say that four living-standard deprivations are unimportant. It shows how classification works under this index. Policy may still need to address them even when the household is not classified as multidimensionally poor. A measurement cut-off is an analytical rule, not a judgment that deprivation below it is harmless or that the household cannot qualify for any public programme.
How are incidence and intensity combined?
The headcount ratio, H, is the proportion of people identified as multidimensionally poor. Intensity, A, is the average weighted deprivation score among those identified as poor. MPI is H multiplied by A. Use proportions consistently: if H is 0.20 and A is 0.40, MPI is 0.08. Multiplying 20 by 40 and reporting 800 as the index mixes percentage notation with proportions.
The average for A is not taken across everyone in the population. Including non-poor people would answer a different question. H and A also describe different routes to improvement. Fewer people may cross the poverty cut-off, reducing H; or people still classified as poor may experience fewer weighted deprivations, reducing A. A headcount-only account misses the latter change even though conditions have improved.
Which survey periods support the 2023 results?
The Progress Review compares NFHS-4, conducted in 2015–16, with NFHS-5, conducted in 2019–21. It reports the national multidimensional-poverty headcount declining from 24.85 per cent to 14.96 per cent. The absolute decline is 9.89 percentage points. These are historical survey-period estimates, not a measurement newly conducted in 2023 merely because the report was published then.
That distinction is especially important when a headline combines a publication date and an estimate. Report year, survey period and policy period need not coincide. A change between two surveys also does not, on its own, establish the causal contribution of one particular programme. Establishing causation requires a suitable evaluation design and attention to other influences, rather than simply placing a policy name next to an improving indicator.
How should the January 2024 discussion paper be read?
NITI Aayog's discussion paper estimates multidimensional poverty at 29.17 per cent for 2013–14 and 11.28 per cent for 2022–23. It derives those estimates through interpolation and extrapolation using the available survey-based information. They are not direct NFHS observations for newly surveyed rounds in those years. The method should accompany the figures whenever they are used.
This does not make an estimate meaningless; it identifies what kind of evidence it is. Observed survey estimates and model-based estimates can both inform discussion, but they should not be relabelled as one another. Nor should the 2024 paper's later estimated endpoint be spliced into a series without explaining the methodological transition. A clear answer names the source, reference period and estimation status together.
What can the index tell a policymaker, and what can it miss?
Indicator and regional breakdowns can identify where different deprivations overlap and help coordinate services. A composite score provides a summary, while the underlying profile shows whether the problem concerns nutrition, schooling or infrastructure. Policy should use both. Two areas with the same MPI can have different deprivation profiles and therefore require different combinations of interventions.
The index is also bounded by its indicators, weights, cut-offs and household-based information. It does not measure every aspect of income insecurity, inequality, service quality or individual experience.
Household classification can conceal differences among members. The appropriate conclusion is to use MPI alongside other evidence and retain its methodological limits. For UPSC, a correct weighted example and the H×A calculation demonstrate more understanding than an isolated headline about how many people left poverty.