Imagine two questions about the same town. The first asks whether a household can obtain enough food, shelter, clothing, care, learning and other essentials to reach an accepted minimum standard. The second asks how unevenly income, consumption, wealth or opportunity is spread across all households in the town.
These questions are connected, but they are not the same. Suppose every household becomes better off, while the richest households gain much more than the rest. Fewer households may fall below the minimum, so poverty falls, even though the distribution becomes more unequal. Now imagine that everyone has nearly the same resources, but those resources are very low. Inequality may be small while severe poverty remains. A third town may improve the minimum and narrow large gaps at the same time.
The first question is about poverty. It looks for a serious shortfall from a chosen standard. The second is about inequality. It looks at dispersion or concentration across a distribution. Neither answer can be read straight from the other.
Measurement begins before any formula. We must decide what we are measuring. Is it income, consumption, wealth, opportunity or a set of deprivations? We must decide whether the unit is a household or a person, which population is covered, which period and prices apply, and which data record those facts. A poverty measure then needs a threshold. An inequality measure usually needs the whole ordered distribution.
These choices do not make measurement useless. They make its meaning precise. A figure becomes informative only when the learner knows the question it answers, the choices that produced it and the limits it carries.
Poverty, inequality, exclusion and vulnerability answer different questions
Poverty is a serious shortfall from an accepted minimum. The shortfall may concern money resources, or it may concern health, learning, living conditions and other deprivations. A poverty measure identifies what counts as the minimum and who falls short.
Inequality describes how unevenly a resource, outcome or opportunity is distributed. It can examine the gap between richer and poorer people, the share held by different groups, or the shape of the full distribution. It does not by itself establish that anyone falls below a minimum.
Social exclusion concerns blocked participation. A household may have some money yet remain unable to use a school, market, bank, workplace or public process because of discrimination, unsafe travel, missing documents or weak legal protection. Exclusion can contribute to poverty and inequality, but a money measure may not fully reveal it.
Vulnerability is the risk of falling into serious deprivation in the future. A household just above a poverty line may face unstable work, illness, crop failure, debt or climate risk. It may not be counted as poor on the survey date, yet a modest shock could push it below the line. Crossing a threshold does not make future risk disappear.
Time creates another distinction. Chronic poverty persists over a long period. Transient poverty arises when a temporary shock or seasonal fall reduces resources. A survey that observes a household once cannot reliably tell which condition it faces. Repeated observations improve the diagnosis, though they too can suffer from missing households and measurement error.
The four ideas therefore belong together without being collapsed. Poverty concerns a current shortfall, inequality concerns a distribution, exclusion concerns barriers to participation, and vulnerability concerns risk over time.
Absolute and relative poverty use different standards
An absolute poverty line represents a chosen minimum standard that is intended to remain broadly stable in real terms. Its money value must change when prices change. If the cost of reaching the same standard differs between two places, the nominal line must also differ. “Absolute” therefore does not mean that nature supplied one perfect currency amount. People still choose the reference needs, valuation method and updating rule.
A relative poverty line moves with a feature of the current distribution, such as a fraction of typical income. It asks whether people have resources far below the living standard common in their society. The concern is not envy. Very low relative resources can restrict social participation, suitable housing, connectivity, education and the ability to meet ordinary social expectations.
The two approaches can move differently. If every income doubles and prices do not change, people may cross a fixed real absolute line. A relative line linked to the middle of the distribution may double too, so the relative-poverty classification need not change. This does not make one measure true and the other false. They answer different questions.
Every poverty line simplifies a continuum. Two households immediately on opposite sides of a line may have nearly identical lives. The line remains useful for consistent monitoring, but it is not a natural wall between the deprived and the secure. Nor is it the amount required for a fully good or dignified life. It is a measurement rule designed for a stated purpose.
Monetary poverty follows a chain of choices
Monetary poverty asks whether command over money-valued resources reaches a chosen line. The calculation can use income or consumption expenditure. Before choosing a line, however, the measurer must build a welfare aggregate: the value that will represent the household's material resources.
That aggregate may include cash purchases, goods produced and consumed at home, gifts, in-kind transfers and the service received from owner-occupied housing. It must decide how to treat occasional durable purchases, taxes, cash transfers and free or subsidised items. It must also choose recall periods for expenses that occur daily, monthly or rarely. Different reasonable choices can produce different aggregates.
Income and consumption show different parts of material life
Income records resources received during a period. It can show wages, self-employment returns, rent, interest and transfers. But income may vary sharply by season or be difficult to separate from business receipts. Informal and irregular earnings are often hard to recall. Some households may also under-report them.
Consumption expenditure records what the household uses or spends during a period. It may change less than income because a household can save in a good month, borrow in a bad month, receive help or sell an asset. This smoothing is one reason consumption has often played a large role in Indian poverty measurement.
Consumption is not automatically superior. A household may maintain spending by taking harmful debt or selling productive assets. Spending does not fully record the quality of publicly supplied health or education. A very ill household may report high medical expenditure without enjoying a high living standard. Income and consumption therefore illuminate different parts of command over resources.
Wealth is different again. Income and consumption are flows over a period. Wealth is a stock of assets minus liabilities on a date. Wealth can provide housing, collateral, security and influence, but it is difficult to value and often poorly captured at the top. A consumption-poverty estimate cannot silently become an income- or wealth-poverty estimate.
A household total is not yet individual welfare
Surveys often collect resources for the household. A simple per-person measure divides the household total by the number of members. This is transparent, but it assumes that every member has the same needs and receives an equal share.
Household composition matters. A household with young children may have different needs from one with the same number of adults. Members may share housing, cooking and durable goods, so doubling household size need not double every cost. Adult-equivalence and economies-of-scale adjustments try to reflect these differences, but they introduce assumptions about needs and sharing.
No household adjustment fully reveals allocation within the home. A household average can hide unequal food, care, leisure or schooling between members. A result that counts people using the household average must not be described as if each person's consumption were directly observed.
Prices and periods must be comparable
One nominal currency amount cannot measure the same real standard everywhere or in every year. Food, rent, transport and services cost different amounts across rural and urban areas, across states and over time. A line and the welfare aggregate must use compatible spatial and temporal prices.
The survey period matters too. Food bought frequently may be remembered differently from clothing, medical treatment or a durable good purchased months earlier. Seasonality can change both income and consumption. A survey conducted in different months, using different recall windows, may capture a different pattern even when the underlying living standard is similar.
The monetary measurement chain is therefore causal. First define the welfare resource. Then select the household or person unit. Adjust for needs, sharing, time and place. Construct a matching line. Only then calculate how many people fall short and by how much.
A poverty line serves a purpose; it does not settle every policy question
A common construction begins with reference needs or reference expenditure. A cost-of-basic-needs approach identifies a food standard and adds an allowance for non-food needs. Both parts require choices. The food component depends on what goods form the reference basket and how they are priced. The non-food component must represent needs such as shelter, clothing, transport, health or education without pretending that one basket describes every household perfectly.
The money value must then be adjusted for regional price differences and updated as prices and consumption patterns change. Old weights or baskets may become less representative. Better price data or a new survey can alter an estimate even when the underlying concern remains poverty.
Different lines can validly serve different purposes. A national statistical line supports monitoring within a country. An international line supports comparison using a common purchasing-power method. A programme eligibility rule selects beneficiaries under a particular law or scheme. None automatically determines the others.
This distinction matters for targeting. A statistical line estimates population poverty; it need not identify every household correctly. A benefit rule may consider disability, age, occupation, location, housing or other conditions. Treating the line as a perfect household list can create two errors. An exclusion error leaves an eligible person out. An inclusion error admits someone who does not meet the programme rule.
Thresholds can also change behaviour. Administrators may focus on moving people just across a cutoff while neglecting deeper shortfalls. Households may have an incentive to report circumstances that preserve eligibility. These risks do not make thresholds unnecessary. They show why a measured category should not become a permanent identity attached to a person.
Counting poverty is not enough; depth and severity matter
The simplest poverty measure asks how many people fall below the line. The headcount ratio divides the population below the line by the total population. Its denominator is people, using the appropriate population weights, not merely the number of surveyed households.
Headcount treats every person below the line alike. If a very poor person's resources rise but remain below the threshold, headcount does not change. It also does not tell us whether people are just below the line or far below it. A falling headcount can occur even while the people who remain poor experience no improvement.
The poverty gap asks about distance below the line. For each person below it, subtract resources from the line. Dividing that shortfall by the line creates a normalised gap. People at or above the line receive a zero gap. Averaging these gaps across the whole population combines incidence and depth.
The population denominator is important. An average gap across all people is not the same as the average gap only among poor people. The first measure becomes smaller when fewer people are poor; the second describes shortfall among those already identified.
A squared poverty gap gives larger shortfalls extra weight. After each normalised gap is found, it is squared before the population average is taken. Squaring a large gap increases its influence relative to a small gap. The resulting index is sensitive to severity, though it is not a percentage of people.
Consider four imaginary households of equal size. Their per-person resource values are 40, 80, 120 and 160 units. Use a fictional poverty line of 100 units. The members of two households fall below the line, so half the imaginary population is poor. The headcount ratio is 50 per cent.
The first poor household is 60 units short per person, giving a normalised gap of 0.60. The second is 20 units short, giving 0.20. The other two households receive zero gaps. Because household sizes are equal, averaging 0.60, 0.20, zero and zero gives a population poverty gap of 0.20, or 20 per cent.
For severity, square the two positive gaps. The squared gaps are 0.36 and 0.04. Their sum is 0.40, and dividing by four gives 0.10, or 10 per cent. This is a severity index, not a claim that 10 per cent of people are poor.
Now raise the first household's fictional per-person resources from 40 to 60 while leaving everything else unchanged. Headcount stays at 50 per cent because two equal-sized households remain below the line. The population poverty gap falls from 20 per cent to 15 per cent. The squared-gap index falls from 10 per cent to 5 per cent. Depth and severity detect an improvement that headcount misses.
No single measure is enough. Headcount describes incidence, the gap describes average population shortfall, and the squared gap gives greater importance to the deepest shortfalls. Disaggregation can also show whether an aggregate improvement hides a region or group that has fallen behind, provided the survey supports reliable subgroup estimates.
Inequality must name the resource and the population
Income inequality, consumption inequality and wealth inequality describe different distributions. Income can fluctuate and is measured over a period. Consumption is often smoother and may look less unequal. Wealth is accumulated over time and can be far more concentrated, especially when surveys miss top assets. None of their measures can substitute for another without warning.
The stage of income also matters. Market income is received before taxes and transfers. Gross income may add some transfers. Disposable income reflects what remains after direct taxes and relevant transfers. A policy can leave market-income inequality unchanged while altering disposable-income inequality. The fiscal mechanisms belong to later public-finance chapters; D06 owns the measurement distinction.
Outcome inequality records realised differences in resources, health, education or another result. Opportunity inequality asks how circumstances beyond a person's reasonable control shape those results. Inherited wealth, place of birth, discrimination and unequal access can affect opportunity. Outcomes also reflect choices, effort, luck and shocks, so a survey cannot always separate opportunity cleanly.
Inequality can also be examined through several social lenses. Vertical inequality compares people ordered by the level of a resource, such as poorer and richer households. Horizontal inequality examines systematic gaps between groups that may have similar economic positions but differ by gender, caste, community or another identity. Regional inequality compares places. Intergenerational inequality asks how advantages and disadvantages pass from parents to children or differ between generations.
These lenses may overlap without being compressed into one score. A national income distribution can improve while one region falls behind. Average gender gaps can narrow while wealth remains highly concentrated. Mobility across generations can remain weak even when current consumption gaps fall.
Absolute and relative inequality must also be separated. An absolute gap measures the resource difference between people. A relative comparison uses ratios or shares. If every household's resources double, the currency gap between richer and poorer households doubles, but each household's share can remain unchanged. A relative measure such as the Gini may then stay the same while absolute inequality rises.
Shares, ratios, Lorenz curves and Gini reveal different features
To study the distribution, first order people or households from the lowest to the highest value of the chosen resource. Percentiles divide the ordered population into hundredths, while deciles divide it into tenths. A bottom-decile share asks what fraction of total resources goes to the lowest tenth. A top-decile share asks the same for the highest tenth.
A percentile ratio compares values at two positions. A share ratio compares the resources received by two groups. The denominator must be explicit because reversing it reverses the interpretation. These measures are easy to communicate, but they ignore variation within the selected groups and say little about the rest of the distribution.
The Palma ratio divides the resource share of the top tenth by the share of the bottom four tenths. A value rises when the top share grows or the bottom share shrinks. It focuses attention on the ends of the distribution, but it leaves the middle and differences within each broad group less visible. It must also name whether the resource is income, consumption or something else.
A Lorenz curve shows cumulative shares
A Lorenz curve uses the entire ordered distribution. Begin with the poorest units and move towards the richest. At each point, add the population share already included and the share of total resources held by that population. Plot cumulative population horizontally and cumulative resources vertically.
The line of equality shows what would happen if every equal-sized population group held the same resource share. The poorest half would then hold half the resources, and every other cumulative point would lie on the diagonal. A Lorenz curve usually lies below this line because lower groups receive less than an equal share. Greater bowing away from the equality line suggests greater relative inequality.
Lorenz curves do not always give a complete ranking. One curve may lie closer to equality among lower groups but farther away among upper groups. The curves then cross. Neither distribution is unambiguously more equal by Lorenz dominance alone. A summary measure may still rank them, but that ranking depends on how it combines differences across the distribution.
Gini compresses the curve into one number
The Gini coefficient summarises the area between the equality line and the Lorenz curve relative to the largest possible area under that comparison. More bowing generally produces a larger Gini.
The scale must be stated. On a zero-to-one scale, zero represents complete equality and one is the limiting extreme. The same scale may be multiplied by one hundred, making the endpoints zero and one hundred. Thus a coefficient written as 0.30 may also appear as 30 on the larger scale. Gini is not a poverty rate or the percentage of people deprived.
A low Gini does not prove prosperity. Everyone can be equally poor. A high Gini does not reveal whether the gap occurs near the bottom, middle or top. Different distributions can produce the same coefficient. Transfers between two people may affect the coefficient differently depending on their positions, even if the transferred amount is the same.
Gini is also variable-specific. Income, consumption and wealth Ginis usually differ. Survey design, household-size adjustments, nonresponse and failure to capture the richest households can affect the result. One Gini should therefore be read with shares, ratios and the underlying distribution rather than treated as a complete inequality diagnosis.
Growth, distribution and poverty interact
Average growth expands total resources, but its distribution determines who crosses a poverty line and how far shortfalls shrink. If the resources of households below the line grow rapidly, poverty can fall even when richer households also gain. If growth is concentrated among people already far above the line, the headcount and gaps may change little.
Redistribution can alter this pattern through taxes, transfers, public services, assets or opportunities. Its effect depends on design, finance, incentives, administration and time. It can protect capability and widen participation. Poor design can also create exclusion, capture or fiscal strain. It is therefore wrong to assume that redistribution always destroys growth or always improves every poverty measure.
The phrase pro-poor growth has two common meanings. Under an absolute interpretation, growth is pro-poor when poor people's real resources rise. Under a relative interpretation, it is pro-poor when their gains exceed the average or the gains of richer groups, so relative inequality narrows. A growth episode can satisfy the first meaning without satisfying the second.
The arithmetic behind poverty change therefore has two parts. Average resources may rise or fall, and the distribution around that average may change. The poverty line and prices determine who crosses the threshold. Headcount shows crossing, while the gap and severity measures show what happened below it. A claim that “growth reduced poverty” is incomplete until the resource, distribution, period, prices and measure are known.
Surveys can disagree without one being fraudulent
Household surveys depend on a sample frame, questionnaire, interview method and recall periods. They can miss homeless or mobile populations, areas outside the frame or households that do not respond. People may forget irregular purchases, understate sensitive income, misunderstand items or report round numbers. Seasonal fieldwork can change the pattern captured.
The top of a distribution is especially difficult to measure. Very rich households may be rare in a sample, less likely to respond or hold complex assets that are hard to value. Household surveys also do not directly reveal how resources are shared among members. These limitations should be investigated, not used to dismiss all survey evidence.
National accounts answer another question. They assemble economy-wide production, income and expenditure from many sources. A household survey estimates the distribution of reported household resources in its covered population. Their concepts, boundaries, imputations and timing differ. A gap between a survey total and a national-accounts aggregate does not automatically prove that either side is fraudulent.
Comparability requires a disciplined check. The welfare variable, distribution stage, household/person unit, geography, price basis, reference period, recall window, sample, line, estimate status and data vintage must align. An observed survey result must not be mixed with an interpolation, projection or nowcast as if all were direct observations.
India’s poverty methods evolved with the measurement problem
Early Indian poverty estimation often linked minimum consumption to calorie norms and valued expenditure needed to reach a reference food requirement. This gave an apparently concrete anchor, but it faced serious limits. People with the same expenditure could consume different foods. Non-food needs changed. Health and education costs mattered. Price indices and consumption patterns differed between rural and urban areas and across states.
Later methods revised baskets, prices, survey reference periods and treatment of non-food spending. The history is therefore not a contest over one perfect line. It is an effort to make the welfare aggregate, price adjustment and minimum standard more suitable for the question and available data.
The Tendulkar Expert Group reported in 2009. It retained household consumption expenditure but moved away from directly anchoring each poverty line to a calorie norm. It used mixed reference periods, relied on a common urban reference basket adjusted for state and rural–urban price differences, and gave broader attention to health and education costs. Official Tendulkar-method estimates were produced for survey year 2011–12.
The Rangarajan group submitted an alternative in 2014. It used separate rural and urban reference baskets, normative food requirements and selected non-food components. It also used a modified mixed reference period. A parliamentary statement in 2014 said the report would be examined after feedback. That historical statement does not prove that the method became a timeless or presently binding official line.
Method names therefore cannot replace a version record. A complete claim must identify the method, survey year, welfare aggregate, prices, estimator and publication status. An estimate written by researchers does not become an official national series merely because an official publication later reproduces it.
HCES is a survey input, not a poverty rate
The Household Consumption Expenditure Survey for 2023–24 was conducted from August 2023 through July 2024. It records household consumption using a modified mixed reference period, three visits and item-specific recall. Monthly per-capita consumption expenditure divides household monthly consumption by household size and assigns that average to each member.
The method produces consumption evidence, not a poverty headcount by itself. A poverty estimate still needs a line, prices, unit, weighting and estimator. The 2023–24 design also differs from the 2011–12 survey in item coverage, questionnaire structure, number of visits, collection mode and treatment of specified free items. Comparisons can be studied, but only with those design changes visible.
International lines require a purchasing-power and data vintage
International poverty lines support comparison across countries. They use purchasing-power parity, or PPP, to translate resources by relative living costs. A market exchange rate measures the price of currencies in financial and trade transactions; it does not by itself equalise what households can buy in each country.
In June 2025, the global comparison method shifted to a 2021 PPP basis and revised historical estimates. This was a methodology and vintage change, not proof that past households suddenly gained or lost resources. Changing the PPP basis, adding surveys or revising welfare data can alter a historical series.
A complete international claim must therefore name the line used, PPP vintage, income or consumption concept, survey or estimate year, estimate state and data vintage. It must also distinguish a survey observation from an interpolation, projection or nowcast. An international line is not a domestic benefit cutoff, and its value must not be imported into programme eligibility.
This durable lesson does not publish a current international threshold or an Indian rate. Those values can change with data and methodology and belong in a dated update.
Multidimensional poverty identifies overlapping deprivations
Money cannot directly reveal every deprivation. A household may have enough measured consumption but lack safe sanitation, schooling or adequate nutrition. Public provision may improve one of these conditions without appearing as household cash. Multidimensional poverty examines selected deprivations together.
It is not automatically a more complete or objective measure. Designers choose dimensions, indicators, deprivation cutoffs and weights. They also choose the unit and the number or weighted share of deprivations required for classification. Important aspects can still be omitted, and data quality still matters.
The method begins with an indicator cutoff. For each indicator, decide whether a household is deprived. Apply the indicator's weight. Add the weights of the household's deprivations. Then compare that total with a multidimensional poverty cutoff. This second cutoff identifies which households face enough simultaneous weighted deprivation to count as multidimensionally poor. The two-cutoff structure is sometimes called a dual-cutoff method.
After identification, incidence, written as H, is the share of people who live in identified poor households. Intensity, written as A, is the average weighted deprivation share among those people. The adjusted headcount, written as M0, multiplies incidence by intensity. Thus M0 equals H times A. The measure falls when fewer people are identified or when the deprivation share among identified people falls.
Return to the same four equal-sized fictional households used in the monetary example. Now assess four equally weighted indicators. Each indicator carries one quarter of the total weight, and the fictional multidimensional cutoff is one half. The households with per-person resources of 40, 80, 120 and 160 units are deprived in three, one, two and zero indicators respectively. Their deprivation shares are therefore 75, 25, 50 and zero per cent.
The first and third households meet the fictional cutoff. Because all four households are equal-sized, incidence is two divided by four, or 50 per cent. Average intensity among the identified households is the mean of 75 and 50 per cent, which is 62.5 per cent. Multiplying 0.50 by 0.625 gives an adjusted headcount of 0.3125, or 31.25 per cent. These numbers illustrate the method; they describe no actual population. Notice that the household with 120 monetary units is multidimensionally poor in this example while the household with 80 units is not. Monetary and multidimensional classifications need not coincide.
India’s national MPI is one dated design
Under India’s 2023 national MPI method, three equally weighted dimensions contain twelve indicators. Indicator-level deprivation cutoffs and weights are combined, and the multidimensional poverty cutoff is a weighted deprivation share of 33.33 per cent. The report identifies deprivation at household level and reports people using those household classifications.
The 2023 report's observed results use NFHS-5, whose reference period is 2019–21. Later projections or modelled estimates do not become survey observations. A current claim must keep the method, indicator design, survey period and estimate status together.
A global MPI and a national MPI may use different indicators, cutoffs, weights, surveys and purposes. Two national editions can also change data or method. Their values should not be casually compared across incompatible vintages.
Monetary poverty, MPI and human-development indices answer different questions. Monetary poverty measures command over income or consumption. MPI identifies simultaneous selected deprivations. Human-development indices summarise selected average achievements. None automatically replaces or converts into the others.
Measures guide policy only when their limits remain visible
Poverty and inequality measures can guide geographic planning, reveal excluded groups, help target support, and monitor change. They can also help evaluate whether a policy reduced incidence, depth, severity or a chosen deprivation. Using more than one measure prevents a single headline from hiding important movement.
Misuse begins when a measure is treated as the person. A household can move around a line because of error, seasonality or a temporary shock. A multidimensional classification can change because one indicator crosses a cutoff. Gini can stay unchanged while absolute gaps grow. None of these labels captures a permanent identity or the full dignity and capability of the people measured.
A careful reader therefore starts with the question. For poverty, identify the minimum, resource or deprivation, unit, prices, period, data and depth measure. For inequality, identify the resource, population, distribution stage and summary used. Then check survey design, geography, estimate status and vintage.
The chapter began with two questions because every later method serves one of them. Poverty asks who falls seriously short and by how much. Inequality asks how resources, outcomes or opportunities are distributed. Sound measurement keeps the questions distinct, then uses them together to understand how growth and policy change people's lives.