Impoverishing Health Expenditure

Impoverishing Health Expenditure and the Poverty Impact of Health Payments

Impoverishing health expenditure is measured by comparing each household's consumption with a poverty line twice, once as reported and once after its out-of-pocket payments for care have been subtracted. The rise in the poverty headcount and the poverty gap between the two comparisons is the poverty impact of health payments, which ties the financial protection goal of universal health coverage to the goal of reducing poverty. The revised Sustainable Development Goal (SDG) indicator 3.8.2 can be split into impoverishing and large health spending, and WHO and the World Bank report the impoverishing part separately.

Why standard poverty counts miss the cost of illness

O'Donnell and colleagues, in a World Bank guide to health equity analysis, note that standard poverty measurement compares total household expenditure with a poverty line that is not sensitive to highly variable health care needs. A household that diverts so much spending to care that its spending on basic necessities falls below the poverty threshold is still not counted as poor. Nor is a poor household that borrows to pay for care and so raises its total expenditure above the line.

Their remedy is to assess poverty on household expenditure net of out-of-pocket costs for health care, since health spending responds to a basic need that the poverty line does not reflect. Not all health spending is for necessary treatment, so subtracting all of it overstates poverty, while ignoring it understates poverty. The guide reports an estimate that 78 million people in Asia were not counted as poor although their per capita expenditure net of health spending fell below the extreme poverty line of one US dollar a day.

Measuring poverty gross and net of health payments

The method computes the two basic poverty measures twice. The headcount ratio is the share of people below the line, and the poverty gap is the average shortfall from the line, counting zero for people above it. Each household is weighted by its size, because poverty estimates are conventionally made for individuals.

$$H^{\text{gross}} = \frac{\sum_{i=1}^{N} s_i \, \mathbf{1}(x_i < z)}{\sum_{i=1}^{N} s_i}, \qquad H^{\text{net}} = \frac{\sum_{i=1}^{N} s_i \, \mathbf{1}(x_i - T_i < z)}{\sum_{i=1}^{N} s_i}$$

where $x_i$ is the per capita consumption (or income) of household $i$, $T_i$ is its per capita out-of-pocket payment for health care, $z$ is the poverty line, $s_i$ is the number of household members, $N$ is the number of households and $\mathbf{1}(\cdot)$ equals 1 when the condition holds and 0 otherwise.

$$G^{\text{gross}} = \frac{\sum_{i=1}^{N} s_i \max(0, z - x_i)}{\sum_{i=1}^{N} s_i}, \qquad G^{\text{net}} = \frac{\sum_{i=1}^{N} s_i \max(0, z - x_i + T_i)}{\sum_{i=1}^{N} s_i}$$

where $G$ is the mean poverty gap in currency units per person. The poverty impact of health payments is the difference between the net and gross values of each measure.

$$NG = \frac{G}{z}, \qquad MPG = \frac{G}{H}$$

where $NG$ is the normalised poverty gap, used to compare countries with different lines and currencies, and $MPG$ is the mean positive poverty gap, the average shortfall of those who are poor. Because $G = H \times MPG$, a rise in the gap can come from more people being poor or from the poor falling further below the line.

Pushed into poverty and further impoverished

Some households counted as poor on gross expenditure appear poorer once health payments are netted out, and some counted as non-poor fall below the line. O'Donnell and colleagues show these as two areas under the poverty line in a Pen's parade of household expenditure. Wagstaff and van Doorslaer, studying Vietnam between 1993 and 1998, found that the poverty impact of out-of-pocket payments came mainly from poor people becoming even poorer rather than from the non-poor being made poor. With 1998 data and the USD 1.08 a day line, O'Donnell and colleagues instead trace the rise in the poverty gap to households brought into poverty, while at USD 2.15 the mean positive gap rises, so the split depends on the line.

The 2025 global monitoring report from WHO and the World Bank uses the same split. People pushed into poverty are above the line on total consumption and below it once out-of-pocket spending is deducted. People further impoverished would be poor whether or not they paid for care, yet their spending further reduces their ability to meet basic needs. In the report, everyone below the line with any out-of-pocket health spending counts as further impoverished, however small the payment.

Worked example with five households and one poverty line

All numbers in this example are illustrative. Five households with 20 members in all report per capita consumption and out-of-pocket payments for one year, and the poverty line is 1,000 currency units per person per year.

1. Headcounts. Households A and B are poor on gross consumption, and C joins them on net consumption.

$$H^{\text{gross}} = \frac{6 + 5}{20} = 0.55, \qquad H^{\text{net}} = \frac{6 + 5 + 3}{20} = 0.70$$

where the numerators are the members of the poor households. Health payments raise the headcount by 0.15, the share of people in household C.

2. Poverty gaps. On gross consumption, A falls 200 short of the line and B falls 50 short. On net consumption, A falls 250 short, B is unchanged and C falls 100 short.

$$G^{\text{gross}} = \frac{6 \times 200 + 5 \times 50}{20} = \frac{1{,}450}{20} = 72.5, \qquad G^{\text{net}} = \frac{6 \times 250 + 5 \times 50 + 3 \times 100}{20} = \frac{2{,}050}{20} = 102.5$$

where each term is members multiplied by shortfall. The gap rises by 30 units per person. Half of the rise, 6 × 50 / 20 = 15, comes from household A sinking deeper, and the other half, 3 × 100 / 20 = 15, from household C falling below the line.

3. Normalised and mean positive gaps. The normalised gap rises from 72.5 / 1,000 = 0.0725 to 102.5 / 1,000 = 0.1025. The mean positive gap rises from 72.5 / 0.55 = 131.8 to 102.5 / 0.70 = 146.4 units, so the poor are on average further from the line after their payments.

4. Reading it in the report's terms. The share pushed into poverty is 3 / 20 = 0.15 and the share further impoverished is 6 / 20 = 0.30, since household B is poor but paid nothing. Impoverishing health spending therefore affects 0.45 of the population. Household D spends 250 of its 300 above the line, or 83% of its discretionary budget, which the revised SDG indicator counts as large but not impoverishing.

5. Interpretation. Health payments add 15 percentage points to the poverty headcount and 30 units per person to the poverty gap. Household A spends only 6.25% of its budget on care, below the common 10% catastrophic threshold, yet it is impoverished, while household D crosses 10% but stays above the line.

Poverty lines and other analyst choices

Every estimate depends on the poverty line. O'Donnell and colleagues distinguish absolute lines, set as an amount of expenditure per person, from relative lines set as a fraction of mean or median expenditure, and expect most interest to lie in poverty against an absolute standard. Wagstaff and colleagues, in a study of 122 countries, used the USD 1.90 and USD 3.10 per day international lines and a relative line of 50% of median consumption per capita. The revised SDG indicator uses the societal poverty line, which in 2017 purchasing power parities is the higher of USD 2.15 and USD 1.15 plus 50% of median consumption or income net of out-of-pocket health spending.

A second choice is whether to lower the line when poverty is assessed net of health payments. O'Donnell and colleagues see no reason to adjust a subsistence food line, which allows nothing for health care, but a higher line may allow implicitly for expected health spending. Lowering it by the mean health spending of households near the line is not advisable for comparisons across countries or over time unless the standard line reflects such differences. Other choices are the welfare measure, per capita or equivalent adult scaling, and survey weights and design in the standard errors.

Impoverishing spending inside the revised SDG indicator 3.8.2

The original SDG indicator 3.8.2 counted people whose out-of-pocket spending exceeded 10% or 25% of total household consumption or income, and earlier reports added complementary trends in impoverishing spending. The revised indicator counts people with positive out-of-pocket spending above 40% of the household discretionary budget, defined as consumption or income minus the societal poverty line.

The impoverishing part is spending above 100% of the discretionary budget, which leaves no discretionary budget once payments are subtracted. People below the line have a negative discretionary budget, so any positive spending counts. The SDG metadata describe the revision as a unified framework for large and impoverishing spending, and the report states that the original and revised trends are not directly comparable.

Global evidence on impoverishing health spending

The 2025 report estimates that the share of the world's population facing impoverishing out-of-pocket health spending fell from 29% in 2000 to 20% in 2022, or 1.6 billion people. Most of it comes from people already poor, since the share further impoverished fell from 26.6% to 18.6% while the share pushed into poverty stayed between 1.9% and 2.8%. Global poverty fell from 43% to 25% over the period, but the share of the poor paying out of pocket for health rose from 64% to 76%, which the report reads as progress driven by poverty reduction more than by better financial protection.

Wagstaff and colleagues, using 516 surveys from 122 countries, estimated that at the USD 1.90 line impoverishment fell from 131 million people (2.1% of the world's population) in 2000 to 97 million (1.4%) in 2010. They found it even where the whole population was officially covered by insurance or national health services, and in some low-income countries it might add as much as four percentage points to the poverty headcount.

Uses in health financing policy and extended cost-effectiveness analysis

The measure is used to judge health system financing arrangements. Wagstaff and colleagues found incidence negatively correlated with the share of health spending channelled through social security funds and other government agencies, and concluded that raising the prepaid share, especially through taxes and mandatory contributions, can help reduce impoverishment. This is the logic of risk pooling, and the measure helps locate gaps left by cost sharing. One policy direction in the 2025 report is to make care free at the point of access for the poor.

In economic evaluation, extended cost-effectiveness analysis reports financial risk protection alongside health gains and costs. Verguet, Kim and Jamison list the number of cases of poverty averted as one metric, counting the people no longer pushed below a national or international poverty line by the costs of care under a policy such as public financing, estimated by wealth group.

How impoverishing spending differs from catastrophic spending

Catastrophic health expenditure asks whether payments are large relative to the household budget or its capacity to pay, while impoverishing expenditure asks whether they leave the household below the poverty line. Wagstaff and van Doorslaer present these as two threshold approaches to the fairness of health care payments, one requiring that payments not exceed a set share of income and the other that they not drive households into poverty. As household A shows, a small payment by a poor household can be impoverishing without crossing a budget share threshold.

What the measure cannot show

O'Donnell and colleagues call the gross and net comparison a rough approximation of the impoverishing effect, valid only if payments are completely nondiscretionary and household resources are fixed. Neither holds exactly. A household that chooses to spend heavily on care is not pushed into poverty by it, and one that borrows, sells assets or receives transfers to pay has gross expenditure that differs from what it would have consumed without the payments. The difference therefore cannot be read as the fall in poverty from a reform that removed those payments, although it indicates the scale of the effect.

The measure also sees only people who pay. The SDG metadata note that people unable to overcome barriers to access report no health spending, which lowers the indicator, so low rates need reading alongside service coverage and unmet need. A headcount of the further impoverished treats a tiny payment by a poor household like a large one, so the poverty gap is needed to show depth.

Sources

• O'Donnell O, van Doorslaer E, Wagstaff A, Lindelow M. Analyzing health equity using household survey data: a guide to techniques and their implementation. Washington, DC: World Bank; 2008. Chapters 18 and 19. https://openknowledge.worldbank.org/handle/10986/6896
• United Nations Statistics Division. SDG indicator metadata: indicator 3.8.2. New York: United Nations; last updated 27 March 2026. https://unstats.un.org/sdgs/metadata/files/Metadata-03-08-02.pdf
• Verguet S, Kim JJ, Jamison DT. Extended cost-effectiveness analysis for health policy assessment: a tutorial. PharmacoEconomics. 2016;34(9):913-923. https://doi.org/10.1007/s40273-016-0414-z
• Wagstaff A, Flores G, Smitz MF, Hsu J, Chepynoga K, Eozenou P. Progress on impoverishing health spending in 122 countries: a retrospective observational study. The Lancet Global Health. 2018;6(2):e180-e192. https://doi.org/10.1016/S2214-109X(17)30486-2
• Wagstaff A, van Doorslaer E. Catastrophe and impoverishment in paying for health care: with applications to Vietnam 1993-1998. Health Economics. 2003;12(11):921-934. https://doi.org/10.1002/hec.776
• World Health Organization, International Bank for Reconstruction and Development / The World Bank. Tracking universal health coverage: 2025 global monitoring report. Geneva: WHO and World Bank; 2025. https://www.who.int/teams/health-financing-and-economics/global-monitoring-report/2025