Out-of-pocket (OOP) health spending is among the most direct pathways through which households — including both female and male members — are pushed into poverty in low- and middle-income countries (LMICs). While the consequences fall on everyone in the family, these are not always evenly spread. When money runs short, it is often women who forgo care; when debts mount, women’s assets may be sold off first. Women are also less likely than men to be covered by contributory health insurance schemes that privilege formal-sector workers. We do not engage gender specifically in this blog, but the broader point is worth stating plainly: without health insurance, both women and men are worse off, and women typically more so. With insurance, at a minimum, all boats rise, and plausibly women stand to benefit more.
In low-income countries, 43% of all health spending by households is OOP — more than twice the rate in high-income countries (World Bank, 2024) — and in 30 LMICs, OOP payments remain the primary source of overall health financing (WHO, 2024). This deters many from using health services altogether, while impoverishing others. In 2022, 2.1 billion people — one in four people globally — faced financial hardship from OOP health costs, including 1.6 billion who were living in poverty or pushed deeper into it on account of health expenses alone (WHO, 2025).
Households in developing countries face several types of risk — from crop failures to job losses to natural disasters — and in a narrow sense, almost any such event may be described as impoverishing. Aggregate shocks such as floods hit entire communities and can, in principle, draw government and external support. In contrast, health shocks strike individual households unpredictably, and the associated financial burden can be a significant proportion of lifetime income. It is this combination of household-specific risk and large losses that health insurance is designed to pool. Recent research shows that informal risk-sharing within families and communities does not fully smooth these shocks (Kinnan et al., 2020), which further strengthens the case for formal insurance.
By pooling risk across households, health insurance is one of the main policy tools available to soften the blow in case of health shocks. In practice, however, making insurance work in developing countries throws up challenges at every stage — from designing a policy, to getting people to enroll, to paying for it, to ensuring quality supply of care for the insured population. In the remaining sections of this blog, we work through some of these questions that the last two decades of research has focused on: Does insurance deliver on its promise? If it does, why do people not enroll? And how do countries pay for it?
Does health insurance deliver on its promise?
Health insurance has three key purposes: to increase access to healthcare by making it affordable; to protect households from the financial consequences of illness; and to improve health outcomes through increased healthcare access. Does insurance achieve these objectives? A review of the relevant literature, discussed below, suggests that the evidence is strong on two of these three dimensions and uneven on the third.
Healthcare utilization
Insurance theory predicts that by lowering the price of healthcare at the point of use, health insurance should raise utilization of care (Banerjee et al., 2021). Recent empirical evidence confirms this strongly and consistently. In a systematic review of public insurance schemes in LMICs, Erlangga et al. (2019) find that 32 of 40 studies report a statistically significant positive effect on utilization of curative care, with larger effects among the higher-quality studies that credibly handle ‘selection bias’1.
Country-specific studies corroborate this pattern. Miller, Pinto and Vera-Hernández (2013) examine Colombia’s Régimen Subsidiado — a subsidized insurance scheme for low-income households — and find that insurance raises the probability of a preventive physician visit by 29 percentage points. In the context of Peru’s Seguro Integral de Salud (SIS), Bernal, Carpio and Klein (2017), document large increases in doctor visits, receiving medication, and diagnostic tests among the informally employed2. In a more recent meta-analysis of community-based insurance across 20 LMICs, Eze et al. (2023) find substantial pooled effects3 on outpatient use and the share of births delivered in health facilities.
These findings align with evidence from the developed world. The Oregon Health Insurance Experiment, one of the best-known studies of health insurance in any setting (Finkelstein et al., 2012; Baicker et al., 2013; Finkelstein et al., 2016), shows that access to insurance increases healthcare use across the board — outpatient care, preventive care, prescription drugs, hospital admissions, and emergency room visits. This is likely to have a positive effect on welfare, particularly in the case of preventive services.
Financial protection
Health insurance is, at its core, a financial product that seeks to protect against the economic consequences of illness. The available evidence indicates that insurance in developing countries reduces OOP spending and the incidence of catastrophic payments — two key indicators of financial protection. Because studies use many different indicators and model specifications, the evidence here is less consistent than for utilization, but the direction is clear.
In a randomized controlled trial (RCT) in rural Cambodia, Levine, Polimeni and Ramage (2016) demonstrate that among insured households, healthcare costs following a major health shock decline by 44%, health-related debt falls by 77%, and distress asset sales drop measurably. Erlangga et al. (2019) find that 9 of 14 reviewed studies report reductions in catastrophic health expenditure. In a similar vein, Fink et al. (2013) conduct a randomized community-based insurance rollout in rural Burkina Faso, and find only limited effects on average OOP spending, but a substantial reduction in the likelihood of catastrophic expenditure — which is, arguably, what insurance is designed to do in the first place.
In the developed-country context, the Oregon experiment exhibits a similar pattern. Medicaid virtually eliminates catastrophic OOP medical expenditures, and lowers the probability that households have to borrow money or skip paying other bills on account of medical expenses (Baicker et al., 2013). A product that makes the median doctor visit somewhat cheaper is not particularly notable; a product that prevents a household from falling into financial ruin once a decade is indeed significant.
Health outcomes
Improving health outcomes is a key objective of insurance. Yet, owing to measurement challenges, the majority of existing literature does not attempt to evaluate physical health outcomes. The link between health insurance and health status in developing countries is therefore unclear. Erlangga et al. (2019) identify only 12 studies evaluating health outcomes among scheme participants in LMICs, with considerable variation in the indicators chosen and no conclusive message.
In a large RCT of India’s RSBY (Rashtriya Swasthya Bima Yojana)4, Malani et al. (2024) find very few statistically significant impacts on health, despite substantial utilization gains. In the developed-country context, the Oregon experiment shows improvements in self-reported health, depression, and diabetes diagnosis and management, but no significant effect on blood pressure, cholesterol, or glycated hemoglobin (Baicker et al., 2013). Fink et al. (2013) report a temporary increase in mortality among older community members5 in their Burkina Faso study — a finding that the broader literature rarely highlights.
Two recent studies suggest that life stages and time horizons matter. Gruber, Hendren and Townsend (2014) analyze Thailand’s Universal Coverage Scheme and find that it equalizes infant mortality across rich and poor provinces, with a suggestive aggregate reduction that the authors treat cautiously. Huang and Liu (2023) track children exposed to China’s New Cooperative Medical Scheme and find meaningful improvements in adolescent health, cognition, and educational attainment — but only for children exposed before the age of five. Hence, insurance may be most valuable during critical developmental windows, with benefits showing up over decades rather than years.
In sum, the literature indicates that health insurance substantially improves access to and use of healthcare, and provides financial protection in most cases. The link between insurance and health outcomes remains unclear and understudied. This is mostly attributable to difficulties in measuring health status in LMICs, quality of available data, and the challenge of establishing a causal link — but some of it also reflects the genuinely long time horizons between insurance and measurable physical effects. Insurance works first and most reliably as a financial product. Whether it works as a health product is a question we still cannot fully answer.
Why don’t people enroll?
Given how well insurance performs as a financial product, one would expect eligible households to sign up in large numbers. They don’t. India’s PMJAY (Pradhan Mantri Jan Arogya Yojana) reached roughly a third of eligible households three years after its 2018 launch. The coverage of Ghana’s NHIS (National Health Insurance Scheme) hovered between 35% and 41% of the population through 2019, before rising sharply to around 55% by 2022. Indonesia’s JKN Mandiri (Jaminan Kesehatan Nasional) enrolled about 20% of its target population in the first year. Vietnam took nearly two decades of gradual expansion — combining voluntary uptake, mandatory enrollment, and government subsidies for the poor — to reach 60% coverage. These are not programs with high premiums; many are free or heavily subsidized for the target population. The puzzle, therefore, is why people fail to enroll even when it costs them little.
The recent experimental literature points to three mechanisms.
Subsidies matter, but less simply than one might expect: In an RCT in Indonesia, Banerjee et al. (2021) find that a time-limited full subsidy raises enrollment by 20.9 percentage points — almost seven times the rate in the ‘control’ group (no subsidy) — and attracts ‘lower-cost’ enrollees, reducing ‘adverse selection’6. Coverage remains higher than in the control group even after subsidies expire, as many households choose to renew the insurance. Malani et al. (2024) study a similar question in India and find that uptake is about 20 percentage points higher when RSBY is offered for free relative to being offered at the full actuarially fair premium (that is, a price equal to the government’s per-person cost), but they also document 60% uptake at a premium equal to the government’s cost of providing insurance. The authors calculate that the welfare-maximizing premium for RSBY is 67-95% of average costs, not zero. Along similar lines, Asuming, Kim and Sim (2024), in a field experiment in Ghana, find that partial and full subsidies produce similar long-run enrollment, but only the partial-subsidy group goes on to use more healthcare. In other words, some cost-sharing may motivate people to engage with the product rather than merely holding on to a card.
Transaction costs are a real barrier: Banerjee et al. (2021) also study the impact of transaction costs directly. They randomly offer some Indonesian households at-home registration assistance, which raises enrollment by about 3.5 percentage points. More telling is the finding that over half of those who attempt to enroll fail to do so, on account of technical and administrative glitches. Malani et al. (2024) observe a similar pattern in India at the point of use: many beneficiaries try to use their insurance and are unable to — hospitals reject cards, patients forget to carry them, or nobody is able to explain the system to them. In essence, the friction runs through the whole chain, from enrollment to utilization.
Target population may attach low value to the insurance product: Olken et al. (2024) exploit an unusual Thai policy that offered informal workers a large, one-time lump-sum incentive to enroll in voluntary social insurance. Coverage jumped from 6% to 73% within two months, and then fell to 13% within a year. Analyzing people’s choices across insurance tiers, the authors observe that those induced to enroll by the incentive valued the insurance much less than those who had enrolled without it. Low take-up in Thailand, they argue, reflects low valuation of the insurance product itself, and not administrative barriers. If this finding holds elsewhere, removing frictions will not fix enrollment where people do not see enough value in the coverage being offered. Low valuation, in turn, often reflects low perceived quality of the care that comes with it — a supply-side problem rather than a demand-side one.
Experience cuts both ways: Indonesian and Ghanaian households that used insurance and benefited from it were more likely to renew after subsidies expired (Banerjee et al., 2021; Asuming et al., 2024). On the other hand, nearly all participants in the Nicaragua experiment by Thornton et al. (2010) dropped out once their coverage ran out. Whether experience builds demand, therefore, depends on what people actually get when they walk into a health facility with an insurance card. Supply-side quality and demand-side uptake are closely tied together.
How do you pay for it?
Designing a workable scheme and getting people to enroll are hard enough. Financing the whole thing is harder. The vast majority of developing-country households cannot pay insurance premiums from their own pockets, and governments therefore have to find the money somewhere. Countries have tried nearly every plausible combination: earmarked taxes (Ghana’s National Health Insurance Levy), reallocation from existing revenue pools (Thailand and India), payroll contributions from formal-sector workers combined with tax or donor funding for the rest (Indonesia, Kenya, the Philippines, Vietnam), and tax-financed coverage for the poor with voluntary enrollment for everyone else. So far, no country has found a clean path.
The empirical case against payroll-tax-financed social health insurance in low-income settings is now quite strong. Yazbeck et al. (2020) argue that in contexts where most workers are informal, payroll taxes cannot raise meaningful revenue; they end up being regressive, as lower-paid formal workers contribute a larger share of their income than higher earners. Kenya’s social health insurance system, established in 1966 and financed primarily through payroll contributions from formal-sector workers, effectively covered only about 17-18% of the population by 2023, despite decades of reform. The constraint is structural: universal coverage cannot be built on a payroll base that draws on only 20-30% of the workforce.
General taxation, on the other hand, is the more promising route, but it requires the kind of political commitment that the economics literature tends to understate. Reeves et al. (2015), modelling 89 LMICs, find that progressive tax structures are associated with higher public spending on health and stronger health-system outcomes. Thailand’s tax-financed Universal Coverage Scheme (Tangcharoensathien et al., 2018) is the clearest case of success, but it required a decisive political choice during the 2001 election campaign to fund coverage from general revenues. The arithmetic alone did not produce Thai Universal Health Coverage (UHC); the politics did. And yet, even where financing is in place, coverage is often unequal. Barasa et al. (2021) document that across 36 countries in sub-Saharan Africa (SSA), average health insurance coverage is just 7.9%, and concentrated in the richest quintiles.
Given the limited tax base and low household capacity, progress towards universal insurance will inevitably be gradual and multi-pronged — combining tax-funded coverage for the poorest, community-based schemes for informal workers, and payroll contributions where a formal sector exists. The immediate priority is ensuring that the poorest and most vulnerable are covered at the start through general tax-financed schemes, while building longer-term plans suited to local social and economic conditions. There is no shortcut, and templates borrowed from high-income countries rarely survive contact with the labour markets and fiscal realities of the developing world.
Table 1. How different countries have approached health insurance
Conclusion
The promise of health insurance in developing countries is real. Major health expenses are infrequent but extremely high as a proportion of lifetime income, and pooling risks across households is welfare-enhancing. The evidence confirms that insurance substantially improves access to and use of healthcare, and provides financial protection in most cases. The link to physical health outcomes remains unclear, although recent work suggests that the effects are tangible over longer horizons and for children who are exposed to health insurance at an early stage. Enrollment, meanwhile, remains low, and the reasons run deeper than price alone. Removing frictions helps, but where people do not value the product, often because they do not trust the care that comes with it, frictions are not the binding constraint. Financing is the hardest of all. Payroll contributions cannot carry the weight in low-income settings, and general taxation requires political choices that are rarely purely technical.
The demand for health insurance and its utilization is inextricably linked to the quality of healthcare that is available. In our next blog, we will look at public investments in infrastructure such as gender-sensitive transport, to complement stronger health systems in LMICs.
FOOTNOTES
1 Selection bias occurs when the group being studied is not representative of the broader population, leading to skewed conclusions. In health insurance research, this typically arises when less healthy individuals are more likely to enroll, making it difficult to isolate the true effect of insurance from pre-existing differences between the insured and uninsured.
2 Bernal, Carpio and Klein (2017) use a regression discontinuity design (RDD), a method that exploits a sharp eligibility cutoff to approximate a randomized comparison. In this case, Peru’s SIS scheme covered households below a specific poverty score threshold. By comparing households just below and just above that cutoff, who are otherwise very similar, the researchers can isolate the causal effect of insurance from other factors that might influence healthcare use.
3 A pooled effect is the combined estimate produced by a meta-analysis — a study that aggregates results across multiple independent studies. Rather than relying on any single country or context, Eze et al. (2023) combine findings from 20 LMICs to produce an overall estimate of the effect of community-based insurance on healthcare use. Pooling increases statistical precision and allows conclusions that are more generalizable than any individual study alone.
4 Launched in 2008, RSBY was India’s national health insurance scheme for below-poverty-line workers and their families. It provided cashless hospitalisation coverage of up to Rs. 30,000 per family per year at empaneled public and private hospitals. The scheme was subsequently subsumed into PMJAY (Pradhan Mantri Jan Arogya Yojana) in 2018, which raised population coverage to the poorest 40% as well as the insurance amount (up to Rs. 500,000).
5 Facilities received a flat fee per enrolled person regardless of services delivered, which lowered the motivation to treat insured patients well. This deterred the elderly from seeking timely treatment for diseases such as malaria.
6 Adverse selection in health insurance occurs when those more likely to need healthcare — typically less healthy individuals — are disproportionately likely to enroll, while healthier people opt out. This skews the risk pool toward high-cost members, pushing up average claims and potentially making the scheme financially unsustainable. Subsidies that attract a broader, healthier pool of enrollees help counteract this dynamic.
REFERENCES
Asuming, Patrick Opoku, Hyuncheol Bryant Kim, and Armand Sim. “Selection and Behavioral Responses of Health Insurance Subsidies in the Long Run: Evidence from a Field Experiment in Ghana.” Health Economics 33, no. 5 (2024): 992–1032. https://doi.org/10.1002/hec.4797
Baicker, Katherine, Sarah L. Taubman, Heidi L. Allen, Mira Bernstein, Jonathan H. Gruber, Joseph P. Newhouse, Eric C. Schneider, Bill J. Wright, Alan M. Zaslavsky, and Amy N. Finkelstein. “The Oregon Experiment — Effects of Medicaid on Clinical Outcomes.” New England Journal of Medicine 368, no. 18 (2013): 1713–1722. https://doi.org/10.1056/NEJMsa1212321
Banerjee, Abhijit, Amy Finkelstein, Rema Hanna, Benjamin A. Olken, Arianna Ornaghi, and Sudarno Sumarto. “The Challenges of Universal Health Insurance in Developing Countries: Experimental Evidence from Indonesia’s National Health Insurance.” American Economic Review 111, no. 9 (2021): 3035–3063. https://doi.org/10.1257/aer.20200523
Barasa, Edwine, Jacob Kazungu, Peter Nguhiu, and Nirmala Ravishankar. “Examining the Level and Inequality in Health Insurance Coverage in 36 Sub-Saharan African Countries.” BMJ Global Health 6, no. 4 (2021): e004712. https://doi.org/10.1136/bmjgh-2020-004712
Bernal, Noelia, Miguel A. Carpio, and Tobias J. Klein. “The Effects of Access to Health Insurance: Evidence from a Regression Discontinuity Design in Peru.” Journal of Public Economics 154 (2017): 122–136. https://doi.org/10.1016/j.jpubeco.2017.08.008
Erlangga, Darius, Marc Suhrcke, Shehzad Ali, and Karl Bloor. “The Impact of Public Health Insurance on Health Care Utilisation, Financial Protection and Health Status in Low- and Middle-Income Countries: A Systematic Review.” PLOS ONE 14, no. 8 (2019): e0219731. https://doi.org/10.1371/journal.pone.0219731
Eze, Paul, Lucky Osaheni Lawani, Ujunwa Justina Agu, Linda Uzo Amara, Chioma Adaeze Okonkwo, and Yubraj Acharya. “Impact of Community-Based Health Insurance in Low- and Middle-Income Countries: A Systematic Review and Meta-Analysis.” PLOS ONE 18, no. 6 (2023): e0287600. https://doi.org/10.1371/journal.pone.0287600
Fink, Günther, Paul Jacques Robyn, Ali Sié, and Rainer Sauerborn. “Does Health Insurance Improve Health? Evidence from a Randomized Community-Based Insurance Rollout in Rural Burkina Faso.” Journal of Health Economics 32, no. 6 (2013): 1043–1056. https://doi.org/10.1016/j.jhealeco.2013.08.003
Finkelstein, Amy, Sarah Taubman, Bill Wright, Mira Bernstein, Jonathan Gruber, Joseph P. Newhouse, Heidi Allen, Katherine Baicker, and the Oregon Health Study Group. “The Oregon Health Insurance Experiment: Evidence from the First Year.” Quarterly Journal of Economics 127, no. 3 (2012): 1057–1106. https://doi.org/10.1093/qje/qjs020
Finkelstein, Amy N., Sarah L. Taubman, Heidi L. Allen, Bill J. Wright, and Katherine Baicker. “Effect of Medicaid Coverage on ED Use — Further Evidence from Oregon’s Experiment.” New England Journal of Medicine 375, no. 16 (2016): 1505–1507. https://doi.org/10.1056/NEJMp1609533
Gruber, Jonathan, Nathaniel Hendren, and Robert M. Townsend. “The Great Equalizer: Health Care Access and Infant Mortality in Thailand.” American Economic Journal: Applied Economics 6, no. 1 (2014): 91–107. https://doi.org/10.1257/app.6.1.91
Huang, Wei, and Hong Liu. “Early Childhood Exposure to Health Insurance and Adolescent Outcomes: Evidence from Rural China.” Journal of Development Economics 160 (2023): 102925. https://doi.org/10.1016/j.jdeveco.2022.102925
Kinnan, Cynthia, Krislert Samphantharak, Robert Townsend, and Diego Vera-Cossío. “Insurance and Propagation in Village Networks.” NBER Working Paper No. 28089, 2020.
Levine, David, Rachel Polimeni, and Ian Ramage. “Insuring Health or Insuring Wealth? An Experimental Evaluation of Health Insurance in Rural Cambodia.” Journal of Development Economics 119 (2016): 1–15. https://doi.org/10.1016/j.jdeveco.2015.10.008
Malani, Anup, Cynthia Kinnan, Gabriella Conti, Kosuke Imai, Morgen Miller, Shailender Swaminathan, Alessandra Voena, and Bartosz Woda. “Evaluating and Pricing Health Insurance in Lower-Income Countries: A Field Experiment in India.” NBER Working Paper No. 32239, 2024. https://doi.org/10.3386/w32239
Miller, Grant, Diana Pinto, and Marcos Vera-Hernández. “Risk Protection, Service Use, and Health Outcomes under Colombia’s Health Insurance Program for the Poor.” American Economic Journal: Applied Economics 5, no. 4 (2013): 61–91. https://doi.org/10.1257/app.5.4.61
Olken, Benjamin A., Rema Hanna, Phitawat Poonpolkul, and Nada Wasi. “Willingness-To-Pay vs Administrative Hurdles: Understanding Barriers to Social Insurance Enrollment in Thailand.” NBER Working Paper No. 33096, 2024. Forthcoming, Review of Economics and Statistics.
Reeves, Aaron, Yannis Gourtsoyannis, Sanjay Basu, David McCoy, Martin McKee, and David Stuckler. “Financing Universal Health Coverage — Effects of Alternative Tax Structures on Public Health Systems: Cross-National Modelling in 89 Low-Income and Middle-Income Countries.” The Lancet 386, no. 9990 (2015): 274–280. https://doi.org/10.1016/S0140-6736(15)60574-8
Tangcharoensathien, Viroj, Woranan Witthayapipopsakul, Warisa Panichkriangkrai, Walaiporn Patcharanarumol, and Anne Mills. “Health Systems Development in Thailand: A Solid Platform for Successful Implementation of Universal Health Coverage.” The Lancet 391, no. 10126 (2018): 1205–1223. https://doi.org/10.1016/S0140-6736(18)30198-3
Thornton, Rebecca L., Laurel E. Hatt, Erica M. Field, Mursaleena Islam, Freddy Solís Diaz, and Martha Azucena González. “Social Security Health Insurance for the Informal Sector in Nicaragua: A Randomized Evaluation.” Health Economics 19, S1 (2010): 181–206. https://doi.org/10.1002/hec.1635
World Bank. “Out-of-Pocket Expenditure (% of Current Health Expenditure).” World Development Indicators. Washington, DC: World Bank, 2024. https://data.worldbank.org/indicator/SH.XPD.OOPC.CH.ZS
World Health Organization. “New WHO Report Reveals Governments Deprioritizing Health Spending.” WHO News Release, 12 December 2024. https://www.who.int/news/item/12-12-2024-new-who-report-reveals-governments-deprioritizing-health-spending
World Health Organization. “Universal Health Coverage (UHC).” WHO Fact Sheet, updated 5 December 2025. https://www.who.int/news-room/fact-sheets/detail/universal-health-coverage-(uhc)
Yazbeck, Abdo S., Agnes L. B. Soucat, Ajay Tandon, Cheryl Cashin, Joseph Kutzin, John Watson, Ayodeji Ajuwon, Justine Hsu, and Feng Wang. “The Case Against Labor-Tax-Financed Social Health Insurance for Low- and Low-Middle-Income Countries.” Health Affairs 39, no. 5 (2020): 892–897. https://doi.org/10.1377/hlthaff.2019.00874