Skip to main content

Health Equity and Financial Protection in Asia

Final Report Summary - HEFPA (Health Equity and Financial Protection in Asia)

Executive Summary:

In much of Asia, accessing healthcare involves incurring high out-of-pocket (OOP) payments. Many sick people forgo needed treatment, get inadequate care or incur medical expenses that threaten their financial security. HEFPA evaluates the effectiveness of various demand and supply side policy interventions intended to increase utilization of appropriate care and/or provide better protection from medical expenditure risks. The project has conducted research in six countries at different stages of socioeconomic and health system development: Cambodia, China, Indonesia, the Philippines, Thailand, and Vietnam. Findings reveal that voluntary insurance enrolment is unlikely to achieve large scale extensions of health coverage; that, when achieved, coverage extensions generally lead to increased utilization but that improved protection from the financial consequences of illness is much harder to achieve and likely depends on complex interactions between the design of insurance and provider payment systems; that provider incentives can be used to improve access to appropriate, quality health care; and that illness-induced income loss continues to pose a substantial financial risk to Asian households even when comprehensive health coverage is achieved.

Project Context and Objectives:
HEFPA Summary

Throughout the developing world, billions of people experience great difficulties in accessing adequate health care when ill. Partly, this is a problem of affordability. Where health insurance coverage is low and insured benefits shallow, the burden of paying for medical care is concentrated on those falling sick rather than being spread across the population. Heavy reliance on OOP payments forces a dire choice upon many households: forgo adequate care or incur high expenditures that crowd out consumption and possibly investments, e.g. in education, that determine long-term income potential. Access to adequate care is also hampered by supply side inefficiencies that compromise care quality. Reducing such inefficiencies is paramount in developing country health systems that typically face tight budget constraints.
The benefits of improved access to medical care go beyond the health sector. Improved population health and reduced exposure to health expenditure risks can directly and indirectly (through investments in income-generating assets) increase human capital and productivity - basic ingredients of economic development.

Health insurance coverage is being extended in a number of countries through a variety of approaches ranging from compensating public providers for granting fee waivers to the poor, through subsidising voluntary insurance enrolment of the informal sector, to universal population coverage financed by taxation. HEFPA used these developments in six countries of East and South-East Asia—Cambodia, China, Indonesia, the Philippines, Thailand and Vietnam— to learn about both the feasibility of approaches to extending coverage and the impact of coverage extensions on access to care and protection from the financial risks associated with illness. Recognising the importance of provider behaviour in making coverage effective, the project evaluates interventions designed to improve the appropriateness and efficiency of care delivery.

Two randomized experiments conducted in the Philippines and Vietnam reveal the difficulty of covering substantial proportions of informal sector populations through voluntary insurance even when premiums are subsidized and information on the operation and benefits of insurance is provided. In the Philippines, a 25-50% subsidy plus an information kit increased enrolment by five percentage points, which is 50% of the (very low) enrolment rate achieved without these inducements. While this is an impressive relative gain, it is insufficient to push enrolment substantially closer to full coverage. Among those initially not induced to enrol, the percentage consenting to sign up was raised by over one third when there was an option to have the application form completed and delivered to the insurance office on behalf of the applicant. Inconvenience and time costs appear to be a greater barrier to enrolment than affordability, which is an important lesson for the design of insurance programs.
The message that raising voluntary insurance coverage is not all, or even mainly, about increasing affordability of the premium is reiterated by the experiment conducted in Vietnam. A 25% premium subsidy offered along with an information pack had no significant impact on insurance uptake. In this case, it appears that it is the low value of the insurance, given low out-of-pocket prices of medical care and low perceived quality of the care accessed through insurance that is responsible for the low uptake.
When insurance coverage is extended, or deepened, utilization of medical care tends to increase. HEFPA studies confirm this with respect to the impact of: i) increased generosity of subsidized insurance for the rural population in two Chinese provinces; ii) entitlement to tax-financed public healthcare for the poor in Indonesia; and iii) entitlement to public care for all citizens not covered by formal sector insurance programs in Thailand.
While full insurance coverage would eliminate households’ out-of-pocket (OOP) payments, the impact on financial protection of incomplete insurance that changes the relative prices of different levels of medical care and interacts with provider payment incentives is more ambiguous. This is illustrated by comparison of the findings from the HEFPA studies.
Health Equity Funds (HEF), which are administered by NGOs and compensate public providers for waiving the fees of poor patients, are found to reduce average OOP payments, among those household incurring them, by a very substantial 35%, with a greater impact on the poor. This is achieved not simply by granting a fee waiver but by providing facilities with the incentive to provide treatment without charging the patient.
By contrast, the program providing tax-financed care for the poor in Indonesia (Askeskin/Jamkesmas) did not reduce OOP expenditure. In fact, OOP payments increased among the urban population. This unexpected result may derive from the increased utilization of partially covered, but more affordable, services, in particular inpatient care. Provider incentives to favour treatments generating revenue from OOP payments may also play a role. This problem is even more evident in the program providing coverage to China’s rural population – the New Chinese Medical Scheme (NCMS). Despite a tremendous expansion of NCMS coverage, tripling of its subsidy between 2008 and 2012 and a reduction in OOP as a percentage of total health expenditure from 50% to 35%, prevalence of catastrophically high household payments for medical care has not been reduced. This appears to reflect the design of insurance and provider payments. Initially, NCMS only (partially) covered expenditures on inpatient care. A HEFPA experiment found that extending coverage to ambulatory care and raising reimbursement for treatment at primary facilities above that at secondary and tertiary facilities reduced the prevalence of catastrophic OOP from one-third to less than a quarter.
The Thai universal coverage reform avoided the possibility of increased OOP payments by providing a near comprehensive benefit package and making coverage effective by accompanying the extension of entitlement with various organizational and provider payment reforms intended to improve cost-effectiveness. The HEFPA evaluation finds that the reform reduced OOP payments by one third on average, and by one half among those with the highest (5%) expenditures.
In another strand of research, HEFPA investigated how changes in provider incentives impact on the appropriateness and quality of care that the population can access. Experiments in two Chinese provinces found that a change in provider reimbursement from fee-for-service to a combination of capitation and pay-for-performance reduced prescription of antibiotics, a measure of quality, in one of the two provinces but prescription rates remained above internationally recommended levels across both.

Spreading the economic burden of illness across the population is a major motivation behind the drive for universal coverage. HEFPA has examined the economic consequences of illness, and how households cope with these in the absence of comprehensive insurance, in a number of Asian countries. This research reveals that the extent to which, and how, households protect their consumption when hit by a health shock varies across countries. One important common finding is that income losses arising from ill-health tend to impose a larger economic burden than medical expenses. In part, this may be because poor households forgo needed, but unaffordable, health care. But it also reflects the near absence of sickness and disability insurance coverage of the majority of the populations of low and middle income Asian countries.

On top of the evaluation of the impact of health financing interventions on access and financial protection and examination of the economic impacts of illness on households, the project has yielded methodological innovations regarding the measurement of inequity in the utilization of health care and the measurement of catastrophic medical expenditure risk. The tools developed offer the potential for richer monitoring of health equity and financial protection across and within countries.

By studying countries at different levels of economic and health system development, the evidence provided can potentially inform health policy not only in these countries but is relevant to other countries striving to achieve better healthcare access and financial protection.
The main lessons for policy are:
• Large scale health coverage extension is unlikely to be achieved through voluntary insurance enrolment
• Increasing affordability of insurance will not necessarily substantially increase uptake. Transactions costs and low (perceived) value of insurance are major disincentives to enrolment.
• While coverage extensions generally lead to increased utilization, improvements in financial protection are much harder to achieve and likely depend on complex interactions between the design of insurance and provider payment.
• Carefully crafted provider incentives can increase the appropriateness and quality of care that can be accessed.
• Illness-induced income loss continues to pose a substantial financial risk to Asian households even if comprehensive health coverage is achieved.

Potential applications:
• Streamline the administration of voluntary health insurance to reduce non-monetary costs of application and raise enrolment.
• Simultaneous design of insurance benefits package and provider payment in order to raise incentives for the utilization and delivery of appropriate, cost-effective and quality medical care within available budget.
• Advance agenda for sickness and disability insurance recognising the difficult of providing this in countries with very large informal labour markets.
• Improved monitoring of health equity and financial protection across and within countries.

Project Results:

Health Equity and financial protection in Asia
(HEFPA Project)
Executive summary

Good evidence on interventions that aim to extend coverage and ensure that effective medical care can be accessed without threatening household financial security can quicken the pace of progress toward Universal Health Coverage (UHC). The HEFPA project evaluates policies with these goals in six countries in East and Southeast Asia.
Findings suggest that even when heavily subsidised, voluntary insurance is unlikely to bring coverage close to universality. When accomplished, increased coverage generally raises the utilisation of healthcare, but does not necessarily reduce the burden of household medical expenditures. The impact on out-of- pocket (OOP) payments depends on the design of provider payments, as well as the benefit package, respectively offering incentives for the delivery and utilisation of expensive treatments that are made affordable by insurance, but still only partially covered. Provider incentives may be just as important to securing access to appropriate treatment and financial protection as the extension of coverage itself.
This brief highlights the main findings of various HEFPA studies and identifies some common messages relevant to the UHC agenda that emerge.

Extension of coverage to populations reliant on the informal labour market – which in Asia remains stubbornly large, despite impressive economic growth – is possibly the greatest challenge to the achievement of UHC. Most countries are either unwilling, or find it economically infeasible, to follow Thailand’s example of using general government revenue to finance coverage of the entire informal sector. Indonesia, the Philippines and Vietnam are attempting to combine tax-financed coverage of the poor and certain demographic groups with voluntary enrolment in social health insurance of the non-poor not covered through formal sector employment. Succeeding in this strategy requires the acquisition and utilisation of a good understanding of the factors that motivate people to insure, or not.
Two randomised experiments conducted by HEFPA in the Philippines and Vietnam reveal that subsidisation of premia by as much as 50 per cent, along with the provision of information on the operation and benefits of insurance, is insufficient to bring enrolment rates anywhere close to the realisation of universal coverage.
In Vietnam, despite affordability being reported as the main reason for not purchasing insurance, a 25 per cent premium subsidy, either exclusively or in combination with information promoting the benefits of insurance, had no impact on uptake.
In the Philippines, a larger subsidy of up to 50 per cent combined with a more detailed information package and the sending of SMS reminders encouraging enrolment succeeded in raising take-up by five percentage points. However, this lifted coverage to only 15 per cent. At-home assistance with completion of the insurance application form and arrangement for its delivery to the insurer’s office also succeeded in raising enrolment, this time by a substantial 36 percentage points. But even after being offered a 50 per cent subsidy and removal of registration costs, three-fifths of the target group declined to enrol.
In both countries, a substantial share of the informal sector population appears to be unconvinced of the gain from purchasing health insurance, even at a greatly subsidised price. Administrative changes that can reduce the non-premium costs of enrolment should certainly be more widely considered, but rather than further reducing the price and/or publicising the benefits of a given insurance package, it would seem essential to raise the value of the product. There is frequently scepticism of the quality of health services that can be accessed through public insurance. Moreover, its impact on OOP spending can be muted by operation of a reimbursement ceiling combined with the freedom of providers to set prices above this, which erodes the financial protection against medical expenditure risks that insurance is intended to provide.
Coverage is not the ultimate goal. Rather it is a means towards improving population health through access to effective medicine, as well as insuring household finances against medical expenses. How effective is coverage in meeting these objectives?

HEFPA studies conducted in a number of countries tend to confirm that increased insurance coverage does raise the utilisation of healthcare. But it is not simply the extension of coverage to a larger proportion of the population that is critical. The depth and structure of coverage, as well as its means of acquisition, also appear to be relevant.
After the relatively straightforward – and presumably politically expedient – task of covering formal sector employees, and often their dependents, attention tends to turn to subsidised coverage of the poor. Indonesia achieved this in 2005, through a fully tax-financed programme initially referred to as Askeskin, which has continued from 2008 under the name of Jamkesmas. HEFPA research reveals that on average, Askeskin raised the probability of accessing outpatient care by five percentage points, and the impact on the poor was more than double that. Utilisation of inpatient care was also increased. While the poor and near-poor were the main beneficiaries, the research reveals considerable leakage to the non-poor.
Targeting subsidies on the poor can cause administrative headaches. HEFPA research carried out in Cambodia suggests that the cost of identifying the poor does not always pay off. Irrespective of whether they are targeted towards the poor, vouchers offered for safe motherhood services at public health centres are found to have significantly raised the probability of poor women delivering in a facility, with no significant impact on non-poor women. In fact, the impact on the poor is greater when the vouchers are universal. In this low-income context, targeting may be unnecessary given that even a full subsidy is unlikely to induce women who can afford to deliver in a private clinic to switch to a public health centre.
Target inefficiency of means-tested subsidised coverage was one factor that contributed to the decision taken in Thailand to attempt tax-financed universal coverage in 2001. A HEFPA evaluation of the landmark Thai UHC reform reveals that it reduced the probability of forgoing formal ambulatory treatment when sick by 3.2 percentage points (or around one tenth) and raised the probability of inpatient admission by one point (almost one fifth).
The impacts are even larger among the poor and elderly – both populations that were covered, in principle, under the welfare scheme that operated prior to UHC. Most likely this reflects both deficiencies in targeting and the deepening of coverage made possible by doubling real health spending over the decade following the reform. There may also be an effect of universal entitlement reducing any stigma of seeking public care through what might previously have been considered the ‘poor man’s scheme’.
Further evidence of deeper coverage raising utilisation of both outpatient and inpatient care is provided by a study conducted in China that exploits variation in the generosity of cover offered by the New Cooperative Medical Scheme (NCMS) across districts in Ningxia – a poor north western province – and Shandong – a much more developed province on the east coast. Experience with the NCMS also demonstrates that the pattern of healthcare utilisation can be sensitive to the structure of insurance. Initially, NCMS primarily covered inpatient care. This, in combination with incentives arising from fee-for-service payment of providers and shallow coverage of the insured, potentially exposed patients to the risk of receiving inappropriately invasive treatment with high associated OOP expenses.
A HEFPA experiment conducted in Ningxia redesigned the NCMS benefit package to increase relative coverage of outpatient services, particularly those delivered at primary care facilities. As a result, receipt of ambulatory care at village clinics increased by around 50 per cent, but there is no evidence of substitution from higher to lower level facilities, or from inpatient to outpatient treatment, except among the poorest.
The evidence emerging from HEFPA suggests that both the level and the pattern of healthcare utilisation depend on who and what is covered. But access to healthcare is not contingent only on demand-side parameters. The project also finds that how providers are paid and managed is highly relevant to what coverage delivers.

UHC aims to ensure that effective medical care can be accessed by all without placing undue strain on household finances. Public insurance will not succeed in achieving this if the method of paying providers does not motivate provision of appropriate care, and perhaps even encourages overprovision of partially insured treatments of dubious medical efficacy that burden household budgets.
A HEFPA experiment conducted in Ningxia finds that moving from fee-for-service (FFS) to capitation plus pay-for-performance (P4P) resulted in more appropriate prescribing behaviour, reducing the probability of antibiotics being issued by about six percentage points, equivalent to a relative decrease of around 15 per cent.
In Vietnam, HEFPA finds that replacing FFS with capitation payment of hospitals by the public insurer resulted in cost savings on the provision of care to insured patients, but hospitals compensated by increasing provision to the uninsured. The latter may have had positive effects on health outcomes if the uninsured were previously being underserved, but paying for the additional treatment could push already fragile household finances to the margins of poverty. There is also evidence from Vietnam that granting hospitals greater autonomy, including wider scope in the utilisation of surpluses from user fees, has resulted in higher household OOP spending.
Attuning provider incentives with the goals of access and affordability is clearly necessary to ensure that extensions of public health insurance have the intended effects.

The information gleaned from monitoring of health equity and financial protection across and within countries is contingent on the richness and validity of the indicators employed. HEFPA has yielded methodological innovations in the measurement of inequity in the utilisation of healthcare and of catastrophic medical expenditure risk.
A new measure makes explicit inequity that derives from the unequal treatment response to variation in medical need. Previously employed instruments are likely to have underestimated pro-rich inequity. This is confirmed using data from Bangladesh, India, Malaysia and the Philippines. In those countries, around one half of the socioeconomic inequality is due to utilisation being more responsive to need among higher wealth and urban-dwelling individuals.
Previous measures identify the average risk of incurring catastrophic health payments in a population. HEFPA has developed a measure of risk exposure at the household level and the associated welfare loss. This measure can be used to identify the households that would gain most from the provision of catastrophic health insurance. Across seven Asian countries, medical expenditure risk is found to be highest in Laos and China, and lowest in Malaysia. Exposure to risk is generally higher for households that have less recourse to self-insurance, lower incomes, wealth and education, and suffer from chronic illness.

HEFPA research reveals that, perhaps contrary to expectations, increased coverage does not always reduce OOP spending on medical care. Depending on the composition of the benefit package and the structure of coinsurance, insurance can reduce the relative price of higher level, more expensive treatments, inducing substitution away from cheaper options. Combined with incentives providers may have to deliver high cost care, this can mute the impact on OOP payments and even result in a rise in uninsured expenditures.
The Askeskin programme of tax-financed care for the poor in Indonesia did not reduce households’ medical expenditures.3 OOP payments even increased among the target urban population. This surprising result may derive from the very low initial level of spending on healthcare among the poor. By making care more affordable – particularly inpatient treatment, which the insurer pays for by FFS – it is possible that spending rises as less needed care is forgone.
The tremendous expansion and deepening of NCMS coverage in rural China, arising from a tripling of the subsidy between 2008 and 2012, provides further evidence that insurance need not reduce OOP spending. In Ningxia and Shandong, more generous coverage is found to raise the OOP spending on an inpatient stay and has no significant effect on the expenditures made for an outpatient visit.6 The interaction of insurance design with provider incentives may be responsible. The experiment conducted in Ningxia finds that extending coverage to ambulatory care and raising reimbursement for treatment at primary facilities above that for treatment at secondary and tertiary facilities reduced the prevalence of catastrophic payments from one-third to less than a quarter.7
Provision of a near comprehensive benefit package with only a modest copayment, which was dropped entirely after four years, made it likely that the Thai UHC reform would reduce OOP payments provided that effective care could be delivered from a tight budget. The HEFPA evaluation finds OOP payments were reduced by one third, on average. The share of households incurring catastrophic payments was decreased by two percentage points, or about one third, and spending at the 95th percentile of the OOP distribution was reduced by half, indicating greatly reduced exposure to devastatingly high medical expenses.
Health Equity Funds (HEF), which are administered by NGOs and compensate public providers for waiving the fees of poor patients in Cambodia, demonstrate the potential to increase the financial protection offered to poor patients in extremely constrained settings. HEFPA finds that
HEFs reduce OOP payments by 35 per cent on average across households with medical expenses, with an even greater impact on the poorest.

Spreading the economic burden of illness across the population is a major motivation behind the drive for UHC. HEFPA has examined the economic consequences of illness, and studied how households cope with these in the absence of comprehensive insurance, in a number of Asian countries. The extent to which, and how, households protect their consumption when hit by a health shock varies across countries.
One important, near-common finding is that income losses arising from ill-health tend to impose a larger economic burden, which is sustained for longer, than medical expenses. In part, this may be because poor households forgo needed, but unaffordable, healthcare. But it also reflects the near absence of sickness and disability insurance coverage of the majority of the populations of low and middle income Asian countries. Ambitious as it is, UHC is only one item on the social protection agenda. Providing disability insurance in economies in which informal sector employment remains high will be even more of a challenge.

Caution is called for in any attempt to draw general conclusions from a set of studies that differ in aims and context. Nevertheless, a few points deserve emphasis.
The impacts of changes in coverage on healthcare utilisation and OOP payments depend on the comprehensiveness of the benefit package, changes in the relative prices of high and low cost treatments, co-insurance rates and the method of paying providers. If increased coverage raises utilisation of more effective, and now more affordable, treatments, then wellbeing may be raised even if households are paying as much OOP as they were before. But if patients are being encouraged to make frivolous, inappropriate use of high cost treatment options that could be substituted by more cost-effective alternatives, then a restructuring of insurance and/or provider payment methods could reduce OOP payments with little or no loss of health.
Aligning demand and supply side incentives to ensure that social insurance offers access to effective medical care and financial protection from medical expenditure risk may be what it takes to encourage the informal sector to enrol.
The evidence from the project is offered in the hope that it may motivate and inform policy actions that can better secure health equity and financial protection in Asia.

Health Equity and Financial Protection in Asia – HEFPA

Final Report
This final report is a summary based on the results obtained in the 23 Working Papers and 16 Publications generated by the HEFPA project over the period Jun 2009 to Dec 2013 and summarized in the 7 Policy Briefs relating to the 8 Work Packages (all available from


The information gleaned from monitoring of health equity and financial protection across and within countries is contingent on the richness and validity of the indicators employed. HEFPA has yielded methodological innovations in the measurement of inequity in the utilisation of healthcare and of catastrophic medical expenditure risk.
A new measure makes explicit inequity that derives from the unequal treatment response to variation in medical need [1]. Previously employed instruments are likely to have underestimated pro-rich inequity. This is confirmed using data from Bangladesh, India, Malaysia and the Philippines. In those countries, around one half of the socioeconomic inequality is due to utilisation being more responsive to need among higher wealth and urban-dwelling individuals.
Previous measures identify the average risk of incurring catastrophic health payments in a population. HEFPA has developed a measure of risk exposure at the household level and the associated welfare loss [2]. This measure can be used to identify the households that would gain most from the provision of catastrophic health insurance. Across seven Asian countries, medical expenditure risk is found to be highest in Laos and China, and lowest in Malaysia. Exposure to risk is generally higher for households that have less recourse to self-insurance, lower incomes, wealth and education, and suffer from chronic illness.


[1] Van de Poel E, Van Doorslaer E, O’Donnell O (2012), ‘Measurement of Inequity in Health
Care with Heterogeneous Response of Use to Need’, Journal of Health Economics 31: 676-689.

[2] Flores G, O’Donnell O, ‘Catastrophic Medical Expenditure Risk’, HEFPA Working Paper 11, Erasmus University Rotterdam.


Cambodia has adopted a number of innovative healthcare financing arrangements in recent years that use incentives to encourage the provision and utilisation of essential health services. This policy brief summarises knowledge emerging from the HEFPA project on the impact of three financing schemes and highlights the significance of the evidence for health policy in Cambodia and elsewhere.

Scheme design: Cambodia has pioneered the use of Health Equity Funds (HEFs) to compensate public health facilities for user fee exemptions granted to eligible poor patients and subsidise hospitalisation-related transportation and food costs. They are mainly funded by donors but the government contribution is increasing. Management is entrusted to a third party, usually a national NGO. By 2012, HEFs were up and running in 48 (out of 79) health districts and in a further 12 districts the Ministry of Health operated a subsidy scheme that was largely similar, only without third party management and featuring a more restricted benefit package. Nearly half of the HEF schemes cover hospital services exclusively, while the others also cover care delivered at health centres.
HEFPA findings: A HEFPA study has revealed that HEFs reduce out-of-pocket (OOP) health spending by 35 per cent on average among households that make any payment [1]. The impact is larger for poorer households and for those mainly using public healthcare. While the effect is smaller and less significant for non-poor households, its extension to this group is consistent with the impact resulting not only from the direct subsidy to the poor, but also from a substantial contribution to the revenue of public facilities that improves the care on offer and attracts non-HEF beneficiaries.
The government subsidy scheme also reduces OOP payments but the effect is less pronounced for the poor. By reducing payments for healthcare, both HEFs and the government scheme enable households to increase consumption of non-medical goods and services, but neither has any significant impact on health-related debt. HEFs reduce the probability of seeking care primarily in the private sector, but there is no significant effect on the use of public care, although this conclusion may reflect limitations of the data.
A prospective study that collected detailed utilisation data in two rural health districts shows that the introduction of HEFs to compensate health centres for user fee exemptions granted to poor patients had no significant impact on outpatient consultations at these facilities, and neither did it affect the propensity to opt for private providers or self-medication as a first choice of treatment [2]. Moreover, there was no significant effect on related OOP payments. The lack of effect may be attributable to the fact that consultation fees at health centres are very low. Perceived low quality of care offered and distance may discourage utilisation more than user fees.

Indicators of maternal and child health have improved dramatically in the last decade. While the causes are likely to be multiple, HEFPA findings indicate that a twin-pronged attack using both demand and supply side incentives may have contributed.

Scheme design: Reproductive health vouchers are 100 per cent subsidies of maternal health services issued by a management agency – usually an NGO – to eligible pregnant women along with related information. Benefit packages vary, but all include safe motherhood services at health centres (antenatal care, delivery and postnatal care). Some also include family planning and safe abortion services. Some target poor women, but most do not. The provider uses the voucher to prove delivery of care and is compensated accordingly. Voucher schemes are fully funded by donors and had been implemented in 27 health districts by 2012.
HEFPA findings: Analysis of the 2010 Cambodia Demographic and Health Survey data showed that vouchers increase the probability of delivery in a public health facility, mainly health centres, by 10 percentage points for all women and 16 percentage points for the poorest women [3]. The increase was mainly due to a shift from home delivery by traditional birth attendants. The effect of universal voucher schemes was found to be larger than that of those targeting poor women. Targeting seems unnecessary since poor women are more responsive to incentives to deliver in public facilities in any case. There is a positive effect of vouchers on postnatal care of non-poor women and universal vouchers have been shown to increase use of antenatal care, but only for poor women.

Scheme design: The Government Midwifery Incentive Scheme (GMIS) aims to boost institutional deliveries – and ultimately reduce maternal mortality – by offering cash incentives to midwives and other trained health personnel for deliveries attended in public health facilities; US $15 for a live birth in health centres and $10 in referral hospitals. Unlike HEFs and vouchers, GMIS was designed, funded and implemented nationwide in late 2007 by the Cambodian Government.
HEFPA findings: Deliveries in public facilities expressed as a percentage of expected births tripled from 19 per cent in 2006 just before GMIS started to 57 per cent in 2011. The increase was yet more pronounced for health centres. While the simultaneous nationwide launch of the scheme precludes the opportunity for a treatment-control comparison, detailed analysis of the trend suggests that GMIS may have boosted facility deliveries by 18 per cent during the first month of operation and by 15 per cent after 12 months [4]. Qualitative analysis also suggests that the introduction of GMIS together with other interventions to remove supply and demand barriers to essential maternal health services has led to considerable improvements in public health facilities and a steep increase in facility deliveries. However, several operational issues still need to be dealt with, including late and incomplete financial incentive disbursements and the absence of an effective monitoring system.

Up to now, there has been belief but little or no proof that HEFs, vouchers and midwife incentives improve access to public healthcare in Cambodia. The HEFPA study confirms that vouchers do indeed increase utilisation of safe motherhood services. Giving women free access to maternal healthcare and ensuring that the facilities have the financial incentive to honour this entitlement serves to substantially raise institutional deliveries. Simultaneous nationwide roll-out of the midwifery incentives scheme makes it more difficult to establish its effect with the same degree of confidence; nevertheless, the circumstantial statistical evidence is consistent with a strong impact on deliveries in facilities and by trained health personnel. Furthermore, the demand and supply of reproductive healthcare does seem responsive to cash incentives.
The substantial reduction in OOP payments brought about by HEFs demonstrates that relieving providers from the responsibility to adjudicate entitlement to fee waivers and compensating them for lost user fee revenue can make nominal exemptions from user fees effective. But the lack of evidence of an effect on healthcare use suggests that fees are not necessarily the main barrier to utilisation of public facilities, in particular of health centres where fees are modest. In Cambodia, as in several other low-income countries, the private sector is preferred by the majority of the population, including the poor, for ambulatory curative care. This presumably reflects the inconvenience, unreliability or perceived low quality of the outpatient care offered by the public sector. Rather than designing elaborate schemes to exempt the poor from modest user fees, an alternative and arguably more appropriate policy response is either to correct the deficiencies in the public services, or engage the private sector in the provision of (partly) publicly financed care.

Contributing to the ongoing global debate on ‘user fees or not’, the HEF and voucher experiences in Cambodia suggest more feasible and effective alternatives to across-the-board user fee removal. Targeting the limited public or international resources to priority groups (the poor, women), the barriers they face (user charges but also transport, information, and accessibility) and priority services (hospital care, institutionalised delivery) may be the right strategy when budget constraints are tight. The feasibility of identifying the poorest is a constraint, but the evidence from the comparison of universal and targeted voucher schemes indicates that this is not necessary for all types of care.
Cambodia continues to be a pioneer in adopting innovative healthcare financing strategies tailored to low-resource settings. Its experience suggests that health financing has a key role in improving the performance of public health systems in this context. Consistent with the evidence emerging on the effectiveness of results-based financing, the HEFPA findings demonstrate that both patients and healthcare professionals respond to incentives. In line with evidence from elsewhere, the voucher and GMIS schemes suggest that financial incentives work particularly well for maternal health – for which procedures, and so prices, can be precisely defined.
In a resource-poor setting like Cambodia, where the health system is relatively weak, a package of interventions is probably the most pragmatic approach to follow in order to supply essential health services and remove barriers to access, including financial ones.


[1] Flores G, Ir P, Men CR, O’Donnell O, Van Doorslaer E (2013), ‘Financial Protection of Patients through Compensation of Providers: The Impact of Health Equity Funds in Cambodia’, Journal of Health Economics, 32(6): 1180-93.

[2] Korachais C, Ir P, Macouillard E, Men CR, Meessen B, ‘Health Equity Funds at health center level in Cambodia – a benefit package too far? Expansion of a waiver scheme for the poorest for curative services at health center level’, HEFPA Working Paper 23, Erasmus University Rotterdam.

[3] Van de Poel E, Flores G, Ir P, O’Donnell O, Van Doorslaer E, ‘Can Vouchers Deliver: An Evaluation of Subsidies for Maternal Health Care in Cambodia’, Bulletin of the World Health Organization, forthcoming.

[4] Ir P, Korachais C, Chheng K, Horemans D, Van Damme W, Meessen B, ‘Boosting ‘facility deliveries’ with results-based financing: an evaluation of the government midwifery incentive scheme in Cambodia’, HEFPA Working Paper 22, Erasmus University Rotterdam.


China’s ambitious 2009 healthcare reform plan set the goal of achieving universal health coverage by 2020. A large part of the challenge involves extending and deepening coverage of the rural population through the New Cooperative Medical Scheme (NCMS). HEFPA research reveals how the NCMS benefit package and its means of paying providers can be designed to ensure delivery of more appropriate, cost-effective care and to better protect household finances from medical expenditure risks.

Faced with a situation in which healthcare had become unaffordable to much of the rural population, or could only be accessed through high out-of-pocket payments that seriously threatened living standards, the Chinese Government began in 2003 to roll out the mainly tax-financed NCMS insurance. By 2011, NCMS coverage had reached 98 per cent and the government subsidy per person tripled between 2008 and 2012. In parallel, out-of-pocket (OOP) expenditure as a share of total health expenditure dropped by over 20 per cent between 2003 and 2011.
Despite the rapid expansion of NCMS coverage and reimbursement rates, and the reduction in the OOP health expenditure share, household medical expenditures have continued to escalate in absolute terms and catastrophically high expenditures have not fallen. Shallow benefit packages and misaligned demand and supply side incentives are likely root causes. Early NCMS plans covered only inpatient services with high deductibles, high co-insurance and low ceilings on reimbursed expenses. While this benefit package design improved access to previously least affordable care, it also created incentives to overuse hospital treatment. Since coverage is shallow and there is a lack of appropriate incentives for hospitals to deliver the most cost-effective treatment options, patients continue to be liable for high OOP expenditures when hospitalised.
At the same time, the quality of care delivered in rural areas has shown little sign of improvement on average and has remained highly variable. Perverse provider incentives are suspected to be largely responsible.
The limited progress made towards the goal of access to appropriate treatment and financial protection against medical expenditure risk since the launch of the NCMS has motivated HEFPA studies in two provinces. Their aim is to identify changes to the design of the benefit package and provider payment mechanisms that can provide better incentives for patients and practitioners to use and deliver care that is both more cost-effective and imposes a lower burden on household finances. Given China’s increasing prevalence of hypertension, diabetes, and other non-communicable diseases, the most cost-effective course of care is often ongoing disease management and treatment on an outpatient basis.

In Ningxia – a poor province in Northwest China – HEFPA conducted a quasi-experiment that involved re-designing the NCMS benefit package in some counties to increase coverage for outpatient services. The new benefit package used a tiered reimbursement structure that indemnified visits to primary care facilities (village posts and township health centre) more generously than those to upper level secondary and tertiary hospitals [1]. Compared to an inpatient-orientated benefit package in comparison counties, the intervention increased the probability of receiving outpatient care at a village clinic in the previous two weeks by 0.7 of a percentage point, equivalent to a 44 per cent increase, on average. The effect is larger for poorer and middle-income individuals, for those living closer to village clinics and migrant workers.
There was no evidence of a significant substitution from treatment at higher- to lower-level facilities. There was also no substitution from inpatient to outpatient treatment. This was partly because coverage for outpatient services increased but that for inpatient treatment was not reduced – the latter being a politically unattractive option for local officials.
The expansion in overall coverage was made possible by continuously increasing government subsidies for NCMS. Evidence from another HEFPA study conducted in Ningxia, and also Shandong – a much more developed province on the east coast – shows that a rise in the depth of coverage is not sufficient to reduce the medical spending of households. More generous coverage was found to increase the OOP spending on an inpatient stay but has no significant effect on the expenditures made for an outpatient visit [2].
The Ningxia experiment demonstrates that changing the structure, as opposed to the depth of coverage, can be successful in reducing OOP payments. Shifting the benefit package toward relatively more generous reimbursement of primary care at lower level facilities reduced the incidence of catastrophically high medical expenditures. The percentage of households spending more than 10 per cent of their budget on medical care was reduced from 33 to 24 per cent.

Expansion of insurance coverage without appropriate supply-side incentives can fuel unnecessary spending and utilisation, with no improvement in quality. Over-prescription of drugs and excessive use of diagnostic tests is a major quality and cost concern in China. The causes are many, but fee-for-service (FFS) payment of providers combined with a fee schedule that ensures a profit margin on diagnostic tests and a 15 per cent mark-up on drugs is most likely critical. For primary healthcare providers, who have limited scope to perform diagnostic tests, maximising drug profits has become a central motivation. Prescription of antibiotics far exceeds rates recommended by the World Health Organization (WHO) and is often utilised for self-limiting illnesses, such as common colds.
Paying providers according to their performance in delivering appropriate care promises to raise quality and contain health spending. To test whether this potential can be realised, the HEFPA experiment in Ningxia included a second component that involved replacement of FFS with capitation and pay-for-performance (p4p) payment of providers in a randomly selected subset of the counties in which the NCMS benefit package was re-designed to promote the utilisation of primary care. The capitation rate was based on past spending with adjustments for inflation and the predicted impact of the change in the benefit package. The rate of antibiotic use was the core performance indicator of the p4p design. Performance was assessed from the directory of all visit records, collected through an electronic management information system.
The change from FFS to capitation with p4p led to a reduction in the probability of antibiotics being prescribed at township health centres and village clinics [3]. However, the post-intervention antibiotic prescription rate remained higher than the internationally recommended level.

In the breadth versus depth of coverage tradeoff, China chose to cover its entire population first with shallow coverage, and then to dig deeper as more resources become available. Ensuring that the additional resources reap returns of improved population health and reduced exposure to medical expenditure risks requires the provision of incentives that motivate patients to demand, and providers to deliver, more cost-effective, higher quality care. The HEFPA experiment in Ningxia shows that if properly implemented, appropriate design of the insurance benefit package and provider payment method can ameliorate problems of overuse of higher level, expensive facilities and over-prescription of drugs and diagnostic tests, both of which place a burden on household budgets while bringing little improvement in health.
In the latest Five Year Plan (2012-2016), the Chinese Government identifies provider payment reform as a top priority for advancing its health reform goal, with a particular focus on public hospitals – which account for over 70 per cent of current national health spending. Coupled with the continued increase in government subsidies for NCMS targeted at primary care and priority diseases through expansion of the benefit package, this is a step in the right direction.
However, improved governance and accountability of public health facilities is also required. Recent policy encourages entry of private hospitals and clinics in the belief that competition will force public facilities to improve quality and efficiency. With international experience demonstrating that market competition in healthcare is difficult to create and put to good effect without a carefully designed and enforceable regulatory system, this policy experiment needs to proceed with caution.


[1] Powell-Jackson T, Yip W, Han W, ‘Re-aligning demand- and supply- side incentives to improve primary health care seeking in rural China’, HEFPA Paper P5, Erasmus University Rotterdam).

[2] Hou Z, Van de Poel E, Van Doorslaer E, Yu B, Meng Q, ‘Effects of NCMS on access to care and financial protection in China’, Health Economics (first published online: 23 AUG 2013, DOI: 10.1002/hec.2965).

[3] Yip WCM, Powell Jackson T, Chen W, Hu M, Fe E, Hu M, Jian W, Lu M, Han W, Hsiao WC, ‘Capitation with pay-for-performance improves primary care providers’ antibiotic prescribing practices in rural China’, HEFPA Paper P3, Erasmus University Rotterdam.


A first step towards meeting Indonesia’s ambition for Universal Health Coverage (UHC) by 2019 was made in 2005, with the introduction of subsidised social health insurance for the poor. With the planned transition to UHC in mind, HEFPA has examined lessons that can be drawn from the experience of extending subsidised coverage to the poor.

The subsidised social health insurance for the poor in Indonesia was first introduced in 2005, initially referred to as Askeskin and subsequently, from 2008 onwards, as Jamkesmas.
At the time, health coverage for civil servants and formal private sector employees had been in place for several decades. However, with roughly 60 per cent of the labour force employed in the informal sector, a large part of the population was thereby left without cover. Askeskin-Jamkesmas was a first attempt to bridge this gap by extending coverage to poor and near-poor households. Unlike formal sector social health insurance schemes, premiums for the poor and near-poor are fully subsidised from a tax-financed health fund. Enrolled households are entitled to comprehensive coverage for public healthcare, including inpatient and outpatient services.
The subsidised programme currently has a target population of about 76 million people (about a third of the population) and a further 25 million are covered by the formal sector schemes. That said, this still leaves more than half the population without any form of health insurance.
Since 2006, in response to this persistent coverage gap, there has been a proliferation of local healthcare financing schemes. Collectively known as Jamkesda, these schemes are implemented and operated by district governments, and are often motivated by the local political context. Since Indonesia’s decentralisation in 2001, districts have been afforded a large degree of autonomy in setting local public health policy. Together with regional variation in financial and human resources, this process has led to substantial differences in the design of local schemes between districts, spanning factors such as coverage, benefit packages and provider contracting.


The experience of the Askeskin programme has shown that subsidised social health insurance can be effective in increasing healthcare utilisation, especially in the poorest households [1]. The bulk of the impact is on the use of public health centres in rural areas, and more expensive public hospitals and contracted private providers in urban areas. A factor that may have inhibited a broader impact among the poorest population has been the programme’s targeting performance. While the poor and near-poor were the main beneficiaries of Askeskin coverage, there was also considerable leakage to the non-poor.
Although the local government Jamkesda schemes have had little overall impact, they do seem to have contributed to closing the coverage gap, by increasing healthcare utilisation of the near-poor falling just outside the Askeskin-Jamkesmas target population [2]. A 25 per cent increase in outpatient visits among the second poorest fifth of households can be attributed to Jamkesda schemes. However, there is no evidence of any effect for other income groups, and nor is there an impact on hospitalisations. Moreover, there is large variation in efficacy across districts, reflecting differences in design. The Jamkesda schemes that tend to increase healthcare utilisation typically rely on external management; they contract both public and private providers and prioritise basic services.

While Askeskin’s subsidised coverage for the poor has improved access, it has had a more ambiguous impact on financial protection [1]. Out-of-pocket healthcare spending was actually increased by the programme, in particular for urban households. This indicates that households may be bearing part of the cost of their increased utilisation of subsidised, but not fully insured, services. Nevertheless, the effect on utilisation is much larger than that on spending, suggesting a strong response of demand to the greater affordability of healthcare. Out-of-pocket spending on healthcare by Indonesian households has remained relatively low, accounting for approximately 2 per cent of household budgets, and the HEFPA study finds no conclusive evidence that Askeskin has led to an increased incidence of potentially financially deleterious healthcare spending. As with Askeskin, HEFPA found no evidence for a reduction in out-of pocket spending among beneficiaries of Jamkesda schemes [2].
The financial protection from the risk of ill health offered by subsidised health insurance coverage of medical expenses is inevitably limited because it does not cover the other main source of economic risk from illness – income loss. Informal sector workers lack not only health insurance, but also go without sickness and disability insurance. The HEFPA study finds that an illness-induced income loss forces poor households to reduced consumption (especially for non-food items) [3]. Although the poor tend to resort to borrowing and the sale of assets and savings as strategies for coping with the economic risk from ill health, these financial buffers are not sufficient to fully protect consumption levels in the short term. In addition, high interest rate borrowing, dissaving and productive asset depletion may negatively affect household consumption in the mid- and long terms.

The next phase of the reforms leading to UHC is to commence in 2014. Existing schemes are to be consolidated and scaled up in one nationwide social health insurance with mandatory contributions for formal sector workers and subsidised premiums targeted to the poor and near-poor.
Based on the experience with the Askeskin and Jamkesda schemes, one may expect that extending social health insurance coverage will improve both access and affordability of healthcare, but that gaps in financial protection may persist. The risk of illness-related income loss falls beyond the reach of social health insurance. Combined with the long-term economic consequences of exercising coping strategies such as borrowing and selling assets, uninsured income risk remains a source of potential impoverishment.
Most income risk is borne by the informal sector that harbours the bulk of the labour force from the poorest half of the population. An additional problem is that the roadmap for achieving UHC by 2019 is currently unclear on how the large share of Indonesians that neither qualify for subsidised premiums nor receive coverage through formal sector employment will be covered. This group comprises a third of the population, largely overlapping with the currently uninsured.
The provision of full financial protection for ill health would therefore require a further expansion of insurance to the informal sector, as well as considering broader social security policies that go beyond the UHC agenda, and comprehensively addressing the multiple sources of economic risk from ill health.


[1] Sparrow R, Suryahadi A, Widyanti W (2013), ‘Social Health Insurance for the Poor: Targeting and Impact of Indonesia’s Askeskin Program’, Social Science and Medicine 96: 264-271.

[2] Budiyati S, Yumna A, Warda N, Sparrow R, Suryahadi A, Bedi A, ‘Sub-national Health Care Financing Reforms in Indonesia’, HEFPA Working Paper 15, Erasmus University Rotterdam.

[3] Sparrow R, Van de Poel E, Hadiwidjaja G, Suryahadi A, Warda N, Yumna A, ‘Coping with the Economic Consequences of Ill Health in Indonesia’, Health Economics (first published online: 8 July 2013, DOI: 10.1002/hec.2945).


Achieving universal healthcare for all Filipinos has been the stated objective of the government since the late 1990s. This policy brief summarises the evidence arising from HEFPA studies addressing the following three questions:
1) How can the informal sector be encouraged to enrol in the social health insurance (SHI) programme?
2) What incentives motivate local governments to extend SHI coverage to the poor?
3) To what extent does SHI protect household wellbeing from health shocks?

Various attempts since the turn of the century to increase SHI coverage in the Philippines have been only moderately successful. In 2012, the SHI agency PhilHealth reported coverage of 70 million individuals, equivalent to 75 per cent of the population. Only 33 per cent of eligible persons in the informal sector, however, were covered under the Individually-Paying Program (IPP). No other group accounts for more of the uninsured.
HEFPA conducted a randomised experiment with the aim of establishing the effectiveness of a premium subsidy and the provision of information on the operation and benefits of insurance in encouraging enrolment [1]. In the main experiment, households within randomly selected municipalities were offered a subsidy worth up to 50 per cent of the annual premium, and given a kit containing an application form and informative leaflets. These households were also sent regular SMS reminders to enrol. In a follow-up experiment, households that had not enrolled during the first experiment were given more time to make use of the subsidy and about half of them were also offered free assistance in filling out the application forms and submitting them to PhilHealth

The combination of the subsidy, information and reminders yielded a 4.9 percentage point increase in enrolment, which is a 50 per cent increase relative to the control group. However, even this substantial proportionate increase only raised the enrolment rate to 15 per cent. The additional offer of free assistance in filling out the application form and submitting it to PhilHealth yielded an increase of 36.3 percentage points in enrolment of households who had not responded to the premium subsidy and information provision alone. While this represents a dramatic effect, around 60 per cent of families still chose not to enrol despite being offered a 50 per cent premium subsidy and the effective removal of all indirect costs of application.
The low take-up suggests low willingness-to-pay (WTP), possibly due to lack of knowledge of health insurance, or bad experience of claiming insurance benefits.
However, the provision of information did not prove particularly effective, and WTP is reported to be higher among those with previous PhilHealth coverage, pointing towards a positive learning effect that dominates any negative experience of claiming benefits [2]. WTP is found to be lower for those who received the premium subsidy, which indicates that there is anchoring on the net of subsidy price that reduces future effective demand for insurance.

The large effect of at-home submission of the insurance application relative to that of the premium subsidy suggests that greater attention should be given to lowering time, as opposed to monetary, costs of insurance purchase in order to raise enrolment rates. But if three-fifths of the target population remains unconvinced of the gains from purchasing insurance even after being offered a generous subsidy and a very simplified enrolment process, then getting coverage closer to universality is likely to require genuine improvements in the value of the insurance as opposed to further reductions in its cost. In the Philippines, SHI can have little impact on out-of-pocket (OOP) spending due to the latitude granted to providers to set prices above a reimbursement ceiling. Enrolment is not an end in itself. Rather, it is financial protection from medical expenditure risks that needs to be secured. If PhilHealth can ensure this more effectively, then the demand for insurance is likely to follow.

Local governments in the Philippines are tasked with financing and delivering basic health services, and, until 2010, they were also mandated to enrol indigent families into SHI, with a subsidy provided by the national government. Two HEFPA studies investigate the underlying political (chances of re-election) and fiscal incentives that local governments have to extend SHI coverage and provide health services [3,4].

Local governments enrol a higher proportion of their constituents in SHI and spend more on social services, including health, when they receive higher lump-sum transfers and premium subsidies from the national government. The effect of lump-sum transfers on SHI enrolment is greater than that of the targeted transfers, suggesting an already significant preference for this health service among local governments. However, they also cover fewer families when the national government or other sponsors enrol from the same target groups without local government participation.
Local governments raise more revenue for health spending and enrol more families when spending and enrolment are higher in neighbouring constituencies. This is consistent with a health policy response to political pressure arising from constituents observing the services offered in neighbouring authorities.

Fiscal incentives matter. Conditional grants (like premium subsidies) can motivate local governments to extend SHI coverage. But lump-sum resource transfers may be a pre-requisite to relax fiscal constraints and make spending on SHI feasible. National government sole sponsorship of the poor in the SHI programme could crowd out local government efforts.
Political incentives also matter. SHI coverage is offered by elected officials to increase the chances of re-election. Increasing the electorate’s awareness of health spending and insurance coverage in the neighbouring localities could be used to better align political incentives with health objectives.

The priority that should be accorded to SHI depends on: the incidence of health shocks relative to other threats to wellbeing; the extent to which households can protect themselves against the economic consequences of health shocks in the absence of SHI; and the effectiveness of SHI in securing financial protection. HEFPA provides evidence on each of these three parameters relevant to the case for further extending and deepening SHI coverage in the Philippines [5-7].

Compared with other causes of economic insecurity, death and health shocks are associated with larger unplanned outlays amounting to up to 80 per cent of total food expenditures, and they result in greater depletion of assets. Recovery from a health shock is slower than is recovery from a natural disaster.
While richer households are more likely than poorer households to report that they have fully recovered from a health shock, SHI coverage is not correlated with full recovery. But securing consumption after a health shock is one of the more immediate benefits of SHI coverage, although this is evident only among the poor. Even then, the SHI-covered poor still resort to dissaving or borrowing. In general, those with SHI coverage employ fewer informal mechanisms to cope with the economic consequences of a health shock.

The findings underscore the importance of extending SHI coverage, especially to households exposed to multiple shocks who will eventually exhaust informal coping mechanisms. There is also a need to improve the financial protection that SHI confers on those already covered so as to minimise their need for costly adjustments, especially those that compromise their children’s education.


[1] Capuno JJ, Kraft AD, Quimbo SA, Tan CAR Jr, Wagstaff A, ‘Effects of Interventions to Raise Voluntary Enrollment in a Social Health Insurance Scheme: A Cluster Randomized Trial’, unpublished MS: The World Bank (see also HEFPA Paper P2, Erasmus University Rotterdam).

[2] Quimbo SA, Kraft AD, Capuno JJ, Tan CAR Jr, Wagstaff A, ‘Will Insurance and Information Improve Willingness-to-Pay for Health Insurance? Evidence from the Philippines,’ unpublished MS: The World Bank (see also HEFPA Working Paper 20, Erasmus University Rotterdam).

[3] Capuno JJ, Quimbo SA, Kraft AD, Tan CAR Jr, Fabella VM, ‘The effects of term limits and yardstick competition on local government provision of health insurance and other public services: The Philippine case’, HEFPA Working Paper 7, Erasmus University Rotterdam.

[4] Capuno JJ, Quimbo SA, Kraft AD, Tan CAR Jr, Fabella VM, ‘Perks and public provisions: Effects of yardstick competition on local government fiscal behavior in the Philippines’, HEFPA Working Paper 13, Erasmus University Rotterdam.

[5] Capuno JJ, Kraft AD, Quimbo SA, Tan CAR Jr, ‘Shocks to Philippine households: Incidence, idiosyncrasy and impact’, HEFPA Working Paper 16, Erasmus University Rotterdam.

[6] Kraft AD, Capuno JJ, Quimbo SA, Tan CAR Jr, Fabella VM, ‘Securing Consumption when Ill of Injured: Does Social Health Insurance in the Philippines Help?’, HEFPA Working Paper 10, Erasmus University Rotterdam.

[7] Kraft AD, Capuno JJ, Quimbo SA, Tan CAR Jr, Wagstaff A, ‘Coping and Consumption Adjustment for Similar Shocks over Time’, unpublished MS: The World Bank (see also HEFPA Working Paper 21, Erasmus University Rotterdam).


Prior to 2002, health insurance systems in Thailand were characterised by fragmentation, duplication and incomplete coverage. The Universal Coverage Scheme was introduced in 2002 with the aim of providing access to comprehensive healthcare for all Thais not covered through formal employment based health insurance. This policy brief provides the background, and assesses the impact.

Ever more countries are proclaiming the goal of universal health coverage – comprehensive, effective and affordable healthcare for all. The World Health Organization (WHO) also champions the cause and maintains that its pursuit is feasible everywhere. Yet, sceptics may doubt whether a middle-income, let alone a low-income, country with a narrow tax base can realistically hope to insure its entire population against the cost of most treatments without substantial cost sharing.
Thailand provides perhaps the best example of whether it is indeed feasible to deliver, and not merely promise, universal coverage on a very limited budget. Since 2002, all Thai citizens not insured through formal sector employment have been entitled to comprehensive curative and preventive care through a scheme that initially charged a copayment of less than US $1 per treatment, and later dropped even this, and had an annual budget of only $30 per enrollee.
The insurance expansion was accompanied by supply side measures intended to control costs and deliver cost-effective care: a single public payer with a fixed global budget; capitation for outpatient services and prospective provider payment for inpatient admissions; gatekeeper access to tertiary care; and movement towards a purchaser-provider split. Yet demonstrably, universal and comprehensive coverage cannot be conjured out of thin air. Total health expenditure per capita increased by one sixth in the year of the coverage extension, and then doubled in the following decade but remained below 4 per cent of GDP.
The HEFPA project has examined whether the Thai reform was able to make good on the promise that universal health coverage would raise access to needed medical care and better protect household finances from medical expenses.


Within one year from April 2001, the 70 per cent of Thai citizens not insured through formal sector employment were given entitlement to comprehensive medical care, including medicines prescribed from an essential drug list and high-cost treatments (eg. open-heart surgery, chemotherapy), at local, mainly public, provider networks. The Universal Coverage Scheme (UCS) raised the percentage of the population covered by some form of health insurance from 71 per cent immediately prior to its introduction, to 95 per cent in 2003. The tax-financed scheme initially charged only 30 Baht ($0.75) per service contact, with exemptions made for the poor, children and the elderly. Even this modest copayment was abolished in 2006 (before being partially reinstated in 2012).

The UCS reform accompanied the demand side extension with various supply side measures to maintain cost control and improve spending efficiency. First, citizens register with a Contracting Unit for Primary Care (CUP) consisting of a district hospital and several health centres through which referral to other CUPs and higher level providers is controlled. Second, the UCS operates under a capped global budget fixed by the number of enrollees multiplied by an annually negotiated capitation rate, which has increased by 71 per cent in real terms over 10 years. Third, within the global budget, ambulatory and preventive care budgets are prospectively allocated to CUPs by a capitation formula and inpatient care is paid according to Diagnosis Related Group tariffs. Finally, the 2001 reform initiated movement towards a purchaser-provider split, with the National Health Security Office, which was fully functional by 2006, exercising the purchasing role.

The HEFPA evaluation indicates that the universal coverage reform reduced the probability of a sick person going without formal ambulatory treatment by 3.2 percentage points (or around one tenth relative to baseline) [1]. The reduction in forgone medical care was even larger among the poor, rural and, in particular, elderly populations. Since the poor and the elderly were, in principle, covered before the reform through a targeted scheme, universal entitlement and the budget increase accompanying the reform appear to have made nominal coverage of these groups effective. The UCS also raised the probability of inpatient admission by 1 percentage point (almost one fifth) in its overall target population, with larger increases among the elderly and urban populations. The scheme however did not have a significant impact on inpatient admissions in rural areas.
Increased utilisation of public outpatient and inpatient care was achieved without any crowding out of private sector care. Within the public sector, the location of care was shifted from provincial to district hospitals consistent with the reform assigning a gatekeeper role to a provider network consisting of the district hospital and surrounding health centres. The likelihood that the highest level of ambulatory care was received at a health centre fell in favour of treatment at a district hospital, which may be due to the increased affordability of the latter.
Universal coverage also greatly improved financial protection against medical expenditure risks. It reduced average household out-of-pocket (OOP) medical spending by one third [1]. Household spending on ambulatory care and on medicines was cut by one quarter and almost two fifths respectively. There was no overall impact on average spending on inpatient treatment. This results from the offsetting effects of a rise in the propensity to make any payment for inpatient care, reflecting its greater affordability, and a one third decrease in the amount paid among those incurring expenses.
The reform was particularly successful in reducing exposure to the risk of incurring extremely high medical expenses that could have devastating, impoverishing consequences. The share of Thai households incurring so-called ‘catastrophic payments’ – namely OOP spending exceeding 10 per cent of the household budget – was decreased from 5.8 per cent immediately before the reform to 3.8 per cent. While spending at the 10th percentile of the distribution of (positive) medical expenditures was reduced by a significant but relatively modest 14 per cent, spending at the middle of the distribution fell by over a quarter and at the very top of the distribution (i.e. the 95th percentile) it was reduced by half. The proportion of households impoverished by medical expenses is also likely to have fallen substantially [2].
These findings confirm that the promise of universal coverage was brought to fruition in Thailand. It gave citizens greater access to treatment when they were taken ill, and better protected them from the burden that medical care can impose on household finances.

While the Thai universal coverage reform has been successful, it has nevertheless left some healthcare problems unaddressed, and has done little to avoid others arising. The extension of coverage did not eradicate geographic differences in access to healthcare. While it greatly improved access to ambulatory care in rural settings, the impact on inpatient treatment in these locations was more muted. This may be partly attributable to the failure of the reform to realise its ambition of redistributing medical manpower to hitherto understaffed areas. An attempt to tie finance for salaries to population size met with fierce resistance from the more generously staffed urban and central areas and was consequently repealed.
Inequity continues to exist not only across locality but also between healthcare schemes. Despite the gradually increasing UCS budget, the spending gap compared to the public employee scheme is large and widening. Per capita spending for the latter was about five times that of the UCS in 2010. While there is no clear evidence to prove it, the large financing gap may very well generate clinically significant differences in healthcare quality. The distribution of healthcare is generally pro-poor in Thailand, even after taking the greater medical need of the poor into account, but better-off do make greater use outpatient care at tertiary hospitals [3]. Potentially, this could serve to undermine support for the UCS, should it come to be seen as a poor man’s service from which the slightly better off prefer to opt out.
Affordability is perhaps the major challenge to universal coverage. The accelerating budget of the public employee scheme is a major contributor to the mounting pressure on government health spending and a priority on the current Thai healthcare reform agenda. At the same time, the UCS spending increase has raised concerns about its financial sustainability and brought ideas of cost sharing back to the policy debate.
The Thai experience proves that effective universal coverage can be delivered on a limited budget. However, achieving this goal is contingent upon a variety of factors: previous investment in facilities and manpower; careful design of the health system architecture to provide incentives for cost-effectiveness; and a sufficiently healthy economy to provide the additional resources that a major expansion of coverage will inevitably require.


[1] Limwattananon S, Neelsen S, O’Donnell O, Prakongsai P, Tangcharoensathien V, van Doorslaar E, Vongmongkol V, Kraft AD, Capuno JJ, Quimbo SA, Tan CAR Jr, Wagstaff A, ‘Universal Coverage on a Budget: Impacts on Health Care Utilization and Out-Of-Pocket Expenditures in Thailand’, HEFPA Working Paper 14, Erasmus University Rotterdam.

[2] Limwattananon S, Tangcharoensathien V, Prakongsai P (2011), ‘Reducing impoverishment from health payments: Outcome of universal health care coverage in Thailand’, Journal of Health Systems Research 5 (1): 25-31 (in Thai).

[3] International Health Policy Programme, ‘Understanding inequity in health care utilization: decomposition of income-related inequality and determination of horizontal inequity’, HEFPA Working Paper 6, Erasmus University Rotterdam.


Like many other countries, Vietnam is striving to achieve universal health coverage (UHC) in the context of an evolving health system. This policy brief summarises HEFPA’s research into the viability of extending coverage and constraints on the achievement of this goal.

Extending social health insurance from coverage mainly of civil servants and formal sector workers to the rest of the population in Vietnam has been pursued by fully subsidising the poor and young children (<6 years). This has contributed to around two-thirds of the population now being covered. Reaching the remaining third dependent on informal sector employment is a major challenge. Despite a premium subsidy of 70 per cent offered to the near-poor, coverage remains very low. HEFPA conducted an experiment designed to establish the extent to which insurance uptake is impeded by affordability, as opposed to a lack of information on the benefits of coverage.
Two other studies reveal potential threats to the universal coverage agenda from seemingly unrelated health system reforms. One assesses whether replacing fee-for-service with capitation payment of providers of insured services results in reduced provision to the insured compensating overprovision of services to the uninsured. The other study suggests that granting hospitals operational autonomy increases their activity but can result in over-servicing of patients and higher out-of- pocket payments.

Affordability is the most obvious deterrent to insurance enrolment. Indeed, it was the reason most frequently given by uninsured participants in the HEFPA study for not taking out cover. Other reasons given included low perceived need, poor quality-insured services and not knowing where to buy insurance.
If households lacking experience of insurance products do not fully understand the concept of insurance, nor appreciate its potential benefits, then lowering its cost will be ineffective in raising participation. A randomised experiment was therefore carried out to provide evidence on the relative responsiveness of enrolment to a 25 per cent premium subsidy and to the provision of information through a leaflet explaining both the operation of the insurance scheme and the benefits of the coverage offered [1].

The 25 per cent subsidy and the provision of information had no significant impact on enrolment, irrespective of whether the two were offered in isolation or in combination. In part, this may be attributable to a small sample size. The likelihood of enrolment was higher in the groups offered the subsidy, but the difference did not reach statistical significance. However, a significant increase in enrolment was found among those reporting poor health that were offered both the subsidy and information. This may be indicative of adverse selection in the response to the incentives by those most likely to claim. In the full sample, despite the offer of a 25 per cent subsidy, coverage remained well below 10 per cent.

The most plausible explanation for the lack of impact of the above strategies on enrolment is the low perceived benefits of health insurance, arising from observed high out-of-pocket spending and low quality of care received by the insured. The difficulty of overcoming these obstacles does not appear to be appreciated by central and provincial level officials, who were optimistic about the potential impact of the subsidy and information provision prior to the experiment. Scepticism voiced by district and commune officials, who are directly involved in the day-to-day operations, proved to be better founded. Policy makers should perhaps turn more to officials on the ground for ideas of future changes that would be more effective in encouraging enrolment. These might include better quality and value of the medical services covered by insurance, reduced out-of-pocket spending of those covered and possibly even larger subsidies to make health insurance more affordable. Of course, this combination would require a very substantial increase in government funding of social insurance.

Setting appropriate provider incentives is an essential component of any social health insurance aiming to achieve efficiency, quality of care and financial protection. Vietnam has begun to move away from fee-for service towards capitation payment of district hospitals treating patients. The uninsured continue to pay on a fee-for-service basis. HEFPA evaluated the impact of capitation on efficiency, quality and equity [2].
Under the new payment policy, a capitation fund is allocated to each district hospital based on the size and composition of the social health insurance members in the locality it serves. These facilities play a fundholding role; they are responsible for reimbursing the cost of services used by the insured at provincial and central hospitals. This policy puts strong pressure on the district hospitals to cut costs while providing no immediate incentives to improve quality of care. The capitation scheme started in a small number of district hospitals in 2006 and is currently being scaled up rapidly, set to include all district hospitals by 2015.

Capitation resulted in a 5 per cent reduction in total recurrent expenditures for a given volume of care, while drug expenditure on both the insured and uninsured was cut by nearly 8 per cent. There was no significant effect on clinical outcomes (deaths and adverse events). Among the insured, outpatient contacts per capita were reduced by 15 per cent and inpatient admissions cut by 16 per cent. The latter, however, was offset by a 16 per cent increase in admissions of the uninsured. Drug expenditure per insured patient was reduced by 21 per cent.

Capitation appears to induce cost savings without negatively affecting clinical outcomes. Hospitals respond to harder budget caps from the insurance fund by reducing the intensity of service provision to insured patients, but compensate by increasing intensity and revenues from uninsured patients. Policy makers need to be alert to potential unintended consequences of changes in provider incentives. Capitation payment can contribute to controlling costs for the insurance fund, thus relieving budgetary pressure, but it may also increase out-of-pocket spending and threaten the affordability of care for those remaining uninsured.

Staggered implementation of both the capitation and autonomisation policies enables the effect of each to be identified, principally by comparing changes in costs and outcomes that occur in hospitals where a policy is introduced to changes in hospitals that continue to be paid for and managed as before. This difference-in-differences strategy eliminates variance in costs and outcomes related to disparity between hospitals, as well as those arising from general changes in the health system that would influence outcomes regardless of whether or not a reimbursement or management control policy is implemented.

Financial and operational autonomy of public hospitals in Vietnam has been increased in an effort to raise efficiency. The decision-making power of hospital staff and managers has been strengthened, and their scope to benefit from surpluses extended. Hospitals have become less reliant on budget allocations and more dependent on user fees and income from the social health insurance agency. Parallel reforms have sought to protect the poor by subsidising their enrolment in the social health insurance scheme, rather than asking hospitals to grant them unreimbursed exemptions. The HEFPA evaluation provides evidence on the extent to which autonomisation has achieved efficiency objectives without jeopardising UHC goals [3].

Autonomisation slightly increased hospital admissions and outpatient visits, but had no effects on bed stocks, or bed-occupancy rates. There is some evidence that the policy led to higher household out-of-pocket spending on hospital care, as well as higher spending per treatment episode. Autonomy did not affect in-hospital death rates or adverse events, but in lower-level hospitals it did lead to a more intensive style of care, with more lab tests and imaging per case.

Greater hospital autonomy appears to have brought little gains in efficiency, and yet has raised household out-of-pocket spending and so threatened the financial protection goal of UHC. Looking forward, incentives need to be tuned more finely to ensure that hospitals operate in a way that is consistent with the affordability, access and quality of care, and not only with maintaining their bottom line.


[1] Wagstaff A, Nguyen HTH, Dao H, Bales S, ‘Encouraging Health Insurance for the Informal Sector: A Cluster Randomized Trial’, Unpublished MS: The World Bank (see also HEFPA Paper P1, Erasmus University Rotterdam).

[2] Nguyen HTH, Bales S, Wagstaff A, Dao H, ‘Getting Incentives Right: An Impact Evaluation of a District Hospital Capitation Payment Scheme in Vietnam’, Policy Research Working Paper 6709, The World Bank (see also HEFPA Working Paper 17, Erasmus University Rotterdam).

[3] Wagstaff A, Bales S, ‘The Impact of Public Hospital Autonomization – Evidence from a Quasi-Natural Experiment’, Policy Research Working Paper 6137, The World Bank (see also HEFPA Working Paper 12, Erasmus University Rotterdam).

Potential Impact:

Please provide a description of the potential impact (including the socio-economic impact and the wider societal implications of the project so far) and the main dissemination activities and the exploitation of results. The length of this part cannot exceed 10 pages.

Having tested different policy approaches to a) increase the share of people with health insurance coverage and to b) make health insurance coverage effective in terms of access to appropriate care and better financial protection against excess healthcare spending, HEFPA contributes to ongoing national and international discussions about health system reform towards the ambitious goal of Universal Health Coverage. The concept of universality is directly related to the debate on ‘Health as a Human Right’ and thus has societal implications that reach beyond the healthcare sector. With increasing awareness among developing and emerging country populations that access to appropriate care is a human/civic right, policymakers are under growing pressure to respond. Thus, given the specific health system requirements and limitations of the developing world, health system research specific to this context is in high demand. HEFPA has interacted with national and regional health system stakeholders from the beginning of the project to warrant that its research caters directly to country needs.
For instance, after greatly expanding health insurance enrollment, China has been struggling with both an expected explosion in overall healthcare spending and, unexpectedly, with stubbornly high out-of-pocket spending. In collaboration with local stakeholders, HEFPA has examined strategies to address both issues and its conclusions provide guidance for Chinese health policy makers on how to increase the efficiency of public spending, limit out-of-pocket expenditure and improve the appropriateness of treatments.
Another example is Thailand’s Universal Coverage Scheme that had previously been championed as a model for reform towards Universal Coverage in the global health policy debate. Robust evidence on its impacts was, however, widely lacking. HEFPA research provided the first robust estimates of the reform’s impact on out-of-pocket medical spending as a key indicator of healthcare coverage. In addition, HEFPA found regional heterogeneities in the reform’s impact on utilization that underscored the need for further reform to achieve true universality. This evidence we hope will contribute to an even more informed debate of the prominent Thai case.
In a different strand of HEFPA research, evidence emerging from policy experiments in the Philippines and Vietnam strongly suggests that voluntary health insurance is not an appropriate instrument to achieve broad health coverage for informal sector households - even if premiums are heavily subsidized and information on insurance benefits is provided. This evidence has been extensively shared with policymakers in Vietnam, the Philippines and beyond and we hope will inform future attempts to extend coverage.
Potentially motivating similar policies in other resource poor countries trying to tackle high maternal and newborn mortality, HEFPA provided important findings on the effectiveness of different demand and supply side interventions to increase the use of antenatal care and skilled birth attendance.
HEFPA research on financial implications of illness beyond out-of-pocket medical spending provides a final example for the timeliness and relevance of the project. For Vietnam, Thailand, Indonesia, and the Philippines, HEFPA finds that illness-related income loss can pose a substantial risk to households that often exceeds the risk of excess medical spending and that households typically apply coping mechanisms that likely harm them in the future – evidence that underscores the need for a debate on disability insurance as the next step on the social protection agenda.
Further policy implications of HEFPA findings can be found in the description of the main S & T results in this report.

Emphasizing policy implications, HEFPA uses various strategies to disseminate its findings both in national and international fora.
Firstly, HEFPA research has been presented on numerous prominent scientific conferences, many of which bring together not only researcher but also practitioners, and policymakers. In this context, the project has, among other things, organized HEFPA sessions at the 2009 7th World Congress of the International Health Economics Association in Beijing, China; the 2011 iHEA conference in Toronto/Canada; the 2011 Health System Reform in Asia conference, in Hong Kong; the 2012 Global Symposium on Health Systems Research in Beijing/China; the 2013 iHEA 9th World Congress in Sydney/Australia; the 2013 Health System Reform in Asia conference in Singapore; and the 2014 3rd Global Symposium on Health Systems Research in Cape Town, South Africa.
Outreach to policymakers and practitioners has been a focus of HEFPA’s dissemination activities from the start. Country teams have organized numerous meetings with health policy makers and insurance providers in Vietnam, China, the Philippines, Indonesia and Cambodia. For instance, in late 2013, a summarizing presentation of HEFPA evidence relevant to Indonesia’s upcoming Universal Coverage reforms was given at Indonesia’s State Ministry of National Development Planning (BAPPENAS) in Jakarta, and evidence on the effectiveness of the Philippines’ National Health Insurance was presented at the country’s Department of Budget and Management Main Office in Manila. In HEFPA’s final workshop held in September 2013 in Yogyakarta, Indonesia, the project team presented and discussed its findings with policymakers from HEFPA countries and several representatives of donor agencies and international organizations. A list of participants is attached to this report.
To disseminate its messages to more policymakers, practitioners, and researchers, HEFPA has developed 7 policy briefs (attached to this report). A summarizing brief outlines the projects main findings and conclusions, and six country-specific briefs present the country/project context, the methods applied, key findings and policy implications for the country under study and further afield. The briefs are currently in print and will be disseminated in both hard-copy and electronic format at upcoming conferences and through newsletters/mailing lists of international organizations and national stakeholders.
Furthermore, and also addressing the general public, HEFPA has created a continuously updated project website ( that provides detailed information on the project background and participating research teams and links to the project’s publications. HEFPA findings have also been presented in several lecture series in global health, among other things in the Global Health Track of Erasmus University’s Master in Health Economics, Policy & Law. Moreover, HEFPA findings have been communicated through social media platforms like e.g. through and
Finally, HEFPA is currently producing several project videos that will be uploaded on online platforms – among others. HEFPA chose this outreach tool over a ‘policy road show’ because of its lasting presence and greater reach to policymakers, practitioners, academic researchers and the general public worldwide. Footage for the videos comes from field visits in Indonesia, Cambodia and Thailand where patients, health staff and health system administrators gave testimony of their experience with the policies HEFPA evaluated. These are combined with interviews with HEFPA researchers, national health policy makers and representatives of international organization and donor agencies. The videos’ aim is to present the project findings in a non-technical way that is accessible to and interests also a non-academic audience. Among other things, showings of a 30-minute video summarizing HEFPA’s main findings and exemplifying them through patient and health staff interviews will be held at the 2014 3rd Global Symposium on Health Systems Research in Cape Town/South Africa and shortly at Erasmus University. Both showings will be combined with panel discussions. In addition to the 30-minute summary, HEFPA is currently producing detailed country specific project videos for Thailand and Cambodia and various excerpts of interviews on HEFPA’s findings will be made available online.

List of Websites: