- Email: [email protected]
Developing health insurance in transitional Asia
PERGAMON
Social Science & Medicine 48 (1999) 871±879
Developing health insurance in transitional Asia Tim Ensor * Centre for Health Economics, University of York, Heslington, York YO1 5DD, UK
Abstract Many European and Asian economies are currently undergoing a process of economic transition away from state based command systems to market led economies. The impact of transition, such as a decline in public expenditure, break up of state enterprises and economic recession, has aected levels of funding available for social sectors. In the health sector, health insurance is being introduced as a way of alleviating the decline in funding arising from these processes. Most of the Former Soviet Union and a number of other Asian transition economies are currently introducing, extending or considering payroll based systems of health insurance. Comparisons with many Latin American countries, where social security based insurance has been encouraged since the ®rst World War, can be illuminating. Experience suggests that, various factors have impeded or permitted development in these countries. General processes of economic change (transition factors) tend to aect all economies attempting to change the basis for public funding of services. Structural factors, such as urbanisation and the level of state or industrial employment, act as longer term inhibitors to the extension of coverage. These factors vary considerably across transition economies. This suggests that while a social security base for insurance may be a viable option for smaller industrialised European transitional economies, this is not the case for many of larger less industrialised economies. It is unclear how insurance will develop in the future. If a separate insurance fund is maintained it is important that its' purchasing function is developed. Otherwise it is not clear what value is added to the current health system. If entitlement is to be based on contribution, with the fund based on geographic or employment groups, systems for ensuring access for those not in employment and not classi®ed as socially protected must be developed. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Transition economies; Health insurance; Social security; Asian transitional economies
Introduction Access to health care in much of transitional Asia has largely depended on publicly funded and provided services. A by-product of the economic transition has been a re-evaluation of the basis of funding and a growing interest in alternative models of ®nance based on enterprises or individual insurance. A recent study (Ensor and Thompson, 1998) found that 23 out of 26 countries in transition were implementing or discussing the introduction of insurance. Yet the vast disparity in
* Tel.: +44-1904-433-639; fax: +44-1904-432-701; e-mail: [email protected]
economic and social circumstances in the transition economies may mean that, while payroll based insurance is a viable option for more industrialised economies, for others it may not be appropriate. This paper examines some of the diculties involved in extending social health insurance in transitional Asia. It draws on, as a parallel, the experience of Latin America countries many of which have implemented social security based systems in the last 80 years. A distinction is made between the individual structural features of a country and the common features of transition. The ®rst section provides a brief description of key features of Asian economies in transition. Section 2 looks at recent experience in introducing social health
0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 8 ) 0 0 3 8 9 - X
872
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Table 1 Basic health and infrastructure indicators (eight Asian transition economies) GDP per capita (1993 US$)
Cambodia China Lao PDR Mongolia VietNam Low income Kazakstan Kyrgyzstan Middle income
Public expenditure on health care, % of GNP carec
Doctors per 10,000
200 490 290 400 170 350
0.5 2.3 0.7 4.3 1.1
0.4 13.7 2.3 27 3.5 1.5
22 26 25
1540 830 2480
2 2.2
41.0 37.0 4.9
Beds per 10,000
DPT coverage
Life expectancya
Infant mortality rate
Maternal mortality (/100000)
33 10
38 95 22 84 83 90
52/2.8 69/3.2 51/3.0 64/2.7 65/4.5 62/2.1
129 31 97 44.6 36 71
460 95 300 200 120
136 120 16
41.7b N/A 82
70/8.8 69/7.7 67/5.9
31 37 38
53 43
Sources: UNDP (1995) and World Bank (1995). aAverage life expectancy at birth/dierence between males and females. b90% for ®rst dose of DPT but drop-out is high. BCG almost universal. cIncludes public insurance and state budget but not user charges.
insurance in a number of these countries. Sections 3 and 4 show how dierent country characteristics may hinder the development of social security based systems of insurance in individual countries. Section 5 discusses how the models of insurance might be re®ned in countries where social insurance is not a viable option for much of the population.
Changing Asia The countries range from the Central Asian of the former Soviet Union (FSU) through the emerging Asian economies of Vietnam and Lao PDR to China where market reforms now have a long history. The common features of these countries are: 1. A historic dependence on state funding and provision of many services including health care. 2. Liberal economic although not always political reforms. 3. Relatively good health indicators for income level. 4. A relatively large medical care infrastructure. In the case of the FSU countries and Vietnam, entitlement to services was universal. In China entitlement was never universal although a combination of labour insurance and the commune insurance system meant 1
Although the rapid increase in medical prices may have meant that the volume of services also declined. What is certainly true is that mix of provision has changed with expenditures even more concentrated in urban areas.
that more than 90% of the population was once covered. A consequence of the transformation process has been for resources available to the state to decline. Market reforms in economies throughout Asia have led to fundamental changes in the structure of public expenditure with implications. This has aected the amount of public funding for the health sector. In the case of Vietnam and China this decline has been relative rather than absolute. In China, for example, the proportion of GNP collected by the state fell from about 30% in 1980 to about 12% today1. In the FSU there has been an absolute decline. In Kazakstan it is estimated that real revenues in 1994 were only about 30% of those in 1990 (World Bank, 1996). Similar patterns are apparent in other Central Asian countries. The eect of the decline in state funding is exacerbated by health service infrastructure that is considerably larger and more complicated than in other countries with similar incomes (see Table 1). Central Asia exceed most other countries in the number of beds and doctors per head of population. Among low income countries, all the transition countries listed in Table 1 have a larger number of doctors per head than the average for the group with the exception of Cambodia. A similar pattern is apparent for the number of hospital of beds. The insurance approach: how much extra funding for the health sector? A number of factors could lie behind a decision to develop a social insurance funding basis for health
T. Ensor / Social Science & Medicine 48 (1999) 871±879
care. There are, perhaps, two most important reasons. The ®rst is the need to provide additional funding for health care to reverse the relative or absolute budgetary decline. The second, more subtle but arguably more important, is the desirability of creating a separate buyer or purchaser for health care. This is a key issue in countries where there is no organisational separation between the public agency that manages the provision of care and the one that ®nances care. Of these two justi®cations, the ®rst appears to be the most important in transition economies. Reduced funding for state facilities has placed this infrastructure under considerable pressure. This is manifested in a number of ways including delays in payment of salaries, dramatic reductions in building and equipment spending and the development of user charges for services in public hospitals. This has led to a phenomena of `creeping privatisation' whereby a reliance on private funding distracts attention away from the provision of publicly funded services and towards more pro®table enterprises. The extreme example of this is China where a hospital may operate a scanner simply to increase billing to an insurance fund (Lok San Ho, 1995). This is because scanning technologies, even if inappropriate, are often attractive to patients. Also the technologies are relatively new and so have not yet been controlled through the national system of taris. Although China has a history of health insurance dating back to the 1950s, the decline of the commune system of agriculture in the early 1980s dramatically reduced the proportion of the agricultural population covered for medical care. From a peak of over 90% the numbers covered by commune schemes declined to less than 10% in the late 1980s (Young, 1989). This trend is also apparent in Vietnam. Despite the universal basis of the Vietnamese system, the introduction of individual farming, which replaced the communal system, reduced resources available to rural communities for social services. The Government has not been able to fund this de®cit and much of health care is now funded by private payments (legalised in 1988). In the FSU this trend is being replicated in the break up of state farms which historically provided resources for building and running polyclinics and rural hospitals. Health insurance in Vietnam was introduced at a national level in 1993 following pilot projects in a number of provinces (Carrin et al., 1993; Ensor, 1995). Two schemes were originally envisaged. A compulsory scheme for those in state and industrial sector employment (enterprises larger than ten employees) and another voluntary scheme mainly aimed at rural areas. Both are managed by the same organisation, Vietnam Health Insurance, at national and provincial level. Although the compulsory scheme has managed to enrol much of the formal sector, this is still only a small proportion of total employment. The voluntary
873
scheme is now being widened into a number of schemes aimed at dierent groups including school children and students, farmers, urban informal sector, dependents of government workers and humanitarian insurance for the poor. The intention is to develop a system that gradually ®lls in gaps in coverage. Currently around two million children and 200,000 adults are covered through the voluntary schemes, between ®ve and six million by the compulsory scheme, leaving about 90% of the population without cover and liable for user charges. In Central Asia health insurance is furthest developed in Kazakstan where a national law took eect in January 1996 requiring all employers to register their employees. This followed pilot programmes in four oblasts (regions). The contribution is paid by the employer for those in employment, by the individual for those in self-employment or without employment and for the socially protected from the state budget. Because the state sector is still large, it has proved relatively easy to register a large part of the workforce. By the middle of the year around 80% of employees had been registered across the country. A weak economy has made it much more dicult to collect contributions. Between 20 and 30% of expected revenues had been collected from enterprises for those covered by insurance with considerable variation across the country. Collections varied from as low as 9% in 1996 to as high as 52% in Almaty city. Income of the oblast explains some, but not all, the variance. Just as important is the individual management ability of each oblast fund and the extent to which they are supported by the local tax collection authorities (Ministry of Health, 1996). It is too early to say how large the uncovered population will be. Early indications are that collections will be con®ned largely to state and large private companies. Although the system is compulsory in theory most of those in small private ®rms or self-employed can avoid paying contributions just as they avoid paying taxes. At present, the self-employed are required to pay a ¯at rate contribution although registrations are currently few. One of the problems is that the perceived bene®ts are unclear, or of little value. Most people already pay informal charges or contribute inkind to their own treatment while formal charges are few. Anecdotal evidence suggests that many expect that they will continue to pay informally whether or not they are insured. This impression is likely to prove dicult to dispel. At present around 25% of the population are outside the formal workforce and do not qualify as socially protected. These include the unregistered unemployed, non-working dependants, self-employed and employees in small non-juridical enterprises.
874
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Other FSU republics are currently considering insurance. Kyrgyzstan has a law but implementation has been slow due to poor results in a pilot oblast. Proposals for a personal health account have been advanced in Uzbekistan although legislation has yet to be prepared. In Turkmenistan a system of voluntary insurance was introduced in January 1996. This covers 90% of the cost of outpatient prescriptions as well as a number of other items such as discounts on dentures. Although popular, the scheme has also proved expensive with far more paid out than collected in contributions. There are proposals for a national system of compulsory health insurance modelled on the system in Kazakstan and the Russian Federation. Preconditions for social insurance and lessons from Latin America The factors making the development of insurance more or less feasible and appropriate are complex. In order to analyse the current context in transitional Asia, it is instructive to examine the development of insurance in Latin America since the end of the First World War. Several aspects of insurance development in Latin America make such a comparison relevant. Insurance was introduced for a geographically proximate, yet heterogeneous group of countries. Latin American countries share certain common characteristics. These include closeness of language and some economic inter-dependence. Most fall into low and upper middle income groups. During the 1980s, many suered substantial and prolonged economic recession. Real GDP in Venezuela, for example, declined by around 20% while in Peru it fell by more than 30%. Many South American and Caribbean countries adopted, on mass, systems of social security based on enterprise contributions. These include pension, unemployment and maternity entitlements in addition to health care bene®ts. Coverage has been extended in the classic way from civil and other public servants, through white and then blue collar industrial workers and later to other urban and then rural workers. The most advanced countries in this respect are Argentina, Brazil and Costa Rica, with 75 to 100% coverage, followed by countries as such as Uruguay and Chile (Mesa-Lago, 1991). Cuba automatically entitles all citizens to health service provision, although it retains a contributory social security approach for other bene®ts. In Latin America the institution acting as the major impetus to insurance development was the International Labour Organisation (ILO). In transitional Asia there has been no one organisation that has actively promoted social insurance. The develop-
ment has been spurred by a combination of factors including the rapid development of Bismarkian models in Eastern Europe (Marree and Groenewegen, 1997) and a need to obtain additional funding during deep economic recession. The World Bank, of all the major donors, has probably had the strongest in¯uence over ®nancing reforms. This has, however, usually been in the form of assistance in the development of best practice for new ®nancing rather than initial promulgation of insurance. The World Health Organisation has, similarly, been involved in helping to develop nascent systems and sometimes even attempting to dissuade countries from developing the system in the ®rst place. The factors that help or hinder the development of a social security approach to health system funding can be divided into three main groups; features associated with transition, structural characteristics of a country and country-speci®c pre-disposing factors. Transition features Transition characteristics are those features that are common to countries undergoing economic change. They are exhibited by most transition countries and may impede the development of new tax structures and social security systems required for the development of public services such as health care. A number of trends can be observed. 1. The process of economic transition has a profound eect on the structure of the economy and employment patterns. State industries are privatised, downsized and closed. A consequence is that employment in public sector enterprises is declining while small private employment is growing. A feature of this process is that much of this new workforce is `informal' in sense of being unregistered for taxation and regulatory purposes. This restructuring has important consequences for raising revenue. 2. Economic transition has a substantial impact on the ability of public providers to deliver aordable health services to paying or insured patients. Increased employment opportunities in these transforming economies makes it progressively more dif®cult to retain a skilled workforce in the health sector. For example, many rural health workers in China have left posts for more lucrative employment (Lok San Ho, 1995). Additionally, the process of creeping privatisation of public facilities leaves patients funded by modest insurance fund payments far less attractive than those willing to pay larger direct payments (both formal and informal). So, for example, in Vietnam it is reported that public hospitals prefer to take fee paying patients and neglect those funded only by compulsory insurance (Thompson, 1995). In Almaty one newspaper survey
T. Ensor / Social Science & Medicine 48 (1999) 871±879
found that almost all patients make additional payments to doctors in order to receive prompt attention or a doctor of choice (Caravan, 1996). Changes in the wider economy, such as greater opportunities for private work in and outside the health sector, are making it more dicult for public health facilities to deliver the bene®ts promised. 3. A feature common to former soviet and East European countries, but not to other Asian transitional economies, has been deep economic recession. Rana (1995) has suggested some of the reasons why this dierence is observed. The includes a much greater dependence on (inecient) heavy industry and a large state sector as well as a much greater level of interdependence between economies, making enterprise restructuring more dicult.Recession makes taxes, social security and insurance contributions much harder to collect. Mesa-Lago (1991) argued that while the Latin America and Caribbean region as a whole would have required growth in GDP exceeding 7% in order to reduce the informal sectors signi®cantly and increase insurance coverage, the actual negative rates of growth made this impossible. Although the need for health care does not diminish and probably increases, the resources available to provide a comprehensive insured package declines. Indeed those most requiring health and other social protection may be unable to participate fully in the social insurance system because they lack formal employment. Systems of provision in Latin America have been criticised because they do not provide coverage for those in extreme poverty (Camcho, 1996). 4. One of the most fundamental diculties is that while many countries have increased coverage and health service access through insurance, transition countries appear to reduce access. The justi®cation for doing so is clear. Although the concept of universality remains, in practice it is a ®ction. Therefore it might be more honest and lead to more eective services if a minimal package of services were oered to all with additions funded from contribution related insurance. Yet there are immense political rami®cations involved in overturning a long history of `free' services. There is also a suspicion that if insurance is introduced then the bene®ts will mean nothing, people may go on paying informally for services just as they did under the old system. In these circumstances, there is little to be gained from voluntarily joining an insurance scheme. The result is that those that are able to pay directly can avoid the new tax/contribution and the overall yield is eroded. The entitlement problem highlights perhaps the most fundamental dierence between the experiences of
875
these two country groups. In Latin America social insurance expanded entitlement to services free at the point of delivery funded on the basis of income related (rather than risk rated, as with private insurance) contributions. In transitional Asia most countries begin from a universal entitlement based on citizenship. Payroll insurance has two possible implications. One is to dis-entitle people from universal coverage and demand that they re-register through their employment, purchase individual cover or qualify as socially protected to be covered from the public budget. The second possibility is to retain universal entitlement and to treat the payroll contribution as a tax on those unfortunate enough to work for enterprises liable (in theory and fact) for payment. However, in this case, there is little incentive to pay the contribution since it provides no additional bene®ts. Structural characteristics Structural characteristics are features of a country that are important in determining the feasibility of collecting revenues through a payroll tax that can be expected to change slowly, if at all. The proportion employed in the industrial sector is a good indication of the ease of registration since most enterprises will have a formal payroll system from which payroll and other taxes will already be paid. Similarly, the size of the urban population and population density will also contribute to the administrative cost of the system. If the population is largely rural and widely scattered the registration and assessment of contributions is made more dicult than if the population is concentrated in urban areas. Finally, the income of the country and growth rate of incomes are indicators of how easy it will be for enterprises and individuals to pay insurance contributions. A growing, prosperous economy will easily absorb contributions which an economy in recession ®nds impossible to pay. Structural factors have been important in determining the speed of social security development in Latin America and the Carribean (Mesa-Lago, 1996). Countries with the most well established systems, such as Argentina, Uruguay and Costa Rica, also have the highest proportion of the population living in urban areas and working in the industrial sector. Countries which established systems later and more slowly tend to have a larger proportion of the workforce living in rural areas (coverage ranges from 14 to about 35%). In general, the results from this and other studies (Jenkins, 1993) suggest that it is very dicult to extend overall coverage beyond the sum of formal urban and modern rural sectors. Using the quantitative indicators described above and following a similar but simpli®ed analysis carried out by Shaw and Ainsworth (1996) for Sub-Saharan
876
T. Ensor / Social Science & Medicine 48 (1999) 871±879 Table 2 Feasibility of introducing insurance in low and middle income countries Score
Countries
3, 4 2
Trinidad and Tobago, South Korea, Mexico, Belarus, Estonia Argentina, Barbados, Uruguay, Chile, Brazil, Czech Republic, Hungary, South Africa, Latvia, Lithuania, Bulgaria, Cuba, Ukraine Slovenia, Saudi Arabia, Malaysia, Panama, Poland, Russian Federation, Colombia, El Salvador, Jamaica, Romania, Philippines, Moldova Turkey, Costa Rica, Peru, Dominican Republic, Ecuador, Morocco, Armenia, Georgia, Kazakstan, Uzbekistan Botswana, Namibia, Paragua, Indonesia, Bolivia, Egypt, Sri Lanka, Albania, Azerbaijan, Kyrgyzstan, China, Pakistan, Ghana, Tajikistan, India Belize, Thailand, Turkmenistan, Senegal, Cote d'Ivoire, Mongolia, Nigeria, Kenya, Bangladesh, VietNam, Burundi Lesotho, Guinea, Zimbabwe, Gambia, Zambia, Guinea-Bissau, Zaire, Uganda, Malawi, Tanzania, Ethiopia, Cambodia, Lao PDR, Niger, Sudan
1 0 ÿ1 ÿ2 ÿ3, ÿ4
Transition countries in italics.
countries, a group of low and middle income countries were ranked according to a composite index of all four variables (population density, percent of the population urbanised, per cent of the workforce working in industry and per capita income)2. A ranking near the top of the list suggests that insurance is relatively easy to implement while a ranking towards the bottom suggests that it would be extremely dicult. Many of the countries with a high level of coverage in Latin American and Asia, such as Brazil, Argentina and Korea, are at the top of the list (see Table 2). Of the transition economies, countries such as Estonia, Czech Republic and Russian Federation with a high concentration of the population in the urban industrialised sector are near the top of the list. Further down is Kazakstan, Kyrgyzstan, Tajikistan and Turkmenistan. At the bottom of the list are very poor African and Asian countries such as Cambodia, Burundi and Uganda. While transition features are important in delaying the process of change, structural factors may be enough to block particular patterns of social security development altogether. Estonia found that the development of insurance in the early stages was impeded but that improving economic circumstances, combined with good structural preconditions, now permit the system to develop quite eectively (see Laur, 1997). 2 For each indicator the value for each country was compared to the mean for all countries. If the value was more than one standard deviation above the mean it scored one, if more than one below it scored minus one and if somewhere in between a score of zero. The scores for each indicator were added together to give a composite score.
It is notable that a number of the Asian transition countries introducing schemes, notably Turkmenistan, Vietnam and Kazakstan, are towards the bottom of the list, suggesting that they will ®nd it dicult to extend the scheme to the majority of their population. Kazakstan, for example, is ranked alongside Ecuador which has achieved around 23% social security coverage (Mesa-Lago, 1996). Turkmenistan and Vietnam are ranked below Bolivia (18% coverage) and alongside countries such as Kenya which has managed to achieve only about 25% coverage. Country speci®c features Apart from general transition and structural factors, features that are country-speci®c are also likely to be important. One is the extent to which insurance funds are well managed which has several dimensions. First are systems for preventing and dealing with fraud. An advantage of the restrictive system of line funding and largely cash free transactions was that the scope for fraudulent activity was quite limited since few people had access to substantial cash balances. A worry expressed in a number or transition countries is that insurance funds not included in general government budget setting often lack a strict system of line control but also have not developed modern accounting systems for audit common in the West. Several transition countries have reported scandals at the national or regional level where poorly regulated fund directors have stolen, or made personal use of, fund balances. A second management issue is where funds, although not used inappropriately are not managed eciently. This may arise in a number of ways.
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Administrative issues There are technical problems related to fund administration (see Normand and Weber, 1994). These include reserves that are not invested in way that ensures a high hedged rate of return, an undeveloped collection system so that the rate of collections is low and excessive levels of administration relative to the size of collections (a particular problem where the total number of enrollers is low). Financial management More fundamentally, many systems are based on inadequate calculations relating to the expenditures and revenues to be generated from a given package of insured bene®ts. The result is that bene®ts promised far exceed contributions obtained. This is a very common problem with the `pay as you go' payroll tax systems being introduced in many FSU countries. Part of the problem arises because of a historic commitment to universal services. Another reason is that the contribution is seen as a tax that must be paid rather than a payment for services rendered (in this case the insured risk rather than the medical services themselves). Another facet of this problem is where extravagant bene®ts are promised in order to increase enrolment. In Uruguay, for example, (pension) bene®ts were permitted to rise faster than contributions. The consequence was the government was unable to maintain the size of bene®ts, the system got a bad reputation and people lost con®dence in its operation (Camcho, 1996). Ineective purchasing A key potential role for an insurance fund introduced into a predominantly integrated health care system (as in much of transitional Asia) is as an independent purchaser-commissioner of health care. This role includes the assessment of treatment, type of services and best-value providers. Yet this role is often underdeveloped for several reasons. One reason is that the insurance fund is seen primarily as a contribution (or tax) collector and little thought is given to the disbursement role. Insurance funds in those countries have primarily been concerned with collecting money from enterprises rather than disbursing to providers. It is not surprising that most of the statistics published report numbers enrolled and contributions collected but not the way money is spent. A second issue relates to a confusion about the precise role of the fund. Should the insurance fund provide for the (constrained) demands or wants of individuals or, alternatively, provide for their needs. The former model is closer to a private insurer that provides limited access for a range of bene®ts desired by the individuals insured. In the later case, the fund
877
will decide, on a population basis, what services best add to the health of all those insured. This model is closer to the purchaser role of, say, a UK health authority or Swedish county. Re®ning the insurance model in transition economies The foregoing discussion suggests that there is considerable diversity across transitional economies. The characteristics of transition, changing structure of the workforce, economic recession and declining public sector expenditure, are common features present across the region from Poland to transitional Asia. Longer term structural features which make it more or less dif®cult to extend social insurance vary enormously. The current characteristics of most Asian transition economies suggest that the model of social insurance will, on its own, fail to provide comprehensive insurance cover. As currently conceived, insurance provides some marginal additional income for the health sector. Those covered still have to pay to receive treatment and there is therefore little incentive to join or pay promptly. Yet it is too early to judge whether the schemes will, in time, develop to oer more comprehensive coverage. There are three main directions in which schemes might develop. The ®rst is to recognise the current reality: that the new insurance schemes are basically earmarked payroll taxes oering little in the way of additional service entitlement. As the economy strengthens in each country this extra source of revenue will grow to form a useful addition to the health service budget. In time the system might lead to comprehensive social security or remain a universal tax funded system. In the short term there is no need for an additional supporting infrastructure (health fund, local oces, overall management board) to collect what amounts to another tax. A consequence is that the insurance system adds little value to the health care system other than the collection of often marginal payroll taxes. If this is the case it is hard to see how the creation of a separately organised insurance bureaucracy can be justi®ed. The second model, the Latin American paradigm, is to view social health insurance as one part of a comprehensive system of social security that is managed by a small number (perhaps only one) of national employment related agencies. Social security can be provided as a package through a combination of `pay as you go' entitlement, more appropriate for health care and some other bene®ts and funded (capitalised) systems for long term bene®ts such as pensions. For transition countries such systems would require a radical change in the way in which social bene®ts are provided. There is evidence that this is being considered in some
878
T. Ensor / Social Science & Medicine 48 (1999) 871±879
countries. There has been discussion in Kazakstan, for example, over the merging of the insurance and social security fund that provides sickness, unemployment and maternity bene®ts. The third model is to view the insurance fund as an independent public purchaser of health care for a de®ned population. This function is made easier if the enrolled population is based on a geographic area rather than employment group. This is certainly true in the case of the developing systems in Central Asia and VietNam. It is true of the rural schemes in China but not of the urban schemes which are still predominantly based on employment groups. In most cases, this has not been the model adopted in Latin America. An exception to this is Brazil which has begun to move towards a system of universal entitlement based on geographic based local purchasers. Until the 1980s it had a well developed system of employment based cover supported by private schemes for the rich and government provision for the very poor. In the late 1980s Brazil introduced a uni®ed system (Kutzin and Barnum, 1992) whereby public funding, from insurance and taxation, was uni®ed giving all citizens of Brazil access to the same providers. Only the private insurance funding remains separate. Large inequalities do, however, remain. This is largely the result of massive disparities in the level of resources available in each state and has promoted a strong interest in methods of geographic resource allocation similar to the RAWP formula used in the UK. Independent planning and commissioning of care, if carried out properly, would have substantial bene®ts. There is growing evidence that much medical care provided in transition countries does not represent good value for money. There is a growing trend of induced demand for pharmaceuticals and diagnostic tests. Many of these services are provided simply to generate revenue and are medically inappropriate (see Ensor, 1998 for a summary of experience). In addition many treatments routinely used in the past are ineective for many of their current uses. A conference in Almaty on quality in health care drew attention to two treatments that are routinely used: hyperbaric oxygenation and low frequency laser therapy (ZdravRefom, 1996). The evidence for both these treatments for many of their current uses is poor, yet hospitals continue to purchase such technology in order to attract paying patients. A role for the insurance fund is to distinguish between treatments that do not work and those that do and to ensure that those that do work are obtained at low cost. These are not easy tasks but are important to develop if the insurance funds are to become more than additional tax collectors. The fundamental issue of equity in the ®nance of services is how those outside the workforce obtain health services.
If the ®rst option is chosen then it is presumed that this will be achieved in a similar way as for those in employment. The issue of ineective treatments and eective treatment provided ineciently still remains and must be addressed through the reform of funding disbursement mechanisms within an integrated health service. A possible variant here is to follow the model favoured by a number of OECD countries and institute a purchaser provider split at the local health department level. The second option requires speci®c identi®cation of those unable to contribute to insurance, who are explicitly covered from the budget, and others who are not in waged employment but can aord to contribute. However, is it is clear from Section 4 that this cannot be achieved through compulsory payroll taxes. Instead, the system might evolve to include complementary systems of funding such as voluntary insurance, that provide valued medical bene®ts for those buying cover. There may be lessons from the experience of community ®nancing in Africa and other parts of Asia (see Shaw and Ainsworth, 1996). This suggests that local communities can be persuaded to buy limited and very well de®ned bene®ts through the purchase of voluntary ®xed premium cards, provided that bene®ts are attractive and delivered when required. Management capability is also required. Evidence suggests that where management support and local motivation is high, schemes are more likely to succeed (Carrin and Vereecke, 1992). In the third option, local health fund-purchasing could follow several routes. One is to maintain, or return, to a universal funding system. Compulsory payroll taxes supplement resources derived from general taxation. In this paradigm it makes sense for the tax service to collect contributions although there may be some suspicion in the health fund that insucient attention is given to health payroll contributions compared to other taxes. The only advantage of the health fund is as a purchaser of care and it is important that this role is developed. The alternative route is for local funds to develop a contribution basis for entitlement. In this case, as with the second option, other ways for covering those outside the salaried sector and socially protected groups must be found. Conclusion A strength of most former centrally planned health systems was that almost universal access to health facilities was ensured. In terms of basic indicators of good primary curative and preventive health care these systems scored well. The central weakness was that the funding system tended to encourage inecient patterns of hospital provision.
T. Ensor / Social Science & Medicine 48 (1999) 871±879
It is currently unclear how the new systems of insurance will develop in most Asian Transition Economies. A number of paths are possible. The main fear is that progressing too far along some type of contribution based system will jeopardise universal access and substitute one inecient funding system (funding according to numbers of sta and bed days) for another (fee for service from the insurance fund). If social insurance is the preferred option, the insurance agency must be capable of bringing about changes that ensure and increase access for those in rural areas and promotes better use of resources within the hospital sector. In pursuing these objectives, transition countries have much to learn, positive and negative, from countries with established insurance systems. Acknowledgements I am grateful to the conference organisers and an anonymous referee for comments on a draft of this paper. References Camcho, L., 1996. Financing social security in Latin America: new perspectives in the light of current developments. International Social Security Review. Caravan, 1996. If you don't grease you won't go. Caravan National Newspaper. Almaty, April 19. Carrin, G., Sergent, F., Murray, M., 1993. Towards a Framework for Health Insurance Development in Hai Phong, Vietnam. Macroeconomics Health and Development Series, Technical Paper 12. World Health Organisation, Geneva. Carrin, G., Vereecke, M., 1992. Main lessons from the practice of community ®nancing schemes. In: Carrin, G. (Ed.), Strategies for Health Care Finance in Developing Countries. Economic Issues in Health Care, Macmillan. Ensor, T., 1995. Introducing health insurance in Vietnam. Health Policy and Planning 10 (2), 154±163. Ensor, T., 1998. Health sector reform in Asian transition countries. In: Brooks, D., Thant, M. (Eds.), Social Sector Issues in Asian Transitional Economies. OUP, Hong Kong, in press. Ensor, T., Thompson, R., 1998. Health insurance as a catalyst to change in former communist countrues? Health Policy, in press.
879
Jenkins, M., 1993. Extending social security protection to the entire population: problems and issues. International Social Security Review 46 (2), 3±20. Kutzin, J., Barnum, H., 1992. Institutional features of health insurance programs and their eects on developing country health systems. International Journal of Health Planning and Management 7, 51±72. Laur, P., 1997. Reforming the health service in Estonia. In: Sorenson, J. Rudbeck, B., Bilde, L. (Eds.), Reforming the Health Services in the Baltic Countries. Danish Institute for Health Services Report 97.08. Lok San Ho, 1995. Market reforms and China's health care system. Social Science and Medicine 41 (8), 1065±1072. Marree, J., Groenewegen, P.P., 1997. Back to Bismark: Eastern European Health Care Systems in Transition. Ashgate, Aldershot. Mesa-Lago, C., 1991. Social security in Latin America and the Caribbean: a comparative assessment. In: Ahmad, E., Dreze, J., Hills, J., Sen, A. (Eds.), Social Security in Developing Countries. Oxford. Mesa-Lago, C., 1996. Comparative study of the development of social security in Latin America. International Social Security Review 2, 127±151. Ministry of Health, 1996. Evaluation of Medical Insurance in the Republic of Kazakstan. Almaty. Normand, C., Weber, A., 1994. Social Health Insurance: a Guidebook for Planning. WHO and ILO. Rana, P.B., 1995. The Asian approach to reforming transitional economies. In: Rana, P.B., Hamid, N. (Eds.), From Centrally Planned to Market Economies: the Asian Approach, vol. 1. Oxford. Shaw, R.P., Ainsworth, M., 1996. Financing Health Services though User Fees and Insurance. World Bank Discussion Paper 294. The World Bank. Thompson, R., 1995. Extending the scope of health insurance in Vietnam. M.Sc. dissertation (unpublished). UNDP, 1995. Human Development Report. United Nations Development Programme, Oxford. World Bank, 1995. Social Indicators of Development. John Hopkins. World Bank, 1996. Republic of Kazakstan: Transition of the State, vol. I. World Bank. Yang, B., 1996. The role of health insurance in the growth of the private health sector in Korea. International Journal of Health Planning and Management 11, 231±252. Young, M., 1989. Impact of the rural reform on ®nancing rural health services in China. Health Policy 11, 27±42. ZdravRefom, 1996. Eectiveness in health care: laser therapy and hyperbaric oxygen. In: Conference Notes for National Conference on Assuring Quality of Care, Almaty, 15±16 May.
Social Science & Medicine 48 (1999) 871±879
Developing health insurance in transitional Asia Tim Ensor * Centre for Health Economics, University of York, Heslington, York YO1 5DD, UK
Abstract Many European and Asian economies are currently undergoing a process of economic transition away from state based command systems to market led economies. The impact of transition, such as a decline in public expenditure, break up of state enterprises and economic recession, has aected levels of funding available for social sectors. In the health sector, health insurance is being introduced as a way of alleviating the decline in funding arising from these processes. Most of the Former Soviet Union and a number of other Asian transition economies are currently introducing, extending or considering payroll based systems of health insurance. Comparisons with many Latin American countries, where social security based insurance has been encouraged since the ®rst World War, can be illuminating. Experience suggests that, various factors have impeded or permitted development in these countries. General processes of economic change (transition factors) tend to aect all economies attempting to change the basis for public funding of services. Structural factors, such as urbanisation and the level of state or industrial employment, act as longer term inhibitors to the extension of coverage. These factors vary considerably across transition economies. This suggests that while a social security base for insurance may be a viable option for smaller industrialised European transitional economies, this is not the case for many of larger less industrialised economies. It is unclear how insurance will develop in the future. If a separate insurance fund is maintained it is important that its' purchasing function is developed. Otherwise it is not clear what value is added to the current health system. If entitlement is to be based on contribution, with the fund based on geographic or employment groups, systems for ensuring access for those not in employment and not classi®ed as socially protected must be developed. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Transition economies; Health insurance; Social security; Asian transitional economies
Introduction Access to health care in much of transitional Asia has largely depended on publicly funded and provided services. A by-product of the economic transition has been a re-evaluation of the basis of funding and a growing interest in alternative models of ®nance based on enterprises or individual insurance. A recent study (Ensor and Thompson, 1998) found that 23 out of 26 countries in transition were implementing or discussing the introduction of insurance. Yet the vast disparity in
* Tel.: +44-1904-433-639; fax: +44-1904-432-701; e-mail: [email protected]
economic and social circumstances in the transition economies may mean that, while payroll based insurance is a viable option for more industrialised economies, for others it may not be appropriate. This paper examines some of the diculties involved in extending social health insurance in transitional Asia. It draws on, as a parallel, the experience of Latin America countries many of which have implemented social security based systems in the last 80 years. A distinction is made between the individual structural features of a country and the common features of transition. The ®rst section provides a brief description of key features of Asian economies in transition. Section 2 looks at recent experience in introducing social health
0277-9536/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 9 5 3 6 ( 9 8 ) 0 0 3 8 9 - X
872
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Table 1 Basic health and infrastructure indicators (eight Asian transition economies) GDP per capita (1993 US$)
Cambodia China Lao PDR Mongolia VietNam Low income Kazakstan Kyrgyzstan Middle income
Public expenditure on health care, % of GNP carec
Doctors per 10,000
200 490 290 400 170 350
0.5 2.3 0.7 4.3 1.1
0.4 13.7 2.3 27 3.5 1.5
22 26 25
1540 830 2480
2 2.2
41.0 37.0 4.9
Beds per 10,000
DPT coverage
Life expectancya
Infant mortality rate
Maternal mortality (/100000)
33 10
38 95 22 84 83 90
52/2.8 69/3.2 51/3.0 64/2.7 65/4.5 62/2.1
129 31 97 44.6 36 71
460 95 300 200 120
136 120 16
41.7b N/A 82
70/8.8 69/7.7 67/5.9
31 37 38
53 43
Sources: UNDP (1995) and World Bank (1995). aAverage life expectancy at birth/dierence between males and females. b90% for ®rst dose of DPT but drop-out is high. BCG almost universal. cIncludes public insurance and state budget but not user charges.
insurance in a number of these countries. Sections 3 and 4 show how dierent country characteristics may hinder the development of social security based systems of insurance in individual countries. Section 5 discusses how the models of insurance might be re®ned in countries where social insurance is not a viable option for much of the population.
Changing Asia The countries range from the Central Asian of the former Soviet Union (FSU) through the emerging Asian economies of Vietnam and Lao PDR to China where market reforms now have a long history. The common features of these countries are: 1. A historic dependence on state funding and provision of many services including health care. 2. Liberal economic although not always political reforms. 3. Relatively good health indicators for income level. 4. A relatively large medical care infrastructure. In the case of the FSU countries and Vietnam, entitlement to services was universal. In China entitlement was never universal although a combination of labour insurance and the commune insurance system meant 1
Although the rapid increase in medical prices may have meant that the volume of services also declined. What is certainly true is that mix of provision has changed with expenditures even more concentrated in urban areas.
that more than 90% of the population was once covered. A consequence of the transformation process has been for resources available to the state to decline. Market reforms in economies throughout Asia have led to fundamental changes in the structure of public expenditure with implications. This has aected the amount of public funding for the health sector. In the case of Vietnam and China this decline has been relative rather than absolute. In China, for example, the proportion of GNP collected by the state fell from about 30% in 1980 to about 12% today1. In the FSU there has been an absolute decline. In Kazakstan it is estimated that real revenues in 1994 were only about 30% of those in 1990 (World Bank, 1996). Similar patterns are apparent in other Central Asian countries. The eect of the decline in state funding is exacerbated by health service infrastructure that is considerably larger and more complicated than in other countries with similar incomes (see Table 1). Central Asia exceed most other countries in the number of beds and doctors per head of population. Among low income countries, all the transition countries listed in Table 1 have a larger number of doctors per head than the average for the group with the exception of Cambodia. A similar pattern is apparent for the number of hospital of beds. The insurance approach: how much extra funding for the health sector? A number of factors could lie behind a decision to develop a social insurance funding basis for health
T. Ensor / Social Science & Medicine 48 (1999) 871±879
care. There are, perhaps, two most important reasons. The ®rst is the need to provide additional funding for health care to reverse the relative or absolute budgetary decline. The second, more subtle but arguably more important, is the desirability of creating a separate buyer or purchaser for health care. This is a key issue in countries where there is no organisational separation between the public agency that manages the provision of care and the one that ®nances care. Of these two justi®cations, the ®rst appears to be the most important in transition economies. Reduced funding for state facilities has placed this infrastructure under considerable pressure. This is manifested in a number of ways including delays in payment of salaries, dramatic reductions in building and equipment spending and the development of user charges for services in public hospitals. This has led to a phenomena of `creeping privatisation' whereby a reliance on private funding distracts attention away from the provision of publicly funded services and towards more pro®table enterprises. The extreme example of this is China where a hospital may operate a scanner simply to increase billing to an insurance fund (Lok San Ho, 1995). This is because scanning technologies, even if inappropriate, are often attractive to patients. Also the technologies are relatively new and so have not yet been controlled through the national system of taris. Although China has a history of health insurance dating back to the 1950s, the decline of the commune system of agriculture in the early 1980s dramatically reduced the proportion of the agricultural population covered for medical care. From a peak of over 90% the numbers covered by commune schemes declined to less than 10% in the late 1980s (Young, 1989). This trend is also apparent in Vietnam. Despite the universal basis of the Vietnamese system, the introduction of individual farming, which replaced the communal system, reduced resources available to rural communities for social services. The Government has not been able to fund this de®cit and much of health care is now funded by private payments (legalised in 1988). In the FSU this trend is being replicated in the break up of state farms which historically provided resources for building and running polyclinics and rural hospitals. Health insurance in Vietnam was introduced at a national level in 1993 following pilot projects in a number of provinces (Carrin et al., 1993; Ensor, 1995). Two schemes were originally envisaged. A compulsory scheme for those in state and industrial sector employment (enterprises larger than ten employees) and another voluntary scheme mainly aimed at rural areas. Both are managed by the same organisation, Vietnam Health Insurance, at national and provincial level. Although the compulsory scheme has managed to enrol much of the formal sector, this is still only a small proportion of total employment. The voluntary
873
scheme is now being widened into a number of schemes aimed at dierent groups including school children and students, farmers, urban informal sector, dependents of government workers and humanitarian insurance for the poor. The intention is to develop a system that gradually ®lls in gaps in coverage. Currently around two million children and 200,000 adults are covered through the voluntary schemes, between ®ve and six million by the compulsory scheme, leaving about 90% of the population without cover and liable for user charges. In Central Asia health insurance is furthest developed in Kazakstan where a national law took eect in January 1996 requiring all employers to register their employees. This followed pilot programmes in four oblasts (regions). The contribution is paid by the employer for those in employment, by the individual for those in self-employment or without employment and for the socially protected from the state budget. Because the state sector is still large, it has proved relatively easy to register a large part of the workforce. By the middle of the year around 80% of employees had been registered across the country. A weak economy has made it much more dicult to collect contributions. Between 20 and 30% of expected revenues had been collected from enterprises for those covered by insurance with considerable variation across the country. Collections varied from as low as 9% in 1996 to as high as 52% in Almaty city. Income of the oblast explains some, but not all, the variance. Just as important is the individual management ability of each oblast fund and the extent to which they are supported by the local tax collection authorities (Ministry of Health, 1996). It is too early to say how large the uncovered population will be. Early indications are that collections will be con®ned largely to state and large private companies. Although the system is compulsory in theory most of those in small private ®rms or self-employed can avoid paying contributions just as they avoid paying taxes. At present, the self-employed are required to pay a ¯at rate contribution although registrations are currently few. One of the problems is that the perceived bene®ts are unclear, or of little value. Most people already pay informal charges or contribute inkind to their own treatment while formal charges are few. Anecdotal evidence suggests that many expect that they will continue to pay informally whether or not they are insured. This impression is likely to prove dicult to dispel. At present around 25% of the population are outside the formal workforce and do not qualify as socially protected. These include the unregistered unemployed, non-working dependants, self-employed and employees in small non-juridical enterprises.
874
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Other FSU republics are currently considering insurance. Kyrgyzstan has a law but implementation has been slow due to poor results in a pilot oblast. Proposals for a personal health account have been advanced in Uzbekistan although legislation has yet to be prepared. In Turkmenistan a system of voluntary insurance was introduced in January 1996. This covers 90% of the cost of outpatient prescriptions as well as a number of other items such as discounts on dentures. Although popular, the scheme has also proved expensive with far more paid out than collected in contributions. There are proposals for a national system of compulsory health insurance modelled on the system in Kazakstan and the Russian Federation. Preconditions for social insurance and lessons from Latin America The factors making the development of insurance more or less feasible and appropriate are complex. In order to analyse the current context in transitional Asia, it is instructive to examine the development of insurance in Latin America since the end of the First World War. Several aspects of insurance development in Latin America make such a comparison relevant. Insurance was introduced for a geographically proximate, yet heterogeneous group of countries. Latin American countries share certain common characteristics. These include closeness of language and some economic inter-dependence. Most fall into low and upper middle income groups. During the 1980s, many suered substantial and prolonged economic recession. Real GDP in Venezuela, for example, declined by around 20% while in Peru it fell by more than 30%. Many South American and Caribbean countries adopted, on mass, systems of social security based on enterprise contributions. These include pension, unemployment and maternity entitlements in addition to health care bene®ts. Coverage has been extended in the classic way from civil and other public servants, through white and then blue collar industrial workers and later to other urban and then rural workers. The most advanced countries in this respect are Argentina, Brazil and Costa Rica, with 75 to 100% coverage, followed by countries as such as Uruguay and Chile (Mesa-Lago, 1991). Cuba automatically entitles all citizens to health service provision, although it retains a contributory social security approach for other bene®ts. In Latin America the institution acting as the major impetus to insurance development was the International Labour Organisation (ILO). In transitional Asia there has been no one organisation that has actively promoted social insurance. The develop-
ment has been spurred by a combination of factors including the rapid development of Bismarkian models in Eastern Europe (Marree and Groenewegen, 1997) and a need to obtain additional funding during deep economic recession. The World Bank, of all the major donors, has probably had the strongest in¯uence over ®nancing reforms. This has, however, usually been in the form of assistance in the development of best practice for new ®nancing rather than initial promulgation of insurance. The World Health Organisation has, similarly, been involved in helping to develop nascent systems and sometimes even attempting to dissuade countries from developing the system in the ®rst place. The factors that help or hinder the development of a social security approach to health system funding can be divided into three main groups; features associated with transition, structural characteristics of a country and country-speci®c pre-disposing factors. Transition features Transition characteristics are those features that are common to countries undergoing economic change. They are exhibited by most transition countries and may impede the development of new tax structures and social security systems required for the development of public services such as health care. A number of trends can be observed. 1. The process of economic transition has a profound eect on the structure of the economy and employment patterns. State industries are privatised, downsized and closed. A consequence is that employment in public sector enterprises is declining while small private employment is growing. A feature of this process is that much of this new workforce is `informal' in sense of being unregistered for taxation and regulatory purposes. This restructuring has important consequences for raising revenue. 2. Economic transition has a substantial impact on the ability of public providers to deliver aordable health services to paying or insured patients. Increased employment opportunities in these transforming economies makes it progressively more dif®cult to retain a skilled workforce in the health sector. For example, many rural health workers in China have left posts for more lucrative employment (Lok San Ho, 1995). Additionally, the process of creeping privatisation of public facilities leaves patients funded by modest insurance fund payments far less attractive than those willing to pay larger direct payments (both formal and informal). So, for example, in Vietnam it is reported that public hospitals prefer to take fee paying patients and neglect those funded only by compulsory insurance (Thompson, 1995). In Almaty one newspaper survey
T. Ensor / Social Science & Medicine 48 (1999) 871±879
found that almost all patients make additional payments to doctors in order to receive prompt attention or a doctor of choice (Caravan, 1996). Changes in the wider economy, such as greater opportunities for private work in and outside the health sector, are making it more dicult for public health facilities to deliver the bene®ts promised. 3. A feature common to former soviet and East European countries, but not to other Asian transitional economies, has been deep economic recession. Rana (1995) has suggested some of the reasons why this dierence is observed. The includes a much greater dependence on (inecient) heavy industry and a large state sector as well as a much greater level of interdependence between economies, making enterprise restructuring more dicult.Recession makes taxes, social security and insurance contributions much harder to collect. Mesa-Lago (1991) argued that while the Latin America and Caribbean region as a whole would have required growth in GDP exceeding 7% in order to reduce the informal sectors signi®cantly and increase insurance coverage, the actual negative rates of growth made this impossible. Although the need for health care does not diminish and probably increases, the resources available to provide a comprehensive insured package declines. Indeed those most requiring health and other social protection may be unable to participate fully in the social insurance system because they lack formal employment. Systems of provision in Latin America have been criticised because they do not provide coverage for those in extreme poverty (Camcho, 1996). 4. One of the most fundamental diculties is that while many countries have increased coverage and health service access through insurance, transition countries appear to reduce access. The justi®cation for doing so is clear. Although the concept of universality remains, in practice it is a ®ction. Therefore it might be more honest and lead to more eective services if a minimal package of services were oered to all with additions funded from contribution related insurance. Yet there are immense political rami®cations involved in overturning a long history of `free' services. There is also a suspicion that if insurance is introduced then the bene®ts will mean nothing, people may go on paying informally for services just as they did under the old system. In these circumstances, there is little to be gained from voluntarily joining an insurance scheme. The result is that those that are able to pay directly can avoid the new tax/contribution and the overall yield is eroded. The entitlement problem highlights perhaps the most fundamental dierence between the experiences of
875
these two country groups. In Latin America social insurance expanded entitlement to services free at the point of delivery funded on the basis of income related (rather than risk rated, as with private insurance) contributions. In transitional Asia most countries begin from a universal entitlement based on citizenship. Payroll insurance has two possible implications. One is to dis-entitle people from universal coverage and demand that they re-register through their employment, purchase individual cover or qualify as socially protected to be covered from the public budget. The second possibility is to retain universal entitlement and to treat the payroll contribution as a tax on those unfortunate enough to work for enterprises liable (in theory and fact) for payment. However, in this case, there is little incentive to pay the contribution since it provides no additional bene®ts. Structural characteristics Structural characteristics are features of a country that are important in determining the feasibility of collecting revenues through a payroll tax that can be expected to change slowly, if at all. The proportion employed in the industrial sector is a good indication of the ease of registration since most enterprises will have a formal payroll system from which payroll and other taxes will already be paid. Similarly, the size of the urban population and population density will also contribute to the administrative cost of the system. If the population is largely rural and widely scattered the registration and assessment of contributions is made more dicult than if the population is concentrated in urban areas. Finally, the income of the country and growth rate of incomes are indicators of how easy it will be for enterprises and individuals to pay insurance contributions. A growing, prosperous economy will easily absorb contributions which an economy in recession ®nds impossible to pay. Structural factors have been important in determining the speed of social security development in Latin America and the Carribean (Mesa-Lago, 1996). Countries with the most well established systems, such as Argentina, Uruguay and Costa Rica, also have the highest proportion of the population living in urban areas and working in the industrial sector. Countries which established systems later and more slowly tend to have a larger proportion of the workforce living in rural areas (coverage ranges from 14 to about 35%). In general, the results from this and other studies (Jenkins, 1993) suggest that it is very dicult to extend overall coverage beyond the sum of formal urban and modern rural sectors. Using the quantitative indicators described above and following a similar but simpli®ed analysis carried out by Shaw and Ainsworth (1996) for Sub-Saharan
876
T. Ensor / Social Science & Medicine 48 (1999) 871±879 Table 2 Feasibility of introducing insurance in low and middle income countries Score
Countries
3, 4 2
Trinidad and Tobago, South Korea, Mexico, Belarus, Estonia Argentina, Barbados, Uruguay, Chile, Brazil, Czech Republic, Hungary, South Africa, Latvia, Lithuania, Bulgaria, Cuba, Ukraine Slovenia, Saudi Arabia, Malaysia, Panama, Poland, Russian Federation, Colombia, El Salvador, Jamaica, Romania, Philippines, Moldova Turkey, Costa Rica, Peru, Dominican Republic, Ecuador, Morocco, Armenia, Georgia, Kazakstan, Uzbekistan Botswana, Namibia, Paragua, Indonesia, Bolivia, Egypt, Sri Lanka, Albania, Azerbaijan, Kyrgyzstan, China, Pakistan, Ghana, Tajikistan, India Belize, Thailand, Turkmenistan, Senegal, Cote d'Ivoire, Mongolia, Nigeria, Kenya, Bangladesh, VietNam, Burundi Lesotho, Guinea, Zimbabwe, Gambia, Zambia, Guinea-Bissau, Zaire, Uganda, Malawi, Tanzania, Ethiopia, Cambodia, Lao PDR, Niger, Sudan
1 0 ÿ1 ÿ2 ÿ3, ÿ4
Transition countries in italics.
countries, a group of low and middle income countries were ranked according to a composite index of all four variables (population density, percent of the population urbanised, per cent of the workforce working in industry and per capita income)2. A ranking near the top of the list suggests that insurance is relatively easy to implement while a ranking towards the bottom suggests that it would be extremely dicult. Many of the countries with a high level of coverage in Latin American and Asia, such as Brazil, Argentina and Korea, are at the top of the list (see Table 2). Of the transition economies, countries such as Estonia, Czech Republic and Russian Federation with a high concentration of the population in the urban industrialised sector are near the top of the list. Further down is Kazakstan, Kyrgyzstan, Tajikistan and Turkmenistan. At the bottom of the list are very poor African and Asian countries such as Cambodia, Burundi and Uganda. While transition features are important in delaying the process of change, structural factors may be enough to block particular patterns of social security development altogether. Estonia found that the development of insurance in the early stages was impeded but that improving economic circumstances, combined with good structural preconditions, now permit the system to develop quite eectively (see Laur, 1997). 2 For each indicator the value for each country was compared to the mean for all countries. If the value was more than one standard deviation above the mean it scored one, if more than one below it scored minus one and if somewhere in between a score of zero. The scores for each indicator were added together to give a composite score.
It is notable that a number of the Asian transition countries introducing schemes, notably Turkmenistan, Vietnam and Kazakstan, are towards the bottom of the list, suggesting that they will ®nd it dicult to extend the scheme to the majority of their population. Kazakstan, for example, is ranked alongside Ecuador which has achieved around 23% social security coverage (Mesa-Lago, 1996). Turkmenistan and Vietnam are ranked below Bolivia (18% coverage) and alongside countries such as Kenya which has managed to achieve only about 25% coverage. Country speci®c features Apart from general transition and structural factors, features that are country-speci®c are also likely to be important. One is the extent to which insurance funds are well managed which has several dimensions. First are systems for preventing and dealing with fraud. An advantage of the restrictive system of line funding and largely cash free transactions was that the scope for fraudulent activity was quite limited since few people had access to substantial cash balances. A worry expressed in a number or transition countries is that insurance funds not included in general government budget setting often lack a strict system of line control but also have not developed modern accounting systems for audit common in the West. Several transition countries have reported scandals at the national or regional level where poorly regulated fund directors have stolen, or made personal use of, fund balances. A second management issue is where funds, although not used inappropriately are not managed eciently. This may arise in a number of ways.
T. Ensor / Social Science & Medicine 48 (1999) 871±879
Administrative issues There are technical problems related to fund administration (see Normand and Weber, 1994). These include reserves that are not invested in way that ensures a high hedged rate of return, an undeveloped collection system so that the rate of collections is low and excessive levels of administration relative to the size of collections (a particular problem where the total number of enrollers is low). Financial management More fundamentally, many systems are based on inadequate calculations relating to the expenditures and revenues to be generated from a given package of insured bene®ts. The result is that bene®ts promised far exceed contributions obtained. This is a very common problem with the `pay as you go' payroll tax systems being introduced in many FSU countries. Part of the problem arises because of a historic commitment to universal services. Another reason is that the contribution is seen as a tax that must be paid rather than a payment for services rendered (in this case the insured risk rather than the medical services themselves). Another facet of this problem is where extravagant bene®ts are promised in order to increase enrolment. In Uruguay, for example, (pension) bene®ts were permitted to rise faster than contributions. The consequence was the government was unable to maintain the size of bene®ts, the system got a bad reputation and people lost con®dence in its operation (Camcho, 1996). Ineective purchasing A key potential role for an insurance fund introduced into a predominantly integrated health care system (as in much of transitional Asia) is as an independent purchaser-commissioner of health care. This role includes the assessment of treatment, type of services and best-value providers. Yet this role is often underdeveloped for several reasons. One reason is that the insurance fund is seen primarily as a contribution (or tax) collector and little thought is given to the disbursement role. Insurance funds in those countries have primarily been concerned with collecting money from enterprises rather than disbursing to providers. It is not surprising that most of the statistics published report numbers enrolled and contributions collected but not the way money is spent. A second issue relates to a confusion about the precise role of the fund. Should the insurance fund provide for the (constrained) demands or wants of individuals or, alternatively, provide for their needs. The former model is closer to a private insurer that provides limited access for a range of bene®ts desired by the individuals insured. In the later case, the fund
877
will decide, on a population basis, what services best add to the health of all those insured. This model is closer to the purchaser role of, say, a UK health authority or Swedish county. Re®ning the insurance model in transition economies The foregoing discussion suggests that there is considerable diversity across transitional economies. The characteristics of transition, changing structure of the workforce, economic recession and declining public sector expenditure, are common features present across the region from Poland to transitional Asia. Longer term structural features which make it more or less dif®cult to extend social insurance vary enormously. The current characteristics of most Asian transition economies suggest that the model of social insurance will, on its own, fail to provide comprehensive insurance cover. As currently conceived, insurance provides some marginal additional income for the health sector. Those covered still have to pay to receive treatment and there is therefore little incentive to join or pay promptly. Yet it is too early to judge whether the schemes will, in time, develop to oer more comprehensive coverage. There are three main directions in which schemes might develop. The ®rst is to recognise the current reality: that the new insurance schemes are basically earmarked payroll taxes oering little in the way of additional service entitlement. As the economy strengthens in each country this extra source of revenue will grow to form a useful addition to the health service budget. In time the system might lead to comprehensive social security or remain a universal tax funded system. In the short term there is no need for an additional supporting infrastructure (health fund, local oces, overall management board) to collect what amounts to another tax. A consequence is that the insurance system adds little value to the health care system other than the collection of often marginal payroll taxes. If this is the case it is hard to see how the creation of a separately organised insurance bureaucracy can be justi®ed. The second model, the Latin American paradigm, is to view social health insurance as one part of a comprehensive system of social security that is managed by a small number (perhaps only one) of national employment related agencies. Social security can be provided as a package through a combination of `pay as you go' entitlement, more appropriate for health care and some other bene®ts and funded (capitalised) systems for long term bene®ts such as pensions. For transition countries such systems would require a radical change in the way in which social bene®ts are provided. There is evidence that this is being considered in some
878
T. Ensor / Social Science & Medicine 48 (1999) 871±879
countries. There has been discussion in Kazakstan, for example, over the merging of the insurance and social security fund that provides sickness, unemployment and maternity bene®ts. The third model is to view the insurance fund as an independent public purchaser of health care for a de®ned population. This function is made easier if the enrolled population is based on a geographic area rather than employment group. This is certainly true in the case of the developing systems in Central Asia and VietNam. It is true of the rural schemes in China but not of the urban schemes which are still predominantly based on employment groups. In most cases, this has not been the model adopted in Latin America. An exception to this is Brazil which has begun to move towards a system of universal entitlement based on geographic based local purchasers. Until the 1980s it had a well developed system of employment based cover supported by private schemes for the rich and government provision for the very poor. In the late 1980s Brazil introduced a uni®ed system (Kutzin and Barnum, 1992) whereby public funding, from insurance and taxation, was uni®ed giving all citizens of Brazil access to the same providers. Only the private insurance funding remains separate. Large inequalities do, however, remain. This is largely the result of massive disparities in the level of resources available in each state and has promoted a strong interest in methods of geographic resource allocation similar to the RAWP formula used in the UK. Independent planning and commissioning of care, if carried out properly, would have substantial bene®ts. There is growing evidence that much medical care provided in transition countries does not represent good value for money. There is a growing trend of induced demand for pharmaceuticals and diagnostic tests. Many of these services are provided simply to generate revenue and are medically inappropriate (see Ensor, 1998 for a summary of experience). In addition many treatments routinely used in the past are ineective for many of their current uses. A conference in Almaty on quality in health care drew attention to two treatments that are routinely used: hyperbaric oxygenation and low frequency laser therapy (ZdravRefom, 1996). The evidence for both these treatments for many of their current uses is poor, yet hospitals continue to purchase such technology in order to attract paying patients. A role for the insurance fund is to distinguish between treatments that do not work and those that do and to ensure that those that do work are obtained at low cost. These are not easy tasks but are important to develop if the insurance funds are to become more than additional tax collectors. The fundamental issue of equity in the ®nance of services is how those outside the workforce obtain health services.
If the ®rst option is chosen then it is presumed that this will be achieved in a similar way as for those in employment. The issue of ineective treatments and eective treatment provided ineciently still remains and must be addressed through the reform of funding disbursement mechanisms within an integrated health service. A possible variant here is to follow the model favoured by a number of OECD countries and institute a purchaser provider split at the local health department level. The second option requires speci®c identi®cation of those unable to contribute to insurance, who are explicitly covered from the budget, and others who are not in waged employment but can aord to contribute. However, is it is clear from Section 4 that this cannot be achieved through compulsory payroll taxes. Instead, the system might evolve to include complementary systems of funding such as voluntary insurance, that provide valued medical bene®ts for those buying cover. There may be lessons from the experience of community ®nancing in Africa and other parts of Asia (see Shaw and Ainsworth, 1996). This suggests that local communities can be persuaded to buy limited and very well de®ned bene®ts through the purchase of voluntary ®xed premium cards, provided that bene®ts are attractive and delivered when required. Management capability is also required. Evidence suggests that where management support and local motivation is high, schemes are more likely to succeed (Carrin and Vereecke, 1992). In the third option, local health fund-purchasing could follow several routes. One is to maintain, or return, to a universal funding system. Compulsory payroll taxes supplement resources derived from general taxation. In this paradigm it makes sense for the tax service to collect contributions although there may be some suspicion in the health fund that insucient attention is given to health payroll contributions compared to other taxes. The only advantage of the health fund is as a purchaser of care and it is important that this role is developed. The alternative route is for local funds to develop a contribution basis for entitlement. In this case, as with the second option, other ways for covering those outside the salaried sector and socially protected groups must be found. Conclusion A strength of most former centrally planned health systems was that almost universal access to health facilities was ensured. In terms of basic indicators of good primary curative and preventive health care these systems scored well. The central weakness was that the funding system tended to encourage inecient patterns of hospital provision.
T. Ensor / Social Science & Medicine 48 (1999) 871±879
It is currently unclear how the new systems of insurance will develop in most Asian Transition Economies. A number of paths are possible. The main fear is that progressing too far along some type of contribution based system will jeopardise universal access and substitute one inecient funding system (funding according to numbers of sta and bed days) for another (fee for service from the insurance fund). If social insurance is the preferred option, the insurance agency must be capable of bringing about changes that ensure and increase access for those in rural areas and promotes better use of resources within the hospital sector. In pursuing these objectives, transition countries have much to learn, positive and negative, from countries with established insurance systems. Acknowledgements I am grateful to the conference organisers and an anonymous referee for comments on a draft of this paper. References Camcho, L., 1996. Financing social security in Latin America: new perspectives in the light of current developments. International Social Security Review. Caravan, 1996. If you don't grease you won't go. Caravan National Newspaper. Almaty, April 19. Carrin, G., Sergent, F., Murray, M., 1993. Towards a Framework for Health Insurance Development in Hai Phong, Vietnam. Macroeconomics Health and Development Series, Technical Paper 12. World Health Organisation, Geneva. Carrin, G., Vereecke, M., 1992. Main lessons from the practice of community ®nancing schemes. In: Carrin, G. (Ed.), Strategies for Health Care Finance in Developing Countries. Economic Issues in Health Care, Macmillan. Ensor, T., 1995. Introducing health insurance in Vietnam. Health Policy and Planning 10 (2), 154±163. Ensor, T., 1998. Health sector reform in Asian transition countries. In: Brooks, D., Thant, M. (Eds.), Social Sector Issues in Asian Transitional Economies. OUP, Hong Kong, in press. Ensor, T., Thompson, R., 1998. Health insurance as a catalyst to change in former communist countrues? Health Policy, in press.
879
Jenkins, M., 1993. Extending social security protection to the entire population: problems and issues. International Social Security Review 46 (2), 3±20. Kutzin, J., Barnum, H., 1992. Institutional features of health insurance programs and their eects on developing country health systems. International Journal of Health Planning and Management 7, 51±72. Laur, P., 1997. Reforming the health service in Estonia. In: Sorenson, J. Rudbeck, B., Bilde, L. (Eds.), Reforming the Health Services in the Baltic Countries. Danish Institute for Health Services Report 97.08. Lok San Ho, 1995. Market reforms and China's health care system. Social Science and Medicine 41 (8), 1065±1072. Marree, J., Groenewegen, P.P., 1997. Back to Bismark: Eastern European Health Care Systems in Transition. Ashgate, Aldershot. Mesa-Lago, C., 1991. Social security in Latin America and the Caribbean: a comparative assessment. In: Ahmad, E., Dreze, J., Hills, J., Sen, A. (Eds.), Social Security in Developing Countries. Oxford. Mesa-Lago, C., 1996. Comparative study of the development of social security in Latin America. International Social Security Review 2, 127±151. Ministry of Health, 1996. Evaluation of Medical Insurance in the Republic of Kazakstan. Almaty. Normand, C., Weber, A., 1994. Social Health Insurance: a Guidebook for Planning. WHO and ILO. Rana, P.B., 1995. The Asian approach to reforming transitional economies. In: Rana, P.B., Hamid, N. (Eds.), From Centrally Planned to Market Economies: the Asian Approach, vol. 1. Oxford. Shaw, R.P., Ainsworth, M., 1996. Financing Health Services though User Fees and Insurance. World Bank Discussion Paper 294. The World Bank. Thompson, R., 1995. Extending the scope of health insurance in Vietnam. M.Sc. dissertation (unpublished). UNDP, 1995. Human Development Report. United Nations Development Programme, Oxford. World Bank, 1995. Social Indicators of Development. John Hopkins. World Bank, 1996. Republic of Kazakstan: Transition of the State, vol. I. World Bank. Yang, B., 1996. The role of health insurance in the growth of the private health sector in Korea. International Journal of Health Planning and Management 11, 231±252. Young, M., 1989. Impact of the rural reform on ®nancing rural health services in China. Health Policy 11, 27±42. ZdravRefom, 1996. Eectiveness in health care: laser therapy and hyperbaric oxygen. In: Conference Notes for National Conference on Assuring Quality of Care, Almaty, 15±16 May.