《卫生经济学》课程参考文献：HealthInsuranceLessonGermany

SOCIAL SCIENCE MEDICINE PERGAMON Social Science Medicine 54(2002)1559-158 www.elsevier.com/locate/socscimed One hundred and eighteen years of the german health insurance system: are there any lessons for middle- and low-income countries? Till Barnighausen* Rainer Sauerborn Department of Tropical Hygiene and Public Health, Medical School, University of Heidelberg, INF 324, 69120 Heidelberg, Germany Abstract A number of low and middle income countries(LMICs)are considering social health insurance(SHD) for adoption into their social and economic environment or striving to sustain and improve already existing SHI schemes. SHI was first introduced in Germany in 1883. An analysis of the German system from its inception up to today may yield lessons relevant to other countries. Such an analysis. however, is largely lacking, especially with regard to LMICs. This paper attempts to fill this gap For each of the following lessons, it considers if and under which conditions they may be of relevance to LMICs. First, small, informal, voluntary health insurance schemes may serve as learning models for fund administration and solidarity, but in order to achieve universal coverage government action is needed to formalise these schemes and to introduce a principle of compulsion. Once compulsory health insurance exists for some people. incremental expansion of coverage to other regions and social groups may be feasible to achieve universality. Second, in order to ensure sustainability of SHi, the mandated benefit package should be adapted incrementally in accordance with changing needs, values and economic circumstances. Third, in a pluralistic SHI system equity, as well as risk pooling and spreading, can be enhanced if funds merge. The optimal number of funds. however, will depend on the stage of development of the Shi system as well as on other objectives of the system, including choice and competition a risk equalisation scheme may prevent the adverse effects of risk selection, if competition between insurance funds is introduced into the system. Fourth, as an alternative to both state and market regulation, self-governance may serve as a source of stability and sustainability as well as a means of decentralising and democratising a health care system inally, costs can be successfully contained in a fee-for-service system, if cost-escalating provider behaviour is onstrained by either political pressure or technical means. C 2002 Elsevier Science Ltd. All rights reserved Keywords: Health care reform; Health care financing: Social health insurance: Universal coverage: Benefit package; Risk equalisation; Self-governance; Cost containment: Low and middle income countries; Germany Introduction (Tangcharoensathien, Supachutikul, Lertiendumror 1999: Nitayarumphong Pannarunothai, 1998; Khe A number of low and middle income man, 1997), Viet Nam (Ensor, 1999), Indonesia, (LMICs) are considering social health insurane Philippines, Bangladesh (Tan, 1998; DSE: SHINE/ for adoption into their political and economic GTZ, 1998), South Korea (Yang, 1995: Shin, 1996) ment or striving to sustain and improve already existing Kazakstan (Ensor, 1999), Russia(Sheiman, 1995), SHI schemes, e.g. China (World Bank, 1997), Thailand Bosnia, Romania (The InterHealth Institute, 1998) Hungary(Donaldson Gerad, 1993: Deppe& Ores- orresponding author. Tel: +49-6221-56-53-44: fax:+49. kovic, 1996), the Czech Republic (Deppe Oreskovic 6221-56-594 1996) E-mailaddresstill-baernighausen(@yahoo.com The main reasons for choosing SHI as the method of health care financing are that SHi can provide a stabl 0277-9536/02/Ssee front matter C 2002 Elsevier Science Ltd. All rights PI:s0277-9536(01)00137X

Social Science & Medicine 54 (2002) 1559–1587 One hundred and eighteen years of the German health insurance system: are there any lessons for middle- and low-income countries? Till Barnighausen*, Rainer Sauerborn . Department of Tropical Hygiene and Public Health, Medical School, University of Heidelberg, INF 324, 69120 Heidelberg, Germany Abstract A number of low and middle income countries (LMICs) are considering social health insurance (SHI) for adoption into their social and economic environment or striving to sustain and improve already existing SHI schemes. SHI was first introduced in Germany in 1883. An analysis of the German system from its inception up to today may yield lessons relevant to other countries. Such an analysis, however, is largely lacking, especially with regard to LMICs. This paper attempts to fill this gap. For each of the following lessons, it considers if and under which conditions they may be of relevance to LMICs. First, small, informal, voluntary health insurance schemes may serve as learning models for fund administration and solidarity, but in order to achieve universal coverage government action is needed to formalise these schemes and to introduce a principle of compulsion. Once compulsory health insurance exists for some people, incremental expansion of coverage to other regions and social groups may be feasible to achieve universality. Second, in order to ensure sustainability of SHI, the mandated benefit package should be adapted incrementally in accordance with changing needs, values and economic circumstances. Third, in a pluralistic SHI system equity, as well as risk pooling and spreading, can be enhanced if funds merge. The optimal number of funds, however, will depend on the stage of development of the SHI system as well as on other objectives of the system, including choice and competition. A risk equalisation scheme may prevent the adverse effects of risk selection, if competition between insurance funds is introduced into the system. Fourth, as an alternative to both state and market regulation, self-governance may serve as a source of stability and sustainability as well as a means of decentralising and democratising a health care system. Finally, costs can be successfully contained in a fee-for-service system, if cost-escalating provider behaviour is constrained by either political pressure or technical means. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Health care reform; Health care financing; Social health insurance; Universal coverage; Benefit package; Risk equalisation; Self-governance; Cost containment; Low and middle income countries; Germany Introduction A number of low and middle income countries (LMICs) are considering social health insurance (SHI) for adoption into their political and economic environ￾ment or striving to sustain and improve already existing SHI schemes, e.g. China (World Bank, 1997), Thailand (Tangcharoensathien, Supachutikul, & Lertiendumrong, 1999; Nitayarumphong & Pannarunothai, 1998; Kho￾man, 1997), Viet Nam (Ensor, 1999), Indonesia, Philippines, Bangladesh (Tan, 1998; DSE; SHINE/ GTZ, 1998), South Korea (Yang, 1995; Shin, 1996), Kazakstan (Ensor, 1999), Russia (Sheiman, 1995), Bosnia, Romania (The InterHealth Institute, 1998), Hungary (Donaldson & Gerad, 1993; Deppe & Ores￾kovic, 1996), the Czech Republic (Deppe & Oreskovic, 1996). The main reasons for choosing SHI as the method of health care financing are that SHI can provide a stable *Corresponding author. Tel.: +49-6221-56-53-44; fax: +49- 6221-56-59-48. E-mail address: till baernighausen@yahoo.com (T. Barnighausen). . 0277-9536/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0277-9536(01)00137-X

1560 source of revenues, a visible flow of funds into the Incremental achievement of universal coverage health sector, and a combination of risk pooling ith mutual support. In addition, a SHI scheme Developments in health care systems can be cate- can be established independently from the gorised according to their scale and the pace at which government, while it can nevertheless operate in line they occur on a continuum between transformational with government health policy. Disadvantages of SHl, and incremental change. Unlike many other health care on the other hand, comprise problems with insuring systems(such as the British NHS), the formation of th informal sector workers and a lack of cost control German system has been characterised by incremental (Normand Weber, 1994) changes and adjustments during both its nascent and its While the experience of Latin American more mature stages. in developing SHI systems has been with respect to transitional economies(Ensor Small, voluntary, informal risk-sharing schemes as the it is striking that literature analysing the lessons starting point learnt from the evolution of the German system d their potential relevance to LMICs is lac Statutory sickness funds evolved out of the relief With this paper, we intend to fill this gap. We will funds that had originated as solidarity-based do so by analysing the long-term evolution of the support systems within the medieval guilds. Since the German health insurance system and examining what, if end of the 17th century, five types of relief funds had EAy. lessons can be derived for the design of SHI in developed in different regions of Germany: relief funds for journeymen, relief funds for craftsmen modelled As Germany has the world's oldest SHI system, it after the mutual support systems of the guilds, factory naturally lends itself to historical analyses. Yet, although relief funds founded by socially-oriented entrepreneurs, the value of diachronic analyses that trace the paths relief funds founded by local authorities for workers or of an insurance systems evolution over time is increas- trades people and community relief funds for people ingly recognised (Field, 1999), most English language who could not otherwise find insurance(Zorn, 1912: articles about the German health care system take a Peters, 1978) synchronic perspective, providing a snapshot, rather When Bismarck reformed the german health care than a longitudinal, overview of the system as it exists at system in 1883, the policies he needed to implement to moment(see, for instance, Brown Amelung, 1999: establish a comprehensive, social insurance system Jost, 1998; Wahner-Roedler, Knuth, Juchems, 1997; were-compared to their far reaching consequences- Lassey, Lassey, Jinks, 1997: Roy, 1993: Von der of relatively m The Schulenburg, 1992: Reinhardt, 1990). The exceptions to work of a century was, in fact, an incremental this rule tend to concentrate on one-often technical rather than a transformational change of structures aspect of the system(for instance, Busse Schwartz, already in existence. The law of 1883 built on,first 1997; Henke, Murray, Ade, 1994; Kirkmann-Lift, experiences gained in the administration regional relief 1990)or confine themselves to a background chronology funds and, second, social change brought about by without working out the more general implications membership in the funds(Caron, 1882; Peters, 1978; flowing from history(for example, Altenstetter, 1999: Abel-Smith, 1992; Herder-Dorneich, 1994). Administra Iglehart, 1991). If one leaves the more short-term tively, the voluntary relief funds had served as an political and statistical vagaries behind and chooses apprenticeship stage for the development of skills in instead to look-where meaningful-with the widest insurance administration and actuarial science at the angle lens that history permits, some sequential level of the fund as well as in insurance regulation at the dynamics and cumulative effects otherwise hidden level of government. More specifically, the basic become apparent which are of potential relevance to principles under which Bismarck's system was to operate the context of lmics had already been tried and proven to work in its B In the following, we will examine how during this numerous, regional predecessors: (1)The support funds history up to date ( universal coverage was achieved, were largely self-governed. (2)Both employers and equal access to a comprehensive benefit package was employees were represented in the bodies of self- established, (ii) equity in financing was improved, (iv) governance in most of the company-based funds. (3) consumer choice and competition were introduced into Company-based funds were financed in part by employ the system, (v) sustainability was ensured and (vi) costs ers, in part by employees. (4)Compulsory insurance had were contained. We will focus our analysis on the mode already been introduced in many municipalities(Alber, of development and the institutional arrangements For 1992) each question analysed, we will consider whether the experiences from the german case may be of use to the In 1876, 869204 people were insured in 5239 officially contexts of lmics recognised regional sickness funds(Peters, 1978)

source of revenues, a visible flow of funds into the health sector, and a combination of risk pooling with mutual support. In addition, a SHI scheme can be established independently from the government, while it can nevertheless operate in line with government health policy. Disadvantages of SHI, on the other hand, comprise problems with insuring informal sector workers and a lack of cost control (Normand & Weber, 1994). While the experience of Latin American countries in developing SHI systems has been analysed with respect to transitional economies (Ensor, 1999), it is striking that literature analysing the lessons learnt from the evolution of the German system and their potential relevance to LMICs is lacking. With this paper, we intend to fill this gap. We will do so by analysing the long-term evolution of the German health insurance system and examining what, if any, lessons can be derived for the design of SHI in LMICs. As Germany has the world’s oldest SHI system, it naturally lends itself to historical analyses. Yet, although the value of diachronic analyses that trace the paths of an insurance system’s evolution over time is increas￾ingly recognised (Field, 1999), most English language articles about the German health care system take a synchronic perspective, providing a snapshot, rather than a longitudinal, overview of the system as it exists at moment (see, for instance, Brown & Amelung, 1999; Jost, 1998; Wahner-Roedler, Knuth, & Juchems, 1997; Lassey, Lassey, & Jinks, 1997; Roy, 1993; Von der Schulenburg, 1992; Reinhardt, 1990). The exceptions to this rule tend to concentrate on oneFoften technical￾Faspect of the system (for instance, Busse & Schwartz, 1997; Henke, Murray, & Ade, 1994; Kirkmann-Liff, 1990) or confine themselves to a background chronology without working out the more general implications flowing from history (for example, Altenstetter, 1999; Iglehart, 1991). If one leaves the more short-term political and statistical vagaries behind and chooses instead to lookFwhere meaningfulFwith the widest￾angle lens that history permits, some sequential dynamics and cumulative effects otherwise hidden become apparent which are of potential relevance to the context of LMICs. In the following, we will examine how during this history up to date (i) universal coverage was achieved, (ii) equal access to a comprehensive benefit package was established, (iii) equity in financing was improved, (iv) consumer choice and competition were introduced into the system, (v) sustainability was ensured and (vi) costs were contained. We will focus our analysis on the mode of development and the institutional arrangements. For each question analysed, we will consider whether the experiences from the German case may be of use to the contexts of LMICs. Incremental achievement of universal coverage Developments in health care systems can be cate￾gorised according to their scale and the pace at which they occur on a continuum between transformational and incremental change. Unlike many other health care systems (such as the British NHS), the formation of the German system has been characterised by incremental changes and adjustments during both its nascent and its more mature stages. Small, voluntary, informal risk-sharing schemes as the starting point Statutory sickness funds evolved out of the relief funds that had originated as solidarity-based support systems within the medieval guilds. Since the end of the 17th century, five types of relief funds had developed in different regions of Germany: relief funds for journeymen, relief funds for craftsmen modelled after the mutual support systems of the guilds, factory relief funds founded by socially-oriented entrepreneurs, relief funds founded by local authorities for workers or trades people and community relief funds for people who could not otherwise find insurance (Zorn, 1912; Peters, 1978). When Bismarck reformed the German health care system in 1883, the policies he needed to implement to establish a comprehensive, social insurance system wereFcompared to their far reaching consequences￾Fof relatively minor immediate impact. The Bismarck￾ian ‘work of a century’ was, in fact, an incremental, rather than a transformational change of structures already in existence. The law of 1883 built on, first, experiences gained in the administration regional relief funds and, second, social change brought about by membership in the funds (Caron, 1882; Peters, 1978; Abel-Smith, 1992; Herder-Dorneich, 1994). Administra￾tively, the voluntary relief funds had served as an apprenticeship stage for the development of skills in insurance administration and actuarial science at the level of the fund as well as in insurance regulation at the level of government. More specifically, the basic principles under which Bismarck’s system was to operate had already been tried and proven to work in its numerous, regional predecessors:1 (1) The support funds were largely self-governed. (2) Both employers and employees were represented in the bodies of self￾governance in most of the company-based funds. (3) Company-based funds were financed in part by employ￾ers, in part by employees. (4) Compulsory insurance had already been introduced in many municipalities (Alber, 1992). 1 In 1876, 869 204 people were insured in 5239 officially recognised regional sickness funds (Peters, 1978). 1560 T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587

T. Barnighausen, R. Sauerborn Social Science Medicine 54(2002)1559-1587 Socially, the voluntary relief funds served at least two analyses start with a dichotomy between real insurance important functions: first, as an opportunity to build and informal risk sharing, but often end up emphasising trust between the management of a scheme and its that many similarities exist. In a study of the traditional participants, and second, as learning models for tontine systems in Africa, Lespes(1990) found that as solidarity, the functioning of which could be experienced the tontines grow they take on many of the formal first hand among people with similar social identities. characteristics of insurance. Platteau(1997). in a review Both a basic trust in risk-sharing schemes and an of the concepts underlying traditional risk sharing, starts understanding of solidarity eventually translated into by pointing out that traditional mutual aid schemes are increased willingness to participate in larger schemes. based on balanced or generalised recipro The German experience during this phase of the insurance is based on conditional reciprocity. Under evolution of the health insurance system suggests the both balanced and generalised reciprocity people expect. following lesson: over time, to receive as much from a scheme as they contributed; under balanced reciprocity there are tight Small, informal, voluntary health insurance schemes rules on how and when a return will be paid, und may serve as learning models for fund administration generalised reciprocity these rules are much looser. and solidarity, both of which will make introduction of Under conditional reciprocity, on the other hand, the larger, more formal, compulsory schemes an easier task. giver will only receive a return if she herself falls victim to the event she insured against by enrolling in the Whether such a development can be repeated in scheme, i.e, income is redistributed between the lucky LMICs may depend on how far the experiences with and the unlucky. Platteau ends his analysis by arguing voluntary schemes in other countries mirror those in that traditional risk-sharing schemes may be able to Germany. From the study of the rural cooperative health serve an insurance function if either some standard of insurance schemes in rural China since the 1960s, evidence balanced reciprocity is upheld or redistribution does exists that the development of good management practice take place, but is not visible to enrolees. Similarily as well as of a trust relationship between administrators Cashdan (1985) argued that generalised reciprocity and beneficiaries are crucial elements for the long-term could act in the same way as conventional insurance. survival of schemes(Feng, Tang, Bloom, Segall,& G In addition, under generalised reciprocity, those whe 1995: Liu, Hsiao, Li, Liu,& Ren, 1995; Khan, Zhy, lave gained most during a certain time period are Ling, 1996: World Bank, 1997). While the existence of expected to give most to those who lose most during the voluntary, non-profit health insurance schemes in itself same period-even if at some unspecified point in the does not guarantee that such developments take place, it future reciprocal action is expected. As such, a higher offers opportunities for learning, which may not be degree of solidarity is realised in the short term than in otherwise provided. The implicit danger is that, if the both risk- and community-rated insurance, in which scheme fails, the patch may be spoilt for more promising contributions are independent of the financial situation efforts at a later time (World Bank, 1997) of the individual. In Thailand. for instance. a number of In many African countries traditional risk-sharing voluntary community-based funds originally founded as schemes exist. For instance, anthropological studies non-insurance schemes to provide loans to mem from Burkina Faso (to be published elsewhere) have bers-have for some years successfully provided health revealed a dense network of traditional mutual aid care insurance coverage as well (Nitayarumphong organisations based on profession or on risks(such as Pannarunothai, 1998 funeral funds). It is intuitively tempting to compare Going back to the origins of the German funds for these schemes to the guild-based relief funds of the pre- mutual aid as they developed within the medieval Bismarckian era and to consider initiatives to integrate miners associations and guilds, especially before the health risk into the cover of these funds. It has, however, 16th century, generalised rather than conditional re- been argued that the development of SHI in 19th ciprocity stands out as the main principle. What is more, century Germany(and Europe) is unlikely to repeat it was under this principle that an understanding of itself in todays Africa. The first argument is that community self-help, social justice and solidarity was traditional informal risk-sharing schemes differ in logic developed, which later formed the conceptual and function from insurance(Criel, Van Dormael, the evolution of more formal insurance funds Lefevre, Menase, Van Leberghe, 1998) Dorneich, 1994). Thus, the European and The different logics underlying informal risk histories of risk sharing, though separated by time d insurance (as understood in developed to be related ir economies)have been analysed from different perspec- The second argument points to the fact that where tives, notably anthropological and economic(see for voluntary health insurance schemes exist, they are instance, Platteau, 1997; Besley, 1995; Coate Ravail mostly initiated by agencies external to African society. lon, 1993; Lespes, 1990: Cashdan, 1985). As a rule, these namely foreign NGOs, and thus lack the dynamic of an

Socially, the voluntary relief funds served at least two important functions: first, as an opportunity to build trust between the management of a scheme and its participants, and second, as learning models for solidarity, the functioning of which could be experienced first hand among people with similar social identities. Both a basic trust in risk-sharing schemes and an understanding of solidarity eventually translated into increased willingness to participate in larger schemes. The German experience during this phase of the evolution of the health insurance system suggests the following lesson: Small, informal, voluntary health insurance schemes may serve as learning models for fund administration and solidarity, both of which will make introduction of larger, more formal, compulsory schemes an easier task. Whether such a development can be repeated in LMICs may depend on how far the experiences with voluntary schemes in other countries mirror those in Germany. From the study of the rural cooperative health insurance schemes in rural China since the 1960s, evidence exists that the development of good management practice as well as of a trust relationship between administrators and beneficiaries are crucial elements for the long-term survival of schemes (Feng, Tang, Bloom, Segall, & Gu, 1995; Liu, Hsiao, Li, Liu, & Ren, 1995; Khan, Zhy, & Ling, 1996; World Bank, 1997). While the existence of voluntary, non-profit health insurance schemes in itself does not guarantee that such developments take place, it offers opportunities for learning, which may not be otherwise provided. The implicit danger is that, if the scheme fails, the patch may be spoilt for more promising efforts at a later time (World Bank, 1997). In many African countries traditional risk-sharing schemes exist. For instance, anthropological studies from Burkina Faso (to be published elsewhere) have revealed a dense network of traditional mutual aid organisations based on profession or on risks (such as funeral funds). It is intuitively tempting to compare these schemes to the guild-based relief funds of the pre￾Bismarckian era and to consider initiatives to integrate health risk into the cover of these funds. It has, however, been argued that the development of SHI in 19th century Germany (and Europe) is unlikely to repeat itself in today’s Africa. The first argument is that traditional informal risk-sharing schemes differ in logic and function from insurance (Criel, Van Dormael, Lefevre, Menase, & Van Leberghe, 1998). The different logics underlying informal risk-sharing and insurance (as understood in developed market economies) have been analysed from different perspec￾tives, notably anthropological and economic (see for instance, Platteau, 1997; Besley, 1995; Coate & Ravail￾lon, 1993; Lespes, 1990; Cashdan, 1985). As a rule, these " analyses start with a dichotomy between ‘real’ insurance and informal risk sharing, but often end up emphasising that many similarities exist. In a study of the traditional tontine systems in Africa, Lespes (1990) found that as " the tontines grow they take on many of the formal characteristics of insurance. Platteau (1997), in a review of the concepts underlying traditional risk sharing, starts by pointing out that traditional mutual aid schemes are based on balanced or generalised reciprocity while insurance is based on conditional reciprocity. Under both balanced and generalised reciprocity people expect, over time, to receive as much from a scheme as they contributed; under balanced reciprocity there are tight rules on how and when a return will be paid, under generalised reciprocity these rules are much looser. Under conditional reciprocity, on the other hand, the giver will only receive a return if she herself falls victim to the event she insured against by enrolling in the scheme, i.e., income is redistributed between the lucky and the unlucky. Platteau ends his analysis by arguing that traditional risk-sharing schemes may be able to serve an insurance function if either some standard of balanced reciprocity is upheld or redistribution does take place, but is not visible to enrolees. Similarily, Cashdan (1985) argued that generalised reciprocity could act in the same way as conventional insurance. In addition, under generalised reciprocity, those who have gained most during a certain time period are expected to give most to those who lose most during the same periodFeven if at some unspecified point in the future reciprocal action is expected. As such, a higher degree of solidarity is realised in the short term than in both risk- and community-rated insurance, in which contributions are independent of the financial situation of the individual. In Thailand, for instance, a number of voluntary community-based funds originally founded as non-insurance schemes to provide loans to mem￾bersFhave for some years successfully provided health care insurance coverage as well (Nitayarumphong & Pannarunothai, 1998). Going back to the origins of the German funds for mutual aid as they developed within the medieval miners’ associations and guilds, especially before the 16th century, generalised rather than conditional re￾ciprocity stands out as the main principle. What is more, it was under this principle that an understanding of community self-help, social justice and solidarity was developed, which later formed the conceptual basis for the evolution of more formal insurance funds (Herder￾Dorneich, 1994). Thus, the European and African histories of risk sharing, though separated by time, appear to be related in concept. The second argument points to the fact that where voluntary health insurance schemes exist, they are mostly initiated by agencies external to African society, namely foreign NGOs, and thus lack the dynamic of an T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587 1561

T. Barnighausen, R Sauerborn/ Social Science Medicine 54(2002 )1559-1587 endogenous social movement. As such, they usually lack structures, to establish new structures of social support some components, which, it could be argued, are or to expand coverage On a more abstract level, three important for scheme performance, notably participa lines of incremental developments in these laws can be tion, accountability and social control(Atim, 1999). In distinguis addition, non-movement-based schemes normally do First, the content of the laws moved from general not pose a threat to government and will thus fail to principles to more and more concrete rules and prompt government regulation or take over of those regulations. While, for instance, the Prussian Common schemes, as in Germany. Without government involve- Land law of 1794- established basic, general tenets of ment, it has been argued, broad social protection will public welfare and officially sanctioned the existing ly remain elusive for a majority of the population chain of subsidiarity (individuah-family-guilds and relief (Criel Van Dormael, 1999) funds-communities-the state), the laws that followed up Atim(1999), in a comparison of voluntary, non-profit on those tenets laid out detailed rules on how sickness health insurance schemes in Ghana and Cameroon, funds should be organised(including provisions about found that the character of social movement was not a contributions, the benefit package, entry conditions and main determinant of scheme performance. In addition, the management of the funds). he argues that, where circumstances suggest that social Second, the character of the laws gradually changed movement characteristics would improve performance, from permissive to obligatory. In 1843, the Comm non-movement-based schemes could over time incorpo- Law of Trade conceded municipal authorities the right rate elements of a social movement to recognise existing voluntary funds and make insur- Government involvement as well does not depend on ance in these funds compulsory-a first, albeit hesitant, a social movement character of voluntary schemes. step away from the liberal principle of the early relief While health insurance legislation based on existing funds towards compulsion. The right was extended in voluntary schemes may, in fact, be intended to gain 1849, when local governments were given permission to support of industrial workers, as in Germany(see make insurance compulsory for certain employment above) and in many Latin American countries(Abel- groups. In 1854, local governments were allowed Smith, 1976), the schemes in themselves do not need to pressure all uninsured into creating insurance funds for be an expression of social dissatisfaction in order for mutual support(Hirsch, 1875; Gladen, 1974: Peters, that goal to be reached. The current Chinese government, 1978; Herder-Dorneich, 1994) for instance, views well-functioning community-based Third, the laws moved from regional to supraregional health insurance as a means of ensuring'social stability in competence. In 1854, compulsory insurance was for the rural areas. It therefore supports still existing scheme first time established on a supraregional level covering (which were originally established by the central govern- the entire territory of Germany for one employment ment through legislation in the 1960s ), while promoting group: all miners were required to join one of the many the re-establishment of schemes in communities where regional miners' insurance funds. a number of non- arrently none exist( Gwatkin, 1999). What is more, it has Prussian states had, at this time, already established been argued that non-movement-schemes may, over time, compulsory health insurance for workers. In some incorporate elements of a social movement and thereby states, compulsion was tied to a specific fund(such as enhance their success(Atim, 1999) in Hannover); in other states, workers had a choice To sum up potential lessons learnt with respect to between different sickness funds(such as in Hamburg) LMICS, the German case, among others, demonstrates(Herder-Dorneich, 1994) that small, informal, voluntary, community-based In 1883, these three lines of incremental development health insurance schemes may serve as crystallisation were brought together in Bismarck's workers'insurance. points from which larger, more formal, compulsor It laid out detailed rules for the provision of health schemes can be developed Countries should investigate insurance including a minimum benefit package, the how to promote such schemes, especially if alternative types of sickness funds, management of the funds and insurances currently do not seem feasible. As an the extension of coverage to family members (Vogel alternative to the de novo creation of health insurance 1951: Laden, 1974; Herder-Dorneich, 1994). It made schemes, consideration should be given to including health insurance coverage a legal obligation for most health in the cover of pre-existing non-health risk- workers and people employed in trade and crafts. And sharing schemes above all, it replaced the existing regional principle of compulsory insurance by a supraregional principle-a Before 1883: incremental legislative changes to achieve epoch-making, but nonetheless incremental step. In supraregional compulsory insurance laying the groundwork for universal Six major laws led up to the reform of 1883, 2Allgemeines preussisches landrecht tempting either te

endogenous social movement. As such, they usually lack some components, which, it could be argued, are important for scheme performance, notably participa￾tion, accountability and social control (Atim, 1999). In addition, non-movement-based schemes normally do not pose a threat to government and will thus fail to prompt government regulation or take over of those schemes, as in Germany. Without government involve￾ment, it has been argued, broad social protection will likely remain elusive for a majority of the population (Criel & Van Dormael, 1999). Atim (1999), in a comparison of voluntary, non-profit health insurance schemes in Ghana and Cameroon, found that the character of social movement was not a main determinant of scheme performance. In addition, he argues that, where circumstances suggest that social movement characteristics would improve performance, non-movement-based schemes could over time incorpo￾rate elements of a social movement. Government involvement as well does not depend on a social movement character of voluntary schemes. While health insurance legislation based on existing voluntary schemes may, in fact, be intended to gain support of industrial workers, as in Germany (see above) and in many Latin American countries (Abel￾Smith, 1976), the schemes in themselves do not need to be an expression of social dissatisfaction in order for that goal to be reached. The current Chinese government, for instance, views well-functioning community-based health insurance as a means of ensuring ‘social stability in rural areas’. It therefore supports still existing schemes (which were originally established by the central govern￾ment through legislation in the 1960s), while promoting the re-establishment of schemes in communities where currently none exist (Gwatkin, 1999). What is more, it has been argued that non-movement-schemes may, over time, incorporate elements of a social movement and thereby enhance their success (Atim, 1999). To sum up potential lessons learnt with respect to LMICs, the German case, among others, demonstrates that small, informal, voluntary, community-based health insurance schemes may serve as crystallisation points from which larger, more formal, compulsory schemes can be developed. Countries should investigate how to promote such schemes, especially if alternative insurances currently do not seem feasible. As an alternative to the de novo creation of health insurance schemes, consideration should be given to including health in the cover of pre-existing non-health risk￾sharing schemes. Before 1883: incremental legislative changes to achieve supraregional compulsory insurance Six major laws led up to the reform of 1883, attempting either to regulate more closely existing structures, to establish new structures of social support, or to expand coverage. On a more abstract level, three lines of incremental developments in these laws can be distinguished. First, the content of the laws moved from general principles to more and more concrete rules and regulations. While, for instance, the Prussian Common Land law of 17942 established basic, general tenets of public welfare and officially sanctioned the existing chain of subsidiarity (individual–family–guilds and relief funds–communities–the state), the laws that followed up on those tenets laid out detailed rules on how sickness funds should be organised (including provisions about contributions, the benefit package, entry conditions and the management of the funds). Second, the character of the laws gradually changed from permissive to obligatory. In 1843, the Common Law of Trade3 conceded municipal authorities the right to recognise existing voluntary funds and make insur￾ance in these funds compulsoryFa first, albeit hesitant, step away from the liberal principle of the early relief funds towards compulsion. The right was extended in 1849, when local governments were given permission to make insurance compulsory for certain employment groups. In 1854, local governments were allowed to pressure all uninsured into creating insurance funds for mutual support (Hirsch, 1875; Gladen, 1974; Peters, 1978; Herder-Dorneich, 1994). Third, the laws moved from regional to supraregional competence. In 1854, compulsory insurance was for the first time established on a supraregional level covering the entire territory of Germany for one employment group: all miners were required to join one of the many regional miners’ insurance funds. A number of non￾Prussian states had, at this time, already established a compulsory health insurance for workers. In some states, compulsion was tied to a specific fund (such as in Hannover); in other states, workers had a choice between different sickness funds (such as in Hamburg) (Herder-Dorneich, 1994). In 1883, these three lines of incremental development were brought together in Bismarck’s workers’ insurance. It laid out detailed rules for the provision of health insurance including a minimum benefit package, the types of sickness funds, management of the funds and the extension of coverage to family members (Vogel, 1951; Gladen, 1974; Herder-Dorneich, 1994). It made health insurance coverage a legal obligation for most workers and people employed in trade and crafts. And, above all, it replaced the existing regional principle of compulsory insurance by a supraregional principleFa epoch-making, but nonetheless incremental step. In addition to laying the groundwork for universal 2Allgemeines Preussisches Landrecht. 3Allgemeines Handelsrecht. 1562 T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587

T. Barnighausen, R. Sauerborn Social Science Medicine 54(2002)1559-1587 1563 coverage, the supraregional principle solved the problem insurance system-from informal to more formal, from of providing insurance coverage for an increasingly voluntary to compulsory, and from small to larger mobile population, for whom -as social ties were schemes--can and should be emulated in LMIC may be severed and regional insurance was lost-social security highly contingent on the context: the power stru had become a more and more elusive concept trust and legitimacy between the different actors. Will In introducing the workers' insurance Bismarck did informal schemes be opposed to government regulation? not primarily intend to further social justice, but to Will the participants and the current non-participants in fortify the state against the threat from a proletariat, the scheme resist being compelled to join? will solidarity which had become both well organised in the trade (or generalised reciprocity) suffer as the average distance unions and politically powerful, as its interests wer physical and social) between members increase? Will represented by the Social Democratic Party. By incor- trust in the scheme and its management remain intact as porating formerly self-administered insurance into the community participation and social control becom state Bismarck hoped to counter the increasing politi- increasingly difficult? cisation of the working class. Indirect evidence supports As in the german case. it has been suggested his view. First, compulsory insurance at the inception of elsewhere for the context of LMICs that incremental the system was limited to workers. Although blue-collar changes may be easier to implement than transforma workers were employed in the formal sector economy, tional changes in developing efficient and equitable they were harder to insure than many other groups, as health insurance( Carrin, De graeve, deville, 1999a) they had low incomes and high risks of work-related In many of the formerly socialist Eastern European illness, accidents, and disability. Better risks in terms of countries a development towards sHI similar to that in income and health care needs such as civil servants were Germany had been taking place, starting in the 18th not included under compulsory health insurance cover century until restructuring of the health care sector by until 30 years later. The likely explanation is that the communist governments(see, for instance, Observa- Bismarck expected civil servants to be naturally loyal to tory, 1999a, b). In Hungary, voluntary self-help funds the state and interested in preserving the status quo. for industrial workers were legally legitimised in 1840: a Second, a number of provisions in Bismarck's "Socialist voluntary General Fund for sick and disabled workers Law, passed in 1878, were intended to obstruct the was established in 1870: a national compulsory insu functioning of those sickness and relief funds that had ance for industrial workers, similar to Bismarcks been founded by workers-the stick preceded the workers' insurance, was established; and, finally, at the carrot. Third, both trade unions and the Social turn of the century a national insurance fund for Democratic Party were openly opposed to Bismarcks agricultural workers was set up(Observatory, 1999b). social insurance, as they-correctly-viewed the pro. It is safe to say that a number of Eastern European gramme as a means to tie the workers to the existing countries picked up threads of development that had state structures(Herder-Dorneich, 1994) been abandoned during the communist era when the (re-)established SHI. The mode of reforms, however, did If informal risk-sharing schemes exist, the creation of not follow the earlier incremental pattern. A crisis in legal frameworks formalising these and health care financing accompanied by a fall in life Itually making them compulsory expectancy in an environment characterised by rapid important step towards establishing un litical and economic changes had opened a window health insurance of opportunity, in which fast and drastic, rather than slow and steady, action was felt to be required. Introducing compulsory health insurance has been Of the three transitions described above the move to part of health care reform in many countries of South compulsion may be the most difficult to achieve even if East Asia, Central and Eastern Europe and has come it is only for one segment of the population and in one under consideration in a number of African countries region of the country. In many contexts, the establish Zwi Mills, 1995; Bennet Ngalande-Banda, 1994). ment of any form of compulsory insurance may be Countries where successful informal voluntary risk- deemed not to be politically feasible. In such situations, sharing schemes for health exist should consider making voluntary schemes may remain a second-best option. them more formal, as without such government action The recommendations regarding rural community-based risk-sharing social protection will remain limited (or cooperative) health insurance schemes in the Peoples contingent on local circumstances( Criel Van Dor- Republic of China in 1998 to promote voluntary mael, 1999). A legal framework may define a minimum community-based health insurance in the countryside benefit package and regulate contributions, provider exemplify this. Although policy makers were aware that payment mechanisms and scheme administration problems with adverse selection could arise and that Whether the three transitions that took place at this willingness-to-join could be generally low because of phase in the development of the German health

coverage, the supraregional principle solved the problem of providing insurance coverage for an increasingly mobile population, for whomFas social ties were severed and regional insurance was lostFsocial security had become a more and more elusive concept. In introducing the workers’ insurance Bismarck did not primarily intend to further social justice, but to fortify the state against the threat from a proletariat, which had become both well organised in the trade unions and politically powerful, as its interests were represented by the Social Democratic Party. By incor￾porating formerly self-administered insurance into the state Bismarck hoped to counter the increasing politi￾cisation of the working class. Indirect evidence supports this view. First, compulsory insurance at the inception of the system was limited to workers. Although blue-collar workers were employed in the formal sector economy, they were harder to insure than many other groups, as they had low incomes and high risks of work-related illness, accidents, and disability. Better risks in terms of income and health care needs such as civil servants were not included under compulsory health insurance cover until 30 years later. The likely explanation is that Bismarck expected civil servants to be naturally loyal to the state and interested in preserving the status quo. Second, a number of provisions in Bismarck’s ‘Socialist Law’, passed in 1878, were intended to obstruct the functioning of those sickness and relief funds that had been founded by workersFthe stick preceded the carrot. Third, both trade unions and the Social Democratic Party were openly opposed to Bismarck’s social insurance, as theyFcorrectlyFviewed the pro￾gramme as a means to tie the workers to the existing state structures (Herder-Dorneich, 1994). If informal risk-sharing schemes exist, the creation of legal frameworks formalising these schemes and eventually making them compulsory, can be an important step towards establishing universal social health insurance. Introducing compulsory health insurance has been part of health care reform in many countries of South East Asia, Central and Eastern Europe, and has come under consideration in a number of African countries (Zwi & Mills, 1995; Bennet & Ngalande-Banda, 1994). Countries where successful informal voluntary risk￾sharing schemes for health exist should consider making them more formal, as without such government action risk-sharing social protection will remain limited and contingent on local circumstances (Criel & Van Dor￾mael, 1999). A legal framework may define a minimum benefit package and regulate contributions, provider payment mechanisms and scheme administration. Whether the three transitions that took place at this phase in the development of the German health insurance systemFfrom informal to more formal, from voluntary to compulsory, and from small to larger schemesFcan and should be emulated in LMIC may be highly contingent on the context: the power structure, trust and legitimacy between the different actors. Will informal schemes be opposed to government regulation? Will the participants and the current non-participants in the scheme resist being compelled to join? Will solidarity (or generalised reciprocity) suffer as the average distance (physical and social) between members increase? Will trust in the scheme and its management remain intact as community participation and social control become increasingly difficult? As in the German case, it has been suggested elsewhere for the context of LMICs that incremental changes may be easier to implement than transforma￾tional changes in developing efficient and equitable health insurance (Carrin, De Graeve, & Deville, 1999a). ! In many of the formerly socialist Eastern European countries a development towards SHI similar to that in Germany had been taking place, starting in the 18th century until restructuring of the health care sector by the communist governments (see, for instance, Observa￾tory, 1999a, b). In Hungary, voluntary self-help funds for industrial workers were legally legitimised in 1840; a voluntary General Fund for sick and disabled workers was established in 1870; a national compulsory insur￾ance for industrial workers, similar to Bismarck’s workers’ insurance, was established; and, finally, at the turn of the century a national insurance fund for agricultural workers was set up (Observatory, 1999b). It is safe to say that a number of Eastern European countries picked up threads of development that had been abandoned during the communist era when they (re-)established SHI. The mode of reforms, however, did not follow the earlier incremental pattern. A crisis in health care financing accompanied by a fall in life expectancy in an environment characterised by rapid political and economic changes had opened a ‘window of opportunity’, in which fast and drastic, rather than slow and steady, action was felt to be required. Of the three transitions described above the move to compulsion may be the most difficult to achieve, even if it is only for one segment of the population and in one region of the country. In many contexts, the establish￾ment of any form of compulsory insurance may be deemed not to be politically feasible. In such situations, voluntary schemes may remain a second-best option. The recommendations regarding rural community-based (or cooperative) health insurance schemes in the Peoples’ Republic of China in 1998 to promote voluntary community-based health insurance in the countryside exemplify this. Although policy makers were aware that problems with adverse selection could arise and that willingness-to-join could be generally low because of negative past experiences with community-based T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587 1563

T. Barnighausen, R Sauerborn/ Social Science Medicine 54(2002 )1559-1587 insurance, the Chinese State Council shied away fro health contributions in the time of economic crisis. In mpulsion. Another compulsory payment to a state addition, the state lacked the capacity to collect organisation, it was feared, would further increase anti- contributions from those companies and workers that government sentiments among farmers( Gwatkin, 1999. were able to pay(Observatory, 1999a-c) Hsiao, 1984). Similarily, in Nigeria a social health n sum, the German case suggests that the central insurance scheme was established on a voluntary basi government plays a crucial role in establishing a SHI, as as a compulsory scheme was judged not to be politically it is the institution best placed to create a legal feasible(Bennet Ngalande-Banda, 1994) framework for SHI. Legislation in Germany incremen In addition, compulsion may prove not to be enforce- tally formalised and expanded insurance as well as made able for administrative or economic reasons. In China, it compulsory. While many LMICs already have under national law all urban workers in state-owned successfully introduced compulsory health insurance companies and their family members fall under the for some segments of the population, other cases from mpulsory cover of the Labour Insurance Scheme LMICs suggest that introducing compulsion -even if company-based Bismarckian health insurance. As more only in the formal sector-may be difficult to enact or to d more state-owned enterprises near bankruptcy. the enforce, if the government is politically or administra ompany-based funds become increasingly insolvent In tively weak or the economy is flagging theory, workers still receive full reimbursement for a omprehensive benefit package, their family members After 1883: incremental expansion of coverage to achieve are covered with 50% of their eligible health care universal coverage xpenditures. Increasingly, however, deficit-running en terprises have been unable to pay contributions to the The incremental approach taken to develop the funds. As a result, in 1998, about one-third of workers in system after the introduction of the workers'insurance state enterprises had no health insurance coverage at all, manifested itself mainly in the expansion of population many others received only marginal health care benefits coverage, the size of the risk pools and the benefits from their insurance(Grogan, 1995, Yip Hsiao, 1997; covered. By some estimates, Bismarck's law doubled u, Ong, Lin, Li, 1999: Center for statistical sickness insurance coverage among workers from information of the Chinese ministry of health, 1999). round 5%o to 10% of the total population. Thereafter In many Eastern European countries, the implemen- coverage in the statutory health insurance grew steadily tation of a compulsory SHI after 1989 has faced serious from 11% in 1885 to 37% in 1910. By 1930 about 50% problems(such as in Bulgaria and Hungary) or even of the total population were covered and by 1950 about failed (such as in Kazakhstan). In these cases, both 70%. Since 1975 more than 90% of the population are employees and employers were unable to pay social enrolled in the statutory health insurance; the remaining 1885189519101914192519341950195519601965196819761980198519901995 9oIO Private health insurance a Statutory health insurance 50 Fig 1. Population coverage, 1885-1

insurance, the Chinese State Council shied away from compulsion. Another compulsory payment to a state organisation, it was feared, would further increase anti￾government sentiments among farmers (Gwatkin, 1999, Hsiao, 1984). Similarily, in Nigeria a social health insurance scheme was established on a voluntary basis, as a compulsory scheme was judged not to be politically feasible (Bennet & Ngalande-Banda, 1994). In addition, compulsion may prove not to be enforce￾able for administrative or economic reasons. In China, under national law all urban workers in state-owned companies and their family members fall under the compulsory cover of the Labour Insurance Scheme, a company-based Bismarckian health insurance. As more and more state-owned enterprises near bankruptcy, the company-based funds become increasingly insolvent. In theory, workers still receive full reimbursement for a comprehensive benefit package, their family members are covered with 50% of their eligible health care expenditures. Increasingly, however, deficit-running en￾terprises have been unable to pay contributions to the funds. As a result, in 1998, about one-third of workers in state enterprises had no health insurance coverage at all, many others received only marginal health care benefits from their insurance (Grogan, 1995, Yip & Hsiao, 1997; Hu, Ong, Lin, & Li, 1999; Center for statistical information of the Chinese ministry of health, 1999). In many Eastern European countries, the implemen￾tation of a compulsory SHI after 1989 has faced serious problems (such as in Bulgaria and Hungary) or even failed (such as in Kazakhstan). In these cases, both employees and employers were unable to pay social health contributions in the time of economic crisis. In addition, the state lacked the capacity to collect contributions from those companies and workers that were able to pay (Observatory, 1999a–c). In sum, the German case suggests that the central government plays a crucial role in establishing a SHI, as it is the institution best placed to create a legal framework for SHI. Legislation in Germany incremen￾tally formalised and expanded insurance as well as made it compulsory. While many LMICs already have successfully introduced compulsory health insurance for some segments of the population, other cases from LMICs suggest that introducing compulsionFeven if only in the formal sectorFmay be difficult to enact or to enforce, if the government is politically or administra￾tively weak or the economy is flagging. After 1883: incremental expansion of coverage to achieve universal coverage The incremental approach taken to develop the system after the introduction of the workers’ insurance manifested itself mainly in the expansion of population coverage, the size of the risk pools and the benefits covered. By some estimates, Bismarck’s law doubled sickness insurance coverage among workers from around 5% to 10% of the total population. Thereafter, coverage in the statutory health insurance grew steadily from 11% in 1885 to 37% in 1910. By 1930 about 50% of the total population were covered and by 1950 about 70%. Since 1975 more than 90% of the population are enrolled in the statutory health insurance; the remaining Fig. 1. Population coverage, 1885–1995. 1564 T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587

T. Barnighausen, R. Sauerborn Social Science Medicine 54(2002)1559-1587 1565 10% are nearly completely covered under private or the population remain uncovered(Ensor, 1997). South other health insurance. Today, less than 0.5% of all America has enjoyed a long tradition of social insurance. 1999a,b)Fig.. than 80% in Costa Rica, although the proportion of the The extension of population coverage in an insurance population covered in most countries is rising. The ystem can be classified based on three principles: the population group least likely to be covered is the regional, the personal, and the place of work principle growing number of urban informal sector workers (Zollner, 1963). According to the regional principle, an Donaldson gerard, 1993). In Africa, a number of insurance scheme is first established in some regions of a countries have established social health insurance country (usually the most industrialised ones),an schemes, such as Cameroon, Ghana and Kenya. Again, gradually extended to cover other (usually less devel- coverage is mostly confined to the formal workforce oped) regions. An extension of coverage according to (Bennet Ngalande-Banda, 1994) the personal principle can be either oriented at horizontal In Chinese cities, up to the market reforms in the criteria such cupation or vertical criteria such as 980s. SHI schemes covered more than 90% of the income. Similarly, extension according to the place of urban population, although only 50% of health care work principle ca along horizontal lines (e.g. costs incurred by spouses and children of the insured economic sector) or vertical lines(e.g. size of company) were covered. Today, the number of urban residents Before the introduction of the workers' health covered has dwindled to less than 50%, as more and insurance, access to insurance had depended either on more state enterprises declare insolvency and people the region of residence or on the place of work increasingly find work in private enterprises or in the Correspondingly, coverage was extended according to informal urban sector(Center for statistical information the regional or the place of work principle. By contrast, of the Chinese ministry of health, 1999: World Bank, once the principle of supraregional compulsory insur- 1997; Hsiao,1995).The or stumbling block to nce was introduced for different occupational groups in universal coverage in these cities is the growing number 1883, coverage was extended according to the personal of informal sector employees and migrants from the principle Along horizontal lines coverage was expanded countryside(both legal and illegal). The central Chinese incrementally to cover more and more occupational government currently attempts to promote a stepwise groups and-in three major shifts of expansion -to expansion of coverage in all cities from state to non-state cover the unemployed, all primary dependents and enterprises to the self-employed and-eventually and retirees(see Table 1)(Wasserrab, 1889: Lang, 1925: perhaps with the help of subsidies-to the urban poor Peters, 1978: Alber, 1989; Manow, 1999). Vertically, (Hu, 1999; Zhu, Zhou, Zhang, Ma, Gao, 1999: Bloom, coverage was expanded by increasing the income ceiling 1998). In accordance with a national policy recommen above which health insurance is no longer compulsory dation, some city governments consider offering volun- as was done, for example, in 1918 when the monthly tary enrolment in the city-wide funds to anybody not yet income limit was doubled from RM 2500 to RM 5000. under mandatory cover, if they are able to contribute as or people who fell under the law of compulsory much to the fund as do workers earning 60% of the citys urance. but who did not have access to a sickness average annual salary (see, for instance, Social health fund through their work and could not insure in a town- insurance administration office of Shenzhen city, 1999: based fund, every municipality had to provide insurance Labour office of Yichang city, 1998) through a municipal sickness fund. The German case suggests that compulsory coverage This second phase in the development of the German an be extended incrementally to achieve universality. health insurance system suggests the following lesson This 'lesson. however, cannot be drawn without some neral qualifications and without If compulsory insurance already exists for some the specific methods used in Germany can and should be people, extending it incrementally to other regions transferred to other countries and times. From an and social groups will--if a number of conditions are ethical point of view, it has to be kept in mind that the met-be a feasible way to achieve universal cover German government in adopting an incremental ap- proach towards universality was motivated by argu- ments of power rather than social justice(Rimlinger, Expanding compulsory insurance Nam. a c is a task 1971: Observatory, 2000b). If one accepts a utilitarian many LMICs face today. In Viet Na ethic that preservation of power may be a legitimate goal SHI scheme was introduced in 1993 covers civ of social policy as long as the ultimate outcome serves servants and workers in larger enterprises, but-in spite social justice, it has to be kept in mind that an of attempts to expand cover to family members, farmers, incremental approach to establishing SHI may, in fact and urban informal sector workers-more than 90% of lead to more inequity

10% are nearly completely covered under private or other health insurance. Today, less than 0.5% of all people living in Germany do not have health insurance (Tennstedt, 1977; Peters, 1978; Neubauer, 1988; BMG, 1999a, b) Fig. 1. The extension of population coverage in an insurance system can be classified based on three principles: the regional, the personal, and the place of work principle (Zollner, 1963). According to the . regional principle, an insurance scheme is first established in some regions of a country (usually the most industrialised ones), and then gradually extended to cover other (usually less devel￾oped) regions. An extension of coverage according to the personal principle can be either oriented at horizontal criteria such as occupation or vertical criteria such as income. Similarly, extension according to the place of work principle can be along horizontal lines (e.g. economic sector) or vertical lines (e.g. size of company). Before the introduction of the workers’ health insurance, access to insurance had depended either on the region of residence or on the place of work. Correspondingly, coverage was extended according to the regional or the place of work principle. By contrast, once the principle of supraregional compulsory insur￾ance was introduced for different occupational groups in 1883, coverage was extended according to the personal principle. Along horizontal lines coverage was expanded incrementally to cover more and more occupational groups andFin three major shifts of expansionFto cover the unemployed, all primary dependents and retirees (see Table 1) (Wasserrab, 1889; Lang, 1925; Peters, 1978; Alber, 1989; Manow, 1999). Vertically, coverage was expanded by increasing the income ceiling above which health insurance is no longer compulsory, as was done, for example, in 1918 when the monthly income limit was doubled from RM 2500 to RM 5000. For people who fell under the law of compulsory insurance, but who did not have access to a sickness fund through their work and could not insure in a town￾based fund, every municipality had to provide insurance through a municipal sickness fund. This second phase in the development of the German health insurance system suggests the following lesson: If compulsory insurance already exists for some people, extending it incrementally to other regions and social groups willFif a number of conditions are metFbe a feasible way to achieve universal cover￾age. Expanding compulsory insurance coverage is a task many LMICs face today. In Viet Nam, a compulsory SHI scheme was introduced in 1993, which covers civil servants and workers in larger enterprises, butFin spite of attempts to expand cover to family members, farmers, and urban informal sector workersFmore than 90% of the population remain uncovered (Ensor, 1997). South America has enjoyed a long tradition of social insurance. But population coverage is highly variable. It ranges from less than 10% in the Dominican Republic to more than 80% in Costa Rica, although the proportion of the population covered in most countries is rising. The population group least likely to be covered is the growing number of urban informal sector workers (Donaldson & Gerard, 1993). In Africa, a number of countries have established social health insurance schemes, such as Cameroon, Ghana and Kenya. Again, coverage is mostly confined to the formal workforce (Bennet & Ngalande-Banda, 1994). In Chinese cities, up to the market reforms in the 1980s, SHI schemes covered more than 90% of the urban population, although only 50% of health care costs incurred by spouses and children of the insured were covered. Today, the number of urban residents covered has dwindled to less than 50%, as more and more state enterprises declare insolvency and people increasingly find work in private enterprises or in the informal urban sector (Center for statistical information of the Chinese ministry of health, 1999; World Bank, 1997; Hsiao, 1995). The major stumbling block to universal coverage in these cities is the growing number of informal sector employees and migrants from the countryside (both legal and illegal). The central Chinese government currently attempts to promote a stepwise expansion of coverage in all cities from state to non-state enterprises to the self-employed andFeventually and perhaps with the help of subsidiesFto the urban poor (Hu, 1999; Zhu, Zhou, Zhang, Ma, & Gao, 1999; Bloom, 1998). In accordance with a national policy recommen￾dation, some city governments consider offering volun￾tary enrolment in the city-wide funds to anybody not yet under mandatory cover, if they are able to contribute as much to the fund as do workers earning 60% of the city’s average annual salary (see, for instance, Social health insurance administration office of Shenzhen city, 1999; Labour office of Yichang city, 1998). The German case suggests that compulsory coverage can be extended incrementally to achieve universality. This ‘lesson’, however, cannot be drawn without some general qualifications and without considering, whether the specific methods used in Germany can and should be transferred to other countries and times. From an ethical point of view, it has to be kept in mind that the German government in adopting an incremental ap￾proach towards universality was motivated by argu￾ments of power rather than social justice (Rimlinger, 1971; Observatory, 2000b). If one accepts a utilitarian ethic that preservation of power may be a legitimate goal of social policy as long as the ultimate outcome serves social justice, it has to be kept in mind that an incremental approach to establishing SHI may, in fact, lead to more inequity. T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587 1565

T. Barnighausen, R. Sauerborn Social Science Medicine 54(2002)1559-1587 Table I Introduction of new types of social insurance, expansion of compulsory health insurance coverage, and extension of the mandated minimum benefit package Year Creation of components Population coverage Scale and scope of the mandated of social security of social health insurance benefit package of social health insurance 1854 Miners 1883 Statu Blue collar workers (in saltworks Minimum benefit package Insur processing plants, factories, metallurgical Sickpay (63% of all benefits) plants, railway companies, shipping Restricted in- and outpatient care trade co es, shipyards, building companies, free pharmaceuticals medical aid devices Craftsmen de Persons employed by lawyers, notaries, maternity support nsurance funds l884 Statutory accident Transport workers Statutory pension 1892 Commercial office workers Increase in the duration of sickpay from 13 to 26 weeks Sickpay extended to cases of sexually transmitted diseases Increase in allowances to family members in case of a hospitalisation of a relative 1911 Agricultural and forestry workers Increase in maternity support Domestic servants Increase in sickpay of high-wage workers Itinerant workers 1914 Civil servants Earlier start of sickpay Family support for spouses and children 1917/18 The unemployed Obstetric services 1919 Persons Persons employed in Wives and daughters without own income 927 Statutory unemployme Persons employed in the educational and ocial welfare sectors All primary dependents 938 Self-employed workers in nursing and child care Full cover of the treatment of all fugees and expellees sically disabled Increase in sickpay for workers 1972 Self-employed agricultural workers workers for sick farmers

Table 1 Introduction of new types of social insurance, expansion of compulsory health insurance coverage, and extension of the mandated minimum benefit package Year Creation of components of social security Population coverage of social health insurance Scale and scope of the mandated benefit package of social health insurance 1854 Miners 1883 Statutory health insurance Blue collar workers (in saltworks, processing plants, factories, metallurgical plants, railway companies, shipping companies, shipyards, building companies, trade companies, power plants) Minimum benefit package: Sickpay (63% of all benefits) Restricted in- and outpatient care free pharmaceuticals medical aid devices Craftsmen deathpay Persons employed by lawyers, notaries, bailiffs, industrial cooperatives, insurance funds maternity support 1884 Statutory accident insurance 1885 Transport workers 1889 Statutory pension insurance 1892 Commercial office workers 1901 Increase in the duration of sickpay from 13 to 26 weeks Sickpay extended to cases of sexually￾transmitted diseases Increase in allowances to family members in case of a hospitalisation of a relative 1911 Agricultural and forestry workers Increase in maternity support Domestic servants Increase in sickpay of high-wage workers Itinerant workers 1914 Civil servants Earlier start of sickpay Family support for spouses and children 1917/18 The unemployed Midwife services Obstetric services Pregnancy allowance Nursing mother’s allowance 1919 Persons employed in public cooperatives Persons employed in private cooperatives Persons who are only partially capable of gainful employment Wives and daughters without own income 1927 Statutory unemployment insurance Seamen Persons employed in the educational and social welfare sectors 1930 All primary dependents 1935 Increase in the duration of maternity support 1938 Midwives Self-employed workers in nursing and child care 1941 Retirees Full cover of the treatment of all notifiable diseases 1953 Refugees and expellees The seriously disabled 1957 The physically disabled Increase in sickpay for workers 1970 Prevention Pediatric screening 1972 Self-employed agricultural workers Salary of a temporary replacement workers for sick farmers 1566 T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587

T. Barnighausen, R Sauerborn/ Social Science Medicine 54(2002 )1559-1587 1567 Table I(continued) Creation of components Population coverage of social health insurance benefit package of social health insurance 1974 Removal of the time limit to in-patient care to compensate for wages lost while for a child ease in the cover of rehabilitation services ease in the cover of dental and orthodontic Students 1978 Contraception consultation Counselling and medical support in cases of legal sterilisation and legal abortion 1981 1995 Statutory nursing Insurance For one, revenues in a SHI system traditionally flow concerns have been raised, as the scheme is subsidised from formal wages and salaries. As a result, the from general government revenues. In addition, bene- population groups likely to be covered last are the most ficiaries were found to use public hospitals at a rate that vulnerable segments of the society: those without was five times the national average( Prescott, 1991). For incomes(the unemployed, retirees, and family depen- these reasons, a fast-track approach to universal health dents)or those with incomes that are variable and hard insurance, may be preferable in some circumstances, to assess(urban informal workers and farmers ). This has although it may require a much larger effort. an important implication. Current members of social In the formerly socialist countries of Eastern Europe health insurance schemes may be opposed to including establishing social insurance step by step, starting other groups in the insurance cover. On the one hand, with partial coverage, would have meant red since those who are as yet without insurance are likely to equity in comparison to the universal access be low income and high risk people, those who are uaranteed before under the command-and-control currently insured would likely pay part of the price of Soviet model of health care(Observatory, 1999a, b, including these groups in the form of higher insurance 2000a, c: Twigg, 1999) contributions. On the other hand. the incremental Whether or not the specific means by which groups approach to include people along employment without formal wages or salaries were integrated under solidarity between current memb, roximity and thus the SHI cover in Germany can be replicated in LMICs regional lines implies that the socia depends on a series of factors. First, the unemployed between members and non-members. Unemployment and the retired are covered through the wider system of and informal sector employment are increasing in many social insurance. The statutory pension insurance and LMICS(ILO, 1999). Thus countrywide solidarity across the statutory unemployment insurance provide the two ployment lines may be increasingly hard to establish groups with regular, taxable incomes from which in LMICs. But continuing commitment to solidarity mandatory health insurance contributions are automa among all people living in a country-as it still exists in tically deducted. Obviously, this is only possible Germany (Hinrichs, 1995)is a basic condition for countries where comprehensive social insurance exists establishing universal ShI is esta blished at the same time as the health insurance oreover, a stepwise pass Second, children and spouses are included under result in decreased access to health care for the cover of the breadwinner. Since contributions are uninsured in the interim periods of partial coverag independent of family size, a re-distribution from singles (which may be quite long, if political will is lacking to families and from families with fewer to families with socio-economic conditions are unfavourable) Resources nore children results. how far a re-distribution may be drained away from the uninsured to provide feasible within a SHI system depends on the dominant health care for the insured(Normand Weber, 1994; hierarchy of values within a society Abel-Smith, 1992). A case in point is a compulsory Third, self-employed farmers were not covered until insurance scheme for Indonesian civil servants. Equity 90 years after the introduction of Bismarck's workers

For one, revenues in a SHI system traditionally flow from formal wages and salaries. As a result, the population groups likely to be covered last are the most vulnerable segments of the society: those without incomes (the unemployed, retirees, and family depen￾dents) or those with incomes that are variable and hard to assess (urban informal workers and farmers). This has an important implication. Current members of social health insurance schemes may be opposed to including other groups in the insurance cover. On the one hand, since those who are as yet without insurance are likely to be low income and high risk people, those who are currently insured would likely pay part of the price of including these groups in the form of higher insurance contributions. On the other hand, the incremental approach to include people along employment or regional lines implies that the social proximity and thus solidarity between current members is higher than between members and non-members. Unemployment and informal sector employment are increasing in many LMICs (ILO, 1999). Thus countrywide solidarity across employment lines may be increasingly hard to establish in LMICs. But continuing commitment to solidarity among all people living in a countryFas it still exists in Germany (Hinrichs, 1995)Fis a basic condition for establishing universal SHI. Moreover, a stepwise passage to universality may result in decreased access to health care for the uninsured in the interim periods of partial coverage (which may be quite long, if political will is lacking or socio-economic conditions are unfavourable). Resources may be drained away from the uninsured to provide health care for the insured (Normand & Weber, 1994; Abel-Smith, 1992). A case in point is a compulsory insurance scheme for Indonesian civil servants. Equity concerns have been raised, as the scheme is subsidised from general government revenues. In addition, bene- ficiaries were found to use public hospitals at a rate that was five times the national average (Prescott, 1991). For these reasons, a fast-track approach to universal health insurance, may be preferable in some circumstances, although it may require a much larger effort. In the formerly socialist countries of Eastern Europe establishing social insurance step by step, starting with partial coverage, would have meant reducing equity in comparison to the universal access guaranteed before under the command-and-control Soviet model of health care (Observatory, 1999a, b, 2000a, c; Twigg, 1999). Whether or not the specific means by which groups without formal wages or salaries were integrated under the SHI cover in Germany can be replicated in LMICs depends on a series of factors. First, the unemployed and the retired are covered through the wider system of social insurance. The statutory pension insurance and the statutory unemployment insurance provide the two groups with regular, taxable incomes from which mandatory health insurance contributions are automa￾tically deducted. Obviously, this is only possible in countries where comprehensive social insurance exists or is established at the same time as the health insurance. Second, children and spouses are included under the cover of the breadwinner. Since contributions are independent of family size, a re-distribution from singles to families and from families with fewer to families with more children results. How far a re-distribution is feasible within a SHI system depends on the dominant hierarchy of values within a society. Third, self-employed farmers were not covered until 90 years after the introduction of Bismarck’s workers’ Table 1 (continued) Year Creation of components of social security Population coverage of social health insurance Scale and scope of the mandated benefit package of social health insurance 1974 Removal of the time limit to in-patient care Sickpay to compensate for wages lost while caring for a child Domestic aid during in-patient or in-patient cures Increase in the cover of rehabilitation services Increase in the cover of dental and orthodontic services 1975 Students All disabled 1978 Contraception consultation Counselling and medical support in cases of legal sterilisation and legal abortion 1981 Artist Publicists 1995 Statutory nursing insurance T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587 1567

1568 T. Barnighausen, R. Sauerborn Social Science Medicine 54(2002)1559-1587 insurance(Holler, 1977). This reflects the difficulty in health insurance can be established in a country with a assessing and taxing farmers' incomes. For many social, political and cultural history which is very countries, especially those where farmers constitute different from that in Germany. It also shows that the large proportion of the total population, such as China, pace of incremental development can be much acceler- Vietnam and Thailand, such a delay may not be ated. It has to be kept in mind, however, that this practicable. The difference in context between 19th ened against a backdrop of fast and sustaine century Germany and todays LMICs is even more economic growth and a shrinking informal sector-two prominent in the case of the informal sector. In conditions that, while neither necessary nor sufficient, Germany, the proportion of informal sector workers are conducive to establishing a SHI declined during the years of the systems existence. In n sum, Germany succeeded in achieving universal many LMIC, on the other hand, the informal sector is coverage following an incremental pattern of expanding expected to continue to grow in the next years (ILo, compulsory insurance. This success has been contingent 1992,1996,1999) n a number of social. economic and institutional Self-employed farmers and informal sector workers circumstances. Yet, as the above cases suggests,a are hard to cover in a SHI system since their incomes similar approach holds promise for countries that have fluctuate and are hard to assess objectively. Still, many already established partial coverage such as many stems have been devised for assessing the incomes of countries in transitional Asia and South america the self-employed and charging contributions(Normand Weber, 1994). These systems, however, may be Incremental extension of the benefit package to attain administratively cumbersome and costly or may be comprehensive coverage extremely crude, such as flat-rate contributions. Conse quently, countries that have large or growing informal The approach to extend the mandated benefit package sectors should consider alternatives to shi unless some was incremental as well. It occurred along three means of including the informal sector in the social dimensions insurance cover has been shown to work well First, the largest changes in the scope of the benefit Yet, even in a context of high formal sector employ- package were brought about by the introduction of ment payroll deductions may prove problematic as the types of statutory social insurance system, such as (sole) revenue base for insurance. In Germany, as in accident, pension and unemployment insurance. Each other developed countries, wages and salaries constitute new type extended the benefit package to an area of a decreasing proportion of total GDP as the contribu- social need, which the social net had not covered before tion of business profits and capital investment to GDP The principle of compulsion applied to the same group growing(Statistisches Bundesamt, 1997: OECD, 1998). of the population as before. Today, all types of statutory As a result, payroll deduction rates to SHI in Germany insurance cover health-related benefits, The latest increased, even in times when sickness fund expenditures addition to the statutory insurance system was long proportion of GDP remained fairly constant term nursing care insurance. Introduced in 1995, it pays Barnighausen, 2000: Barnighausen et al., 1999; Braun, for ambulatory as well as in-patient nursing care(Bloch, Kuhn, Reiners, 1998) Hillebrandt, Wolf, 1997) A Bismarckian health insurance system can, in fact Second, the benefit packages of already existing types be implemented in a MIC following an incremental of the statutory insurance were gradually extended to process very similar to that in Germany. In 1965, the additional disease groups and services. Examples en first voluntary health insurance fund was organised in compass occupational diseases(which were added to the Korea. By 1977, when compulsory insurance was first coverage under the statutory accident insurance in 1925 ntroduced, there were ll voluntary funds, which and 1929), the treatment of sexually transmitted diseases covered about 0.2% of the population. Compulsory and a broad spectrum of preventive measures(which insurance was expanded vertically step by step to were added to the benefit package of the statutory health companies with 500, 300, 100 and finally 16 employees 1952 and 1955, respectively). over the following six years. Similarly, coverage was Third, already existing benefits were more or less expal ally to government gradually increased in amount or durati private school teachers(1979)and families of military stance, amount and duration of sick pay were increased servicemen and employees of private school foundations in 1957; the time limit on coverage of in-patient care was (1980). Universal compulsory coverage was achieved 26 eliminated in 1974(Lang, 1925; Peters, 1978; Winte years after the establishment of the first stein, 1980a, b). through schemes covering the rural and the urban sel The expansion of both coverage and benefits has lead employed (in 1988 and 1989)(Moon, 1998: Peabody to a gradual transformation of the statutory health Lee,& Bickel, 1995: Anderson, 1989). The achievement insurance system. On the one hand, as more and more f universality in South Korea shows that a Bismarckian groups of society fell under the laws of compulsory

insurance (Holler, 1977). This reflects the difficulty in assessing and taxing farmers’ incomes. For many countries, especially those where farmers constitute a large proportion of the total population, such as China, Vietnam and Thailand, such a delay may not be practicable. The difference in context between 19th century Germany and today’s LMICs is even more prominent in the case of the informal sector. In Germany, the proportion of informal sector workers declined during the years of the system’s existence. In many LMIC, on the other hand, the informal sector is expected to continue to grow in the next years (ILO, 1992, 1996, 1999). Self-employed farmers and informal sector workers are hard to cover in a SHI system since their incomes fluctuate and are hard to assess objectively. Still, many systems have been devised for assessing the incomes of the self-employed and charging contributions (Normand & Weber, 1994). These systems, however, may be administratively cumbersome and costly or may be extremely crude, such as flat-rate contributions. Conse￾quently, countries that have large or growing informal sectors should consider alternatives to SHI, unless some means of including the informal sector in the social insurance cover has been shown to work well. Yet, even in a context of high formal sector employ￾ment payroll deductions may prove problematic as the (sole) revenue base for insurance. In Germany, as in other developed countries, wages and salaries constitute a decreasing proportion of total GDP as the contribu￾tion of business profits and capital investment to GDP is growing (Statistisches Bundesamt, 1997; OECD, 1998). As a result, payroll deduction rates to SHI in Germany increased, even in times when sickness fund expenditures as a proportion of GDP remained fairly constant (Barnighausen, 2000; B . arnighausen et al., 1999; Braun, . Kuhn, & Reiners, 1998). . A Bismarckian health insurance system can, in fact, be implemented in a MIC following an incremental process very similar to that in Germany. In 1965, the first voluntary health insurance fund was organised in Korea. By 1977, when compulsory insurance was first introduced, there were 11 voluntary funds, which covered about 0.2% of the population. Compulsory insurance was expanded vertically step by step to companies with 500, 300, 100 and finally 16 employees over the following six years. Similarly, coverage was expanded horizontally to government officials and private school teachers (1979) and families of military servicemen and employees of private school foundations (1980). Universal compulsory coverage was achieved 26 years after the establishment of the first voluntary fund through schemes covering the rural and the urban self￾employed (in 1988 and 1989) (Moon, 1998; Peabody, Lee, & Bickel, 1995; Anderson, 1989). The achievement of universality in South Korea shows that a Bismarckian health insurance can be established in a country with a social, political and cultural history which is very different from that in Germany. It also shows that the pace of incremental development can be much acceler￾ated. It has to be kept in mind, however, that this happened against a backdrop of fast and sustained economic growth and a shrinking informal sectorFtwo conditions that, while neither necessary nor sufficient, are conducive to establishing a SHI. In sum, Germany succeeded in achieving universal coverage following an incremental pattern of expanding compulsory insurance. This success has been contingent on a number of social, economic and institutional circumstances. Yet, as the above cases suggests, a similar approach holds promise for countries that have already established partial coverage, such as many countries in transitional Asia and South America. Incremental extension of the benefit package to attain comprehensive coverage The approach to extend the mandated benefit package was incremental as well. It occurred along three dimensions. First, the largest changes in the scope of the benefit package were brought about by the introduction of new types of statutory social insurance system, such as accident, pension and unemployment insurance. Each new type extended the benefit package to an area of social need, which the social net had not covered before. The principle of compulsion applied to the same groups of the population as before. Today, all types of statutory insurance cover health-related benefits. The latest addition to the statutory insurance system was long￾term nursing care insurance. Introduced in 1995, it pays for ambulatory as well as in-patient nursing care (Bloch, Hillebrandt, & Wolf, 1997). Second, the benefit packages of already existing types of the statutory insurance were gradually extended to additional disease groups and services. Examples en￾compass occupational diseases (which were added to the coverage under the statutory accident insurance in 1925 and 1929), the treatment of sexually transmitted diseases and a broad spectrum of preventive measures (which were added to the benefit package of the statutory health insurance in 1952 and 1955, respectively). Third, already existing benefits were more or less gradually increased in amount or duration. For in￾stance, amount and duration of sick pay were increased in 1957; the time limit on coverage of in-patient care was eliminated in 1974 (Lang, 1925; Peters, 1978; Winter￾stein, 1980a, b). The expansion of both coverage and benefits has lead to a gradual transformation of the statutory health insurance system. On the one hand, as more and more groups of society fell under the laws of compulsory 1568 T. Barni . ghausen, R. Sauerborn / Social Science & Medicine 54(2002) 1559–1587