Suppose you were located behind a veil of ignorance and given the choice as to whether to receive your health care in a Bismarck social insurance system or a tax-based Beveridge system. How much confidence could you have in your ability to choose prudently, as judged by the health outcomes you could rationally expect to enjoy, based solely on the knowledge of which type of system you would be treated in? If Or et al. are right, the answer is ‘not very much’. A simple rank ordering of Bismarck and Beveridge systems would be of limited use in terms of predicting health outcomes, especially if your choice were limited to one between the social insurance countries of France and Germany or the tax-based systems of Denmark, Sweden and the United Kingdom. In particular, if your ranking were based solely on the health outcomes associated with each type of system, you would rationally opt for either France or Sweden, a preference that – unhappily for health policy comparativists – straddles the Beveridge/Bismarck divide.
Now suppose that your veil of ignorance were lifted somewhat, and you knew that you would be a man in his fifties. In that case, you would have good reason to choose either of the social insurance systems over any of the others, as five-year survival rates for prostate cancer are higher in France and Germany – 73.7 and 76.4%, respectively, as reported – than for any of the three tax-based systems, the best of which is in Sweden (66.0%) and the worst in Denmark (38.4%). On the other hand, as a woman concerned about breast cancer, by the same test of survival rates, you would opt for Sweden (82.0% compared with the next best of 79.8% in France). Suppose we imagine the thought experiment of asking rational decision makers still in the womb to make a prudent choice of country, then, given live birth survival rates, they would also opt for Sweden, with France second. Finally, for those on the verge of retirement, life expectancy at 65 years is pretty evenly split between France (20.4) and Sweden (19.2). Putting these results together, the overall result of all these possible comparisons is that, as Or et al. summarise, there is “no clear-cut distinction between Beveridge and Bismarck-type countries. France and Sweden consistently outperform the UK and Denmark while Germany is typically placed in between” (Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010: 272).
This conclusion is also notable for the fact that it is broadly consistent with an earlier study by Figueres et al. (Reference Figueras, Saltman, Busse and Dubois2004), which compared the performance of eight social insurance countries (Austria, Belgium, France, Germany, Israel, Luxembourg, the Netherlands and Switzerland) against a range of tax-based systems. Figueres et al. do not use the same measures of health outcomes as Or et al., focusing instead on life expectancy and avoidable mortality. However, by their measures Germany is located below France and Sweden, the two top performers, and above the United Kingdom and Denmark, the relative laggards.
Comparison between Beveridge and Bismarck schemes becomes even more complex once the effects of country selection are taken into account. Consider the question of waiting times, which is one variable for which Or et al. do note a systemic difference between their social insurance and tax-based schemes. The clarity of this difference disappears in the Figueres et al. (Reference Figueras, Saltman, Busse and Dubois2004: 100) study, because the Netherlands shows comparatively long waiting times. Similarly, the better performance of tax-based systems on the question of cost control, another systematic difference noted by Or et al., is thrown into doubt by the comparison between Spain and Portugal. Both spent a similar proportion of GDP on health in 1980 (around 5.5%), but Portugal increased its spend to 9.0% in 2000, whereas Spain increased its spend only to 7.5% of GDP (Figueres et al., Reference Figueras, Saltman, Busse and Dubois2004: 118). Even in 2011, after the effects of the recession and austerity budgets, Portugal still recorded a spend of 10.2% of GDP compared with 9.3% for Spain (OECD, 2013, table A.6).
In short, for our hypothetical decision maker located behind a veil of ignorance, the bare knowledge that one system fell into the social insurance category and another system fell into the tax-based category would be little help in making a prudent choice between them. As far as health system performance goes, within group variance is large relative to between-group variance. Given these findings, is there any point in continuing to think about the difference between Bismarck and Beveridge systems? Even if we suppose that we are not in the game of ranking countries, with the folly that such exercises involve (Oliver, Reference Oliver2012), there is still the question of whether the distinction between Bismarck and Beveridge systems remains useful, a view that some have doubted on the grounds that social insurance schemes have come to rely more heavily on tax-based funding (Saltman, Reference Saltman2012: 12). If outcomes overlap and sources of funding are becoming more similar, does the distinction still serve any useful purpose?
One possible reason for a positive answer to this question relates less to health outcomes, which given the range and diversity of factors that bear on health status will always be hard to relate to organisational differences, and more to the norms of governance characteristic of each system. From a narrow health economics viewpoint, such matters are relatively unimportant. Governance structures are just different forms of production that matter only insofar as they yield different outcomes. As Alexander Pope once put it: “For forms of government, let fools contest. Whate’re is best administered is best”. However, health care institutions are an important element in a society’s basic institutional structure, and there are typically good historical and cultural reasons why a system takes the particular form it does and remains relatively stable in its essentials over time. For example, the decentralised nature of German policy making is not just a policy preference, but is built into Germany’s Basic Law and the constitutional norms that Law embodies. In this context, pressures to centralisation are likely to be met by considerable resistance on the part of independent bodies (Kieslich, Reference Kieslich2012: 375). By the same token, political developments of a general kind outside the health care sector may well have considerable implications for health care governance. In the United Kingdom in particular, the devolution of power to the constituent nations of the Union since 1997 has had important consequences for the organisation of health care, leading, as Or et al. note, to different decisions between Scotland and Wales on the one hand and England on the other about the purchaser–provider split. (In this context, however, I note that Or et al. flip-flop a number of times in the reporting of their data between ‘England’ and the ‘United Kingdom’, suggesting that there are nuances to the issues of centralisation and decentralisation that have not found their way into the analysis.)
Questions about the centralisation or decentralisation of decision making are not simply technical matters of decision structures, but they have an importance of their own. Observers have noted, for example, that the German sickness funds attach great weight to their autonomy and independence. The effect seems to be to diffuse controversy about health policy, by contrast with the UK’s National Health Service, in which relatively detailed matters of service provision and performance can suddenly become political hot potatoes. I have not seen a thorough comparative content analytical study, but those who know both the United Kingdom and Germany are struck by the contrast between the political prominence of health policy in the former but not in the latter. (It is, of course, an interesting question as to whether or not policy making is better when conducted in the glare of political controversy.)
Questions of organisational and constitutional structure are closely related to the values that different systems pursue. One of the most interesting, but underdeveloped, aspects of the analysis by Or et al. (Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010, table 1) is their discussion of the social values and animating principles of the different types of system. According to their analysis, Beveridge systems are based upon the principles of universality and equity, whereas the social insurance systems are based on the principles of plurality, liberty and solidarity. However, earlier in the paper they write that the “principal objectives of the healthcare system in all countries are to optimise health outcomes, quality of care, ease and equity of access, subject to the constraints of containing costs” (Or et al., Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010: 271). A way of reconciling these two sets of statements is to assume that there is a list of objectives that all systems might be thought to pursue, but in practice, given organisational constraints, some will score well on some values, whereas others will score well in different terms.
However, this way of thinking ignores the extent to which values are endogenous to political organisation and culture. Consider the contrast between the principles of equity and solidarity, for example. Both principles underwrite universal access, understood as a reduction in the financial and other barriers to access that citizens confront. From this point of view, the claim that one type of system favours universality, whereas another type does not involves some category mistake. However, the meaning of universal access is different in the two conceptions. The principle of equity makes sense in a system in which the policy-making problem is one in which the central dilemma for the government is to allocate resources efficiently between competing uses. In a fixed budget system, like that of the United Kingdom, the equity principle is there to guide the way in which the government, as a benevolent central social planner, allocates its resources. By contrast, the principle of solidarity makes sense in a system in which the fundamental purpose of social insurance is financial risk pooling, and in which the normal actuarial principles of risk discrimination are not allowed to operate for reasons of communal cohesion (Saltman, Reference Saltman2004: 5). This does not mean that the principle of solidarity has no conceivable role in UK health policy. Policies on blood and organ donation may well be thought of in these terms. However, when thinking about financial provision, solidarity as a principle does not make sense outside of a system in which autonomous funds act on the principle of pooling their own funds across social groups. Conversely, equity, when it means more than a principle of universal access, has no meaning in a system in which there is a plurality of payers responsible for their own funds, rather than a central social planner seeking to allocate resources.
To note these points might seem to preclude any comparative analysis. If concepts are only meaningful within an institutionally embedded policy paradigm, we cannot generalise even within the class of Beveridge- or Bismarck-type schemes. However, this would be too strong a conclusion. The members of each class might have strong family resemblances to one another, even when they were not prepared to speak the language of other families. Our hypothetical agent behind the veil of ignorance will require the art of translation as well as the ability to rank alternatives.
Suppose you were located behind a veil of ignorance and given the choice as to whether to receive your health care in a Bismarck social insurance system or a tax-based Beveridge system. How much confidence could you have in your ability to choose prudently, as judged by the health outcomes you could rationally expect to enjoy, based solely on the knowledge of which type of system you would be treated in? If Or et al. are right, the answer is ‘not very much’. A simple rank ordering of Bismarck and Beveridge systems would be of limited use in terms of predicting health outcomes, especially if your choice were limited to one between the social insurance countries of France and Germany or the tax-based systems of Denmark, Sweden and the United Kingdom. In particular, if your ranking were based solely on the health outcomes associated with each type of system, you would rationally opt for either France or Sweden, a preference that – unhappily for health policy comparativists – straddles the Beveridge/Bismarck divide.
Now suppose that your veil of ignorance were lifted somewhat, and you knew that you would be a man in his fifties. In that case, you would have good reason to choose either of the social insurance systems over any of the others, as five-year survival rates for prostate cancer are higher in France and Germany – 73.7 and 76.4%, respectively, as reported – than for any of the three tax-based systems, the best of which is in Sweden (66.0%) and the worst in Denmark (38.4%). On the other hand, as a woman concerned about breast cancer, by the same test of survival rates, you would opt for Sweden (82.0% compared with the next best of 79.8% in France). Suppose we imagine the thought experiment of asking rational decision makers still in the womb to make a prudent choice of country, then, given live birth survival rates, they would also opt for Sweden, with France second. Finally, for those on the verge of retirement, life expectancy at 65 years is pretty evenly split between France (20.4) and Sweden (19.2). Putting these results together, the overall result of all these possible comparisons is that, as Or et al. summarise, there is “no clear-cut distinction between Beveridge and Bismarck-type countries. France and Sweden consistently outperform the UK and Denmark while Germany is typically placed in between” (Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010: 272).
This conclusion is also notable for the fact that it is broadly consistent with an earlier study by Figueres et al. (Reference Figueras, Saltman, Busse and Dubois2004), which compared the performance of eight social insurance countries (Austria, Belgium, France, Germany, Israel, Luxembourg, the Netherlands and Switzerland) against a range of tax-based systems. Figueres et al. do not use the same measures of health outcomes as Or et al., focusing instead on life expectancy and avoidable mortality. However, by their measures Germany is located below France and Sweden, the two top performers, and above the United Kingdom and Denmark, the relative laggards.
Comparison between Beveridge and Bismarck schemes becomes even more complex once the effects of country selection are taken into account. Consider the question of waiting times, which is one variable for which Or et al. do note a systemic difference between their social insurance and tax-based schemes. The clarity of this difference disappears in the Figueres et al. (Reference Figueras, Saltman, Busse and Dubois2004: 100) study, because the Netherlands shows comparatively long waiting times. Similarly, the better performance of tax-based systems on the question of cost control, another systematic difference noted by Or et al., is thrown into doubt by the comparison between Spain and Portugal. Both spent a similar proportion of GDP on health in 1980 (around 5.5%), but Portugal increased its spend to 9.0% in 2000, whereas Spain increased its spend only to 7.5% of GDP (Figueres et al., Reference Figueras, Saltman, Busse and Dubois2004: 118). Even in 2011, after the effects of the recession and austerity budgets, Portugal still recorded a spend of 10.2% of GDP compared with 9.3% for Spain (OECD, 2013, table A.6).
In short, for our hypothetical decision maker located behind a veil of ignorance, the bare knowledge that one system fell into the social insurance category and another system fell into the tax-based category would be little help in making a prudent choice between them. As far as health system performance goes, within group variance is large relative to between-group variance. Given these findings, is there any point in continuing to think about the difference between Bismarck and Beveridge systems? Even if we suppose that we are not in the game of ranking countries, with the folly that such exercises involve (Oliver, Reference Oliver2012), there is still the question of whether the distinction between Bismarck and Beveridge systems remains useful, a view that some have doubted on the grounds that social insurance schemes have come to rely more heavily on tax-based funding (Saltman, Reference Saltman2012: 12). If outcomes overlap and sources of funding are becoming more similar, does the distinction still serve any useful purpose?
One possible reason for a positive answer to this question relates less to health outcomes, which given the range and diversity of factors that bear on health status will always be hard to relate to organisational differences, and more to the norms of governance characteristic of each system. From a narrow health economics viewpoint, such matters are relatively unimportant. Governance structures are just different forms of production that matter only insofar as they yield different outcomes. As Alexander Pope once put it: “For forms of government, let fools contest. Whate’re is best administered is best”. However, health care institutions are an important element in a society’s basic institutional structure, and there are typically good historical and cultural reasons why a system takes the particular form it does and remains relatively stable in its essentials over time. For example, the decentralised nature of German policy making is not just a policy preference, but is built into Germany’s Basic Law and the constitutional norms that Law embodies. In this context, pressures to centralisation are likely to be met by considerable resistance on the part of independent bodies (Kieslich, Reference Kieslich2012: 375). By the same token, political developments of a general kind outside the health care sector may well have considerable implications for health care governance. In the United Kingdom in particular, the devolution of power to the constituent nations of the Union since 1997 has had important consequences for the organisation of health care, leading, as Or et al. note, to different decisions between Scotland and Wales on the one hand and England on the other about the purchaser–provider split. (In this context, however, I note that Or et al. flip-flop a number of times in the reporting of their data between ‘England’ and the ‘United Kingdom’, suggesting that there are nuances to the issues of centralisation and decentralisation that have not found their way into the analysis.)
Questions about the centralisation or decentralisation of decision making are not simply technical matters of decision structures, but they have an importance of their own. Observers have noted, for example, that the German sickness funds attach great weight to their autonomy and independence. The effect seems to be to diffuse controversy about health policy, by contrast with the UK’s National Health Service, in which relatively detailed matters of service provision and performance can suddenly become political hot potatoes. I have not seen a thorough comparative content analytical study, but those who know both the United Kingdom and Germany are struck by the contrast between the political prominence of health policy in the former but not in the latter. (It is, of course, an interesting question as to whether or not policy making is better when conducted in the glare of political controversy.)
Questions of organisational and constitutional structure are closely related to the values that different systems pursue. One of the most interesting, but underdeveloped, aspects of the analysis by Or et al. (Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010, table 1) is their discussion of the social values and animating principles of the different types of system. According to their analysis, Beveridge systems are based upon the principles of universality and equity, whereas the social insurance systems are based on the principles of plurality, liberty and solidarity. However, earlier in the paper they write that the “principal objectives of the healthcare system in all countries are to optimise health outcomes, quality of care, ease and equity of access, subject to the constraints of containing costs” (Or et al., Reference Or, Cases, Lisac, Vrangbæk, Winblad and Bevan2010: 271). A way of reconciling these two sets of statements is to assume that there is a list of objectives that all systems might be thought to pursue, but in practice, given organisational constraints, some will score well on some values, whereas others will score well in different terms.
However, this way of thinking ignores the extent to which values are endogenous to political organisation and culture. Consider the contrast between the principles of equity and solidarity, for example. Both principles underwrite universal access, understood as a reduction in the financial and other barriers to access that citizens confront. From this point of view, the claim that one type of system favours universality, whereas another type does not involves some category mistake. However, the meaning of universal access is different in the two conceptions. The principle of equity makes sense in a system in which the policy-making problem is one in which the central dilemma for the government is to allocate resources efficiently between competing uses. In a fixed budget system, like that of the United Kingdom, the equity principle is there to guide the way in which the government, as a benevolent central social planner, allocates its resources. By contrast, the principle of solidarity makes sense in a system in which the fundamental purpose of social insurance is financial risk pooling, and in which the normal actuarial principles of risk discrimination are not allowed to operate for reasons of communal cohesion (Saltman, Reference Saltman2004: 5). This does not mean that the principle of solidarity has no conceivable role in UK health policy. Policies on blood and organ donation may well be thought of in these terms. However, when thinking about financial provision, solidarity as a principle does not make sense outside of a system in which autonomous funds act on the principle of pooling their own funds across social groups. Conversely, equity, when it means more than a principle of universal access, has no meaning in a system in which there is a plurality of payers responsible for their own funds, rather than a central social planner seeking to allocate resources.
To note these points might seem to preclude any comparative analysis. If concepts are only meaningful within an institutionally embedded policy paradigm, we cannot generalise even within the class of Beveridge- or Bismarck-type schemes. However, this would be too strong a conclusion. The members of each class might have strong family resemblances to one another, even when they were not prepared to speak the language of other families. Our hypothetical agent behind the veil of ignorance will require the art of translation as well as the ability to rank alternatives.