Article Text

other Versions

PAPER
Income-based equity weights in healthcare planning and policy
1. Anders Herlitz1,2
1. lDepartment of Philosophy, Linguistics and Theory of Science, University of Gothenburg, Gothenburg, Sweden
2. 2Department of Philosophy, Rutgers University—the State University of New Jersey, New Brunswick, New Jersey, USA
1. Correspondence to Dr Anders Herlitz, Department of Philosophy, Linguistics and Theory of Science, University of Gothenburg, Box 100, Gothenburg SE-405 30, Sweden; andersherlitz{at}gmail.com

Abstract

Recent research indicates that there is a gap in life expectancy between the rich and the poor. This raises the question: should we on egalitarian grounds use income-based equity weights when we assess benefits of alternative benevolent interventions, so that health benefits to the poor count for more? This article provides three egalitarian arguments for using income-based equity weights under certain circumstances. If income inequality correlates with inequality in health, we have reason to use income-based equity weights on the ground that health inequality is bad. If income inequality correlates with inequality in opportunity for health, we have reason to use such weights on the ground that inequality in opportunity for health is bad. If income inequality correlates with inequality in well-being, income-based equity weights should be used to mitigate inequality in well-being. Three different ways in which to construe income-based equity weights are introduced and discussed. They can be based on relative income inequality, on income rankings and on capped absolute income. The article does not defend any of these types of weighting schemes, but argues that in order to settle which of these types of weighting scheme to choose, more empirical research is needed.

• Allocation of Health Care Resources
• Political Philosophy
• Health Economics
• Health Care Economics
• Distributive Justice

Statistics from Altmetric.com

A recent study found that, in 2001–2014, the gap in life expectancy between the richest 1% and the poorest 1% in the USA was 14.6 years.1 This actualises the question: when we assess and compare alternatives with the purpose of making benevolent interventions, should we on egalitarian grounds weight individual health benefits with reference to income, so that benefits to the poorer count for more? Many proponents of cost-effectiveness analysis recognise that one needs to introduce equity weights to take the distribution of health into account.2–8 Yet, what to base these weights on is a contested issue. Some propose that the age should be taken into account.9 ,10 Some propose that weights should be based on proportional shortfalls in health outcomes.7 ,11 Some propose that non-health equities in life should ground the equity weights.12 In this paper, I will argue that we sometimes have good reasons to introduce and use income-based equity weights in healthcare planning. I present three types of income-based equity weights and suggest that all of these might be reasonable in some circumstances. I will not argue that we ought to take egalitarian concerns into account and give extra weights to benefits to the worse off. There is a large body of literature defending this, and there is no space to repeat these arguments here.5 ,13–15 Rather, I will address the question: if we care about equality, is it reasonable to use income-based equity weights?

Before addressing this, a note on ‘income’ is warranted. The study mentioned above looks at tax records to measure income.1 This might be problematic if we are interested in income in a wider sense, for example, ‘accessible economic resources’. Wealthy individuals sometimes pay low taxes, and there might be large fluctuations in how much individuals pay in taxes from year to year. It is not certain that tax records provide us with the best indicator for individuals' economic situation, and there are other possible indicators that might do better. In what follows, I will, for reasons of space, sidestep these problems. I will assume that we have an adequate income indicator that approximates the overall economic situation (ie, accessible economic resources) of different individuals during a relevant period of time. Tax records might be the best (useable and roughly accurate) way of estimating economic situations, but it is conceivable that some other indicator could be better. It is beyond the scope of this paper to settle this issue. Thus, when I speak of income-based equity weights, I will refer to equity weights that are based on a roughly accurate indicator of the overall economic situation of individuals in a well-defined and relevant period of time.

Equity weights are weights that are applied to units of individual benefits in order to account for their distribution. For example, if we let the numbers represent the amount of a good and the brackets indicate different individuals, many hold that A is worse than B:13–14

• A: |100|200|300|

• B: |200|200|200|

Without equity weights, a cost-effectiveness analysis cannot discern between these outcomes. The total sum of benefits is the same in both outcomes. The judgement that B is better than A can, however, be easily established if we introduce equity weights. For example, if we ascribe weights in accordance with the priority view, so that benefits count for more the less an individual has, we can establish that B is better than A, since 300–200 would count for <200–100.4 ,13 ,15 Essentially, equity weights can thus adjust or qualify summations with reference to considerations other than the total sum of the good.

Equity weights can be plausibly based on the unit of measurement in some situations. This seems to be the case when the unit of measurement is extensionally equivalent to a unique good (eg, well-being), and this good's distribution matters. In some situations, it is problematic to base equity weights on the unit of measurement because what matters is not reflected in the unit of measurement and its distribution. Consider the following two outcomes, P and Q, in a situation in which we can distribute good health across a population and use this as the unit of measurement, and also care about and have data on well-being (let the order of the individuals be the same under both ‘Good health’ and ‘Well-being’):

• Good health         Well-being

• P:|250|250|200|     |110|150|190|

• Q:|200|250|250|     |100|150|200|

In light of good health only, these outcomes are equivalent (they are equally unequal distributions of 700 units of health). No weights that are based only on good health would be able to differentiate between P and Q. Yet, we might have egalitarian reasons to give extra weight to benefits that befall individuals who fare worse in terms of well-being, that is, favour P. In order to get that result, we must base the equity weights on something other than the unit of measurement of that which we distribute, that is, something different than good health. The question that I will give an assertive answer to below is: do we sometimes have reason to use income as a basis for equity weights when we distribute good health?

In the ‘Arguments for income-based equity weights’ section, I introduce three arguments for using income-based equity weights in healthcare planning and policy. In the ‘Types of income-based equity weights’ section, I discuss three different types of income-based equity weights that can be used. I end the paper with a brief discussion on issues that need to be studied further.

Arguments for income-based equity weights

Consider two intervention alternatives, X and Y, that generate the same total sum of good health, 100, but the distribution of which differs. This might concern a choice between two different drugs that we can subsidise. Let each bracket demarcate the health benefits that will befall a specific income quartile. The poorest are at the extreme left and the richest at the extreme right:

• X: |10|20|30|40|

• Y: |40|30|20|10|

Income-based equity weights can help us put into practice the judgement that Y is better than X. I will introduce three egalitarian reasons for using such weights. First, weights can be grounded in the fact that income differences sometimes correlate with health inequalities. Second, weights can be grounded in the fact that income differences correlate with differences in opportunity for health. Third, weights can be grounded in the fact that income differences can be expected to correlate with inequalities in well-being under standard interpretations of what well-being is. Importantly, I will not argue that income-based equity weights should be used because there is a causal relation between income and health, opportunity for health or well-being. Such connections might or might not exist, but none of my arguments depends on it.16

First, there are egalitarian reasons to hold that inequality in health should be mitigated.5 ,17 When income differences correlate with inequalities in health, we have reason to use income-based equity weights. The study cited above found a strong correlation between longevity and income. The richest 1% live a whole 14.6 years longer than the poorest 1%.1 This, as well as other research that point to correlation between health aspects and income, gives us reason to seriously consider income-based equity weights.3 Longevity is on the one hand an indicator of health.18 On the other hand, a longer life is, everything else equal, a life with more health in it. Insofar as we care about the equal distribution of good health, we should care about the distribution of life expectancy. If we can allocate a certain amount of good health to either of two individuals of equal age, and one of the individuals has a greater life expectancy, we have some egalitarian reason to give priority to the individual with the lower life expectancy. If income is an indicator of life expectancy or any other health aspect, we have egalitarian reason to give extra weight to health benefits that befall individuals with lower income. On this view, it is in some respect better to provide a certain amount of good health to an individual in the poorest 1% than to an individual in the richest 1%, because the richer person can be expected to have better health.

Second, it is often suggested that equality of opportunity for health matters.3 ,19 Income is, plausibly, often correlated with opportunity for health. It is the privilege of the rich to make use of private health clinics that provide health services that are not accessible to all (eg, rehab clinics for addicts, high-end surgical clinics), to jump queues (eg, to see specialists), to travel across the world to bypass local legislation (eg, to access healthcare services that might be illegal where the person lives) and so on. Thus, insofar as we care about equality of opportunity for health, and insofar as income is a decent indicator of inequality in opportunity for health, we have reason to use income-based equity weights. On this view, it is better to provide a certain amount of good health to an individual with a lower income, because he/she has fewer opportunities.

Third, it is widely held that inequalities in well-being matter.15 ,20–21 If income differences correlate with such inequalities, we have reason to use income-based equity weights. Philosophers disagree about what well-being is. Some hold that it is happiness, some that it is preference fulfilment, and some hold that well-being consists of a list of things, such as happiness, friendships and family relations.22 If income correlates with differences in life expectancy or other conditions that are needed for human flourishing, it also plausibly correlates with well-being, regardless of which view one takes of what well-being is. The longer a person lives, the more happiness he/she can experience, the more of his/her preferences can be fulfilled and the more of the items on an objective list can be achieved. This is not to say that there is always a correlation between income and well-being, a proposition that Daniel Kahneman and Angus Deaton have refuted.23 Yet, the correlation plausibly holds for at least some income inequalities (which the study of Kahneman and Deaton also shows). Insofar as we care about equality of well-being, and insofar as differences in income correlate with differences in conditions for living an active life, we seem to have reason to use income-based equity weights. On this view, it is better to allocate a certain amount of good health to a poorer individual, because we have reason to believe that he/she has a lower well-being level.

One could object to using income-based equity weights by pointing to how it conflates different ‘spheres of justice’.24 When allocating health resources, it could be argued that we should only care about health-related issues, and income is something else; the purpose of healthcare planning and policy is after all to provide health. I believe that this objection fails to invalidate income-based equity weights for two reasons. First, it is not clear that we should accept the idea that the different spheres should be kept separate. Some consequentialists, for example, would object to separating any spheres. Yet, even if one accepts that spheres of justice sometimes should be separated, it is not clear that this is an instance where this is the case. Healthcare policy and planning, in particular at macro levels, is part of a general social apparatus where spheres cannot be easily separated. Second, this objection only applies to the reasons to use income-based equity weights to mitigate non-health-related inequality. The first two arguments above belong to the health sphere; so, even proponents of the view that we should keep spheres separated should see these as at least relevant.

A second objection to using income-based equity weights points to how it is problematic to use income information in this way. We might not have this information, and we might not want decision-makers to have the information. This is a valid worry when we think about clinical decision-making, and more generally decision-making where decision-makers are close to the patients. Hospitals and doctors do not keep records of patients' income, and we do not want them to because it might distort the mission of doctors and violate patient privacy. However, at macro levels, things are different. We know that certain benevolent interventions will benefit poorer income groups more (eg, because of the geographical distribution of the benefits, or since certain diseases are more prevalent in certain income groups). Decision-makers are in these situations far from the beneficiaries, do not have personal information about them, and it might even be impossible to have an efficient system where decision-makers are unable to predict the extent to which interventions benefit different income groups differently.

Finally, it could be the case that income is a too coarse indicator of what matters. Perhaps, it is the case that income inequalities correlate with health-related and well-being inequalities, but this is only because there is a correlation at subgroup levels (eg, the correlation appears for women only). This is clearly a theoretical possibility. However, it does not invalidate income-based equity weights as such. Rather, it shows that we might discover that we have reason to qualify the income-based equity weights in certain ways.

Given certain empirical conditions under which income is an indicator of health, opportunities for health and/or well-being, we have egalitarian reason to introduce income-based equity weights. Since these arguments rely on income as an indicator of something (health, opportunity, well-being), these reasons are unrelated to whether the income is fairly distributed. It could even be suggested that unfair distributions of income provide us with an additional reason to use income-based equity weights, since that would be a way to correct for some unfairness.

Yet, it is not clear what sort of income-based equity weights we should consider. In the next section, I will discuss three different types of income-based equity weights. Before that, two points should be stressed. First, accepting that income-based equity weights should be used does not mean that one has to be committed to the view that they are all the equity weights that matter. It might well be that other equity weights are justified too, and that we ought to ideally combine a plurality of different types of equity weights. For example, it might be the case that we have special reasons to protect the worse off in health, opportunity and well-being, and income might not be a good indicator to capture this. Second, income-based equity weights might not be applicable in every situation. I have argued that we have egalitarian reasons to sometimes use equity weights based on income. This does not mean that we should always use income-based equity weights.

Types of income-based equity weights

When ascribing different weights to different benefits, depending on how the beneficiaries do in terms of income, for example, in a cost-effectiveness analysis, how should we design the weighting scheme? Some questions relating to this cannot be answered without empirical data. Yet, we can make some conceptual progress without such data. I will discuss three different types of equity weights that are based on income:

• Relative income-based equity weights

• Rank-based equity weights based on income

• Capped absolute income-based equity weights

These are not the only types of weighting scheme that we can construe, but they seem to be plausible in some situations.

Relative income-based equity weights

Income-based equity weights can be based on relative income inequality, so that the size of the equity weight represents the relative income inequality. For example, in a population of two individuals where one person earns US$100 000/year and the other earns US$50 000/year, such an approach could provide us with equity weights that directly correspond to this: the latter person earns half as much, so the equity weight that qualifies the value of a benefit to him/her should be twice as big. This might look like a neat approach that simply translates relative income inequality to equity weights. Yet, there are at least two problems with this approach. First, that income inequality might correlate with inequality in health, opportunity for health or well-being does not mean that the magnitude of income inequality correlates with the magnitudes of these other inequalities. Second, when we consider larger populations, this type of approach faces theoretical complications. The philosophical literature on inequality has unveiled that inequality is a highly complex notion and that there are many different aspects of inequality.14 There are different ways to measure inequality. One can plausibly measure inequality in terms of, for example, total sum of distance from mean, total sum of distance from the best of and total sum of distance to every individual who is better off. Consider the following population:

• |US$100 000|US$100 000|US$20 000| Should the equity weight that we apply to benefits to the individual with income US$20 000 be based on the distance to the mean (ie, US$73 333–20 000), on the distance to the best off (ie, US$100 000–20 000) or on the total sum of distances to everyone who is better off (ie, US$100 000–20 000 +US$100 000–20 000)? This needs to be settled before we can use equity weights that are grounded in relative inequality. Income-based equity weights that are inferred from relative inequality seem reasonable in circumstances where there is a correlation between relative income and the magnitude of inequality in health, opportunity for health and/or well-being, but more work is needed to figure out which aspects of inequality to take into account in this context.

Rank-based equity weights based on income

A different way of construing income-based equity weights looks at the income ranking and establishes weights based on this. Perhaps, we should not care about the size of income differences when we construe the equity weights, but instead look at the income ranking. This sort of income-based equity weight could be based on precise places in the ranking, so that benefits that befall the individual with the 537th largest income receive larger weight than benefits that befall the individual with the 536th largest income, and lower weight than benefits that befall the individual with the 538th largest income. Yet, more plausibly, this type of income-based equity weights could be based on income percentiles. For example, one could introduce equity weights, so that health benefits to individuals in the top 1% receive the same weight, health benefits to individuals in the second top 1% receive the same weight and so on. One reason for using such an approach is that the income indicator is likely to be imperfect, and one way of reducing the impact of these imperfections is to base equity weights on more coarse data. This might be the most practical way to address income-based equity weights, and it is the approach that in the most obvious way finds support in the findings from the study on income and life expectancy in the USA. This type of income-based equity weights seems reasonable in circumstances where there is a correlation between income rank and health, opportunity for health and/or well-being, and where the differences between different income ranks correlate with significant differences in health, opportunity and/or well-being.

Capped absolute income-based equity weights

Finally, we could base equity weights on absolute income and use caps above and/or below which we let the weight be the same for each individual. Using upper caps is similar to sufficientarian approaches to distributive fairness that hold that we only have reason to give extra weight to benefits to individuals who are worse off as long as these individuals are below a certain threshold.20 ,25 This might be a plausible approach to income-based equity weights in healthcare planning and policy. Why should we give different, penalising, weights to benefits that befall individuals in a group of very wealthy people who all do very well in terms of health, opportunity for health and well-being just because their income might differ? This type of income-based equity weights seems reasonable in circumstances where there is a correlation between income and health, opportunity for health and/or well-being, as well as levels above or below which this changes.

There are different ways to introduce income-based equity weights. These generate different results when we assess outcomes. Establishing which of the different types of income-based equity weights to use must be connected to what sort of correlation we can expect between income and health, opportunity for health and well-being. As an illustration, imagine (which seems likely) that the marginal effect of income on health declines. Such a relation indicates that a cap should be introduced to fixate the income-based equity weights when the effects of income plan out, and also that the equity weights should increase in a non-linear way, the lower the income is.

Discussion

In this paper, I have argued that we have good egalitarian reasons to use income-based equity weights in healthcare planning and policy. I have not argued that there is a causal relationship between income and any of the aspects that matter when we assess outcomes. Such relationships might exist, but they cannot be established with purely theoretical arguments. Rather, I argued that under certain conditions, it makes sense to use income-based equity weights on egalitarian grounds. The conditions are the following: (1) income correlates with life expectancy and/or other aspects of health; (2) income correlates with opportunity for health and/or (3) income correlates with well-being. More research is needed in order to establish if, when and where these correlations obtain.

More research is also needed in order to establish what type of income-based equity weights we should use. Above, I introduced three broad categories of income-based equity weights. We need to examine which of these makes most sense in light of the reasons to use income as a basis for equity weights. This research needs to be conducted in relation to empirical studies. Here, we should also not be myopic. It is not clear that we should restrain ourselves to one of these types. It might be most reasonable to construe weighting schemes that have elements taken from all of the types discussed above.

A different issue relates to how large income-based equity weights should be. When we have established when to use income-based equity weights and what shape the weighting scheme should take, we need to ask: How large should these weights be relative to other normative considerations? What is a justified trade-off between efficiency and egalitarian concerns? What is the trade-off between income-based equity weights and other equity weights that might be justified, such as weights that protect the worse off?

These are difficult questions that require significant amount of research to answer. Yet, I believe that we have good reasons to make this investment, and I believe that already very rough approximate answers to these questions will enable us to design better health policy than approaches that ascribe no importance to income inequalities.

View Abstract

Footnotes

• Funding Forte, the Swedish Council for Health, Working Life and Welfare (2014-2724).

• Competing interests None declared.

• Provenance and peer review Not commissioned; externally peer reviewed.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.