This issue of the journal contains papers presented to the first meeting of the International Society for Equity in Health, Havana, Cuba, June 2000.
- health inequality
Statistics from Altmetric.com
As this is the inaugural conference of the International Society for Equity in Health, I thought it would be an appropriate occasion to reflect on some foundational questions in my keynote address. Why are we concerned with equity in health, and what is its relation to equity in general? Should we be more concerned about inequalities in health than about inequalities in other dimensions such as income? Should we be more concerned with some types of health inequalities than with others? Should we be less tolerant of inequalities across certain population groups than across others? Attempting to answer these questions might help sharpen our understanding of the special priority we accord to combating inequalities in health.
Let me start with the welfare-economic approach to assessing the distribution of a good—for simplicity, let us call this good “income”. A positive value attaches to higher total or average income, and a negative value to inequality of incomes around the average. The trade off between these two attributes of the distribution—sometimes labelled “efficiency” and “equity” by economists—is inferred from the society's social welfare function, which explicitly incorporates its distributional values.
I think it makes much sense to treat the distribution of health outcomes in a similar fashion. More aggregate or average health is positively valued as a good thing, and inequality of health around the average is negatively valued as a bad thing. Again there is a normative trade off where we might, if necessary, be willing to sacrifice some aggregate health for more equality of health. Of course, in particular empirical situations we may not be faced with a trade off: there may be policies that permit the achievement of both a higher average and more equality.
As a matter of valuation, however, we do need to acknowledge the existence of a trade off. As health egalitarians, we should not be evaluating health distributions solely in terms of inequality and without regard to the average. Consider a distribution of two groups of equal size, each of which has a life expectancy at birth of 50 years—so there is perfect equality in health achievement of the two groups. Now suppose that one group's life expectancy increases to 55 years while the other group's life expectancy increases to 65 years. In the new situation, average life expectancy has gone up from 50 to 60 years, but there is inequality now in health achievement between the two groups. Much as we might be concerned with health inequality, it would be difficult for us to judge the old situation of a 50 year life expectancy for each group as better than the new situation of a 55 year life expectancy for one group and a 65 year life expectancy for the other. Of course, what egalitarians would prefer is a distribution with an average life expectancy of 60 years where both groups have the same life expectancy of 60 years, instead of one having 55 and the other 65 years. Compared with the latter, we would even be willing to accept an equal distribution with both groups having a life expectancy lower than 60 years (but more than 55 years). (The amount of sacrifice of “efficiency” for “equity” that we are willing to accept—in proportionate terms—is the definition of the Atkinson index of inequality 1).
The trade off between average achievement and relative equality around the average will be dictated by our aversion to inequality, or concern for equality. The terms of this trade off—indeed our aversion to inequality—may well be different in the health space compared with the income space. In the economic inequality literature the trade off has been formalised by use of a parameter ε of the social welfare function, which measures society's aversion to inequality.1 The value of ε varies from zero, where there is no concern for inequality and a distribution is assessed entirely by its (arithmetic) average value, to infinity where there is an extreme concern for inequality and the distribution is assessed solely by its minimum value (the so called Rawlsian case)—see Anand and Sen.2 As ε increases, the weight in the social welfare function on someone who is less well off increases relative to the weight on someone who is better off.
I want to argue that we should be more averse to, or less tolerant of, inequalities in health than inequalities in income. The reasons involve the status of health as a special good, which has both intrinsic and instrumental value. Income, on the other hand, only has instrumental value. Health is regarded to be critical because it directly affects a person's wellbeing and is a prerequisite to her functioning as an agent. Inequalities in health are thus closely tied to inequalities in the most basic freedoms and opportunities that people can enjoy. In contrast, there are sometimes reasons to tolerate income inequalities.
There are economic reasons why we may be willing to accept certain income inequalities. Economists often assert—with some justification—that income incentives are needed to elicit effort, skill, enterprise, and so on. These incentives—and the resulting income inequalities—have the effect of increasing the size of total income (or the “cake”) from which, in principle, the society as a whole can gain (through taxation and possibly trickle down). Thus the increase in the size of the cake has to be balanced against the income inequalities that must be tolerated to provide the appropriate incentives for “efficiency”. Furthermore, effort, skill, enterprise, and so on are regarded as legitimate and fair reasons for some people to earn—perhaps even to deserve—more than others.
But this incentive argument would not seem to apply in the case of health. Inequalities in health do not directly provide people with similar incentives to improve their health from which society as a whole benefits. There thus seem to be no incentive reasons for accepting inequalities in health, other than those that might be derivative on tolerating income inequalities. As the empirical literature demonstrates, inequalities in income do produce inequalities in health—with richer people generally having better health. I will presently argue against tolerating inequalities in health for this derived reason.
Our willingness to accept some inequality in general incomes must, I believe, be tempered by what the Nobel laureate James Tobin 3 called “specific egalitarianism” some 30 years ago. This is the view that certain specific goods—such as health and the basic necessities of life—should be distributed less unequally than people's ability to pay for them. (Indeed, I regard this to be a central reason why many of us are concerned with socioeconomic gradients in health.) We are more offended by inequalities in health, nutrition, and health care than by inequalities in clothes, furniture, motor cars or boats. We should somehow remove health and the necessities of life from the prizes that serve as incentives for economic activity, and instead let people strive and compete for non-essential luxuries and amenities. In other words, we would like to arrange things so that crucial goods such as health are distributed less unequally than is general income—or, more precisely, less unequally than the market would distribute them given an unequal income distribution. This idea is the basis of specific—in contrast with general—egalitarianism.
WHY IS HEALTH A SPECIAL GOOD?
The rationale for specific egalitarianism in the health space rests on the premise that health is a special good. There is a related notion in public economics, that of a merit good—whose distribution, it is argued, should not be determined according to people's income.
That health is a special good has been recognised through the ages. We find this view in ancient Greek poetry, and in the Hippocratic texts. According to the author Democrit writing in the 5th century BC, he states in his book On Diet that:
“ (w)ithout health nothing is of any use, not money nor anything else.”
Some 2000 years later, René Descartes 4 asserted that health is the highest good. In Discours de la Méthode published in 1637, Descartes writes:
“... the preservation of health is ... without doubt the first good and the foundation of all the other goods of this life.”
The reason that health is so important is that (a) it is directly constitutive of a person's wellbeing, and (b) it enables a person to function as an agent—that is, to pursue the various goals and projects in life that she has reason to value. This view deploys the notion of health as “well functioning”, but it is not grounded in notions of welfare that are based on utility or some other consequential good, such as enabling the person to increase his or her “human capital” and hence “income”. It is, rather, an agency centred view of a person, for whom ill health reduces the full scope of human agency. In the terminology of Amartya Sen, health contributes to a person's basic capability to function 5—to choose the life she has reason to value.
If we see health in this way, then inequalities in health constitute inequalities in people's capability to function or, more generally, in their “positive freedom” (in the language of Isaiah Berlin 6). This is a denial of equality of opportunity, as impairments to health constrain what people can do or be. The principle of “fair equality of opportunity” is one of three principles of John Rawls' “justice as fairness”.7 Rawls assessed opportunity in terms of people's holdings of “primary goods”—or resources such as income, wealth, and so on. In his book Just Health Care, Norman Daniels extended the principle to deal with fair access to health care 8 (see also Daniels et al 9 and the commentary by Anand and Peter10). However, opportunity is best seen directly in terms of the extent of freedom that a person actually has—that is, by one's capability to achieve alternative “beings” and “doings”11—most of which depend critically on one's health. Moreover, the capability to lead a long and healthy life must itself be regarded as a basic capability, as our ability to do things typically depends on our being alive. Thus if we apply Rawls' “fair equality of opportunity” principle in the space of (basic) capabilities, the reduction of inequalities in health will follow as a direct requirement of justice.
DIMENSIONS OF HEALTH
I have ducked any attempt to define health and do not propose to offer a definition here. Earlier, I used a particular measure of health, namely life expectancy in years, to illustrate the equity-efficiency trade off in health. There are, of course, many different aspects or dimensions of health and ill health, captured by various different measures. The reasons we adduce for disvaluing inequalities in health more than inequalities in income will also direct us to pay more attention to inequalities in some dimensions (measures) of health than to inequalities in others. Thus, equality of opportunity reasoning may lead us to be more averse to a twofold (that is, a 2 to 1) disparity in the infant mortality rate (IMR) or the child mortality rate (CMR) between groups than to a twofold difference in adult or old age mortality rates. The reasoning may also lead us to be especially concerned about disabilities in health (physical or mental) that prevent a person being mobile or gaining employment.
THE UNIT OF ANALYSIS
Before closing, I would like briefly to address the question of the unit of analysis of inequality—in other words, the question of “inequality among whom?”. This is distinct from the question we have been considering so far, which is “inequality of what?”—income, health, or specific dimensions of ill health.
Much of the existing empirical literature on health inequalities—undertaken largely by epidemiologists—has been concerned with differences in health across socioeconomic groups, typically defined by occupation, education, or income. Thus, social class “gradients” have been estimated for Britain and several other European countries. Some researchers have tried to understand these gradients by controlling for factors such as smoking behaviour. Yet the gradients persist, and much research is underway attempting to understand the social causes and pathways that produce them.
There is much merit in analysing differences in life expectancy, mortality, and morbidity among socioeconomic groups. The classification by groups helps to explain how they might be generated. As tools for understanding the determinants of population health, the categories should obviously be extended to include not just socioeconomic status but also gender, race, and geographical location. In many developing country contexts, these latter variables have been found to be powerful in identifying inter-group inequalities—for example, race in South Africa, region in China, gender in Bangladesh. Moreover, cross classifications of socioeconomic and other variables often provide further epidemiological clues.
Apart from explanation, there are at least two other reasons for investigating inter-group inequalities in health. Firstly, it allows us to identify groups that are at high risk or suffer particularly poor health. Public policy and public health policy may thus be able to target them directly in order to improve their health. This is the case with the United Kingdom government's current initiative on inequalities in health.
Secondly, and perhaps more importantly, it allows us to uncover those inequalities in health that we regard as particularly unjust. In the language that I have been using, we will be more averse to—or less tolerant of—certain inter-group inequalities in health, such as racial or gender inequalities, than to inequalities where the groups are randomly defined (say by the first letter of a person's surname). Likewise, we will be more averse to socioeconomic inequalities in health than to inter-individual inequalities in health that are undifferentiated, or unconditional on information about individuals.
Group inequalities give rise to the suspicion that they derive from social rather than natural (for example, genetic) factors—and may thus be avoidable through public intervention. Moreover, health inequalities stratified by relevant variables often reveal a compounding of disadvantage—to wit, the observation of a positive correlation between (low) socioeconomic status and (poor) health. Such inequalities will typically be less tolerable than health inequalities observed across randomly defined groups or across undifferentiated individuals. In identifying inequity or injustice, we must take into account—or stratify by—those categories across which we are most averse to health inequalities.
Any approach to conceptualising and analysing inequality must confront two fundamental questions: (1) inequality of what?, and (2) inequality among whom?
On the what question, I have tried to argue that our aversion to inequality in health is likely to be greater than our aversion to inequality in income. And within different dimensions of health, I have tried to suggest that our aversion to inequality in some dimensions of health—such as infant and child mortality—is likely to be higher than it is for others (namely, those that do not constitute as serious a denial of lifetime opportunity).
On the whom question, I have tried to suggest that our aversion to inequality across certain population groups is likely to be greater than it is across others—in particular across undifferentiated individuals (who are not identified by systematic differences in opportunity).
In all of this I have tried to adapt and extend the framework and language of welfare economics to illuminate the study of equity in health.
Research support from the Rockefeller Foundation is gratefully acknowledged. Thanks are also due to Timothy Evans, Sanjay Reddy, Amartya Sen, and Barbara Starfield for their comments.
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.