Letters

Mortality and distribution of income

BMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7144.1611a (Published 23 May 1998) Cite this as: BMJ 1998;316:1611

Societies with narrower income distributions are healthier

  1. Stephen Senn, Professor of pharmaceutical and health statistics
  1. Department of Epidemiology and Public Health, Department of Statistical Science, University College London, London WC1E 6BT
  2. Trafford Centre for Graduate Medical Education and Research, University of Sussex, Brighton BN1 9RY
  3. Nuffield Institute for Health, Leeds LS2 9PL
  4. National Primary Care Research and Development Centre, Centre for Health Economics, University of York, York YO1 5DD

    EDITOR—Gravelle's recent contribution to the debate on the relation between income distribution and health may have been difficult reading for some.1 I offer a simple analogy by way of explanation.

    Imagine two identical fields, both of which are manured with equal quantities of a fertiliser. In one field the fertiliser is spread evenly; in the other some plots receive far more fertiliser than others. At harvest, the yield is considerably higher in the evenly treated field. One possible explanation of this result is that, in the unevenly treated field, plants growing faster and higher in plots receiving more fertiliser have drawn moisture away from those less well treated and have also shaded them from the sun. This is analogous to the relative income hypothesis.

    An alternative explanation is simply that the first kg of fertiliser spread on a given plot produces a greater increase in yield than does the second. The optimal application of fertiliser is therefore to spread it as evenly as possible. This is analogous to the absolute income hypothesis. This second explanation works at the level of individual plants and does not need to consider interactions between plants. In the same way, the absolute income hypothesis for health considers only the direct effect of income on the health of an individual. On the other hand, the relative income hypothesis requires that the incomes of others affect the health of an individual through complex societal mechanisms.

    Gravelle's examination is important. The absolute income hypothesis is a simpler explanation than the relative income hypothesis for the observation that, other things being equal, societies with narrower income distributions are healthier. The more complex hypothesis should not be asserted until the claims of the simpler one have been exhausted. In saying, however, that “studies using population level data.. cannot distinguish between the absolute income and relative income hypotheses,” he goes too far. Some reasonable modelling may be attempted using aggregate data, provided enough is known about income distribution. For example, if the richest people in society A are poorer than the poorest in society B but B has a much greater income inequality and also worse health, this will cast doubt on the absolute income hypothesis.

    References

    1. 1.

    Low relative income affects mortality

    1. Richard G Wilkinson, Professorial research fellow
    1. Department of Epidemiology and Public Health, Department of Statistical Science, University College London, London WC1E 6BT
    2. Trafford Centre for Graduate Medical Education and Research, University of Sussex, Brighton BN1 9RY
    3. Nuffield Institute for Health, Leeds LS2 9PL
    4. National Primary Care Research and Development Centre, Centre for Health Economics, University of York, York YO1 5DD

      EDITOR—Gravelle's explanation of why more egalitarian societies tend to have lower mortality rates is neither new nor true.1 The fact that an additional income of £1 per person made more difference to the health of the poor than that of the rich was initially my stated reason for looking at income distribution. Twelve years ago2 and twice since I showed data on the curvature of the relation between individual income and health. Gravelle does not provide data and ignores the published reasons for abandoning this explanation of the income distribution relation.3

      His view requires a strong, and strongly curvilinear, relationship between absolute (not relative) income and mortality among rich developed countries. The evidence contradicts all of these. The relation between gross national product per head and life expectancy puts the rich Organisation for Economic Cooperation and Development (OECD) countries all on the near linear, near horizontal part of the international curve.3 Even at half the average income of the United States (the threshold defining relative poverty) the relation is still weak and almost linear. Likewise, the correlation between median income and mortality among the 50 states of the United States is only —0.26. Rather than a stronger relation with median income being, as Gravelle suggests, confounded by income distribution, controlling for income distribution reduces this correlation to a non-existent —0.06.

      The same is true across developed countries. Contrary to what Gravelle's model predicts, health inequalities have not been reduced during the past half century of economic growth, even when growth has “trickled down.” Nevertheless, the only study using matched data shows that health inequalities are closely related to income inequalities internationally.4

      Although there is no strong curvilinear relation with absolute levels of income between countries, poor people in the United States often have death rates comparable with people in Bangladesh. Their high death rates are not so much a product of their absolute living standards (with freezers, central heating, CDs, and sometimes air conditioning) but reflect their low relative incomes and social status untouched by economic growth. Since mortality is associated with relative income the inequality relation is not an ecological fallacy.

      Gravelle misrepresents my argument.3 Rather than asserting that someone with a given income will be healthier in a more egalitarian society, I suggest that the main effect is through individual relative income. Someone with an absolute income that equals half the United States average income might do better to be moderately well off in Greece or Spain than poor in the United States. Mortality is lower in more egalitarian societies because the burden of relative deprivation is smaller. This ties in with the evidence that psychosocial effects of low social status that damage health and that have been show in experiments among non-human primates, are also found among human beings.5 That the psychosocial effects are widespread is shown by the growing evidence that more egalitarian societies are more cohesive.3

      References

      1. 1.
      2. 2.
      3. 3.
      4. 4.
      5. 5.

      Widening income inequalities cause poorer health

      1. Graham C Sutton, Senior clinical lecturer
      1. Department of Epidemiology and Public Health, Department of Statistical Science, University College London, London WC1E 6BT
      2. Trafford Centre for Graduate Medical Education and Research, University of Sussex, Brighton BN1 9RY
      3. Nuffield Institute for Health, Leeds LS2 9PL
      4. National Primary Care Research and Development Centre, Centre for Health Economics, University of York, York YO1 5DD

        EDITOR—If the rich get richer their health will improve only slightly, whereas if the poor get poorer their health will suffer greatly. Widening income inequalities will therefore worsen overall health. This is, however, far from being an artefact as Gravelle suggests1; rather, it actually reveals the causal mechanisms.

        Consider the privatisation of the water industry. Many wealthy people profited from share dealings, either directly or indirectly through holdings such as personal equity plans. A few water company directors became wealthy—some even fabulously wealthy. At the other end of the scale, however, thousands of people were made redundant and impoverished. Widening inequality does not just mean the flattening of a symmetrical bell shaped curve, but the aggrandisement of the few at the expense of the many.

        When this is translated into health behaviours, the equations of price, income, and consumption for the bulk of the population break down at the extremes. Those enriched might buy more expensive wines and entertain more lavishly, but their personal intake of alcohol is unlikely to increase. Those cast into poverty may actually drink more and in a more uncontrolled and self damaging fashion. In other words, alcohol may be an example of negative elasticity or Giffens paradox, as potatoes in the Irish famine were said to be.

        The health impact may be amplified by a multiplier effect. The wealthy directors of the water companies found their new situation delightful and pursued policies to entrench it. They sought to disconnect late payers or make them pay more through metering. They became obsessed with financial deals and lost sight of their core business—water supply—so that parts of the country teetered on the edge of drought. They blocked fluoridation and thereby disadvantaged the poorest people, who have the worst dental health.

        Similar examples can be drawn from transport, housing, mining, and energy. An underpinning theme is the loss of the “hidden wage” or “social income,” so that the poorest people are impoverished even if their money wage is unchanged.

        I therefore believe that widening inequalities cause poorer health, although I accept Gravelle's caution about the strength of the effect. Indeed, there have been times in the past decade when I have wondered if poorer health was not the object of the exercise.

        References

        1. 1.

        Author's reply

        1. Hugh Gravelle, Professor of economics
        1. Department of Epidemiology and Public Health, Department of Statistical Science, University College London, London WC1E 6BT
        2. Trafford Centre for Graduate Medical Education and Research, University of Sussex, Brighton BN1 9RY
        3. Nuffield Institute for Health, Leeds LS2 9PL
        4. National Primary Care Research and Development Centre, Centre for Health Economics, University of York, York YO1 5DD

          EDITOR—My purpose was not to propose a new theory but to draw attention to a problem in the interpretation of population level studies that are used to support the argument that the health of a person is influenced by his or her relative income. Senn's analogy is a helpful illustration of the artefactual problem.

          Wilkinson makes a number of points in saying that the argument is incorrect.

          Firstly, he states that it requires a non-linear aggregate relation between population health and mean income, and then he says that across developed countries the relation is linear. The argument requires, however, that the relation between health and income in individual people is non-linear.

          Secondly, he says that the international association between inequalities in health and in income1 supports the relative income hypothesis. In fact, this association is irrelevant for distinguishing between the relative and absolute income hypotheses. It would arise if health were positively related to income, in a linear or non-linear way. Hence both the relative income and the non-linear absolute income hypothesis would predict the reported association.

          Thirdly, Wilkinson says that I misinterpreted his version of the relative income hypothesis. Since he still posits, however, that more equal societies will have better health, the artefactual argument still applies. Both the relative income hypothesis and the non-linear absolute income hypothesis yield the same prediction about the correlation of health and income inequality in populations.

          Recent evidence suggests that the aggregate level correlations between population health and income inequality may be very weak.2 A study that used data on individual people showed that aggregation can lead to misleading conclusions.3 It found that aggregating the individual data to area level produced negative correlations between population health and income inequality but that area income inequality had no effect on individual health once individual income was allowed for.3

          Relative income or social position could well have an effect on the health of individual people, as Wilkinson and Sutton suggest. We require more studies that combine individual cohorts and careful specification of the hypothesised relation to take account of the influences of lifestyle, consumption, environment, and income on health, as well as of the effect of health on income.4

          References

          1. 1.
          2. 2.
          3. 3.
          4. 4.
          View Abstract

          Sign in

          Log in through your institution

          Free trial

          Register for a free trial to thebmj.com to receive unlimited access to all content on thebmj.com for 14 days.
          Sign up for a free trial

          Subscribe