The Spirit Level: Why Greater Equality Makes Societies Stronger

by Richard Wilkinson

  Another similar study compared infant mortality rates in Sweden with England and Wales.318 Infant deaths were classified by father’s occupation and occupations were again coded the same way in each country. The results are shown in Figure 13.4. Deaths of babies born to single parents, which cannot be coded by father’s occupation, are shown separately. Once again, the Swedish death rates are lower right across the society. (Note that as both these studies were published some time ago, the actual death rates they show are considerably higher than the current ones.)

  Figure 13.3 Death rates among working-age men are lower in all occupational classes in Sweden compared to England and Wales.317

  Figure 13.4 Infant mortality rates are lower in all occupational classes in Sweden than in England and Wales.318

  Comparisons have also been made between the more and less equal of the fifty states of the USA. Here too the benefits of smaller income differences in the more equal states seem to spread across all income groups. One study concluded that ‘income inequality exerts a comparable effect across all population subgroups’, whether people are classified by education, race or income – so much so that the authors suggested that inequality acted like a pollutant spread throughout society.319 In a study of our own, we looked at the relationship between median county income and death rates in all counties of the USA.8 We compared the relationship between county median income and county death rates according to whether the counties were in the twenty-five more equal states or the twenty-five less equal states. As Figure 13.5 shows, in both the more and less equal states, poorer counties tended – as expected – to have higher death rates. However at all levels of income, death rates were lower in the twenty-five more equal states than in the twenty-five less equal states. Comparing counties at each level of income showed that the benefits of greater equality were largest in the poorer counties, but still existed even in the richest counties. In its essentials the picture is much like that shown in Figures 13.3 and 13.4 comparing Sweden with England and Wales. Just as among US counties, where the benefits of greater state equality extended to all income groups, so the benefits of Sweden’s greater equality extended across all classes, but were biggest in the lowest classes.

  Figure 13.5 The relation between county median income and county death rates according to whether the counties are in the twenty-five more equal states or the twenty-five less equal states.

  Figure 8.4 in Chapter 8, which compared young people’s literacy scores across different countries according to their parents’ level of education (and so indirectly according to the social status of their family of upbringing) also showed that the benefits of greater equality extend throughout society. In more equal Finland and Belgium the benefits of greater equality were, once again, bigger at the bottom of the social ladder than in less equal UK and USA. But even the children of parents with the very highest levels of education did better in Finland and Belgium than they did in the more unequal UK or USA.

  A question which is often asked is whether even the rich benefit from greater equality. Perhaps, as John Donne said, ‘No man is an Island’ even from the effects of inequality. The evidence we have been discussing typically divides the population into three or four income or educational groups, or occasionally (as in Figure 13.4) into six occupational classes. In those analyses it looks as if even the richest groups do benefit. But if, when we talk of ‘the rich’, we mean millionaires, celebrities, people in the media, running large businesses or making the news, we can only guess how they might be affected. We might feel we live in a world peopled by faces and names which keep cropping up in the media, but such people actually make up only a tiny fraction of 1 per cent of the population and they are just too small a proportion of the population to look at separately. Without data on such a small minority we can only guess whether or not they are likely to escape the increased violence, drugs or mental illness of more unequal societies. The lives and deaths of celebrities such as Britney Spears, John Lennon, Kurt Cobain, Marilyn Monroe, the assassinated Kennedy brothers, Princess Diana or Princess Margaret, suggest they might not. What the studies do make clear, however, is that greater equality brings substantial gains even in the top occupational class and among the richest or best-educated quarter or third of the population, which include the small minority of the seriously rich. In short, whether we look at states or countries, the benefits of greater equality seem to be shared across the vast majority of the population. Only because the benefits of greater equality are so widely shared can the differences in the rates of problems between societies be as large as it is.

  As the research findings have come in over the years, the widespread nature of the benefits of greater equality seemed at first so paradoxical that they called everything into question. Several attempts by international collaborative groups to compare health inequalities in different countries suggested that health inequalities did not differ very much from one country to another. This seemed inconsistent with the evidence that health was better in more equal societies. How could greater equality improve health unless it did so by narrowing the health differences between rich and poor? At the time this seemed a major stumbling block. Now, however, we can see how the two sets of findings are consistent. Smaller income differences improve health for everyone, but make a bigger difference to the health of the poor than the rich. If smaller income differences lead to roughly the same percentage reduction in death rates across the whole society then, when measured in relative terms, the differences in death rates between rich and poor will remain unchanged. Suppose death rates are 60 per 100,000 people in the bottom class and only 20 per 100,000 in the top one. If you then knock 50 per cent off death rates in all groups, you will have reduced the death rate by 30 in the bottom group and by 10 at the top. But although the poor have had much the biggest absolute decline in death rates, there is still a threefold relative class difference in death rates. Whatever the percentage reduction in death rates, as long as it applies right across society, it will make most difference to the poor but still leave relative measures of the difference unchanged.

  We can now see that the studies which once looked paradoxical were in fact telling us something important about the effects of greater equality. By suggesting that more and less equal societies contained similar relative health differentials within them, they were telling us that everyone receives roughly proportional benefits from greater equality. There are now several studies of this issue using data for US states,8, 319,320 and at least five international ones, which provide consistent evidence that, rather than being confined to the poor, the benefits of greater equality are widely spread.152, 315, 317, 318,321

  CAUSALITY

  The relationships between inequality and poor health and social problems are too strong to be attributable to chance; they occur independently in both our test-beds; and those between inequality and both violence and health have been demonstrated a large number of times in quite different settings, using data from different sources. But association on its own does not prove causality and, even if there is a causal relationship, it doesn’t tell us what is cause and what is effect.

  The graphs we have shown have all been cross-sectional – that is, they have shown relationships at a particular point in time rather than as they change in each country over time. However these cross-sectional relationships could only keep cropping up if somehow they changed together. If health and inequality went their separate ways and passed by only coincidentally, like ships in the night, we would not keep catching repeated glimpses of them in close formation.

  There is usually not enough internationally comparable data to track relationships over time, but it has been possible to look at changes in health and inequality. One study found that changes between 1975 and 1985 in the proportion of the population living on less than half the national average income among what were then the twelve members of the European Union were significantly related to changes in life expectancy.81 Similarly, the decrease in life ex
pectancy in Eastern European countries in the six years following the collapse of communism (1989–95) was shown to be greatest in the countries which saw the most rapid widening of income differences. A longer-term and particularly striking example of how income distribution and health change over time is the way in which the USA and Japan swapped places in the international league table of life expectancy in developed countries. In the 1950s, health in the USA was only surpassed by a few countries. Japan on the other hand did badly. But by the 1980s Japan had the highest life expectancy of all developed countries and the USA had slipped down the league and was well on the way to its current position as number 30 in the developed world. Crucially, Japanese income differences narrowed during the forty years after the Second World War. Its health improved rapidly, overtaking other countries, and its crime rate (almost alone among developed countries) decreased. Meanwhile, US income differences widened from about 1970 onwards.

  In Chapter 3 we provided a general explanation of why we are so sensitive to inequality, and in each of Chapters 4–12 we have suggested causal links specific to each health and social problem. We have also looked to see whether there might be other obvious cultural links between countries that do well or among those which do badly. But what other explanation might there be if one wanted to reject the idea of a causal relationship? Could inequality and each of the social problems be caused by some other unknown factor?

  Weak relationships may sometimes turn out to be a mere mirage reflecting the influence of some underlying factor, but that is much less plausible as an explanation of relationships as close as these. The fact that our Index is not significantly related to average incomes in either our international test-bed or among the US states almost certainly rules out any underlying factor directly related to material living standards. Our analysis earlier in this chapter also rules out government social expenditure as a possible alternative explanation. As for other possible hidden factors, it seems unlikely that such an important causal factor will suddenly come to light which not only determines inequality but which also causes everything from poor health to obesity and high prison populations.

  That leaves the question of which way causality goes. Occasionally when we describe our findings people suggest that instead of inequality causing everything else, perhaps it all works the other way round and health and social problems cause bigger income difference. Of course, in the real world these things do not happen in clearly defined steps which would allow us to see which comes first. The limited evidence from studies of changes over time tells us only that they tend to change together. Could it be that people who succumb to health or social problems suffer a loss of income and that tends to increase inequality? Perhaps people who are sick or very overweight are less likely to have jobs or to be given promotion.

  Could this explain why countries with worse health and social problems are more unequal?

  The short answer is no – or at least, not much. First, it doesn’t explain why societies that do badly on any particular health or social problem tend to do badly on all of them. If they are not all caused at least partly by the same thing, then there would be no reason why countries which, for instance, have high obesity rates should also have high prison populations. Second, some of the health and social problems are unlikely to lead to serious loss of income. Using the UNICEF index we showed that many childhood outcomes were worse in more unequal countries. But low child wellbeing will not have a major influence on income inequality among adults. Nor could higher homicide rates be considered as a major cause of inequality even if the numbers were much higher. Nor for that matter could expanding prison populations lead to wider income differences – rather the reverse, because measures of inequality are usually based on measures of household income which leave out institutionalized populations. Although it could be argued that teenage parents might increase inequality because they are often single and poor, some more equal countries have a high proportion of single parents but a generous welfare system which ensures that a very much smaller proportion of them are in poverty than in more unequal countries. And when the unemployed and the children of single parents are protected from poverty, they are also protected from the human damage it can cause.

  However, there is a more fundamental objection to the idea that causality might go from social problems to inequality. Earlier in this chapter we showed that it was people at almost all income levels, not just the poor, who do worse in more unequal societies. Even when you compare groups of people with the same income, you find that those in more unequal societies do worse than those on the same income in more equal societies. Though some more unequal societies have more poor people, most of the relationship with inequality is, as we pointed out earlier, not explained by the poor: the effects are much more widespread. So even if there is some loss of income among those who are sick or affected by some social problem, this does not begin to explain why people who remain on perfectly good incomes still do worse in more unequal societies.

  Another alternative approach is to suggest that the real cause is not income distribution but something more like changes in ideology, a shift perhaps to a more individualistic economic philosophy or view of society, such as the so-called ‘neo-liberal’ thinking. Different ideologies will of course affect not only government policies but also decisions taken in economic institutions throughout society. They are one of very many different factors which can affect the scale of income differences. But to say that a change in ideology can affect income distribution is not at all the same as saying that it can also affect all the health and social problems we have discussed – regardless of what happens to income distribution. Although it does look as if neo-liberal policies widened income differences (see Chapter 16) there was no government intention to lower social cohesion or to increase violence, teenage births, obesity, drug abuse and everything else. So while changes in government ideology may sometimes be among the causes of changes in income distribution, this is not part of a package of policies intended to increase the prevalence of social problems. Their increase is, instead, an unintended consequence of the changes in income distribution. Rather than challenging the causal role of inequality in increasing health and social problems, if governments understood the consequences of widening income differences they would be keener to prevent them.

  Economists have never suggested that poor health and social problems were the real determinants of income inequality. Instead they have concentrated on the contributions of things like taxes and benefits, international competition, changing technology and the mix of skills needed by industry. None of these is obviously connected to the frequency of health and social problems. In Chapter 16 we shall touch on the factors responsible for major changes in inequality in different countries.

  A difficulty in proving causality is that we cannot experimentally reduce the inequalities in half our sample of countries and not in the others and then wait to see what happens. But purely observational research can still produce powerful science – as astronomy shows. There are, however, some experimental studies which do support causality working in the way our argument suggests. Some of them have already been mentioned in earlier chapters. In Chapter 8 on education we described experiments which show how much people’s performance is affected by being categorized as socially inferior. Indian children from lower castes solved mazes just as well as those from higher castes – until their low caste was made known. Experiments in the United States have shown that African-American students (but not white students) do less well when they are told a test is a test of ability than they do on the same test when they are told it is not a test of ability. We also described the famous ‘blue-eyes’ experiments with school children which showed the same processes at work.

  Sometimes associations which are only observed among human beings can be shown to be causal in animal experiments. For instance, studies of civil servants show cardiovascular health declines with declining social status. But how can we tell wh
ether the damage is caused by low social status rather than by poorer material conditions? Experiments with macaque monkeys make the answer clear. Macaques form status hierarchies but with captive colonies it is possible to ensure all animals live in the same material conditions: they are given the same diet and live in the same compounds. In addition, it is possible to manipulate social status by moving animals between groups. If you take low-status animals from different groups and house them together, some have to become high-status. Similarly, if you put high-status animals together some will become low-status. Animals which move down in these conditions have been found to have a rapid build-up of atherosclerosis in their arteries.322 Similar experiments also suggest a causal relationship between low social status and the accumulation of abdominal fat.323 In Chapter 5 we mentioned other animal experiments which showed that when cocaine was made available to monkeys in these conditions, it was taken more by low social status animals – as if to offset their lower dopamine activity.59

  Although we know of no experiments confirming the causality of the relation between inequality and violence, we invite anyone to go into a poor part of town and try randomly insulting a few people.

  We have discussed the reasons for thinking that these links are causal from a number of different perspectives. But as philosophers of science, such as Sir Karl Popper, have emphasized, an essential element in judging the success of any theory is whether it makes successful predictions. A successful theory is one which predicts the existence of previously unknown phenomena or relationships which can then be verified. The theory that more equal societies were healthier arose from one set of international data. There have now been a very large number of tests (about 200) of that theory in different settings. With the exception of studies which looked at inequality in small local areas, an overwhelming majority of these tests confirmed the theory. Second, if the link is causal it implies that there must be a mechanism. The search for a mechanism led to the discovery that social relationships (as measured by social cohesion, trust, involvement in community life and low levels of violence) are better in more equal societies. This happened at a time when the importance of social relationships to health was beginning to be more widely recognized. Third, the theory that poor health might be one of a range of problems with social gradients related to inequality has been tested (initially on cause-specific death rates as described earlier in this chapter) and has since been amply confirmed in two different settings as we have described in Chapters 4–12. Fourth, at a time when there was no reason to think that inequality had psychosocial effects, the relation between health and equality seemed to imply that inequality must be affecting health through psychosocial processes related to social differentiation. That inequality does have powerful psychosocial effects is now confirmed by its links (shown in earlier chapters) with the quality of social relations and numerous behavioural outcomes.