Saturday, January 8, 2011

Doubts about Equality

Early last year I wrote for the Irish Medical Times about the impact of poverty on health, exploring the large differences in life expectancy between the richest and poorest in modern Ireland. The correlation between wealth and health was such that I wondered if the best way to improve the latter was to aim at full tilt, hell-for-leather economic growth - to increase wealth as quickly as possible.

But several times while researching for this article, Irish health experts disagreed, emphasising instead the negative role not of material poverty, but of inequality, of the great differences in wealth between rich and poor. Health experts like Dr Jane Wilde, chief executive of the Institute of Public Health in Ireland, pointed me towards a new book by Kate Pickett and Richard Wilkinson, The Spirit Level: Why Equality is Better for Everyone.

The Spirit Level argues that material poverty is highly detrimental to health in poor countries, but once countries become reasonably developed, continued economic growth makes little difference to health. Instead it is inequality itself, not poverty, that needs to be tackled. Pickett and Wilkinson then list a great number of social problems and claim that each is strongly correlated with the extent of income inequality in a given society.

Some of the ideas are intuitively pleasing. The authors argue, for example, that people tend to interact and empathise with others of a similar socioeconomic class. In regions with high income inequality, there are great sections of the population with incomes much greater or lesser than one's own, making it harder to identify with - and trust - these other groups. They say that unequal societies tend to have low levels of trust, suggesting that the social breakdown of New Orleans after Hurricane Katrina was a result of the distrust caused by a highly unequal society.

There are a lot of important ideas here that I may come back to, yet I find puzzling weaknesses in Pickett and Wilkinson's analysis. The book is peppered with graphs comparing income inequality in various countries with levels of various social ills. Some of the measures chosen, however, seem doubtful.

Mental health
One section relates inequality to mental illness. The authors show a graph of developed countries listed by mental illness and inequality: equal Japan with low mental illness, unequal US with high mental illness.

Yet Japan has a well known problem with high suicide rates, almost double the suicide rate in the US. Pickett and Wilkinson address this briefly at a later point in the book, arguing first that suicide rates don't usually increase down the social gradient (most of the social problems they say are worse in unequal countries are those which lower economic classes tend to suffer from more than the rich) and also that negative impulses may tend to go outwards (into homicide) in unequal societies and inwards (in suicide) in equal societies.

Yet there is some evidence that suicide is indeed worse among vulnerable, poorer people, i.e. that suicide increases down the social gradient. In any case it seems odd to conclude that a society with very high suicide rates is healthier than another.

Yesterday's mood
World surveys by Gallup help us to see how insecure or content people feel in various countries. According to The Spirit Level, people in unequal countries have greater feelings of anxiety and unhappiness. Instead Gallup shows rather less obvious results. When asked if respondents would like to experience more days like yesterday, the most positive results came from the following:

1 Costa Rica
2 Singapore
El Salvador
4 Paraguay
8 Ecuador

Yet most of these are highly unequal societies. Here is the list again, along with their ranking for inequality based on CIA World Factbook's Gini Index:

1 Costa Rica: 30
2 Singapore: 29
El Salvador: 18
4 Paraguay: 17
Kuwait: (not listed)
Iceland: 123
Guatemala: 13
Ecuador: 31
Panama: 12

A world map for this Gallup result shows that people were happier with yesterday in relatively unequal Ireland and US than in highly equal Sweden or Germany:

Other questions show vague results that don't support the idea that inequality increases stress, anxiety and unhappiness. When asked if respondents felt they were treated with respect yesterday, there are no clear differences between equal social democracies like Germany (90% yes) or Sweden (93% yes) and less equal peripheral EU states like Greece (92% yes) or Spain (97% yes):

Let's get more specific. Gallup asked respondents if they felt stress for a lot of yesterday. Result:

We see no correlation between income inequality and stress among developed countries. Cyprus has the third highest positive response for stress (56% say they felt stress a lot yesterday), yet they have the 18th lowest level of income inequality in the world, just after Denmark. Highly unequal Mexico is less stressed than the US.

Gallup continues with questions about boredom, sadness, depression, anger and so on. Few show the kind of correlation we would expect if inequality was really the driving factor behind these attitudes. In the case of anger, for example, unequal Portugal is far more chilled out than equal France! Likewise, unequal Americans were much more likely to volunteer time to some organisation in the previous month (43% yes) than equal Swedes (13% yes), despite claims in The Spirit Level that 'with greater inequality people are less caring of one another, there is less mutuality in relationships, people have to fend for themselves and get what they can.'

I have similar concerns with other indications of societal wellbeing. Pickett and Wilkinson use an "index of health and social problems" that correlates health with equality. Yet when I look at a particular indicator - life expectancy - compared with inequality in OECD countries I get the following graph:

There is only very weak correlation between income equality and life expectancy. Mexico is many times more unequal than Slovakia, yet Mexicans still live longer (or did in 2005, OECD's latest figures).

Since Pickett and Wilkinson focus on developed economies, next I zoomed in on the richest OECD countries only. I have labelled a few countries and coloured the dots to indicate GDP per capita (size of each dot indicates population size). Again we see weak, if any, correlation between income inequality and longevity.

Japan here is more unequal than most of the European social democracies yet has higher life expectancy than any of them.

The Japan question
This brings another puzzling point: Pickett and Wilkinson consider Japan a highly equal society while OECD's Gini Index shows them more unequal than almost any European country. This may be partly due to the authors' choice to use the Gini Coefficient for exploring income inequality between states of the US, but to base international comparisons on inequality between the richest 20% and poorest 20%. If both are useful indicators of inequality then they should show similar scores, so it is confusing that other studies describe Japan as a rather unequal society.

My concern is that Pickett and Wilkinson emphasise specific indicators that tend to support their thesis. For their thesis to be correct, we should be able to predict that increases in inequality will show worsening social problems of various kinds, but I can't seem to find these results in independent studies. If inequality causes social decline and violence, why do murder rates in the US dramatically fall during the 1990s while inequality continued to rapidly rise?
I will keep reading. There may be more to The Spirit Level and its ambitious attempt to explain the world. But since its claims are being taken seriously by influential people already, it matters deeply whether The Spirit Level is right or not.


  1. The Spirit Level assumes that people in less equal societies suffer from more stress than more equal ones. It also assumes that people are less happy and less involved in the community. In fact, the evidence shows that these factors either have a negative association (volunteering, community involvement) or no association (happiness, stress) with income equality.

    You are quite right to bring up the issue of Japan. Japan is important because it provides the only example of a low tax, equal society in The Spirit Level. Wilkinson and Pickett (W & P) repeatedly use Japan as evidence that they are not necessarily espousing high tax, socialist policies. So long as a country achieves greater income equality, they don’t care how it’s done, or so they say. This is very disingenuous, not least because Richard Wilkinson is a prominent member of the Socialist Health Association and is a long-standing campaigner for wealth redistribution. In reality, Japan is the one and only example of a ‘more equal’ society that doesn’t involve big government and high taxes.

    Or perhaps not. As you say, the OECD’s figures show that Japan is actually quite unequal. No one seems to know why the OECD estimates vary so much from the Luxembourg study’s estimates but W & P accept (on their website) that there is no reason to prefer one estimate over the other. Nonetheless, they use the Luxembourg estimates throughout The Spirit Level, and this has the effect of showcasing Japan as the most equal society. Such is the importance of Japan as an outlier in several graphs that if the OECD figures are used instead, a number of major correlations disappear.

    In fact, if you look at Singapore, Hong Kong and South Korea, it’s obvious that wealthy Asian countries all perform well under such criteria as obesity, crime, teen births, infant mortality etc. Japan’s good performance has nothing to do with income equality and everything to do with being a rich, traditional Asian country. W & P get around this by excluding Singapore from several graphs and excluding Hong Kong and South Korea from all the graphs.

    If you think about it, Japan does not really fit the ideal of The Spirit Level at all. W & P say that the problems of unequal countries stem from being hierarchical and materialistic. Few countries are more hierarchical or materialistic than Japan, but this is never mentioned in The Spirit Level.

    On the subject of mental health, it should be noted that there is no social gradient to mental illness. W & P admit this when they fail to find an association between mental illness and inequality in the USA (p. 68). It is one of the more blatant contradictions in The Spirit Level that they say this having just purported to find a correlation between inequality and mental illness between countries (p. 67). In truth, they are only able to find that correlation between countries by combining studies that even they admit are “not strictly comparable”. The reason they can’t find a correlation from the USA is that the evidence there all comes from one study and is, therefore, strictly comparable.

  2. Thanks very much Chris, lots of food for thought there.

    Speaking of food, The Spirit Level also claims that obesity is associated with inequality, pointing to Japan's lean population. I spent a year living in Japan and found that it was much easier to eat healthy out there simply because the diet was already tilted towards less sugary food. Coming back to Ireland I fell back to bad habits because I was surrounded by unhealthy snack food.

    In other words, my own dietary behaviour worsened in Ireland, not because I was more stressed, but simply because my access to healthy, delicious and cheap food had vanished! I'm not sure what makes Japan's diet differ in the first place, but it seems unlikely to be linked to stress. I still ate snack food in Japan, just as I did in Ireland, but in Japan the options were simply healthier to start with.


Note: Only a member of this blog may post a comment.