Wilkinson & Pickett: Are they right?

Rob de Vries (Imperial College London) finds reasons to be uncertain about The Spirit Level’s claims on inequality.

In his earlier post discussing The Spirit Level, Ben said that we’d return to the subject of whether the books’ claims were actually true or not. It’s a highly fraught question, but with Wilkinson and Pickett saying they’ll no longer respond to non-peer reviewed critics, I may fearlessly address it, secure in the knowledge that they won’t contradict me.

The bold subtitle of The Spirit Level is ‘Why more equal societies almost always do better’. The most obvious first question to ask of this is ‘Do they really?’: do more equal societies really have better health and social outcomes than less equal ones? The book makes a persuasive case that this is so. Unequal societies are shown to perform worse on measure after measure . But is this a real phenomenon, requiring an explanation? Or is it some trick of the data or the way they are presented?

Some critics have made it quite clear that they think it’s a deliberate trick, that Wilkinson and Pickett have carefully limited themselves to a selection of countries that best show the relationship they want. The authors are trying to show that inequality is bad, so they choose equal countries with good outcomes like Sweden and Norway, and compare them to unequal countries with poor outcomes like the U.S. and the U.K. This is a fair criticism up to a point. Whether deliberately intended to or not, which data you choose to include can have a big effect on the results you eventually find. Having said that, it doesn’t seem to explain what’s going on here. Wilkinson and Pickett have always been consistent in the criteria they use to select countries (see the second to last paragraph on page 2)[1] , and studies by other authors on other countries and on U.S. states confirm the basic relationship.

If more equal places really do tend to do better, the next question is ‘Why?’. Is it, as Wilkinson and Pickett claim, because inequality in society causes stress and social breakdown? Or is there another explanation?

It’s been almost 20 years since Wilkinson first suggested that inequality was harmful in and of itself.  In that time many (many) counter explanations have been suggested, some more plausible than others. There are certainly more than I can address here, but most cluster around three main arguments.

I’ll start with the simplest but the most charged – race. This is an argument that I imagine is more often thought than voiced but its impetus is the fact that the more equal, better faring countries are also the least ethnically diverse. Countries like Sweden, Norway and Japan have quite low populations of ethnic minorities. Whereas countries like the U.K. and the U.S. are much more ethnically mixed. The same is true of U.S.  states that are socially better off . The conclusion that follows that it is ethnic diversity, not inequality, that is the cause of the social ills highlighted in The Spirit Level. Wilkinson has recently labelled this argument ‘racist’, and that it may be. Happily it is also probably wrong. A series of papers over the last few years have debunked the idea to (almost) everyone’s satisfaction. A fact to which the experts at the Policy Exchange seem surprisingly oblivious.

The next argument might be the one I hear most often from people who’ve just read (or heard about) The Spirit Level. I’ll also admit it was my first thought when I heard the theory. The argument goes that of course more unequal countries have worse health and more social problems because they have more poor people, who get sick more often, commit more crimes, and so on. This is plausible and isn’t quite as prejudicial as I’ve made it sound. Lower income groups do tend to shoulder the greatest burden of crime, drug addiction, disease and most other social ills. So it would stand to reason that the countries with the biggest populations of poorer people would also have worse overall figures for these problems. The difficulties arise when trying and define what you mean by ‘poor people’. There is no simple box to be drawn and labelled ‘Poor’, within which all society’s health and social problems occur. Any delineation must be arbitrary to an extent and herein lays the problem. Draw the box too small and there is no way that the properties of the people inside it could drive large scale differences in the wellbeing of whole nations. Draw the box a bit larger and you’re talking about something that is part of the effect of inequality, rather than counter to it.

Let’s say you’re persuaded by that, and you’re satisfied that poverty isn’t the answer. There is still one glaring problem with The Spirit Level’s thesis, and that is culture. Wilkinson and Pickett claim that inequality makes people less cooperative, less trusting of each other, less generous and altruistic. In effect they are saying that inequality is a driver of national culture. It is the agent and changing cultural norms are the illness. Take a healthy society, they are saying, and introduce some inequality-causing agent (call it Thatcherium) and it will become unhealthy. The problem is that it could easily be the other way around. A change in national culture, like an increased emphasis on selfishness and individuality, could cause increased inequality through a diminished public appetite for egalitarianism. In grossly exaggerated terms, people in places like Japan and Scandinavia are not trusting and selfless because their countries are equal; their countries are equal because the people there are trusting and selfless.

Wilkinson and Pickett would ask why people in such places are the way they are, and point the finger back to inequality, rather than historical accident, as the root cause. They might be right to do so. There may even be a vicious cycle: from inequality to selfishness and through anti-egalitarian policies back to more inequality. The problem is that we can’t tell with the data we have. For all the relationships in The Spirit Level, both hypotheses work equally well. That’s why Wilkinson is wrong to compare his work to that on smoking or climate change. The data are in on those topics and there is no big alternative explanation that fits them.

Whether it’s inequality changing culture, culture changing inequality, or both, is not a fundamentally irresolvable question. It’s just that the data we have just can’t resolve it conclusively. Unless a more elegant test can be devised, what we really need is good data on social attitudes alongside good inequality data, both over long periods of time. That’s why it has been possible for this debate to go on for the last 20 years, and why it will probably continue for 20 more.


[1].  There’s no need to simply take their word on this as most of the data they use in their analyses are publicly available, for example from the World Bank or the U.S. Census Bureau

About Robert de Vries

I'm an Early Career Research Fellow in the Sociology department at the University of Oxford. I'm mainly interested in how people are affected by concerns about their social status; how it colours the way they think, feel, and behave. I try and contribute here regularly, but my addiction to writing excessively long posts keeps getting in the way.
This entry was posted in Articles and tagged , . Bookmark the permalink.

10 Responses to Wilkinson & Pickett: Are they right?

  1. Rob,

    This is a tremendous piece. I wanted to push you a bit on the question of whether race is a potential confounder. A friend of mine pointed me to some work by Angus Deaton on this issue.* Here’s what Deaton had to say in the 2009 article. First he quotes another review of 98 studies that concludes:

    ‘‘there seems to be little support for the idea that income inequality is a major, generalizable determinant of population health differences within or between rich countries.’’

    The he pushes on the irreducible link between racial inequalities and mortality:

    “the effect of racial composition on mortality reflects an inequality of some kind, though not an income inequality. In our original paper, we discussed several of these provision of public goods, segregation, quality of health services, inequality of political representation, but we failed to find convincing evidence for any of them.” (p. 1917)

    I don’t have a strong prior on this question, so I’m curious to see what you think about Deaton’s arguments.

    *(Deaton, “Health, Inequality, and Economic Development” in the 2003 Journal of Economic Lit.; Deaton and Lubotsky, “Income inequality and mortality in U.S. cities: Weighing the evidence. A response to Ash” 2009 in Social Science and Medicine)

    • Robert de Vries says:

      Hi Brendan.

      First thing I would say is that the 2003 Deaton paper (the one the 2009 paper is defending) is on American cities rather than states. It’s a long running finding that the ‘inequality effect’ is far weaker at smaller aggregations like cities or local authorities. Whether this is a problem for the Wilkinson hypothesis I’ll leave until another time, but what it does mean is that it’s not really surprising that the inequality effect at this level is not robust to confounding by race… or by anything at all really. Wilkinson has never claimed that it was or should be.

      When you move up to the level of U.S. states the evidence that race may be a confound just isn’t there, as the Subramanian and Backlund papers I link to in the piece show. This is something that even reviews by Wilkinson’s critics accept, for example Lynch’s highly critical 2004 review in The Millbank Quarterly. (I wanted to link to this in the piece too but it’s behind a paywall)

      • John says:

        I would love a citation or some evidence of Wilkinson’s failure to find an effect at the city and local level. That seems like a major problem given the causal explanation.

  2. Paul says:

    Good stuff. The real money quote from Deaton, aside from the reportage on 98 studies, is this:

    “Mortality rates for metropolitan statistical areas and for states were taken from vital registration data, and were matched to data on income inequality constructed between the 1980 and 1990 censuses…Our regressions showed that, once the fraction black was controlled, income inequality as measured by the gini coefficient was no longer a risk factor for mortality.” (Deaton & Lubotsky, p. 1914)

  3. Pingback: Winner-Take-All Politics: Why the Ultra Rich are Getting Ultra Richer | Inequalities

  4. Pingback: Highlights so far… | Inequalities

  5. Hi, a possible problem with the reverse causality argument is that it may be feasible to say lower levels of trust lead to higher levels of inequality, but the same logic isn’t as easily applied to other problems such as obesity.

    I think there is a more convincing mechanism from inequality to culture; that we have evolved to behave differently in different environments, and that inequality places a premium on competition with others in the social hierarchy (see my latest post if you’re interested). This argument seems to come from the same area as your interests on your profile, I come from an evolutionary psychology perspective so the importance of adaptive behaviour in climbing the social ladder seems like the strongest explanation to me!

  6. Pingback: A few things that inequality causes | Inequalities

  7. Pingback: Are Fragmented Cities Making us Unhealthy? | Inequalities

  8. Pingback: Inequality and civic morality | Inequalities

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s