One of my summer holiday reads was a fine polemic, The Spirit Level, by Richard Wilkinson and Kate Pickett. It's a blast against inequality, no nonsense about obfuscatory academic language or presenting counter-arguments and nuanced points. This partisanship is a strength and a weakness. The book is an invigorating read – inspiring, even, if you start out agreeing with the authors – but on the other hand is a very long way from being convincing for those who don't start out with the same point of view.
The hypothesis is simply that inequality is causally responsible for many of the ills of society, from poor health, mental illness, teen pregnancy, crime, imprisonment, not to mention unhappiness and low growth. The book takes it as read that there's no point in further economic growth on the average, based on the now-received wisdom (although I believe it's incorrect – see the terrific recent Wolfers and Stephenson paper) that rises in GDP do not increase happiness. As Wilkinson and Pickett put it: “Having come to the end of what higher material living standards can offer us… the evidence shows that reducing inequality is the best way of improving the quality of the social environment, and so the real quality of life, for all of us.”
As I agree at heart with the view that some societies have become too unequal, it's disappointing that the evidence presented here isn't more convincing. The evidence consists of repeated scatter plots showing correlations of the right sign between income inequality and X, where X is the relevant economic or social indicator. Once or twice causality tests are mentioned but not described in detail. Some of the scatter plots are quite tight, others show much greater variation, and some scream out that a different partition of the data, or a multivariate relationship, would deliver different results. Perhaps the original papers contain much more convincing empirical work but if so it's a shame there isn't a better description of it here.
Furthermore, there is a strong claim that more unequal societies deliver worse outcomes not only on average but also for the richest cohorts than do more equal societies. However, the causal mechanism is not clearly described, and I found it hard to understand how the result could hold for indicators such as health which are related to low social status. In other words, stress arising from being low in the pecking order contributes to heart disease – how then can high-status people in an unequal society have more stress-related heart disease than high-status people in a more equal society? Clearly, the causes have to be more complicated than inequality alone.
Having made this empirical gripe, there are many parts of the story the authors tell which are convincing. The work on health certainly falls into this category, and it is a good causal story too. Being poor and lacking control over one's life clearly must contribute to a range of illnesses, physical and mental. I also found the chapter on trust and social capital persuasive in terms of the argument as opposed to the scatter diagrams. Other chapters were less convincing. On crime, for example, although inequality – tied up with racial and social prejudice and other markers of deprivation – will obviously play a part, it is equally clear that other factors must play a role. Glaeser, Sacerdote and Scheinkman have shown that crime patterns are affected by social norms. From this side of the Atlantic, gun ownership seems worth highlighting too.
Still, all in all, the book makes a point that the UK and US certainly need to take seriously. The same global trends in technology and trade affect income disparities much less in some countries than others, and the effects of too much inequality are corrosive – politically as much as socially and economically. The challenge posed by structural change can be addressed. However, the argument is weakened by the poor empirical presentation. I'd have much preferred one which claimed less but made a more convincing case.
PS: For anyone who thinks I'm being too pernicketty about the statistical evidence, John Kay's FT review of The Spirit Level makes a similar point.