The Spirit Level: Why Greater Equality Makes Societies Stronger
Richard Wilkinson and Kate Pickett
Bloomsbury Press, $28 (Hardcover)

The strong version of Richard Wilkinson and Kate Pickett’s argument in The Spirit Level implies that President Obama’s fight to reform health care was pointless. Extending the availability of health insurance cannot substantially improve Americans’ health. Instead, the president would make us all happier, healthier, and longer-lived, their logic suggests, if he could get the richest, say, 5 percent of Americans to leave the country.

Wilkinson and Pickett, eminent health scholars from the United Kingdom, present considerable evidence correlating unequal incomes in nations or American states with negative outcomes in physical health, mental balance, levels of violence, social integration, teen births, school performance, and just about everything else. Inequality, they explain, makes people focus on status and their relative positions on the prestige ladder. Such obsessions, in turn, create anxiety, distrust, and social isolation, which raise people’s level of physiological stress. Finally, stress, as we all now know, exacts high costs. It weakens the immune system, for example, and drives people to poor coping behavior such as overeating and lashing out at others. Through these steps, The Spirit Level argues, economic inequality becomes bad for everyone’s health.

But does this psychological explanation really account for the harms of inequality? And just how sure are we that the social ills Wilkinson and Pickett canvass are even caused by inequality? Whether we accept their psychological framework determines to some extent how we will respond to problems of inequality, and in hewing to it, the authors generate some pretty tepid solutions.


Measuring inequality

The Spirit Level does not argue simply that being poor is bad for people. Indeed, in developed societies, the authors insist, an individual’s wealth is not critical. It is of course healthier to be rich than poor, but what matters most for Westerners are the gaps between the rich and the middling and the poor in their societies. Wilkinson and Pickett reject economic growth as a public-health policy, in part because such growth might benefit the affluent as much or more than those of lesser income. Income differences would not necessarily shrink, and it is these gaps that we must mind.

For proof, the authors present dozens of similar, paired graphs. Across the bottom of each graph is a scale running left to right from low to high income inequality. On the vertical axis is a measure of the prevalence or intensity of a social problem, such as obesity or depression. The authors plot the locations of several Western countries and Japan, and the dots typically line up such that the more inequality, the worse the problem. In the international comparisons, Japan almost always falls in the bottom left corner of the space—low inequality, few problems—while the top right of the space—high inequality, many problems—mostly is reserved for the United States. In the U.S.-focused versions of these graphs, states take the place of nations. The states do not line up quite as neatly, but the pictures convey the same message: more inequality, more bad stuff.

Lest these graphs seem intimidating, understand that one of this book’s virtues is how straightforward and reader-friendly its prose and figures are. (There are even cartoons.) Wilkinson and Pickett also crisply and lucidly summarize research drawn from nearly 400 scholarly references. The authors anticipate criticisms, take pains to explain complex issues, and respect the reader. And as someone who many years ago coauthored a book on the hazards of inequality, I am sympathetic to their project. But are their numbers right? Is there an association between inequality and bad outcomes—do the graphs tell the whole truth? And, if there is a correlation, is inequality really the major cause of all those problems?

One concern is how we measure inequality. Researchers often use a measure of the distribution of income—usually, the “Gini coefficient”—or compare the income of the richest 10 or 20 percent of the population to that of the poorest 10 or 20 percent.

But different metrics produce different results. Good and comprehensive measures of inequalities in accumulated wealth rather than annual income show much greater inequality: in 1999 an American family at the 80th percentile of income made about two times what a family at the 50th percentile did, but the family at the 80th percentile in wealth owned about six times the assets of the 50th-percentile family. On the other hand, good and comprehensive measures of consumption indicate less inequality: a family at the 80th percentile of spending paid only 1.5 times as much for food and clothing as did a family at the 50th percentile. American families almost all the way down to the very poorest own cars, televisions, and the like, and some commentators point to such consumption numbers to dismiss the concern about income inequality.

Using a metric other than money also changes the picture. For instance, while inequality in how much people earn has widened considerably in the United States for about four decades, inequality in how long people live has narrowed somewhat.

A second question about the empirical basis for the connection between inequality and well-being is whether the authors have fairly examined all the bad outcomes. A major omission in their graphs is the suicide rate, which is considerably lower in more unequal countries. The authors try unpersuasively to explain this away, contending that in unequal societies people project their status anxiety outward, blame others rather than themselves, and thus end up killing others rather than themselves. Another outcome Wilkinson and Pickett ignore is the rate of births to unwed mothers, which also trends downward as inequality rises, especially if one brackets anomalous Japan. (By the way, anyone can easily play this game of chart-your-bad-outcomes by ransacking the Web sites of the U.N. Human Development Report, the Organization for Economic Co-operation and Development, and the U.S. Census Bureau, and then copying the tables into spreadsheets.) The authors may have overreached by implying that virtually every social ill can be blamed on inequality.

Researchers have put much of the data Wilkinson and Pickett use onto statistical torture racks to extract confessions, but often elicit only garbled croaks.

Finally, there is the problem of what the aggregate numbers mean for any given person. Wilkinson and Pickett’s graphs are displays of what are called ecological correlations, that is, they represent the connection between the income inequality of a country (or state) and some average outcome—say, average life span, or average risk of being obese. Such averages hide huge variations within countries and states, variations that overlap. The average Japanese man will live four years longer than the average American man, but many millions of American men will outlive many millions of Japanese men. Ecological correlations based on averages vastly overstate the actual connection between inequality and individuals’ life spans. This is not just a technical quibble. A substantive implication of this distinction is that it is better for your health to be rich in America than to be poor in Japan, no matter what the average differences are.

Wilkinson and Pickett would respond that it is still healthier for both the rich and for the poor to live in Japan than in the United States. Whether that is so gets yet more complicated. But, even if the graphs exaggerate the implications of national inequality for individuals, we can allow the authors this: in a Rawlsian sense, if you did not know how rich you would be, then choosing to be born in a more equal society would, all else staying constant, decrease your risks of many bad outcomes. How much of a decrease is difficult to estimate.


Sven versus Jack

Grant that inequality is often correlated with bad outcomes. Is inequality therefore the cause? With overly bold claims such as, “we have shown that reducing inequality leads to a very much better society,” Wilkinson and Pickett assert that there is more than a correlation here, that inequality is a—perhaps the primary—cause of bad outcomes such as violence, short lives, repression of women, psychological depression, and so on. Here is where most of the academic controversy focuses: is there some other factor that is really at work, such that income inequality is just a side issue? Researchers have put much of the data Wilkinson and Pickett use onto statistical torture racks trying to extract truthful confessions, but often elicit only garbled croaks.

Some critics argue that these ecological correlations between inequality and average outcomes are just a statistical illusion arising from the fact that the health benefits of each additional dollar are greatest for people of low incomes and marginal for people of high incomes.

Others hold that some X factor, perhaps as yet unidentified, explains The Spirit Level’s graphs. My own candidate, which I invite others to test, is cultural—the Sven versus Jack factor. If you look at most of the book’s graphs of nations, you will see that the “good” quadrant—low rates of inequality, low rates of problems—is largely composed of Nordic and northern European nations (and non-Western Japan, which should be bracketed). The “bad” quadrant is largely composed of the United Kingdom and its former colonies. Continental European nations fall into a mushy middle. If you look at most of the U.S. state graphs, you generally see in the good quadrant northern tier states, such as Minnesota, which were heavily settled by descendants of Scandinavia, and see in the bad quadrant southern states, which were much more intensively settled by highlanders from the British Isles. Thus, the Sven versus Jack factor.

The Nordic-British contrast also corresponds to the difference between social democratic and neoliberal states, which can confuse cause and effect even more. Is there something about the Nordic region’s history or culture that leads those nations to be welfare states, relatively equal, and healthy, and something about Anglo-Saxon history or culture that does the opposite—with varying levels of inequality being simply a byproduct?

Wilkinson and Pickett well understand these sorts of objections and have responses, both technical and logical. One strategy for handling the correlation-is-not-causation issue is to look at historical change: in cases where inequality has dramatically risen or fallen, what consequences followed? Unfortunately, here one starts cherry-picking examples. When East Germans were integrated into the rest of Germany, they joined a more economically unequal society, and their young people got more obese. Score one against inequality. On the other hand, between 1970 and 2005, income inequality in the United States, as measured by the Gini index, grew about 20 percent, but homicide rates dropped 30 percent. More systematic studies of what follows from changes in inequality tend to be more equivocal.

Researchers have not sorted out the causal issue yet, but the best provisional judgement is probably that economic inequality contributes something, albeit much less than the authors claim, to some health and social problems, but, again, fewer than the authors claim. Even the skeptics, however, do not argue that inequality is good for anyone but those on the top of the pyramid.


Psychology, politics, and solutions

If inequality does, to some degree, cause social problems, why? Wilkinson and Pickett emphasize that the mechanism here is social psychological: inequality creates anxiety about status and feelings of unfairness that eat at people. In the words of a chapter title, “inequality gets under the skin.” Unlike the volume of studies on the correlation between inequality and health, there is little research that directly tests this proposition. The authors collect a variety of suggestive evidence, such as laboratory studies on how people react to being put in low-status positions and primate studies on what happens when rankings among apes are messed with. But a lot of the case is built by argumentation and inferential stretch.

One recurrent issue in trying to explain any causal factor concerns the geographical level at which inequality operates. In the research literature, the strong correlations between inequality and bad outcomes tend to be seen when comparing nations, but when researchers compare smaller units, towns or neighborhoods, the connection between inequality at the local level and outcomes is considerably weaker. This is puzzling for the psychological analysis: wouldn’t people be more psychologically affected by their neighbors’ wealth than by the wealth of folks far away, say, in Malibu or on the Vineyard? The authors firmly argue that, no, what matters is where you—and your neighbors—fit in the national hierarchy; people know their national rank, and that is what generates the angst. Perhaps.

In a status-riven society, the winners fear that their perches are insecure, and they know that there is a long way to fall.

Even if people who feel they are at the bottom pay a psychological and health price for being down, are they not balanced out by those at the top who gain psychologically from being up? (In spite of the authors’ claims to contrary, wealthier people are, according to available metrics, happier.) Shouldn’t these two reactions balance each other out nationally? Wilkinson and Pickett would insist, in response, that everyone suffers psychologically from inequality, those at the bottom but also those at the top. In a status-riven society, the winners fear that their perches are insecure, and they know that there is a long way to fall. Besides, there’s always someone to envy on a yet higher branch.

Is this psychological mechanism necessary to explain the bad outcomes of inequality? One alternative, which the authors reject, is that it’s really all about material disadvantages, not psychological angst. Wilkinson and Pickett say ‘no,’ and point to statistical studies suggesting that international variations in average income make less difference to outcomes than do international variations in the inequality of incomes. But the results are not all consistent. The authors also point to examples: low-income Americans are richer than low-income people in other societies, but the Americans’ health is worse.

A different explanation, recently suggested and documented by many scholars, invokes politics. They find that more heterogeneous societies and states—those highly divided by race, religion, language, or income—under-produce “public goods” such as community health care, safety, and education. For example, the higher the proportion of African Americans in a state, other things held constant, the lower the public welfare expenditures in that state. People in diverse nations or states may have greater trouble building the trust necessary for public action. Or perhaps the reason is that majorities in diverse nations or states resist spending their tax money on “those people.” Income inequality, then, may produce bad outcomes because class divisions in a nation or state lead to political paralysis or to unconcern by the wealthy about the fate of the less well-off. If the politics of inequality account for poorer health, then one might focus on politics as the route to fixing the problems. But Wilkinson and Pickett do not.

Their discussion of solutions dwells mostly on promoting employee-owned businesses, an odd focus. Such enterprises pay their executives less than typical corporations do, and Wilkinson and Pickett believe that their workers therefore have lower status concerns and less stress. Such workers may be more sympathetic to economic redistribution. But there is no logical reason why such businesses would beggar their neighbors any less than other businesses do, and this program hardly seems sufficiently muscular to bring American inequality down to Finland’s level.

As I pointed out at the top, if the authors took their analysis literally, they might suggest direct manipulations of inequality: send the richest people—or, probably more efficiently, the poorest people—out of the country or the state. Inequality would go down and well-being would go up. Alternatively, leave the inequalities as they are, but devise ways to hide them from people—censor the media, say (no more Lifestyles of the Rich and Famous)—so that people do not know their relative positions. That should, according to The Spirit Level, bring down crime, disease, obesity, and so forth. The authors do not go in these directions, and these are, of course, not plausible solutions in a democratic society. But they are the logical implications of The Spirit Level’s explanation.

There are more productive avenues they might have considered. The authors eschew economic growth to lift the poor because their data suggest that national wealth is not as critical as national inequality in affecting health, because growth might preserve or even expand inequality, and because growth violates their green principles. Further economic development in developed nations, they assert, is an exhausted route to greater well-being. Most economists, I am sure, would disagree. Most politicians, I suspect, would consider the dismissal of economic growth a wrong-headed strategy for electoral victory.

Similarly, Wilkinson and Pickett pay little attention to Robin Hood–like redistribution, which would attack inequality more directly (although not as directly as exiling the rich). Maybe they consider that program too obvious to expound upon, or perhaps too politically difficult to attain, or too tied into the very status concerns and materialism that explain why inequality gets “under the skin.”

And there is little, if any, consideration in The Spirit Level for another strategy, one that tackles the specific difficulties of heterogeneous societies straight on: providing public goods in lieu of directly reducing inequality. National health insurance is one such public good; universal pre-school is another. The public, universal provision of water and sewer systems about a century ago did more than any other program to extend Americans’ life spans in the last several generations. Proponents argue that such universal entitlements—Social Security and Medicare being the major examples—evade Americans’ resistance to “hand outs” and to explicit “leveling,” and therefore have the highest chances of political success.

There is probably no way to avoid the heavy political lifting involved in channeling economic growth—yes, it will return—more equitably. Given this, and The Spirit Level’s occasional overconfidence, it might seem easy to discount the book. But Wilkinson and Pickett make a valuable contribution in enthusiasm and evidence, both of which will help fuel any effort to squeeze down the widening inequalities of our era.