review of Wilkinson & Pickett's "The Spirit Level"

Wilkinson and Pickett's book, The Spirit Level is a favorite of my friend/colleague, Sam Sloss. I wouldn't have heard about the book without his repeated reference to it. (One of the problems is that the title doesn't prompt its potential audience to think of its primary topic: income inequality!) I enjoyed the book, so I'm thankful to Sam for pointing it out to me. In a word, the book is helpful in its primary task, but not-so-helpful in an assortment of other ways.



In this, WP is probably similar to Piketty's recent famous/infamous book. As such, if you're interested in the general topic-- and want to jump into some reading-- I'd start with two more-helpful, less-statistical, and easier-to-read books: Tyler Cowen's Average Is Over and especially Charles Murray's Coming Apart. (Interestingly, Piketty and Cowen are both economists; Murray and WP seem to be like sociologists in their approach.)



WP are quite helpful in laying out certain income inequality data (as far as their chosen data go)-- as well as data on a wide (awesome!) array of correlated social problems/issues. They reference at least some of the relevant theories and some of the relevant literature. (On the latter, I don't know enough to say for sure, but would infer that they're selective in a biased sense-- from the areas I do know well.) Then, they combine this mixture with some just-so stories, resulting in inferences somewhere between solid ground and an odd narrative that fits various aspects of their worldview.



Most of the time, WP are noticeably and appropriately reticent in making policy prescriptions (contra Piketty). At the end of the day, their data do not imply much about how to achieve their ends. Sometimes, they deviate a bit and get themselves into contradictory positions. And unfortunately, they are not more forceful in insisting on humility among their readers-- and so, careless proponents of their work could easily end up being less reticent than would be wise.



1.) Income inequality measured
"There are lots of ways of measuring income inequality and they are all so closely related to each other that it doesn't usually make much difference which you use." (p. 17) WP then refer to the choice of quintiles and deciles for an income measurement and Gini coefficients. Among *those* choices, they're correct; it doesn't matter much, particularly for a longitudinal study. But there are other key variables/questions which aren't even raised here: Are we looking at individual or household or family measures of income? Is income pre- or post-tax? Is income pre- or post-transfer? (They touch on these three, very briefly, on p. 243-- largely to dismiss them.) More broadly, what constitutes "income" in these measures? What about the impact of cost-of-living on unweighted measures of income? What about the impact of hours worked? (They have a brief mention of the aggregates on p. 228, but do not discuss work differences between income classes.) What about much-more-equal measures of consumption (vs. "income")? (UPDATE: For new, terrific research on this question, check this out.) And so on. The immense failure to discuss this could be-- by itself-- sufficient to reject their work out-of-hand as naïve or biased. But let's assume the best and move along...



WP focus on averages and aggregate measures. It would be interesting to know how much the numbers (and their implications) would change with significant changes in the "super-rich"-- e.g., if one were to triple or halve the number and dollars of those in the top .1%. My guess is that the numbers would barely move. Later, they take some shots at CEO pay (250), but unless this is extraneous, they're assuming that CEO pay is a significant contributor to the underlying problem (such as it is). This seems remarkably fanciful. In any case, it'd be nice to see them play with this, at least through simulations, since it is a significant part of their narrative.



WP discuss social mobility/dynamics as their final applied topic (in chapter 12). In part, this is reasonable: it's arguably the most important consideration, so why not save it for last? On the other hand, it's curious to wait so long, because it is so important-- AND so important to the foundational topic of (true vs. measured) income inequality. More troubling, they only discuss a narrow slice of a complex/mixed literature on income/class mobility-- and a slice that favors their worldview. Again, this is a lack of scholarship or an unfortunate bias. (The reference to Jargowsky on geographical mobility was impressive and balanced; I wrote a journal article / book review on his research awhile back.)


For a discussion of all of the above and more, check out this set of articles from AEI.


WP are quite careful at times in avoiding the common conflation of income inequality with poverty. Even so-- and perhaps even more worrisome given their apparent efforts to avoid this problem-- they still occasionally stumble. (E.g., see: p. 195's "most deprived neighborhoods"; p. 264's "denying adequate incomes".)



On the flip side, their passing references to the problems of poverty underline another point-- that poverty is as pressing or moreso than income inequality per se. But they rarely seem to recognize this-- even implying that poverty should be considered in strictly relative terms (15), rather than strictly absolute or the more moderate position that poverty is both relative and absolute.


Many on the Left really do care about absolute poverty. (Some are more concerned about income inequality-- out of envy and here, from concerns about its impact on society.) But the Left also prefers static analysis of simple statistics, given a lack of empirical or theoretical rigor-- or because it fits their penchant for emotional appeals. This approach works reasonably well for income inequality. But it works very poorly for the poverty rate. In a word, a reliance on the poverty rate would indicate a massive failure for the War on Poverty. Although the poverty rate is deeply flawed-- and masks many of the problems of the poor-- the Left has wedded itself to stats like this and the limited/flawed analysis that follows. So, it makes more sense for them to talk about inequality than poverty per se.





2.) Social implications
This is probably the most impressive and useful section of the book: chapters 4-12 on community life, mental health and drug use, health, obesity, and life expectancy, education, teen births, crime and punishment, and social mobility. WP present a series of simple correlations, mostly longitudinal. The approach is wide-ranging and impressive. (They even make connections to Frans de Waal and research on chimps and bonobos [204]!)

Arguably, their most powerful point is that income inequality correlates with a number of troubling social indicators-- for both the poor AND the rich! It's one thing to say that inequality (or poverty) hurts the poor. But it's much more provocative to say that it even impacts the wealthy.



Even so, WP ignore or too easily dismiss the potential influence of many other factors: religious belief and practice, family structure and stability, globalization and other economic factors, etc. They spend a lot of time on Robert Putnam's "social capital" (54-56, 79, 140), but don't give it much (any?) weight as a causal factor.

In this, it is odd to see WP be so selective "sociologically", focusing almost exclusively on income inequality. They seem oddly passionate about looking for "a common underlying cause" (186b)-- and casting income inequality in that role-- as if a single cause would likely explain something so complex.

Implicitly, they hold genetics and culture constant-- a weird position for those enamored by science and into sociology. (They try to dismiss the former with a quick sentence on p. 185: "This must be rejected because it is simply an expression of racial prejudice." Reject scientific investigation of the topic or you're a bigot?! Huh? They extend the discussion briefly, to acknowledge that race correlates *strongly* for the results within the U.S., but walk away after that.) Beyond the wisdom of this approach, in trying to describe complex realities, the result must be a lot of just-so stories to make the data fit the simple analysis.


WP also ignore the challenges of large-scale *federal* policies, given poverty, existing inequality, and all sorts of diversity. (This is the case, even with longitudinal data, but even more important when one is looking at times-series data.) In a word, one would expect federal policy to be more beneficial with smaller and more homogeneous populations-- and more damaging as population and diversity increase.




In Coming Apart, Charles Murray provides a very useful complement to WP's discussion. Like WP (44), but in far-greater depth, Murray discusses the role of "sorting" by class and education in marriage-- on measured income inequality. Like WP (164), but again in far-greater detail, Murray describes the cultural barriers increasingly implied by class differences. Like WP (ch. 9 and p. 137-138), Murray talks about family structure and stability-- but in a more comprehensive and coherent manner. (For example, WP argue, at great length, that fathers and family structure matter-- and there's more to life than material goods. But then, apparently conflicted and contrary to the bulk of the available research, they argue that it seems "possible to safeguard children against most of the adverse effects of being brought up by lone parents" [187].)




3.) Policy RX's
Even assuming away the above concerns, now what? In most cases, WP are noticeably and appropriately reticent in making policy prescriptions (contra Piketty). Other times, they are mysteriously optimistic: "There are good reasons to feel confident that we can create a society in which the real quality of life...is far higher." (241) At the end of the day, their data do not have much to say about how to achieve the implied and desired ends. So, reticence is the wise approach.


WP are correct in pointing to the limits of economic growth-- at least on average, in developed countries (5). This is a point made nicely in Charles Murray's indispensable, generalist, public policy book, In Pursuit of Happiness and Good Government. Murray uses Maslow's Hierarchy to note that material needs are the first but not the only policy goal. And in developed countries, material needs have been-- and can easily be met.


That said, this is only true for averages/aggregates. As always, too much excitement about groups will overlook the impact on individuals. And we see this misplaced excitement in their admiration for a "steady-state economy" (220). Yes, if you're in good shape, then the status quo will work quite nicely! This is the sort of thing one often hears from elites in developed countries about those in less-developed countries: don't use your resources, cut down your forests, burn your fuel, etc.-- so we can help humanity. Isn't it great when rich people tell poor people not to use their resources?



At times, WP deviate a bit and get themselves into contradictory positions. Early on, they trot out items from the usual laundry list of policy prescriptions: housing, health care, work/life balance, child support payments, and the "provision of high-quality early childhood education" (112). (As an aside, they seem blissfully unaware of the U.S. government's expensive and largely-failed attempts to do the latter through Head Start.) But later, they argue that "Attempts to deal with health and social problems through the provision of specialized services have proven expensive and, at best, only partially effective." (238) Then, as examples, they cite a *wide* range of such efforts-- police, medical care, social work, drug rehab, obesity-- concluding that they're often "a form of window-dressing, a display of good intentions" (238). They leave unanswered or even unasked-- to determine how other government action could be better motivated, informed, and executed.

WP point to the potential value and positive impact of community-- and by extension, some aspects of public policy. But one should not (easily) extrapolate from what works in a small, homogeneous population-- to what will not work nearly as well with a large, heterogeneous population. (See: p. 208-209 for the most interesting discussion.)  

Among the many reasons cited by economists in the academic literature (193)-- for growing income inequality and struggles for some in the middle class over the last 40 years-- WP focus on Paul Krugman (!), "the Nobel Prize winning economist"-- even though he has no particular expertise in this narrow arena. (You gotta love that ever-so-popular fallacy-of-authority in trotting out Krugman!) He "argues that" inequality has been driven by reduced unions, a lower real minimum wage, and changes in taxes and benefits (even while his fans typically focus on pre-tax/transfer measures of income inequality). WP conclude by constructing a strawman: "changes in income distribution are almost never attributable simply to market forces influencing wage rates"-- to move to the claim that the key is "something much more like the changes in institutions, norms and the use of political power which Krugman describes in the U.S." (243). Nice try.

A few miscellaneous policy notes:
1.) Unions get some play, but without any explanation of how private-sector unions can rebound in an increasingly-competitive, globalized economy (245). 
2.) Without any apparent sense of irony, they cite NH and VT as examples of low/high tax burdens-- with the same results and presumably-similar cultures/populations (246). 
3.) The long discussion of non-profits (252ff) is somewhere between interesting and naïve. (I don't know enough to say where it fits on that spectrum. In any case, it's difficult to imagine how we would get to that world, practically.)
4.) They critique CEO pay on occasion (e.g., 250). But interestingly, they use Robert Reich to write their foreword-- and he provides a useful corrective (xi) on what turns out to be a complicated theoretical issue with a mixed bag of empirical results. (Punchline for the former: it's not the most competitive market, but its flaws are often exaggerated. Punchline for the latter: it's really difficult to measure well-- and even so, you can find results all over the board.)
5.) They have a weird focus on AGW throughout (but especially in chapter 15)-- at least in their efforts to connect it to the thesis of their book. When authors go that far afield, it raises alarm bells about objectivity and worldview. (That said, they have an innovative/impressive reference to "tradeable carbon quotas [222]-- the sort of approach that would usually be bothersome to those concerned with AGW.)
6.) They have an odd claim about public school spending-- not overall spending or spending inequality, but public spending as a percentage of total spending as the key variable (161). On what theoretical basis would one choose that variable?! That's the sort of thing that springs from data-miners or ideology-- neither is a good sign.
7.) They want to avoid "concentrating power in the hands of the State" (255), but they say nothing about doing that in the context of health care or, more important here, K-12 education. (The easiest way to help inequality: fix the jacked-up K-12 education system by reducing "the power in the hands of the State" and empowering those in lower classes!)
8.) Here is the same sort of analysis applied to weight inequality and to higher education.