(Before It's News)
I view the papers as raising two main topics of interest to colloquium participants. First, how should we evaluate the papers’ main causal claim? Second, what follows if this claim is accepted?
The papers examine evidence regarding the empirical relationship between income inequality on the one hand, and health/ behavioral ills that have a social gradient on the other hand. (“Social gradient” means that, within a given society, they tend to be negatively correlated with income.)
The following ills are positively correlated with a society’s measured income inequality (e.g., based on Gini coefficients or the level of the 80th
percentile versus the 20th
percentile, among alternative measures that can be used): lower life expectancy, higher infant mortality, worse educational performance by children, and the frequency of teenage births, homicides, other violence, imprisonment rates, mental illness, drug and alcohol addiction, and obesity.
Likewise, measures of social trust and economic mobility are negatively correlated with income inequality. All this, of course, is based on adjusting for income levels. And the ills are positively correlated with income inequality even among the wealthy in a given society.
An initial question is: should we accept the correlation? It’s based on an enormous weight of evidence from different societies and time periods (all involving recent decades, however). It really seems to far too much evidence of correlation to ignore or reject.
Second question: how do we explain it? When A is shown to be correlated empirically with B, we can posit either that A causes B, or that B causes A, or that something else causes both.
Wilkinson and Pickett posit that income inequality causes the social bads
. They argue that this is intuitively plausible, reflecting that we are competitive social beings, equipped for both egalitarian and hierarchical relationship patterns, but happier and less anxious in the former (which seems to have prevailed in real evolutionary time from the very distant past until the rise of agriculture only a few thousand years ago). But they also argue that it best fits the evidence.
What about the theory that the bads cause income inequality
? Among the counters to such a view is the fact that, in the data, there tends to be lag from a given society’s rise in income inequality to its experiencing greater social gradient health and behavioral ills.
What about the theory that something else explains both
? Suppose we point the figure at culture – e.g., positing that something about U.S. and U.K. culture make us prime hosts both for income inequality and for social gradient ills (e.g., from having highly competitive cultures). Wilkinson and Pickett note, however, that the evidence fits their causal theory pretty well, whereas it’s difficult to make it work in terms of cultural fits. For example, why should Scandinavian countries resemble Japan, and why should Spain be so unlike the (historically far more unequal) Portugal when their cultures are so different?
I am strongly inclined to accept both the claimed correlation and the causal explanation that Wilkinson and Pickett offer. But there is going to be an ongoing social science debate about this, which I believe their side will decisively win.
2. Issues raised by accepting the causal claim
a) Is it culturally or ideologically specific to the present?
I wonder if inequality today is different from inequality in the past. The data obviously doesn’t go nearly far back enough to test this as one might like,
Consider the medieval epoch in Europe, during which inequality and hierarchy were considered entirely natural, by analogy to the family. Familial hierarchy running from parents to children is indeed a mental module that humans appear to have. And might it be generalized under propitious (for it) circumstances? Consider cats, naturally feral yet able to adapt their kitten-to-mother mental module to the circumstances of modern pet culture.
Not to be too whimsical here, but could today’s problems with inequality – not by any means to idealize it in the past – have something to do with the difficulty of adapting the familial module to a mass society, plus the consequences of experiencing it in a modern capitalist context where it has become intertwined with meritocratic ideology?
b) What is the “bad” that affects people?
The paper discusses “status threat.” This brings to mind nature documentaries and books about baboon society, full of violence from above and the continual need to defend oneself from rivals below.
Maybe not entirely a bad metaphor – baboon society is said to have considerable resemblances to ours – but in a modern social context, would the “bad” result more from vertical status differences, or from uncertainty and ambiguity in relative rankings?
Robert Frank would be inclined to shelve “status threat” explanations that seem to sound in bullying and resentment, in favor of positional externalities from consumption levels. There verticality might potentially make things worse, depending on how people judge the relevance for themselves of consumption levels way above their own. But this need not be a case of either-or.
c) Different kinds of inequality
I’ve been emphasizing the differences between high-end and low-end in equality (plutocracy versus poverty). The evidence adduced concededly does not as yet do much to address that issue.
Thus, for example, when one of the papers notes that high-inequality U.S. states have more social gradient ills than their peers, the former are a mix of the likes of New York and the District of Columbia, with the likes of Mississippi and Louisiana. These issues are important to address when one is thinking about policy responses to the negative social effects of inequality.
i) The pollution analogy
Inequality appears to generate negative externalities, in a manner analogous to that caused by pollution. How should this shape our responses? Suppose we are thinking in terms of a Pigovian pollution tax.
In the canonical Pigovian case (with the full information that’s possible in a textbook example), we know what marginal disvalue, expressed in monetary terms, the people adversely affected by pollution face on each additional quantum of it. And while of course we won’t precisely know this in any real world example, at least it gives us a useful framework for thinking about the proper design of a pollution tax.
In the case of inequality, we don’t know which types matter the most, whether the marginal harm rises in any sort of continuous fashion, or really anything about how best to price it. Also, whereas in the Pigovian pollution tax we have consumers whose current utility functions and circumstances determine the cost, in the inequality case we’re asking how people would relatively value their wellbeing under two completely different scenarios. They cannot easily judge this, even in theory.
So even if one comes out in favor of fiscal measures addressing inequality due to its negative externalities, we don’t have a great model available regarding how best to do this.
Of course, this is not the first time that we have faced such a problem in instrument design – as discussed, for example here
(in an article on taxation and the financial sector that I coauthored with Doug Shackelford and Joel Slemrod).
ii) High-end versus low-end inequality
I have been arguing for some time that we need to think about high-end versus low-end to a degree distinctly.
One reason is that the policy tools one would use to address them differ. Low-end inequality may be addressed in large part through public spending on public goods, along with consumer goods that are necessities and/or potential areas for market failure. For example, as Ed Kleinbard
and others have noted, the tax system is generally less on point here than ensuring people of good quality healthcare, education, childcare, transportation options, etcetera.
For high-end inequality, by contrast, we may be inclined to think primarily in terms of high-end marginal tax rates (and the tax base), inheritance taxation, the scope and term of IP protection, corporate governance, rules for the financial sector, etc.
A second reason for distinguishing between high-end and low-end inequality is that our rationales for wanting to address them may be quite different.
With respect to poverty, anyone equipped with basic beneficence should care about alleviating it, in order to make people better off. But concern about plutocracy requires further motivation, since the aim is surely not to make a bunch of people worse-off.
Under standard public economics and optimal income tax models, in which people care only about own consumption, the declining marginal utility of material resources provides the only rationale for concern about plutocracy (although there might also be ad hoc egalitarian weighting of a social welfare function). But if harms are being caused, the analysis may change dramatically – including by rebutting the view that one would never want to set a high-end tax rate above its revenue-maximizing level.
Against this background, the Wilkinson-Pickett evidence, in my view, does more to influence how we should think about high-end than low-end inequality, even though it is pertinent to both.