Yesterday at the colloquium, Gabriel Zucman presented the above paper, coauthored by Emmanuel Saez. This is a forthcoming Journal of Economic Perspectives paper that discusses increasing the menu of available approaches to measuring material inequality by linking up national macroeconomic accounts to information about individuals and households. Its being written for JEP not only increases its accessibility to a legal audience, but also makes it a great vehicle for teeing up questions that interest me about how to conceptualize measuring material inequality. (I would not be the right person to try to add value regarding various of the technical issues that pertain to using different kinds of data sources.
I will use this blog entry (and the next one) to tee up, without purporting to resolve, two types of conceptual issues in the area. The first is how our reasons for being interested in inequality might affect how to measure it. The second, which I will save for a follow-up blog entry, is how one might think about a number of the particular issues that are raised by efforts to measure inequality.
Why does inequality matter? - A lot of the concern that would motivate measures such as those in this paper pertains to high-end inequality, or wealth or income concentration at the top. Thus, a lot of the content relates to the top 1, .1, .01. .001, or .0001 percent.
Even just under utilitarianism, which is but one of many normative theories regarding just distribution (and which incorporates no inequality aversion as such), the problems one might have in mind include the following:
Declining marginal utility - If richer people derive less marginal utility from a dollar's worth of resources than poorer people, the marginal welfare payoff to wealth and income held at the top may be extremely low.
Stress, discord, and social gradient ills - A lot of work in both the social sciences and the humanities suggests that extreme concentration at the top can undermine trust and social concord, not to mention democratic institutions, while also increasing psychological stress at all levels. Richard Wilkinson's and Kate Pickett's The Spirit Level is an important example.
Positional externalities and top-down consumption cascades - People who are not willfully psychologically naive generally agree that it's embedded in our human nature to care about relative position, and to be very prone (at least in a society like ours) to evaluating this in part materially. Robert Frank has applied this analysis to argue that top-down "consumption cascades," a kind of keeping-up-with-the-Jones on steroids, is triggered by substantial and rising high-end inequality.
Top-end political and cultural domination - Anyone who has lived in the United States for any fraction of the last few decades knows what I am talking about here. Political scientists such as Lawrence Bartels and Martin Gilens have documented the degree to which the interests and policy viewpoints of the bottom 99% of the U.S. distribution fail to influence policy outcomes.
Macroeconomic issues (growth, stability, and social mobility) - There is also evidence, albeit some of it still disputed, suggesting that high-end wealth and income concentration reduce economic growth, macroeconomic stability, and also social mobility.
Each of these aims might importantly affect how one measures inequality. For example, an item might increase the lifetime welfare of people at all income and wealth levels, thus making the measure look more equal insofar as it is enjoyed on a roughly per capita basis, yet have no effect whatsoever on one or more of these issues.
Relevance of non-standard consumer choice - This potential gap or even gulf between relative lifetime material welfare at different levels and the evils resulting from extreme high-end inequality is especially enhanced by the frequent failure of standard neoclassical / price theory assumptions about consumers to apply in particular real world settings.
Suppose, for example, that the people in a society exercise consistent rational choice, and have access to perfect and complete financial and other markets. Then everyone's behavior and welfare today would reflect the expected level of their expected retirement benefits, from Social Security and otherwise. For example, even if I were 25, my expected lifetime income, reflecting the value of such benefits, would be allocated by me between periods depending on my preferences. The benefits would be no different than an equal-value cash deposit in my bank account. (Given the assumptions here, I would also be able to swap out the risk.)
In that scenario, not including in one's measure those benefits at their expected value would verge on being malpractice. But once we have illiquidity, incomplete markets, possible myopia, etcetera, the case for inclusion becomes much less clear - and depends on which issues pertaining to inequality one is interested in at the moment.
Likewise, the textbook consumer, operating in complete markets, chooses between commodities with the aim of equalizing the marginal utility derived from the last unit of each. So getting more of anything moves one to a higher indifference curve, and there's no difference between cash and particular commodities that, by assumption, could immediately and costlessly be swapped for other commodities.
This model does not apply, for example, to individuals who get medical treatment that is funded by Medicare or Medicaid. Still less does it apply, for example, to, say, one's per capita share of the benefit (if any) derived from one's government's national defense spending. (This is an item that, for good logical reasons, the paper discusses allocating between individuals at different income levels.)
A final general conceptual point here, before I close out this blog entry so as to discuss particular issues in Part 2, is that a lot of different types of considerations go into designing a measure of a given society's material inequality at a given time. There are, for example:
(1) theoretical issues, such as those noted above,
(2) issues of computational tractability,
(3) issues of comparability, since it is desirable to be able to compare different countries, along with the same country in different eras, using a consistent methodology, and
(4) issues of clear presentation and public / political impact.
Given all these different considerations, it's vital to keep in mind that measures are just a means to greater understanding, and not themselves the end. Thus, for example, even if we rightly decide that a given item must be excluded from the measures, all things considered, we should not back or fool ourselves into ignoring the systematic influence that it might have had if included.
Also, there is no straight line link between how a given measure comes out, and how severe a given problem that is associated with high-end inequality might actually be in a particular social context.
For example, suppose counterfactually that high-end wealth and income concentration were identical as between the U.S. in 1960 as compared to 2020. (In fact, all of the reputable measures agree that it is significantly higher today.) The fact that consumption norms have apparently shifted both at the very top and further down, in favor of greater and more conspicuous display, might nonetheless make a lot of the high-end inequality problems that we face today significantly worse than those from 60 years ago.
On the other hand, other things that interact with inequality may have gotten better. For example, racism, anti-Semitism, and sexism, each of which interacts toxically with material inequality, surely were in many ways worse in the US in 1960 than 2020, however bad they continue to be today.