Does redistribution even increase utility?
Greg Mankiw has an essay forthcoming in the Journal of Economic Perspectives that purports to defend the one percent. Mankiw contrasts the standard “social insurance” view of redistribution with his own “just deserts” theory. He makes a few good points against leftist caricatures of the one percent, such as:
Many of the gains to the one percent in recent decades are not due to rent-seeking. To the extent “that inefficient rent-seeking is a driving force behind rising inequality, the appropriate policy response is to address the root cause.” Tyler Cowen makes some related points at length in his essay, “The Inequality That Matters.”
CEO pay has grown in both closely-held and publicly-traded companies, so high compensation at the top is probably not a principal-agent problem. CEOs are most likely adding real value, at least in expectation, in many cases.
It’s absurd, within the framework of “utilitarianism,” to focus so much attention on domestic redistribution, when global inequality is far more severe. Most Americans are in the global one percent; calculate where you stand here.
After making these and other points, Mankiw assumes his conclusion that people should be compensated according to their contributions to society, which is exactly what is at issue when we argue for or against “social insurance.” This argument has been predictably panned by the blogosphere, including memorably by Noah Smith.
In spite of Mankiw’s weak argument, I think the case for “social insurance” is not very strong, even on its own terms. The standard argument says that there is diminishing marginal utility of wealth, and therefore, after some hand-waving about interpersonal utility comparisons that I will ignore, total utility will be higher if we redistribute from the very wealthy to the very poor. I’m not a philosopher, so I will put aside ethical and meta-ethical arguments to focus instead on how wealth and utility actually seem to interact. What do our utility functions actually look like with respect to wealth and other inputs?
The first thing to recognize about wealth and utility is that we probably overestimate how happy wealth really makes us in the long run. If I experienced an exogenous doubling of my wealth, I would be really happy about it. But over time, I would become habituated to my new lifestyle, and it’s not clear that I would be significantly happier. The short run gains to utility are not a reliable indicator of long run happiness equilibrium. This point is self-evidently true if we think about the extremes. If you’ve spent any time at all in poorer countries and gotten to know the locals, you will have observed that many relatively poor people are about as happy as you are, even though you might be suicidally unhappy if you suddenly experienced a reversal of fortune and ended up in their shoes. Humans evolved in poverty, so it makes sense that poor humans are not necessarily, by virtue of being poor, suicidally unhappy—otherwise the human race would never have survived. Instead, we experience happiness and unhappiness relative to the situation to which we are habituated.
Second, much of our happiness (and unhappiness) is driven by social comparisons of wealth, not by absolute levels of wealth. These social comparisons are not society-wide; envy is local. Most people do not really resent the British royal family, or Kobe Bryant, or Warren Buffett. Instead, they feel envy toward their slightly-more-successful co-workers, their sister-in-law’s husband, and so on. Conversely, most successful Americans do not gloat about how much better or more successful they are than sub-Saharan Africans, but they have been known to take some satisfaction at gains in status within a local community. Relatively small social communities provide a frame of reference for the understanding of one’s wealth, not a community at the scale of a nation-state or globe, and not an absolute standard of wealth.
Finally, even if we could somehow put those issues to the side, wealth explains only a tiny fraction of the variation in happiness. If you regress happiness on wealth (in microdata), you don’t get a high R2. There is a lot of variation in happiness among people at the same income levels. Much of this variation is due to internal attitudes, mental health, relationship status, and other life circumstances mostly unrelated to income. One pair of psychologists estimates that therapy is 32 times more effective at producing well-being than the equivalent cash as compensation for “pain and suffering” in legal cases (caveat lector). Even much of the measured effect of wealth on happiness is likely due to the fact that wealth correlates with success in carrying out one’s own chosen plans, which probably has a causal effect on happiness. Given the high variation in happiness among equally wealthy people, it seems that wealth on its own is not that important of an input into our utility functions.
These arguments, on their own, would suggest that redistribution doesn’t do much good from a “social insurance” perspective, but that it doesn’t do much harm either. After all, the rich will become habituated to the loss of income from redistribution, people in their social circles will be taxed heavily as well, and in any case, other factors matter a lot more than their income for their happiness. Nevertheless, redistribution involves a number of costs. Most obviously, there is a deadweight loss of taxation. If this deadweight loss only represented lost income to the rich, again, my arguments above would tend to diminish its importance. But the deadweight loss doesn’t merely represent lost income—it also represents distortions of activity (incentivizing leisure), a diminution in the opportunity to make and carry out one’s own plans (which may generate happiness of its own accord), and lost employment of rich and poor alike (which has psychological costs beyond the loss of income). Taxation and redistribution also imposes high implicit marginal tax rates on the poor. Again, this distorts activity, and disincentivizes pursuing goals that are correlated with income, which may matter for their own sake if not for the sake of the income that they would generate. Some people argue that welfare has led to the breakdown of two-parent families among the poor; I don’t know for certain if that is true, but if it is, then one can easily see how redistribution could be harmful if wealth itself does not generate long-term utility.
These arguments also suggest scenarios when redistribution is likely to improve welfare: when financially vulnerable people find themselves suddenly and unexpectedly with a problem that money can quickly solve. For example, if you suddenly become very ill and can’t afford the treatment, that can lead to a lot of anxiety. It’s easy to see how redistribution in that situation could be welfare-improving even if you buy all the arguments I have made so far; the sick and anxious person suddenly has a very high marginal utility of wealth, notwithstanding all of the above. However, this is not an argument for more general redistribution on the basis of income, or even for public provision of routine medical care, which is not sudden or as anxiety-inducing.
The bottom line is that the “social insurance” case for mass redistribution is not very strong even on its own terms, which I will emphasize are not my preferred terms on philosophical grounds. Utility functions simply don’t operate the way most MU-of-income arguments presume. Given our actual utility functions, it seems better to focus on the rate of growth, which will maximize our welfare from unhabituated income, and on policies that maximize the opportunities of rich and poor alike to formulate and achieve their own goals. “Fairness” or “just deserts” has little to do with it.