Yesterday at the colloquium, Jason Furman, who is now at the Harvard Kennedy School, presented Should Policymakers Care Whether Inequality is Helpful or Harmful for Growth? Here are some of my thoughts about this very interesting paper.
In common parlance there are these 2 things, “growth” and “inequality,” that often are discussed without the speaker being very precise about what exactly either of them means.
The old conventional wisdom held: growth is good, inequality is bad, but there is a tradeoff between them. Not only they are empirically correlated, but more of either tends to result in more of the other.
There is an emerging new conventional wisdom in some circles holding that one can indeed have it all, i.e., greater growth plus lesser inequality, again with not just correlation but causal arrows running both ways. Hence, directly addressing either can be win-win, improving the other as well.
The paper says: Not so fast. Jason suggests that, if he were a betting man, he would put his chips down in favor of “win-win” if the betting odds were 50-50, but not if they were much tilted the other way. This of course is just a description of his personal subjective probability for the causal relationship. But in part by reason of the relatively close odds, he says to those who favor addressing higher inequality: Don’t bet the house on this being true. After all, even if it’s true that reducing inequality could increase growth. And that’s not likely to be the reason why you care about inequality. So don’t unduly play down the other concerns by making the inequality debate one that is instead about growth.
Abroader point that the paper makes is conceptual: We need better-defined, more normatively meaningful, and more precisely differentiated categories than those that are offered by the general terms “growth” and “inequality.”
I will herein further discuss these issues in 3 parts: first growth, then inequality, then causal theories and takeaways from the topic and the paper.
In the literature that Jason has in mind, “growth” is typically defined as the increase over time in GDP, either absolute or per capita. The higher future GDP is relative to current GDP, the better.
To dramatize the argument that he’s making, let’s start by abstracting from time. What would we do if our policy goal was that current GDP be as high as possible, full stop (i.e., not just, all else equal)?
As a tax person, I naturally think of making tax changes first. So Step 1 might be to replace all taxes on income, consumption, wealth, etcetera, with lump sum taxes, such as uniform head taxes. Hence we would wholly avoid discouraging productive economic activity.
Why stop there, however? We could also, at gunpoint, force people of all ages to work long hours. After all, this would increase GDP, and this by hypothesis is our sole policy aim.
Or if that’s too radical, we could further raise lump sum taxes, such as uniform head taxes, in order to fund income subsidies, under which, the more you earn, the more the government pays you (instead of you paying it).
Something else we might do, if all we cared about was increasing GDP, was to confiscate people’s wealth (while somehow credibly promising that we would never do it again). The income effect of wiping out people’s savings would be to induce them to work more, so as to start replacing it.
By the way, lest this last idea seem too fanciful, it’s worth noting that, in the dynamic growth model that the Joint Committee of Taxation used with regard to the 2017 tax act, one of the sources of GDP growth within the budget window was the assumption that, with the fiscal gap being reduced outside the window with lump sum takeaways from people who had, say, expected retirement benefits under present law, such individuals would farsightedly respond by working more in response to the expected calamity for them. Not too much was said publicly about this supposed cause of “dynamic” growth effects.
Does this sound like an appealing policy proposal? I figured it wouldn’t. But what makes it so unappealing is that we don’t actually care just about GDP. The measure ignores the value of leisure, distributional considerations (i.e., who gets $$ or leisure), and all other relevant amenities and disamenities.
Given the obviousness of the point that maximizing current GDP is not a well-stated policy goal – except as modified to take account of a whole lot more – why would growth proponents state the long-term social goal as maximizing future GDP? The short answer is that they’re being foolish or myopic insofar as they focus just on GDP, without reference to distributional considerations, how much people have to work if they’d rather not, and a wide array of relevant amenities and disamenities. But I think their doing has been encouraged by a couple of things:
(1) Since we know less about future distribution than current distribution, people who are seeking a rhetorical edge as they urge the enactment of what they assert are pro-growth policies, have a degree of freedom simply to assume or assert that a rising tide lifts all boats (in tension with the actual facts about rising US GDP over the last 20 years, which has featured about a 0% share at the bottom).
(2) There are multiple narratives associated with comparing the future to the present that can lead to treating GDP growth as something to be welcomed more unconditionally and unreservedly than just good old GDP itself. For example, there are:
–Biological narratives: We like to think of our own lives as improving over time. And parents often want their children to have better lives than they are having themselves.
—Psychological narratives: Habituation to one’s current material circumstances may prompt wanting them to improve. And, by dreaming of a better future, one may sometimes soothe one’s discontent about the present.
—Historical narratives: Humanity’s economic rise from the Stone Age to the dawn of civilization to the Industrial Revolution and beyond has not gone unnoticed. We may also have examples of mind of countries that “succeeded” versus “failed” from common starting points, with the former experiencing far higher GDP growth. Examples might include the U.S. versus Argentina (which were on a par, as to per capita GDP, in the 19th century, or West Germany versus East Germany between the end of World War II and 1989 unification, or South Korea vs. North Korea. But in each of these examples GDP growth might be better seen as a consequence of greater success, rather than itrself an independent cause.
But whatever the force of these narratives, they don’t support ignoring that, for the future just like the present, GDP and social welfare are not equivalent. So I commend the paper for suggesting that we should be skeptical about just maximizing future GDP, just as we would not treat maximizing current GDP at all costs as a plausible summum bonum.
The paper doesn’t interrogate “inequality” to the same extent that it does “growth.” But it could!
For example, I frequently emphasize the important differences between high-end inequality (e.g., plutocracy) and low-end inequality (e.g., poverty), notwithstanding that they are commonly blended together in a single term (or in a composite measure, such as the Gini coefficient). They may matter for different reasons, and have different effects.
Thus, suppose one thinks inequality may reduce growth because the super-rich capture the political process and engage in rent extraction. That’s about the high end. Or suppose one thinks that poverty leads to a failure to develop children’s human capital. That’s about the low end.
In my view, the typical welfare economics maxim that the main reason for aversion to inequality is that material consumption has declining marginal utility does a better job of capturing the main issues by low-end inequality, but much less with respect to high-end inequality.
Even if high-end and low-end inequality were effectively the same, a given Gini measure that equates them could be under-informative regarding how the composite actually affects people’s wellbeing in one society, as compared to another. It may matter, for example, whether a given society features high or low social and economic mobility. Or it may matter whether (a) old elites are being challenged by new ones, or (b) it’s just new people not much different than the old.
Then of course there are such questions as “equality of what?” Typical candidates might include wealth, consumption, personal lifetime income, dynastic lifetime income, status, legal rights, political power, opportunity, etcetera.
One may also subscribe to an ethical theory under which it matters whether, or to what extent, economic success and failure are thought to be deserved. Meritocracy, for example, can be thought of as a theory of distributive desert. A meritocrat might ask: To what extent do people’s success and failure in my society depend on what I define as merit?
While there is no tension between any of this and the paper, it suggests an arena in which the paper’s deconstructive exercise could further be pursued.
3. CAUSAL THEORIES AND TAKEAWAYS
Sometimes we say: What we need in Area X is a good theory. That is not the issue when we’re considering the relationship between inequality and growth. Rather, there are if anything too many good theories. And, in at least some cases, they may be inconsistent, rather than complementary or offsetting.
Here are just a few:
(1) High-end inequality leads to greater growth, perhaps because the rich save and invest more (Kaldor).
(2) High-end inequality leads to lower growth, because (perhaps via its effect on the fiscal self-interest of the median voter), it prompts the adoption of higher capital income taxes that are anti-growth (Alesina-Rodrick).
(3) High-end inequality leads to lower growth, because the rich use their greater sway to increase rent-seeking (Acemoglu et al). Note that this theory, while having the same causal relationship as Alesina-Rodrick, bases it on a view of the rich as politically strong, rather than politically weak. Hence, one might expect some tension or even incompatibility between the two theories.
(4) Higher growth leads to greater high-end inequality, perhaps because technological transformations proceed via tournaments with concentrated mega-winners.
(5) Higher growth leads to lower high-end inequality, perhaps under a Kuznets model in which it eases (from the diffusion of new knowledge and production methods) as the society grows richer.
(6) Low-end inequality leads to higher growth, perhaps via the deployment of a mass low-wage workforce.
(7) Low-end inequality leads to lower growth, perhaps from wasted human potential as children in poor households suffer from under-privilege.
Each of these theories might at least sometimes be true, and several could be true (perhaps even offsetting each other) at the same time. But the plethora of causal pathways undermines thinking that there will be a stable relationship between inequality and growth, even disregarding all the issues raised by too simplistically deploying either of these two terms.
The paper urges of thinking in terms of a 2 X 2 grid, which might (under a progressive’s view of the issues) look like this:
PRO-EQUALITY Education, aid poor children, Capital income taxation
pro-competition (antitrust, Redistributive taxation?
weaker IP), min wage/unions?
ANTI-EQUALITY Opposite of Box 2? Opposite of Box 1?
Needless to say, there is considerable controversy regarding the assignments above of particular items to particular boxes. But insofar as something does indeed belong in Box 1, it would be dismaying, albeit unsurprising, to see it being rejected by prominent political actors.
A key argument of the paper is that, in economically advanced countries that have been politically stable and considering a relatively limited policy spectrum, there should be a “lexicographic” preference for Box 2 policies (upper right) and against Box 3 policies (lower left). The rationale is as follows. Suppose we look at advanced and (heretofore) stable countries with relatively pro-market policies, such as the US and the UK, and compare them to countries with very different, more pro-regulatory and redistributive policies, such as France or the Nordic nations. The growth differences between them over time have been so small that surely the distributional differences are more consequential. Hence, in a country like the US we should start by ranking our policy choices based on their distributional effects, and only use growth effects as a tiebreaker. The paper agrees that this approach is generally not well-suited to poorer countries with still-developing (or undeveloped) economies, in which basics such as the rule of law may be in question.
Adopting this lexicographic preference for looking at distributional effects first, and growth (or efficiency) effects only secondarily, would be a rather large change in U.S. policy debate. Consider how it compares to consideration of the 2017 tax act. Or consider the Kaplow-Shavell proposition, much debated in the law schools, that distribution issues should be left purely to the tax and transfer system, with all other legal issues (e.g., concerning corporate governance, torts, contracts, intellectual property, etcetera) being analyzed purely on efficiency grounds.
I’m reminded of Boris Bittker’s gibe, from the 1970s, to the effect that the Yale Law School faculty was a mix of young fogies (who cared only about efficiency) and old Turks (who cared only about equity or fairness). The young fogies prevailed for decades, but might the tide be turning again?