The debate about income inequality seems to be happening at two levels, which I'm going to label "first-order" and "top-end" inequality.
- First-order inequality is visible in standard measures such the gini coefficient, and shows up as an increase in the gap between average and median incomes.
- Top-end inequality refers to the share of income that goes to those whose income puts them above the 99th percentile and beyond.
I think it's important to make this distinction: top-end inequality generally doesn't show up in first-order measures, and vice-versa. And each type of inequality has its own challenges for policy.
Here is an exerpt from a very nice column by Pete McMartin, in which he quotes Kevin Milligan:
In the aftermath of the HST referendum, Kevin Milligan, a University of B.C. associate professor of economics, came across a poll that caused him pause. Milligan, who specializes in tax policy, had campaigned hard for the harmonized sales tax, believing it to be good for the economy.
The majority of respondents in that poll agreed with Milligan. They, too, felt that, overall, the HST would benefit the economy.
But oddly — and here is where the respondents’ opinions diverged from Milligan’s — while they recognized the HST would benefit the economy, they also felt that, on a personal level, none of the HST’s benefits would accrue to them.
A message resided in that distinction, Milligan felt, one that perhaps spoke to a greater malaise.
“With all the growth in the economy the HST promised,” Milligan said, “these people couldn’t bring themselves to vote for it because they didn’t see that growth benefiting them.”
(Here is the poll Kevin was talking about.)
This is a first-order problem. Economists will often concentrate their attention on the question of whether or not a policy will increase total - and hence average incomes. Behind this choice is an implicit assumption that most people should see some benefit, or at least that the number of those who benefit outnumber those who lose out. But if this isn't the case, then we end up with what we saw in BC: people rejecting a policy that they believe will increase average incomes.
First-order inequality is - in my view - a far more important problem than top-end inequality. It also turns out that in Canada, it's one that seems to have stabilised, albeit at levels that we may not find acceptable. Here is a graph from an earlier post:
Here is a graph of the ratio of the 80th and 20th income percentiles:
The divergence between mean and median incomes is a first-order concern:
Here is the same graph for incomes after transfers and taxes:
The story before the mid-1990s is pretty grim: stagnant incomes and increasing inequality. But first-order measures of inequality have levelled off since then, and median incomes - in particular, median incomes after transfers and taxes - have grown appreciably. To a very great extent, the belief that most people have not benefited from economic growth is not consistent with data from the last fifteen years. Moreover, the recipe for reducing first-order inequality is fairly straightforward: strengthen the system of transfers.
Top-end inequality is very different. Firstly, it shows no sign of slowing. And since we're not entirely sure why it's happening, it's not immediately obvious that measures such as a high-income surtax will have any more than a symbolic effect. For that matter, it's not clear how improving transfers would affect top-end inequality.
More fundamentally, the questions that are raised by top-end inequality quite different from those raised by first-order inequality. Top-end inequality raises the spectre of a tiny economic super-elite using its income and wealth to establish its dominance in other areas of society.
A solution for one sort of inequality is unlikely to do much for the other. We've already seen a levelling off of first-order inequality, and that didn't stop the worsening of top-order inequality. Nor does it seem likely that a solution to top-end inequality would do much for the first-order problem; the numbers involved are simply too small.
I think this distinction is important:
- Arguments to the effect that inequality is stable because the Gini index has levelled off are addressing first-order inequality, not top-end inequality.
- An excessive focus on top-end inequality - at the expense of first-order inequality - means that too much attention is paid to taking from the rich, and not enough to the other half of Robin Hood's agenda.