Worthwhile Canadian Initiative

Canada's income redistribution strategy: take from the rich, give to the median

There have been any number of MSM stories based on StatsCan's recent release on earnings and income. Median earnings from market income for individuals in 2005 are pretty much the same as they were back in 1980, and market income inequality has - by any measure - increased over the past 25 years.

This isn't really news - at least, not in the sense of 'revealing previously unknown facts'. Here are plots of average and median market income for unattached individuals and for for economic families of two or more people; the data are taken from Statistics Canada's Cansim Tables 202-0202 and 202-0203:

Although median market income in 2005 was the same as it was in 1980, it hasn't remained constant: the recessions of 1983 and especially 1991 reduced real median incomes considerably, and it's taken ten years to recover. And the widening gap between the average and the median indicates increasing inequality during this time; something we knew about already (see this post, and this one, among others).

Inequality in market income doesn't bother me much in itself; what really matters is inequality in income after taxes and transfers. If policy-makers are responding to rising inequality by improving its programs for redistributing income, then the effect on inequality of disposable income will be a wash.

Unfortunately, that's not what has been happening in Canada. Here is how net transfers have changed; the data are taken from Cansim Table 202-0704:

Net transfers to the highest quintile have decreased; income growth has been concentrated at the top end of the income distribution, so their tax payments have grown in proportion. But the main beneficiaries are not those with the lowest incomes; the increase in net transfers has been concentrated on the middle income groups.

Here are how the shares of (gross) government transfers have changed over the past generation:

The share of transfers to the lowest quintile has decreased since 1980, as has - albeit to a lesser extent - that of the second quintile. The winners in this reallocation are the middle and fourth income quintiles.

It's not hard to imagine an explanation for this; the median income group is likely to include the median voter. So every political party will be happy to sacrifice the lowest income group's interests (that group is either taken for granted or written off entirely) in order to gain popularity at the centre.

On the political economy of a basic income

The idea of a basic income - it's generally referred to as a Guaranteed Annual Income in Canada - has been floated again. My initial reaction was that this is such a good idea that it's hard to figure out why we don't have it already.

For some reason, there doesn't seem to much in the way of a formal modeling exercises that work through the general equilibrium implications of a BI (I will be happy to be corrected on this point), so I decided to set up a blog-sized version. It turns out that although the BI is still a good idea, it's not quite the slam-dunk I thought it was.

The gory details are below the fold.

Gini coefficients in Canada, the UK and Sweden: 1980-2005

The national statistics agencies of Canada, Sweden and the United Kingdom have annual series for gini coefficients going back to 1980 or so. Although cross-country comparisons of the levels should be taken with a grain of salt, the trends in these three countries are remarkably similar:

It is a Well-Known Fact^TM that inequality has been increasing over the past couple of decades, but it would appear that the greatest part of this increase took place before 1995 or so. Over the past ten years, inequality in market income has remained fairly stable.

Update: My initial, off-the-top-of-my-head conclusion was that these inequality patterns were driven by technology (eg., the familiar skill-biased technical change story) rather than country-specific things such as trade patterns, labour market structure, etc. But since I didn't have anything stronger than an off-the-top-of-my-head impression to back it up, I didn't mention it. But this study offers some support to this notion:

[T]he causes of rising inequality are primarily structural and related to new technology rather than to trade or institutions.

Cross-country comparisons of inequality in market and disposable income: Policy matters

This graph is taken from a recent Luxembourg Income Study (LIS) working paper (45-page pdf):

The countries are arranged in ascending order of inequality in disposable income, and  the Nordic countries take four of the top five positions. What strikes me is the extent to which this is due to government policy: the Gini coefficient for market income in Canada is the same as Denmark's, and is quite a bit lower than in Sweden. Indeed, Sweden is closer to the US than it is to any of the other Nordic countries.

A recurrent theme in discussions of the Nordic model takes the form of "That's all very well, but those policies won't work here without [insert some feature of Nordic countries here]." Libertarian types who would otherwise approve of the free market dynamism of the Nordics assert that the Nordic model can only work in small, homogeneous countries. As a general argument, I'm not convinced - but I can see why it would be hard to export the Nordic model to the US.

At the other end of the spectrum - those who would otherwise approve of Nordic levels of spending on social programs - some (eg: this commenter) point to the role of trade unions. But it's hard to conclude from this chart that union density matters much when it comes to reducing inequality. For example, look at Germany (where unions play a crucial role in setting wages) and the US (where they are decidedly less important): both have identical levels of inequality of market income. The distribution of disposable income is lower in Germany because of its redistributive policies, not because unions are more powerful.

That's not to say that cross-country institutional/cultural idiosyncrasies aren't important; they are. But there's little reason to believe that these factors have to be  changed before the Nordic model can work.

Corporate profits are not driving the increase in the top-fractile income shares. So how would higher corporate taxes reduce inequality?

Jason Furman, guest-blogging at Free Exchange, advocates higher corporate taxes. He gets at least two important things wrong:

1. He seems to think that corporate taxes will be entirely paid by owners of capital. It's more likely that investors will bear approximately none of the tax.
2. He suggests that corporate taxes will help slow the growth of the income share that goes to the very top end of the income distribution.

I've already dealt with the first point. As for the second, let's take a look at the data so helpfully provided by Emmanuel Saez (excel file). Between 1980 and 2005, the income share of the top 0.1% of the distribution increased from 2.23% to 7.45%. But more than 90% of that income growth came in the form of wages, entrepreneurial income and capital gains - none of which are affected by corporate taxes. Dividend income paid after corporate taxes accounts for only about 5% of the income growth of this group.

It's hard to see how higher corporate taxes could help slow the concentration of income at the top.

Why focus on progressive taxes and not on progressive transfers?

Many posts in the economics blogosphere on the subject of progressive taxation today: Greg Mankiw discusses this paper, Mark Thoma points us to a WSJ article, and Brad DeLong  links to Mark's post twice (here and here).

Inequality - both its level and the rate at which it has been increasing in Canada and the US - is a serious issue. But taxing the rich (and dealing with the resulting incentive effects) isn't the only policy instrument at our disposal; we can also reduce inequality by directed transfers to low-income households.

A couple of weeks ago, at the meetings of the Canadian Economics Association, Andrew Jackson presented a paper with the following table:

I've just ordered the Pontusson reference upon which this table was based, so I'll come back to this point as soon as I can. But it looks very much as though the Nordic countries - which are rich, devote a large fraction of their income to social programs and which do fairly well in reducing inequality - seem to have given up on tax policy as an instrument for income redistribution.

Homogamy, inequality and social mobility

A couple of recent studies by Statistics Canada on the phenomenon of 'assortive marriage' have generated some comment. The point is easy enough to explain and it's not really hard to understand:

Changing role of education in the marriage market in Canada and the United States:

...54% of Canadian young couples had the same educational level in 2001. However, 33% of couples differed by one educational level, while only 12% of couples differed by two educational levels. Less than 2% of couples differed by more than two educational levels.

Educational homogamy has been rising mainly because intermarriage between adjacent education levels has been declining both at the top and bottom of the educational hierarchy.

Declining intermarriage between those with university degrees and those with less education was a major factor in both countries. In Canada, the rate of intermarriage between the university educated and those with only some post-secondary education fell by 38%; in the United States, the rate fell by 45%.

Similarly, the intermarriage rate between high school graduates and those who had completed less than high school fell by 30% in the United States and by 58% in Canada.

In another study, the changing composition of families is offered as an explanation for some of the increase in income inequality:

While this study does not investigate why family market-income inequality rose, one factor which likely plays a role in this is a widening inequality in family earnings (from wages, salaries and net self-employment income). A key driver of this is the rising earning power of the two-earner family, especially when both earners are highly educated. (Preliminary results suggest that individual earnings inequality is not driving this trend.)

While homogamy - I've never seen this word before, but it works for me - has no doubt been a contributing factor to income inequality, it's far from being even a partial explanation for the increase in inequality that we've seen in the past couple of decades. Those gains are concentrated in the top 1% of the income distribution and higher, and I doubt very much that there are that many power couples where each partner is pulling down $1m/yr.

But the longer-run implications for social mobility are disquieting. Parents' education levels are a good predictor for post-secondary education levels, and there's a strong link between education and income. This trend of increasing education-based homogamy will likely increase the correlation of income levels across generations.

Income distributions and Mandelbrot sets

Berkeley's Emmanuel Saez and Mike Veall at McMaster University have a paper (AER version; NBER version) on how the top end of the income distribution in Canada has evolved over time. (Saez has of course been involved with many other projects using US data.) Happily for me, they've posted what is now my favourite excel file so that I can get a better feel for a fascinating piece of work.

I was aware that the share of income at the top end of the distribution had risen in recent years, but what I hadn't realised was just how concentrated those gains were. In fact, they're so strongly concentrated that the income distribution is self-similar: it looks the same regardless of how closely you zoom in. It's not quite as fascinating as all those cool Mandelbrot sets (example here) that popped up when chaos theory was fashionable, but it's still worth looking at.

Why free tuition is like the Bush tax cut

While doing the research for my recent posts on tuition fees, I came across something from the Canadian Federation of Students called 'Myth or Fact: A guide to common myths about the importance of reducing tuition fees' (9-page pdf). Here's an extract:

MYTH: “Tuition fee freezes unnecessarily subsidise the cost of postsecondary education for those who can afford it.”

FACT: Disgraced former Ontario Premier Bob Rae and conservative researcher Alex Usher promote this fallacy in order to popularise the notion that a “one-size-fits-all” tuition fee (also known as regulation) is obsolete. Instead, Rae and Usher champion fully deregulated tuition fees cushioned by a tuition fee waiver for a tiny sliver of the population. The argument is this:
• every student (poor, rich, or in-between) pays roughly the same tuition fee and receives equal benefit from freezes and reductions;
• low-income Canadians are under-represented in universities;
• low-income Canadians pay taxes that support public universities and colleges; therefore
• low-income families are subsidizing the participation of higher income families.
The facts do not support Rae’s and Usher’s tuition fee campaign... [E]conomist Hugh MacKenzie recently examined the issue and found no evidence that low tuition fees result in a net transfer of resources from low-income households to high-income households.

As polemics go, it's not particularly compelling: the 'myth' is ascribed to people who are described as 'disgraced' and 'conservative', and it is countered by an unsupported assertion that someone has disproved it elsewhere.

But I was curious enough to try to track it down, and I eventually came across a paper with the title "The Tuition Trap" (31-page pdf). Chart 5 from this paper presents data that are consistent with the stylised facts from this post:

University students are twice as likely to come from the top income quartile than from the lowest income quartile - which means that twice as much public money will go to tuition subsidies for students from high-income families.

Here is Hugh Mackenzie's explanation for why this is okay:

[T]he fact that young people from high-income families are overrepresented in the student population and young people from low-income families are underrepresented does not mean that subsidizing tuition from general government revenues results in poor families subsidizing rich families.

One cannot make that claim without knowing the distribution of the taxes levied to provide the subsidies.

We know from other studies that overall, provincial taxes are distributed approximately in proportion to income [see this post - ed]. This means that income groups in Ontario contribute to general revenue roughly in proportion to their share of total income...

Referring to chart 5, where an income quartile group’s share of the college and/or  university student population is greater than its share of the total income of all families with children, a tuition subsidy paid for from general government revenues amounts to a net income transfer in favour of that group. Where an income quartile group’s share of the college and/or university student population is less than its share of total income, a tuition subsidy paid for from general government revenues amounts to a net income transfer from that group to other groups.

The first and second quartile groups (incomes below $56,800) make up a larger proportion of the student population than of family income. Subsidized tuition provides a net transfer in favour of families in the lower half of the income distribution. Families in the third income quartile account for roughly the same share of students and of income. There is essentially no cross-subsidy either in favour of or against families in the third income quartile. The fourth (highest) quartile accounts for a smaller proportion of college and university students than it does of total income, so the highest-income 25 per cent of families in effect subsidizes tuition of the lowest-income half of families, through their contributions to the tax system.

Given the overall pattern, one would expect that families in the top half of the third quartile would be underrepresented among students relative to their share of income, and that families in the bottom half of the third quartile would be overrepresented among students relative to their share of income – producing a rough balance within the quartile.

This means that, to the extent that tuition does pose an economic barrier to college and university participation by people from lower-income families, substituting tuition for public funding will tend to reduce the net transfer from higher-income families to lower-income families; replacing tuition with increased public funding will tend to increase the net transfer. More than 60 per cent of families with children are net beneficiaries of the transfer inherent in subsidizing tuition from general government revenues.

The claim that subsidized tuition amounts to an unfair, regressive income transfer from poor families to middle- and upper-income families is simply not true.

My original point about the regressive nature of the tuition subsidy did not take the tax system into account: I deemed it regressive because it spent more money on students from families with high-incomes than it did on low-income students. To my mind, a progressive system would direct more money to low-income students.

The effect of bringing the tax system into the analysis is to greatly lower the progressive bar. In order for a program to conform with this new standard for progressive policies, all you have to do is make sure that the benefits are distributed in a way that is less unequal than the underlying income distribution. This is a remarkably easy test to pass - even George W. Bush's tax cuts satisfy this criterion.

Although free tuition does satisfy this incredibly weak definition of progressive, it just doesn't pass the smell test:

Low-Income Kid: "Hey, you guys from high-income families get twice as much public money as we do. Do you think that's fair?"

High-Income Kid:  "Sure it is - our parents make more than five times as much as yours do."

Increasing tuition subsidies is a regressive policy

It's budget season at the federal and provincial levels, and in its ongoing campaign to improve access to post-secondary education (PSE), the Canadian Federation of Students is campaigning for reduced tuition fees, and is taking to the streets to make its opinion known.

Given what we know about this issue, it's pretty clear that the most important obstacle to increasing access to PSE is the Canadian Federation of Students. Reducing tuition fees amounts to giving scads of free money to upper-income households, and will improve PSE access hardly at all.