On trends in inequality in Canada, the United Kingdom and Sweden

The Conference Board released a study on trends in inequality, so I decided to update the data from this old post. Here's what I found: if you measure inequality using the Gini index, you'd conclude that after increasing during the 1980's, inequality levels had stabilised since 1995.

Here is the updated graph from that post:

(The break in the black line is due to a data redefinition in Sweden.)

These sets of numbers are all generated by the various countries' statistical agencies, so I really shouldn't be plotting them together like this - cross-country differences in methodology likely make direct comparisons a somewhat dodgy exercise. But even if you don't think that the Gini levels have converged to the same level in all three countries, you would still have profiles that are relatively flat since 1995. Current Gini indices are what they were 15 years ago. (You see the same pattern in the US as well.)

In this paper, Lane Kenworthy offers this remarkable graph, which suggests that there really isn't much in the way of cross-country variation in market income inequality:

If there are countries that have succeeded in addessing inequality, it is certainly not by means of arranging more equal market incomes. (See also my series of posts that note that initiatives to use minimum wages to reduce inequality are at best pointless: [1], [2], [3].)

Of course, what really matters isn't inequality in market incomes; we're mainly concerned with inequality in disposable incomes after transfers and taxes. If increasing inequality in market incomes is offset by increased redistribution, then the effect on disposable incomes can be cancelled out.

Here are the Gini coefficients for disposable incomes:

Some observations:

1. These lines are all below those in the first graph: disposable income is more equally distributed than is market income.
2. The 0.08-point increase in Canadian market income Ginis shows up as a 0.04 increase in disposable income Ginis. In other words, half of the increase in inequality in market incomes was offset by the tax and transfer system.
3. In the UK, only about one-quarter of the increase was offset by government policy.
4. The increase in the Gini coefficient in Swedish disposable incomes was greater than the increase in that for market incomes. Swedish policy had the effect of amplifying the increase in inequality.
5. Notwithstanding, Swedish redistribution policy remains the strongest of the three.

Here is a graph of the difference between Gini coefficients for market and disposable incomes:

Although Sweden's system is still more progressive, it has become less so.

Marc Frenette, David Green and Kevin Milligan have an article in Canadian Public Policy on the evolution of Canada's tax-and transfer system:

In this paper, we investigate the relationship between the substantial changes in tax and transfer programs and the movements in after-tax income inequality over the 1980s and 1990s. We show that in the 1980s, tax and transfer programs became more redistributive, offsetting substantial increases in market income inequality. In the 1990s, the tax and transfer system stopped undoing the increases in market income inequality, leading after-tax income inequality to rise. Even so, tax and transfer programs were more redistributive in 2000 than in the 1980s. Much of the changes occurred at the provincial level, with social assistance payments first increasing (in the late 1980s) then decreasing (in the late 1990s) and with surtaxes on high-income earners being first imposed and subsequently removed.

The graph is consistent with this story. They also note that much of the movement in the redistributive nature of the tax and transfer system comes from transfers, particularly at the provinicial level. The Swedish data I have don't break out the roles of taxes and transfers, so here are their respective contributions to the reduction in Gini coefficients in Canada and the UK:

 

Here is a point that cannot be repeated enough: most of the reduction in inequality in Canada and elsewhere is due to the system of transfers. This is a very general result; here is another graph from that Lane Kenworthy article I mentioned:

The countries that have had most success in reducing income inequality have done so with their system of transfers, not with the tax system.

A final technical note: regular readers may be wondering about posts like this, which document the increasing concentration of income in an increasingly smaller number of high earners. Why aren't we seeing this trend in the Gini coefficients? The answer is that Gini coefficients don't do a very good job of capturing this trend - which is why there's such an active research agenda on the question of how to measure inequality.

Comments

[Livio Di Matteo]

Income inequality is only part of the picture. If you look at the Morrisette and Zhang (2006) Statistics Canada study, wealth inequality is also on the rise and it's Gini is higher than income- values above 0.7.

[Stephen Gordon]

Do they correct for age? Because even if income is perfectly equally distributed, financial wealth will be concentrated among those who are older.

[Livio Di Matteo]

They do take population aging into account but conclude in the absence of aging the wealth distribution would have been a bit more unequal.

[Kevin Milligan]

I echo Stephen's comment about wealth distribution. An economy where everyone goes through the exact same pattern of earnings and wealth accumulation (ie no lifetime inequality) would yield big wealth inequality because of differences across age, in the way Morisette / Zhang do it. Within-age inequality of wealth is much more interesting to me, not across age.

[Pdt]

Income inequality is only part of the picture. And transfers are of different kinds. Some relate to essential functions that the market does not perform at all or performs badly (such as providing the next generation or providing for the old or disabled). Some have the larger purposes of reducing poverty or inequality as such. And the difference between tax and transfer and a wage is that a wage is earned - it is seen to be a return for effort and skill, which belong to the person who gets it. A transfer belongs to the state, and the state can impose what conditions it likes. Bit like a coupon from a company store. So unemployment benefits (and increasingly others) come with all kinds of conditions - which have steadily become more onerous and petty over the years. So felt inequality, and the rough equality of relations between citizens essential to a democracy, is much better supported by a minimum wage than by tax and transfer. Examples are not hard to find.

[Patrick]

Sounds more like an argument for universal redistribution: tax progressively, throw the money in big pot, and cut everyone a check. Hard to attach a stigma to that.

[Brett]

Does a growing population of retired people contribute to the growing income inequality?

[Stephen Gordon]

That's a really good question. I'll have to think about that one.

[Frances Woolley]

Brett: "Does a growing population of retired people contribute to the growing income inequality?"

It depends a lot on whether you measure income before or after tax, the kinds of households that people live in, how you adjust for household size, and whether or not you take into account the imputed value of owner-occupied housing.

But it probably would make a difference. Some people argue that two-income families contribute to income inequality because they tend to have higher than average incomes -- that's the flip side of the greater risk of poverty faced by by single parent families.

As important going forward is that today's old are primarily from a generation that had a low divorce rate. Tomorrow's old are going to be from generations with much higher divorce rates/lower marriage rates. And divorce is brutal from a financial asset accumulation perspective, while marriage is excellent.

The high poverty rates experienced by recent immigrants also contribute to income inequality.

[Evan]

Isn't the reason that the Gini coefficient doesn't capture the concentration of income at the top of the distribution the same reason that transfers have a bigger impact than taxes?
i.e. if we use different measures of inequality the difference between taxes and transfers will not be as stark?

[Chris Auld]

Why should we care about wealth---or even income---inequality? We should care about consumption inequality.

[Kosta]

Great post Stephen, thanks!

[Stephen Gordon]

Good point. I've not seen much in the way of inequality measures for consumption spending.

[Frances Woolley]

Krishna Pendakur at SFU has done work on consumption inequality. The big difference is in the situation of the elderly once you taking into account owner occupied housing. Also taking into account value of household non-market production makes a huge difference.

[K]

Chris Auld: "Why should we care about wealth---or even income---inequality? We should care about consumption inequality."

Consumption is not what makes me happy. It's permanent income or *future* consumption. I *live* for future consumption. Maybe the rich all plan to consume massively in the future. Only considering current consumption ought to be of appeal to accountants, not economists.

[Bob Smith]

K,

That's a fair comment, I suppose (although I'd suggest to you that current consumption, or the lack thereof, is of some considerable importance to people other than accountants). But that's equally a fair criticism of most (all?) measures of income (and wealth) inequality.

[Chris Auld]

K, one of the reasons to examine consumption inequality rather than (or in addition to) income inequality is transient income shocks are smoothed out of current consumption but not current income---permanent income is a reason to examine consumption and not income, not the other way around. There is a fairly large literature on consumption inequality, which was authored by labor econometricians and not, surprisingly enough, accountants.

[Frances Woolley]

Though one argument against considering consumption inequality (or for, I guess, depending upon your perspective) is that focussing on consumption brings down the top end of the distribution. Think, for example, of Warren Buffet. One of the wealthiest people in the world, but from a consumption point of view, he's not so much different from an average member of the upper middle class.

Is that a reasonable way of capturing Warren Buffet's economic position?

[Paul Friesen]

According to "The Spirit Level" by Richard Wilkinson and Kate Pickett there is a country where the market inequality outcome is very different. That is Japan. Japanese inequality is comparable to Sweden's, but it is almost all due to market outcomes, with very little from redistribution. I notice that Japan is not in the countries you look at.

This book is well worth reading for anyone interested in inequality.

[Stephen Gordon]

Hmm. I just checked the OECD site. Here's what they gave for Gini indices before and after transfers and taxes:

Canada: 0.44/0.32
Japan: 0.44/0.32
Sweden: 0.43/0.23
UK: 0.46/0.34
US: 0.46/0.38

[Patrick]

Isn't it the impulse response that matters? WB's income could be cut in half and it would have zero effect on his consumption. WB could become permanently unemployed tomorrow and apart from being bored, it would have no effect.

Cut my income in half permanently and I'd be in serious trouble. Reduce my employment income permanently to zero and I'd be on the street within a year.

[Paul Friesen]

I'll check the book when I get a chance. Their data may be old but I think they used a different inequality measure. I think the story was that the very top of the distribution does not get the huge amounts seen elsewhere.

[Paul Friesen]

I looked up some data in the book, although I don't have time to investigate properly. It has a graph of the ratio of incomes between the top 20% and bottom 20%. By this measure, Japan is the most equal country. They don't give the date of their data unless it's hidden somewhere in a footnote or something. The book was first published in 2009.

Reading off the graph, the ratio is about 4 for Japan and just slightly higher for the Scandinavian countries. Canada is around 5.5. Portugal, USA, and Singapore are the least equal of the countries shown, with ratios from 8-10.

I am surprised that the Gini gives such different results. Perhaps things have changed that much since this data. Also, I think the Gini is more sensitive to what's happening in the centre of the distribution than to the tails.

I'll try to find the time to investigate this more, but it will be a few days at least.

[Mike Moffatt]

That's one of the criticisms I've read of "The Spirit Level" - their findings don't hold up if you use more robust measures of inequality than the simple ones they present in the book.

[Mike Moffatt]

Wikipedia has a pretty good collection of various inequality measures here:

http://en.wikipedia.org/wiki/List_of_countries_by_income_equality

[Determinant]

According to "The Spirit Level" by Richard Wilkinson and Kate Pickett there is a country where the market inequality outcome is very different. That is Japan. Japanese inequality is comparable to Sweden's, but it is almost all due to market outcomes, with very little from redistribution. I notice that Japan is not in the countries you look at.

This book is well worth reading for anyone interested in inequality.

Not so fast. Japanese corporations carry a much heavier burden of public welfare than is typical in Western countries. Salarymen are promised job-for-life in return for their undivided and utter loyalty. It is welfare capitalism taken to its extreme. Company towns are common in Japan.

This trend to welfare capitalism is reinforced by Japanese corporate structure which relies heavily on cross-holding of share positions, interlocking directorships which results in a network of seemingly independent companies actually acting as a single company in a common interest.

The downside of this is that employment dynamism, the ability and desire to change employers at will and for companies to shed personnel is significantly reduced compared to Western countries. In fact when the Lost Decade first set in there were lots of stories out of Japan of salarymen who had lost their job, were living in the street and were seeing the reality of a curtailed government welfare state in Japan.

The public welfare choices that in Western countries are most frequently delivered through the means of the state are more often delivered through the employer in Japan.

A state-provided welfare system does have the advantage of not picking winners and losers and letting everyone make a free choice about employment that best suits their abilities. Employers in the West are not burdened with unpooled welfare costs and the choice of not just letting people go but ruining their lives as well.

A broad welfare states does encourage employment dynamism and facilities risk-taking by entrepreneurs. It also helps the formation of small businesses. The implication that the "market" is somehow the best choice in this case has no support in fact, in my opinion.

[Jacques René Giguère]

Determinant: In the 90's early 00's I was involved in a telecom start-up. An american friend was in the habit of lectiring us on the evils of socialist health care. Till he realised we didn't have to manage a health plan, could hire whomever we needed without worryinig about "preexisting conditions", could hire students part-time without disentangling who is responsible (university or us). Employer-provided welafare is like most other forms of vertical integration, that is a sign of undevelopped markets or bad-quality suppliers.

[Jacques René Giguère]

Determiniant: in the 90's early 00's, I was involved in a telecom start-up. An Ammerican friend was constantly lecturing us about the evils of socialised health care. Til he realised that we didn't lost our time managing a health plan, could hire whoever we wanted without wprrying about "preexisting conditions" etc.
Employer -provided welfare is like other forms of vertical integratio0n, that is a market failure, mostly a sign of bad qulity suppliers.

[Determinant]

This is why I want drugs taken in to the public health system in Canada. I have strongly suspected that several employers have taken a hard second look at me when they found out I had Type I Diabetes (injections and frequent testing). I never mention it interviews.

This is also why I say that the business of business is to make money, not be a nanny or operate a shadow welfare state. I have long been aware that employment can end but my Diabetes doesn't. I also know it shouldn't really be an employer's business.

I have friends who have owned their own businesses who agree with everything I just said and it looks like you have come to the same opinion through the employer's perspective, Jacques.

It seems to me that many economists confuse form with function. The fact that an employer provides welfare arrangement over and above straight pay does not mean that no redistribution is taking place, that these arrangements are not a burden to the public (who else pays but the customers?) and that they have negative impacts of their own through the encouragement of dependence and over time a reduction in worker desire and ability to change employment.

Why is it so controversial to have a welfare state to take care of costs that significantly impact employment yet have no direct cost relationship to employment income like health care? The welfare state the canvas against which we paint our market economy.

[Andrew jackson]

Stephen, I thinks there is an important distinction to be made between market income inequality at the family level and the inequality of the wage structure at the individual level. The latter still varies a great deal between countries ... Lots of data on the OECD site. I entirely take your point re the importance of transfers as opposed to taxes, but we should also be looking at ways to equalize wage income to a greater degree, especially as between the bottom and the middle, and the middle and the very top. I think your differences between countries would be greater using the d9 to d1 ratio as opposed to ginis. The odd make that readily available as well.