Are the rich capitalist? Are capitalists rich?

A short answer to both of these questions is "typically not," and an even shorter answer is "no". A longer answer is below the fold.

Many discussions about income inequality suppose that capital's share of income goes to the rich, while labour's share goes to the not-rich. If this is the case, then shifting income from capital to labour will reduce inequality, and there are any number of policy initiatives aimed at bringing about this transfer.  

But just how solidly-rooted is this assumption in the evidence? One of the more interesting results to come out of the recent empirical literature on high income earners is that wages are the biggest and fastest-growing source of income for the very rich. As I noted in this earlier post on the Saez-Veall AER paper,

[i]n 1946, capital holdings accounted for 53% of the income of those in the top 0.01%, and 27% was wage income; the other 20% was entrepreneurial income from self-employment. In 2000, the very rich generated only 25% of their income from capital, and 74% was in the form of wage income; entrepreneurs who work for themselves have almost completely disappeared from the ranks of the very rich.

If wages are the source of growing inequality, then shifting income from capital to labour isn't going to do much to slow this trend. But even if being very rich doesn't mean that most of your income comes from capital, that doesn't mean the reverse is true. If capital income is concentrated among high earners, then it could still be argued that increasing labour's share of income will reduce inequality.

And this is the question that I'm going to look at in this post. The data are from the publicly-available tax file data for 2008 here on the web site of the Canada Revenue Agency.

It turns out that the distinction between what is wage income and what is investment income is somewhat blurry. My guiding principle was that wage income is generated by, well, work: people had to actually do something in order to get paid. Investment income, on the other had, is passive: income is derived from past actions (i.e., the acquisition of a productive asset) and generated without the direct involvement of the capitalist.

The problem is with to do with with the income of the self-employed. It isn't passive, but since they own their own capital, part of their income should be considered a return on past investment. The tax data don't make the distinction, so I can't, either. In the end, I put them in a separate category.

Here is the classification I finally went with:

Wages (66.9%)	Employment income (64.6%)
	Commissions from employment (1.2%)
	Other employment income (1.1%)
Investment income (19.0%)	CPP/QPP benefits (3.5%)
	Other pensions or superannuation (6.5%)
	Dividends (4.1%)
	Interest and other investment income (2.2%)
	Registered Disability Savings Plan income (0.0%)
	Net rental income (0.3%)
	Capital gains (1.4%)
	Registered Retirement Savings Plan income (1.1%)
Business income (10.0%)	Business income (2.2%)
	Professional income (1.9%)
	Commission income (0.3%)
	Farming income (0.2%)
	Fishing income (0.0%)
	Tax exempt income (2.1%)
	Other income (3.1%)

 

A couple of important items in the CRA table don't appear in that list:

• Old Age Security pension payments (2.5%). This isn't wage income, but neither is it a return on asset holdings.
• Employment Insurance (1.4%). I was going to put this in the investment income category, because insurance is an asset that pays off in certain states of the world - in this case, losing your job. I still think that this is the right way to think of it, but arguing the point would be a distraction.

Readers may disagree with some of these judgment calls, but it turns out that the results below are fairly robust. Classifying these last two items as labour or investment income doesn't change the qualitative conclusions.

For each of the 18 income categories, I added up the shares of the various types of income that went to those whose total income put them in that class, and took the cumulative sums. The result is the following graph:

Here's how to read it. The horizontal axis is the total income reported by tax filers, with a log scale. The vertical scale is the cumulative share. Moving horizontally along a given percentage, we can see that (say) 30% of the tax files reported total incomes under $15,000, 30% of business income went to those earning $25,000 or less, 30% of investment income went to those earning less than $40,000, and 30% of wage income went to those earning less than $50,000. Moving vertically, we see that those earning (say) $30,000 or less accounted for 50% of all tax files, 35% of business income, 20% of investment income and about 12% of wage income.

At the top end, the numbers are consistent with my priors. Those earning more than $250,000 - 0.75% of filers - account for 9% of all wage income and 17% of all investment income. The concentration of investment income among high earners is greater than the concentration of wage income. In the graph, this result shows up as the red line being below the blue line at the $250,000 income threshold.

But this ordering doesn't hold throughout the income distribution: at around $80,000, the red and blue lines cross, so the share of capital income going to tax filers in the lower part of the income distribution is greater than its share of wage income. Half of investment income goes to those earning $55,000 or less, but this group receives only 44% of total wage income. It's hard to see why lower-income Canadians would summon much enthusiasm for the idea of shifting income away from capital to labour: their share of the capital pie is larger than their share of the labour pie.

Lower-income filers receive an even larger share of self-employment income. But since this is a relatively small category, the results don't change even if we ignore their investment income and classify the earnings of the self-employed as wages:

 

Not surprisingly, pensions play a crucial role here. If pensions are excluded from investment income, then the resulting cumulative distribution lies below the blue wage line at all income levels.

So the answer to the questions in the title is "typically not". The empirical foundations for the traditional equation capitalist = rich are surprisingly weak:

• The rich are typically not capitalists. To a large and increasing extent, high earners are generating their income from their labour, not their capital.
• Capitalists are typically not rich. Although high earners do receive a disproportionate share of capital income, this pattern holds only for the top 10% of tax filers. Capital income is more concentrated among middle- and lower-income groups than is wage income.

The capitalist-worker classification may have once been a useful way of summarising the distinction between the rich and the not-rich that was in the data. But these days, it's an unhelpful distraction.

Comments

[Frances Woolley]

Stephen - terrific post! Thanks so much for doing this.

A couple of caveats to add:

The mid-income people with significant investment income are typically retired. When they were working, they would have been in the top half of the distribution.

There's a minority of people who spend all of their lives either right at the top or right at the bottom of the income distribution. More typically, most of us start off towards the lower middle when we're children of younger parents, gradually move up the distribution as our parents get more established in the labour market, move down again when we're students/young adults, move up again as we get better/more secure jobs, and then down again as we start to slow down and cut back on our labour force participation.

So "rich" and "poor" have different meanings in a life cycle v. a point-in-time context.

It's also important to remember that this is individual, not household data (even though it's about the best that's publicly available). There are more opportunities for splitting business and retirement income ("income sprinkling" as tax advisers put it) - and that will tend to shift business and retirement income down the income distribution.

[Luis Enrique]

aren't the rich able to choose how they report their income - as capital or wages? If taxes on wages were higher relative to taxes on capital income, we'd still be concluding the rich are capitalists I suspect, but "who the rich are" is not a function of the prevailing tax regime, unlike the choice of how to recognize income for tax purposes.

[marcel]

Two questions, that may be less relevant for Canada than the US.

1) What happens to the results if you put into a separate category those who work in the financial sector?

2) More generally, what happens to the results if you put into a separate category those for whom bonuses are a major part of compensation? Definition of major is important, so it would be interesting to play around with this, but possibilities would be at least 50%, at least 75% and at least 90%.

Another way of cutting the data that might be both more useful and impossible given what it contains is to distinguish rich and poor not by income but by wealth.

I recall seeing on more than one occasion a definition of the boundary that goes something like this: if you are dependent for your income on your occupation, than you are not rich; or if your are dependent for your standard of living on income earned from your occupation, than you are not rich. This definition may beg the question, however.

[Frances Woolley]

Luis, generally the taxation of investment income is more favourable than the tax treatment of wage income - which is why executives like stock options. So the effect that you're identifying would tend to push investment income down the income distribution.

A more interesting complication is the treatment of business income. A lot of self-employed professionals (e.g. doctors) use their businesses as a retirement savings plan, that is, all of their fee income flows into their business. They then withdraw from the business as much as is needed for current consumption, and invest the rest.

What this means is that business income is, for the self-employed, as mix of labour and investment income.

[Wendy]

Fascinating. I'm wondering how this might translate into public policy. Would an NDP party, or other group traditionally oriented toward the needs of labour need to re-think their idea of taxing corporate profits if their goal is a "more equitable society" and instead tax the higher wage earners more? or the lower wage earners less?

[K]

Frances: "generally the taxation of investment income is more favourable than the tax treatment of wage income"

Yes but "capitalists" (business owners) consider wage income vs corporate tax *plus* capital gains.  Those ought to be pretty similar.  I think Luis has a good point.

"executives like stock options"

Income from option exercise is regular income.  Subsequent gains in the stock are capital gains, but it's no different from the exec getting paid the exercise value and then investing in the stock.  Also, execs who aren't owners don't have to weigh corporate taxes in their compensation.

[Patrick]

Employee stock options get a major tax break in Canada: you're only taxed on half the difference between the option strike price and the stock price when the option is exercised. So if you're an exec your incentives have your golf buddies on the compensation committee award you options for a "pump and dump" are pretty strong.

[Patrick]

Above, I mean to write the option strike price. Not the market price of the option.

[Marc]

I think this post totally misses the point -- being rich is about wealth not income. You're are measuring the flows -- some of which are in the form of traditional returns to capital ownership, some in the form of salaries (Galen Weston and Ed Rogers get paid exceptionally well for running the family business, as well as owning it) -- not the stocks.

And key point missing here is the stunning inequality of wealth distribution, as in the bottom 50% having about 6% of total net worth, and the top 10% having more than half of the total.

So not all capitalists, ie small business owners, are rich. But yes, the rich are almost by definition capitalists because you need to have big assets to be rich, which means possession of land and in particular equities.

[Nick Rowe]

This is very good and very important, Stephen.

My prior is always: wage income goes to the young; capital income goes to the old. Capitalists are retired workers! (Some weren't too happy with me on PEF when I suggested that!)

For example, if interest rates stay low in future, this will affect the distribution of income. But it may harm the poor, rather than the rich.

Minor quibble: pension income is partly/mostly capital income, but partly a return of principal.

[Frances Woolley]

K: Yes but "capitalists" (business owners) consider wage income vs corporate tax *plus* capital gains. Those ought to be pretty similar. I think Luis has a good point.

I was going to write a long reply describing how the corporate income tax is to a large extent a withholding tax - it's collected from firms, and then refunded to taxpayers in the form of a dividend tax credit once corporations distribute their income to shareholders.

But have to go into the office and do other worthwhile things.

[Nick Rowe]

Marc: "I think this post totally misses the point -- being rich is about wealth not income."

The value of an asset is the present value of the flow of income it generates. Why exclude the present value of labour income from wealth?

[Jim Rootham]

The real problem with this post is that it assumes no one successfully avoids (or evades) taxes.
That is why the wealth measure is better if you want to see who is rich.

The other problem with looking at this question has been pointed out earlier. We don't know who is getting the big salaries. I have to believe some of them are captured from returns to capital in large corporations.

[Marc]

Good point, Nick, but then let's count all income and parse out some differences as per the argument Stephen advances. I think the way rich is being defined (high income rather than high wealth) is problematic in several ways.

Counted above in wages is labour income capitalists pay to themselves (eg Weston or Rogers) in addition to what they get from dividends and capital gains. This assumption needs to be treated differently as it drives the conclusion.

Also, what is measured, I think, is realized capital gains not accrued capital gains; the latter is more consistent with wealth. Another big piece is the use of company assets for personal benefit; employer-provided benefits are a form of income not usually measured or counted as income (and certainly not taxed). Why buy the jet with after tax personal income when you can use the company's (i.e. your's). Finally, inheritances and gifts would round out the list in terms of including all income.

That gives you "broad income", which might be the flow analogue to wealth that would be useful. But given what's constructed above, it is better to just think of the rich in wealth rather than income terms.

[Kevin Milligan]

Marc: Wealth inequality without conditioning on age is pretty useless. Even more than what Frances notes for income, there are huge life-cycle patterns to wealth accumulation. An economy with identical (but different aged) agents would exhibit a lot of wealth inequality.

[Kevin Milligan]

I think it is great that Stephen (and Armine last week, all building on Mike Veall's work) are moving the debate toward the current reality that high-income people are wage-slaves.

This is really very important when you want to consider what reaction you might have (if any) to rising inequality.

For example, if the high income guys are mostly capitalists, then I would be quite worried that their investment dollars would be fairly elastic wrt changes in tax rates. On the other hand, if they are hyper-competitive CEOs and VPs, I would be less concerned that they would cut back their labour supply in a response to higher tax rates--in fact I might guess the response would be close to zero.

[Stephen Gordon]

Thanks for the various comments. In another time, I might have tried to make a paper out of this (I may yet), but it's fun to be able to get the main result out quickly for feedback.

Frances: Yes, I agree that a life-cycle perspective is how we should be looking at many of these issues. For example, many high earners are artists and athletes who will do extremely well for a couple of years, and then try to live off whatever they have managed to save for the rest of their lives. Many don't manage their short-lived good fortune very well.

Wendy: I don't know. But yes, I would hope that the NDP would focus its attention on income, and not exclusively on wages. The 'profits not people' line is particularly unhelpful in light of these numbers.

Marc: You're focusing on a small fraction of wealth. Available estimates of total wealth in Canada ( [1], [2]) suggest that human capital is about four times as big as non-human capital.

Kevin: This is really the $64k question: who *are* these high earners?

[Jim Rootham]

We have enough data to know that they are not athletes. I made a comment in an earlier thread that worked out that there are only ~300 highly paid athletes in the country.

Especially knowing what I do about the lives of artists, I would expect those numbers to be even smaller.

[Patrick]

"who *are* these high earners?"

CEO's and COO's in companies with 500+ employees? And a few from medium size firms.

Read'em and weep:

Executive salary survey

Assuming these numbers more or less hold for the whole country, when I look at Industry Canada stats for 2009, I see 550 goods producing firms that are over 500+ employees, and 5,127 that are 100-499. And in service producing sector there are 14808 100-499 and 2435 at 500+ employees. So that could be 22K * 2 (one CEO, one COO) or ~44K people earning *at least* in the 200K range, with many in the 7 digits. Throw in a few thousand people from the financial sector in Toronto and is that close to the numbers we'd need to explain who they are?

[K]

From my own limited sample... There seem so be an awful lot of senior lawyers, executives, brokers, traders and investment bankers, i.e. mostly wage slaves, among those whose wealth would appear to be of the order of 10 million or so. The range above that (private jets, golf courses, etc) seems to consist of senior financial markets execs, CEOs of mostly public companies, but also a significant concentration of owners of private Canadian firms. There can't be any serious doubt that of those at the very top, there is a very high concentration of capitalists.

And I checked the effective top marginal tax rates for corp tax + dividends vs wages. From what I can tell in Ontario in 2010 it works out to about 46% for wages and 48% for dividends. I may have missed some things, but they seem to be very close. So it wouldn't be very surprising if even the wealthiest capitalists earn a lot of wages.

[Determinant]

"Minor quibble: pension income is partly/mostly capital income, but partly a return of principal."

Even more minor quibble: pensions are life annuities which pay out over life the annuitant/pensioner. The capital return period is tied to the life expectancy of the pensioner/annuitant instead of being fixed.

You calculations must proceed from there. :devilishsmile:

[hosertohoosier]

If you look at the bigger components of investment, it isn't clear to me that this should be a surprising result. Consider the following items:
-CPP/QPP
-pensions
-RRSP's

Since the CPP and RRSP's max out at a set number we should expect them to be lower for wealthier people, as a percentage of total income. With pensions we should see low numbers for the poorest (generally part-time workers), and we do. However, there are plenty of high-income jobs that do not give workers pensions, either (eg. in management). Also, other high-income people may be paid more precisely because a firm wants to avoid paying benefits (eg. consultants).

Should the NDP push for a capital gains tax cut? It isn't clear. Capital gains represented a small proportion of investment income, and it is possible that its distribution still leans toward the wealthy.

What is more interesting is the dramatic shift in the richest .01%. Perhaps the impact of a shift from firms dominated by entrepreneur-owners to well-paid managers.

[Too Much Fed]

I think you need to look at "retirement inequality" and "retirement date inequality" even if it extends beyond the current generation.

[Frances Woolley]

hosertohoosier: "there are plenty of high-income jobs that do not give workers pensions, either (eg. in management)."

Poor example. E.g. the professions, such as law. Anyone who is self-employed or in a partnership. Managers are, if anything, more likely than the average worker to get pensions.

[Michael Smart]

Very useful analysis, which I will no doubt steal for teaching.

In some respects this is another post in Steven's long campaign to help the Left learn to love the consumption tax. (Good luck with that!) But I don't think these graphs really make the case that lower capital taxes are progressive. In particular:

- Capital income of the poor is in the form of RRSP/RPPs which is already tax exempt, so capital income taxes remain progressive. And raising contribution limits is regressive at least in the short run.
- Something like 50% of RRSP/RPP income as reported in tax data is return of principal, i.e. deferred labour income. (Nick Rowe already made this point.)
- Though it's not directly at issue here, I would point out that the corporate tax would be a tax on excess profits rather capital, if it were better designed (more like a cash flow tax). Excess profits are not going to these pensioners, they're going to Galen Weston et al.

For all these reasons I think there is room for capital and corporate taxes in an efficient, equitable tax system.

[Too Much Fed]

I also think you need to look at whether upper management is paid (overpaid) to exploit an oversupplied labor market to keep as much as possible for themselves (salary and stock options) and the capital owners.

[Deputy Dawg]

Stephen: I agree with other commentors that this post has methodological issues. Your data set excludes unrealized capital gains, the largest part of investment returns. And on what basis can you include rental income, but ignore real estate appreciation and building depreciation?

(For that matter, CPP/QPP is still largely pay-as-you-go, that's not true investment income)

A bad data set will lead to bad policy analysis. I think you'd be better off deleting the post. No information is better than misinformation.

Frances: I agree that the Corporate income tax is mostly a witholding tax -- but then why do some economists argue that it changes incentives for corporations? Since the CIT and PIT systems are integrated (roughly), shouldn't corporations be indifferent to higher corporate income taxes, as shareholders are compensated by a higher dividend tax credit. Something doesn't add up.

[Stephen Gordon]

Your data set excludes unrealized capital gains, the largest part of investment returns.

I stopped paying attention after that bit. Try again.

[Bob Smith]

Deputy Dog

The corporate income tax system is, generally, integrated IF (big IF) you're a taxable Canadian resident who can claim a dividend tax credit. Where it is a problem is if your shareholders are either (i) not Canadian residents or (ii) a Canadian resident that is not taxable (pensions, RRSPs, etc.). For those groups, corporate income tax isn't a withholding tax (since they wouldn't pay Canadian income tax directly) but a real cost. The big push for conversions to income trusts in the late 1990's and early 2000s was largely driven by demand from these groups of investors to avoid corporate level taxation. I don't have the numbers off hand, but I'd bet those two groups (non-residents and non-taxables) make up a hefty chunk of the Canadian capital market.

[Bob Smith]

"I also think you need to look at whether upper management is paid (overpaid) to exploit an oversupplied labor market to keep as much as possible for themselves (salary and stock options) and the capital owners."

If anyone's been exploited by upper management in recent years it has been the capital owners. Indeed, there's a school of thought which suggests that many of the problems with corporate governance in the US (amongst other countries) has come about because too many senior corporate officers end up enriching themselves at the expense of their investors and that corporate governance tools are not effective at controlling this sort of abuse.

[Bob Smith]

Following up on that last point, I wonder if we aren't seeing an odd reverse-Marxist scenario, where workers (managers, skilled athletes, etc.) control the means of the production and exploit the capitalists. In an era where human capital is often of greater significant than physical capital, why couldn't that happen?

Capitalists of the world unite, you have nothing to lose but your chains!

[Erin Weir]

Stephen, did you just take “Taxable Capital Gains” from the Canada Revenue Agency table and label them as “Capital Gains”?

$14 billion / $995 billion = 1.4%

Since the Canada Revenue Agency’s inclusion rate for capital gains is 50%, realized capital gains are actually double “Taxable Capital Gains”:

$28 billion / $1,009 billion = 2.8%

(Of course, even that figure does not reflect unrealized capital gains.)

[Stephen Gordon]

Here's a graph with capital gains removed and with what I've labeled "full" capital gains, which is simply two times the number in the CRA tables. Using 'full' capital gains shifts the red line down a bit, but the main conclusion doesn't change. Even if self-employed income is classified as wage income.

[Andrew]

Do stock options count as wages, or as capital income diverted from outside shareholders to corporate insiders?

Interesting post to wrap my head around though. I suspect the very affluent do work very hard, but the rewards for effort still are hugely excessive.

[RSJ]

"Do stock options count as wages, or as capital income diverted from outside shareholders to corporate insiders?"

Yes. The granting of the option is considered wage income.

Effectively, when top earners (e.g. top executives, management) grant themselves stock, then their wages are being funded by capital losses of everyone else.

This creates a distortion in the income tax data, as it represents a shift from the investment income of non-top earners towards the wage income of top earners. I think you need to interpret the charts in the context of this transfer.

It would also be more helpful if the probability density, rather than the distribution, was shown.

[Erin Weir]

Stephen, thanks for the revised graphs, but I think that further revisions are warranted.

[Marc]

I also forgot to mention that the income measured above should include the imputed rental of owner occupied housing, and this is pretty important in wealth as it is the single largest asset of most households. One could argue that buying a home is a form of capitalism ...

But I actually wanted to check in with Stephen on his reference to the wealth value of human capital. It is a really interesting point but looks like the estimates made count lifetime earnings as their measure. Like labour productivity, this can be an interesting measure in one dimension.

But to compare those estimates with measures of physical capital is problematic since income is a function of physical capital. Auto workers can make big bucks because they work in such capital-intensive plants. Even an economist writing a blog post such as this one: how much is the author's contribution, how much the software and hardware that enabled him to produce it?

And then there is all of that public/infrastructure capital, social capital, natural capital, etc -- the notion that they are somehow comparable in dollar terms is flawed. It is capital controversy enough just to look at physical capital (remember Cambridge?). Statements like "human capital is four times larger than physical capital" need to dismissed entirely, even though I take the point you were getting at.

[Stephen Gordon]

Houses are an asset, and housing services are its return. Housing is pretty widely held, so I don't think including it would materially affect the conclusion.

And calculating the value of assets is indeed fiendishly difficult, which is why I stuck to income flows in the OP. But the fact that wage income is about 4 times as big as investment income is going to end up with an estimate for the value of human capital that is of a similar order of magnitude greater than estimates for nonhuman capital.

[Barry]

"My guiding principle was that wage income is generated by, well, work: people had to actually do something in order to get paid. Investment income, on the other had, is passive: income is derived from past actions (i.e., the acquisition of a productive asset) and generated without the direct involvement of the capitalist."

I think that you have a very old-fashioned idea of the world, where (example) the President of US Steel serves at the whim of the billionaire who owns US Steel. I would not see the top several people in a corporation as being wage earners, but rather gathering income from controlling the assets.

As for people like senior laywers (I'm asssuming it works similarly as in the USA), they are living quite a bit off of hiring others to do the work, and then padding the bill.

[Stephen Gordon]

That's sort of the point I was trying to make. In many important dimensions, the old capitalist-worker classification has lost much of its relevance.