A short history of the Canadian one per cent

Any sensible policy designed to address the concentration of income at the top of the Canadian income distribution over the past thirty years or so has to be based on a working model that explains how and why it happened in the first place. And we still don't have one. We still don't even have a good handle on the basic facts that this model should explain.

I think that one basic fact that such a model should explain is how the composition of the top earners group has evolved over time. So I'm going to present here some numbers from the Public Use Microdata Files for the censuses between 1981 and 2006.

There are severe limitations with what can be done with the PUMFs. The first challenge is top-coding: incomes above a certain threshold ($100,000 in the first few years, $200,000 later on) are truncated at the threshold. Happily, this threshold is below the 99th percentile, but I can't look at higher percentiles or retrieve average incomes. All I can get are the number of people in the top one per cent.

First up are occupations. The challenge here is that Statistics Canada changed its occupational classifications about a kazillion times between the 1981 and the 2006 censuses, so I was only able to track two categories that are important to what's going on in the one per cent: managment occupations and health professionals. I would have liked to track occupations such as lawyers and finance professionals, but I couldn't.

One of the stories we tell about the one per cent involves corporate executives: for one reason or another, managers have been able to increase their bargaining power and leverage higher salaries. But it turns out that the managers account for only about 35% of the top one per cent, and this share has remained fairly constant since 1981:

(The National Household Survey numbers are from here. Make of them what you will.) 

If the surge in incomes at the top were due to increased salaries for executives, I would have expected that managers would have occupied an increasing share of the one per cent. That doesn't seem to have happened.

The other dimension where you can break down high earners is the industry where they work. I was able to track four important industries here:

Apart from the drop in the manufacturing sector, I can't pick out any particular trend. For example, if the surge in top incomes were due to an increasing 'financialisation' of the economy - in which finance professionals find themselves in a position to skim off a small fraction of ever-increasing transactions - then you'd expect that people in the FIRA sector would take up more place in the top one per cent. That hasn't happened, either: an increase from 10% to 12% is a week reed.

Based on a first rough pass with PUMF data, I think it is reasonable to conclude that the one per cent today more or less resembles what the one per cent looked like before the surge in top incomes. They just have higher salaries.

 

Comments

[Frances Woolley]

How do you define top 1%, that is, top 1% of individuals (including children), families, earners?

[JKH]

Apart from distribution:

Public disclosure of top level corporate compensation (5 highest paid officers with policy responsibility) starting in the early 90’s must have been a big factor in the change in levels.

And the widespread use of stock option and restricted share compensation which started around the same time.

Still, the threshold income level for the top 1 per cent is about half or even less of what it is in the US. I’d like to see somebody do a study of how that happened.

Distribution:

People tend to be surprised by the financial industry share (lower than expected).

[autocorrelation]

A thought related to the management/health professionals split - has the data been adjusted to take personal businesses into account (a la Wolfson Brooks Veall)?

It is much more common for medical professionals to have a personal professional corporation for tax purposes, so one might expect to see less reported income relative to other high remuneration occupations, unless CCPCs are taken into account.

[Stephen Gordon]

Frances: I used the thresholds in the Cansim table.

The Cansim numbers are based on all tax files, so I thought they'd be more reliable than the subsample in the PUMF. I did check, and the proportion of PUMF incomes above the Cansim thresholds were all close enough to 1% that I'm pretty sure I'm talking about the same people.

[Two Hats]

Both graphs would benefit from showing an "other" category, where [other = 100%-sum of identified groups]. That effectively creates a 3rd/5th category for you.
The data is not what I would have expected. But then, my expectations are not based on any special insight or information. I'm looking forward to the continued discussion. Thanks.

[Leo]

Krugman often differentiates between the Top 1% and the Top 0.1% for the US data. If it's available, does the Top 0.1% data for Canada have a useful pattern?

[Dylan Gowans]

"If the surge in incomes at the top were due to increased salaries for executives, I would have expected that managers would have occupied an increasing share of the one per cent. That doesn't seem to have happened."

I don't accept this argument. If I understand the data correctly, what the first graph shows is the percentage of persons within the 1% group who are managers and health professionals. If the percentage of income flowing to managers increases, this should not inflate the percentage of *persons* who are managers (at least not a priori), although it may inflate the percentage of top 1% income going to managers.

One could, I suppose, make the argument based on simple labour market supply and demand that an increase in managerial pay would induce a greater labour supply for managerial positions. However, it seems likely that managerial positions are demand constrained, being given by the number of existing firms. Even if an increase in managerial supply lead to a decrease in managerial compensation, we would still assume that this would not induce a greater demand for managers. No matter how cheaply one can acquire a captain for a ship, there is no reason to have more than one captain. Instead, the number of captains should be given by the number of ships.

[Colin Percival]

Dylan, the argument is that if executive salaries are increasing, then executives who were not in the top 1% previously will move into the top 1% thanks to their increased salaries. You're right that the statistics ruled out do not exclude the possibility that high-paid executives are increasing their salaries while lower-paid executives' salaries remain unchanged.

[Stephen Gordon]

Possibly, but management isn't just the CEO. There are lots of managers below the 99th percentile, and one might have expected that some of them would have moved above the threshold.

Unfortunately, because of the top-coding, I can't check to see if managers' salaries increased more than those in the one per cent.

[eta: I see I cross-posted with Colin]

[Stephen Gordon]

Leo: No, I can't. The data I have are top-coded, and the truncation is below the 99.9 percentile.

[Jacques René Giguère]

Discussing the 1% in Canada is like discussing it in Montana. It should take North America as a whole, not some uninteresting norther satrapy. There is no inequality among the village peasants. Only if you include the castle in Wall Street.

[Allan Pollock]

just a thought starter:

Would it be informative to look at how income is distributed across occupations rather than how the top 1% of earners is distributed across sectors? Ie: has the share of income earned in management occupations increased significantly?

[Stephen Gordon]

That's in the first chart, isn't it?

[hix]

I dont understand. Progressive taxes work just fine no matter what the cause is. Besides, i dont trust the data at all in this case. A recent estimate put the share of capital hidden in tax haven at 8%. There should also be opportunities to lower ones "taxable income" compared to actual income in legal ways that are rather unequal distributed.

[mick marrs]

The age old story of greed and envy. These statistics and data mask the dynamics of people entering and leaving the "one percent". Its just a convenient story to justify class struggle. If this was a really important problem we would have set up a sysyem to track people's incomes over their lifetimes and incomes and wealth across generations.