The hollowing-out of the middle class: It's a guy thing

I've been having a hard time getting my head around the 'hollowing out of the middle class' theme that were seeing so much of. What, exactly, does that mean? And is it actually happening? In a Maclean's post a couple of months ago, I tried looking at it in terms of the proportion of the income distribution that was within a certain distance of the median, and this is what I got:

If the middle class were hollowing out, I would have expected to see a trend to smaller proportions of incomes 'close' to the median. But I don't see that trend in the data.

Soon after that post, Kevin Milligan brought this paper to my attention, which includes these remarkable graphs:

The lower right-hand graphs tell a pretty stark tale: for both sexes in the 25 years between 1981 and 2006, the higher you were in the income distribution, the faster your wages were rising. For men, the picture is particularly striking: the only income growth occurred for those above the median. The dynamics are pretty clear: these income distributions were spreading out.

So where does that first graph - which is also based on the same census Public Use Micro Datafiles - come from? The answer seems to lie in the relative positions of the bottom-right profiles: for a given percentile, the increase in women's wages was systematically higher than for men. Basically, the men who drifted below the median were replaced by women whose incomes were initially lower, but had grown faster.

As I typed that last sentence, it occurred to me that I should go back and check the gender composition of the definitions of the middle class in that first graph. Sure enough, the female share of the middle class (as defined in the graphs) has increased over time. What surprised me is that the female share of the middle class was always greater than 50%. 

I'm not sure what to make of this. There is definitely something going very badly wrong with low-skilled men, but I don't think it's helpful to think of it as the hollowing-out of the middle class.

Comments

[rsj]

Stephen,

Can you change "median" to "mean" and provide the data?

No one is saying that laws of arithmetic are changing, or that somehow we have a smaller "middle", in terms of rank ordering. We have a smaller "middle", in the sense of people earning the average (mean) income is a smaller and more elite group, and it is harder to get into this group.

Here is the relevant data:

"Median earnings of Canadians employed on a full-time basis for a full year changed little during the past quarter century, edging up from $41,348 in 1980 to $41,401 in 2005 (in 2005 constant dollars)."
http://www12.statcan.gc.ca/census-recensement/2006/as-sa/97-563/p1-eng.cfm

So effectively zero growth in median income over a 25 year period, when real GDP per capita increased by about 150% from 27,999 to 42,518. (calcwww.indexmundi.com/facts/canada/gdp-per-capita).

[Patrick]

Off the cuff (so maybe it doesn't make sense) this is the story that comes to my mind: robots/computers are to men what tractors were to horses. And the guys (literally) who now manage the robots are appropriating (to varying degrees) the robots' marginal product to themselves as wages and bonuses, rather than paying it to shareholders.

[Stephen Gordon]

rsj: AFAICT, most analysts use the median to define the 'middle' of the income distribution. In 2006, 74% of reported incomes were below the average. I wouldn't call that the 'middle'.

[Jeffrey Johnson]

I think the term "hollowing out" means that the middle class are not sharing equally in income and productivity growth. In the US wages are lagging expenses, particularly with respect to the very important categories of housing costs, education, and health care. I don't know what the numbers show, but my experience is that people feel like they have less disposable income, less to save, and prior to the recession felt more pressure to borrow in order to maintain their lifestyle. I think these are the factors ordinary people feel are "hollowing out" the middle class, which could be interpreted as meaning that the people remain middle class in name only by median position, but their purchasing power has faltered, and they are being left behind by the staggering growth in wealth and income of a small percentage of people at the top, which explains why median by income would not reveal the phenomenon people are complaining about. It's not about proportions of numbers of people, it's about proportions of purchasing power, lifestyle, upward mobility, and economic security.

[Chris S]

To try and clarify - if every single person's wage, from 1981 to 2006, had increased by exactly the same percentage, would these graphs then show a horizontal line at some constant - but non-zero - level?

[Jeff Fisher]

The percentage of population with income within a certain distance of the median would show a certain kind of shape change in the income curve. If incomes were clustered around the median and were becoming less clustered around the median, or more clustered. However I don't think "incomes clustered around the median" is a good definition of "middle class". A country with mostly poor people of very low income could easily show tight clustering around a low median. A country where the bottom 90% had stagnant or decreasing incomes, as long as the 25th to 75th percentiles were moving largely together would also show no change.

That last sentence strikes me as a pretty good description of reality (in Canada or the US).

[Stephen Gordon]

Chris S - yes. But clearly, the reverse is not true.

[Bob Smith]

"[M]y experience is that people feel like they have less disposable income, less to save, and prior to the recession felt more pressure to borrow in order to maintain their lifestyle."

I think that's true that people FEEL that way. On the other hand, looked at from a different perspective, one would be hard-pressed to argue that lifestyles have stayed the same over the past 30 years. It's been a few years, but my recollection is that by many measures of material well-being (housing size, number of bed rooms, number of cars, colour TV's per capita, air conditioning, whatever) lifestyles have materially improved over the past 30 years. (Indeed, at one point, the lifestyles of the American poor, by those measures, were not necessarily all that different from the American middle-class of an earlier generation - that was pre-recession though). I'd have a lot more money in my pocket at the end of the month if I lived the way my parents or grandparents did. So, to the extent the people feel that they can't maintain their lifestyles, is that a testament to the dying middle-class or to lifestyle inflation? (Maybe people can't have it all?)

The other concern with wage data is that it's heavily dependent on the accuracy of inflation numbers. To the extent that rich and poor face different prices (and arguably they do), you'll tend to overstate the income of the rich and understate the income of the poor. There have been somewhat credible arguments that the impact of the "Wal-Mart" effect has been to drive down prices at the bottom end of the consumer market. That doesn't affect high-end retailers whose customers wouldn't be caught dead at Wal-Mart, but it has decimated the traditional "mid-market" retailers who used to appeal to everyone (Eatons?). You could probably tell a similar story with housing (yes, a fixer-upper in downtown Toronto can run you $1.2 million, but that's not where Mr. Median Income is buying a house). The other problem with inflation numbers is that they do a lousy job of taking into account the impact of technological change. A top of the line computer in 1984 cost more or less the same as a top of the line computer today (say $3,000), but a $200 tablet today can do infinitely more. I suppose that's always a problem, but I'd suggest it has become aggravated by the digital revolution of the 1970's and 80's.

[Jacques René Giguère]

For anyone who lived through the '50's and 60's, the sense of stagnation is palpable. I am richer now than my family was then. But we had a sense of progress. And that has an impact on what the numbers mean.
And the fact that median up to the 90% are currently being cleansed out of every major city downtown in the western world adds to the unease.
I am in the 90th percentile of income. There is essentially nothing in the G&M or even La Presse real estate ads that I can buy. They have turned into a kind of porn: they excite desire without the means to satisfy it.

[Jacques René Giguère]

Forgot the link

http://krugman.blogs.nytimes.com/2013/10/28/sheiks-and-princelings-and-beeznessmen-oh-my/

[Jeffrey Johnson]

@Bob Smith,
You make a good point about technological advances improving quality of life. Mass produced consumer electronic goods have improved vastly and prices for new features continue to decrease.

But two points on that:

1. In a way if you compare an individual's change in lifestyle due to increasing power and decreasing price of flat widescreen HD TV, smart phones, and computing technology, it's in a way akin to comparing nominal wages changing over time. If we compare the degree of expanding luxuries of the top earners over that same time period the comparison gets closer to comparing "real" inflation adjusted values, where in this case tech innovation serves as the analog to inflation. Heritage foundation had some papers a few years back making an argument like this, that American poor aren't really poor because they have refrigerators and TVs. But see the next point.

2. The crucial areas where this argument about tech progress fails are in housing, education, and health care. Not so much in Canada perhaps, but very much so in the US. The prices have way outpaced inflation, and televisions and smart phones aren't going to provide opportunities to get out of poverty the way good health and education can. The goal of a good college education for your children is seeming more and more distant to middle class parents even if they manage to keep incomes at the same percentile or manage moderate gains. The crushing amounts of debt required in student loans were unheard of thirty years ago. Employers who pay for health insurance have steadily shifted more and more of the cost burden onto employees. The middle class is bogged down and sliding backwards when it comes to trying to keep up with the costs in these three essential areas of housing, education, and health care.

[Neil]

"Hollowing out of the middle class" is usually a turn of phrase used by people who don't follow the data in any empirical way and just FEEL more stretched to meet their lifestyle expectations (ignoring that their expectations are substantially higher than previous generations).

But the feeling, I think, is driven at least in part by the gap between the middle class and the outlier class (it's probably not even fair to call them upper class anymore, since most of the top 1% are still miles behind this group). Basically, rsj's number that shows that despite producing more wealth (presumably by working ever harder) the middle class has just treaded water in real terms over the past quarter century, while the benefits of increased productivity have accrued only to the very rich.

[rsj]

Stephen,

The decline of the middle class is about the declining share of total output available to the middle class, which is exactly what the relationship between median and mean tells you.

Yes, if you have three people, each earning $100, then one person is in the middle. And if next year, the income distribution is $10, $20, $270, then still one person is in the middle, so according to your logic the middle class has not "shrunk", as it still consists of one person.

But the argument is not that the middle is shrinking, but that it is earning a smaller share of the pie. So you are either arguing against a strawman, or failing to make your point about the real concerns of the middle class.

[Stephen Gordon]

If that's your definition, the 'middle class' is 'everyone below the 80th percentile'; see this old post.

[Bob Smith]

"The crucial areas where this argument about tech progress fails are in housing, education, and health care. Not so much in Canada perhaps, but very much so in the US."

But you have to be careful when talking about the impact of health care and education costs on living standards, especially in the US. Although education pricing in the US is ridiculous (and, as an aside, a testament to the perverse effects of government guaranteed lending. Universities would never be able to charge what they do if students had to borrow to fund it without the good faith and credit of Uncle Sam to back them up), it may well illustrate the proposition that real pricing can differ across income groups, to the extent that universities use high tuition fees to cross subsidize different groups (as they do). The real cost of attending university is going to be a lot higher for Richie Rich than for Sammy Slum. Plus, since Richie Rich is much more likely to consume post-secondary education than Sammy Slum (at any price) education inflation may, perversely, hit the rich harder than the poor as a group.

You could tell a similar story about health care to the extent that the poor obtain health care under Medicare (though that doesn't take away from the middle-class story).

In any event, in the Canadian context of Stephen's post, I don't think either of those observations take away from my original comment.

With respect to housing, though, one of the criticisms of CPI is that it does a lousy job of measuring housing prices. The CD Howe Institute recently raised this point in the context of Canada about how Canada's CPI does a lousy job of measuring housing prices and failing to catch the impact of housing booms (rather it focuses on a index of housing costs, which reflects, amongst other things, interest rates, depreciation, etc.). Apparently, the US has a similar problem, since it uses rental prices to approximate housing costs. In theory that should work in practice. In practice it doesn't (or at least not always).

Mind you, to some extent the CD Howe's observation that the CPI does a lousy job of catching housing booms in Canada reflects that the recent housing boom has been driven by LOWER housing costs, notably record low interest rates. To that extent, the CD Howe misses the point that housing prices, per say, aren't really relevant to consumers, it's the cost of carrying a house that matters.

[Bob Smith]

"The decline of the middle class is about the declining share of total output available to the middle class, which is exactly what the relationship between median and mean tells you."

The problem with that conception of a declining middle class is that it defines "middle class" purely by relative share of national income. But it isn't obvious while the fact that Richie Rich is getting richer means that the "middle class" is "declining" (if Richie Rich was getting poorer, would we say that the middle class is rising?).

To my mind, a "declining middle class" can mean one of two things. Either absolute incomes of the middle of the income spectrum are falling - i.e., the middle class is staying the same size, but it's income is falling - or incomes are polarizing (i.e., the number of a given income range in the middle of the income spectrum is falling. Neither of those appear to be happening.

[Bob Smith]

"In a way if you compare an individual's change in lifestyle due to increasing power and decreasing price of flat widescreen HD TV, smart phones, and computing technology, it's in a way akin to comparing nominal wages changing over time. If we compare the degree of expanding luxuries of the top earners over that same time period the comparison gets closer to comparing "real" inflation adjusted values, where in this case tech innovation serves as the analog to inflation."

I'm not sure I understand your point. Comparing real, material consumption (colour TVs, air conditioners, cars, etc.) is a measure of "real" consumption (and, to the extent that consumption translates into wellbeing - admittedly a contestable notion, although that caveat applies equally to income - welfare). It's not at all akin to comparing nominal wages.

Similarly, it's a little odd to be discussing the "expanded luxuries" of the top earners, because part of the digital revolution has been to rapidly disseminate "luxury" technology to everyone. I'm old enough to remember the clunky and pricy "cell-phones" used by the "Masters of the Universe" (read Tom Wolfe) of the mid-1980s. These days that same technology (well, in an infinitely improved and cheaper form) is readily available to denizens of the most impoverished slums in the world. You see that in other areas too. Cars, for example, are much bigger and more powerful than they were in the 1970s (and filled with all sorts of accessories that would have been luxuries, if they existed at all, a decade ago). Whereas, once upon a time, an automatic transmission was a luxury feature, now it's, well, standard (see today's Star lamenting the fact that Young Drivers of Canada no longer has any standard transmission training cars in Toronto). Back in the day, you knew you made it when you had a big honking satellite dish in your back yard. Now, that's a sign that you're a sucker.

In fact, I think a case could be made that the scope of uniquely upper-class luxuries has narrowed, rather than expanded, over the past few decades. While the Foodie trend of the last decade or so probably owes a great deal to the Food Network, I can't help but think that part of it is owing to the fact that exotic food (which are, of necessity, not susceptible to technological change) remains one of the few remaining true "luxury goods" - since a big car, fancy phone or satellite dish won't serve as much of a status symbol, or at least not for long.

[Patrick]

FWIW, I don't think that comparing consumption options from bygone era's to the present is really all that useful. And Bob may not be a car guy, but I think everyone who cares about such things knows exactly what my 7 year old Mazda 6 wagon signals relative to e.g. a Bugatti Veyron Super Sport. In the '70's there was nothing even close to the gulf separating a regular family car and the super cars of today. Even with stuff like consumer electronics, there are options for signaling. Sure, you may have a big screen TV, but do you have a 50 seat theatre complete with TFW to serve you your popcorn?

[Bob Smith]

"FWIW, I don't think that comparing consumption options from bygone era's to the present is really all that useful."

Surely it says something when yesterday's "luxuries" are today's commonplace household item. Certainly, if we're going to talk about the declining middle class, it would help to keep in mind that the middle-class of yesteryear would swap their lives for ours in a heartbeat.

"Even with stuff like consumer electronics, there are options for signaling. Sure, you may have a big screen TV, but do you have a 50 seat theatre complete with TFW to serve you your popcorn?"

As for signalling, hey, it's always possible to signal ridiculous wealth - buy yourself a herd of giraffes and keep them in your front yard. That's an effective signal (of wealth, if not wisdom), but I don't know how much utility you'd derive from it. The fact that someone can drop a few million on a luxury car or a home theatre is an effective signal of wealth, but surely what we care about it utility. Sure, having a 50 seat home theater may be better than having a 50-inch flat screen TV, in each case with access to hundreds of digital channels, Netflix, etc. but I'd suggest that the utility gap is less than that between the guy with a 36-inch colour TV and cable television and a 13-inch black and white TV with rabbit ears in 1980.

"In the '70's there was nothing even close to the gulf separating a regular family car and the super cars of today."

Really, the gap between a Lamborghini Countach (to name the quintessential supercar of the 1970a) and a Ford Pinto or AMC Gremlin was less than that between today's supercars and a modern subcompact? It's a judgment call, I suppose, but not a self-evidently true proposition. Granted, modern cars are better all along the spectrum than they were in the day (the 1970's was truly a horrid era for cars - even the luxury cars were crap), which again, surely means something, but I'd suggest that the gap in terms of utility derived by their various owners has probably narrowed. To paraphrase the old MasterCard ad: "Bugatti Veyron Support Sport: $4,000,000. Not dying in a flaming wreck because your otherwise shitty sub-compact is better build than it would have been 30 years ago, has ABS, seatbelts, airbags, stability control, yada, yada, yada: Priceless".

[Bob Smith]

In any event, whether the consumption of the "super-wealthy" is what drives perception of the good life. Surely it's about keeping up with the Jones', not keeping up with the Onassis'.

[Patrick]

"Surely it says something when yesterday's "luxuries" are today's commonplace household item"

Meh. It says technology progresses. So?

In the 1930's one could have made the argument "well, we're so much better off than the serfs in the time of the plague ... ", but that would have been missing the mark for the same reasons that I think making comparisons to e.g. the 1970's is missing the mark.

FWIW, what worries me about inequality, as seen in the Stephen's graphs, is if those lines are getting steeper over time. I think relative wealth differences matter enormously for the future of civilization's institutions. If you can have a herd of giraffe in your front park, then why pay taxes for the park for the serfs' kids? If your own kids have private tutors, then why pay taxes for the serfs' kids school? If you have your own private security, then why pay for regular police. If your patronage decides who wins the election, then why bother having the election? Just appoint whomever you wish!

And I don't necessarily come at this as a raging lefty (well, not entirely). I think the median (middle class) capitalist, who saved and invested and has their hard earned money at risk, is getting screwed. I think there's mounting evidence to suggest that a good many of the Richie Rich's aren't really earning their wealth (in the marginal product sense), but rather they've used their wealth to rig the system and they're allocating the fruits of other people's productivity - a good deal of it attributable to capital owned by other people - to themselves, and are thus accelerating the process of generating inequality. See here (pdf), for one example.

I disagree about the Countach (it wasn't advanced at all, just lots of heavy iron), but that's a debate better had at the pub than here, so I'll let it be.

[Jeffrey Johnson]

From Bob Smith: "I'm not sure I understand your point. Comparing real, material consumption (colour TVs, air conditioners, cars, etc.) is a measure of "real" consumption (and, to the extent that consumption translates into wellbeing - admittedly a contestable notion, although that caveat applies equally to income - welfare). It's not at all akin to comparing nominal wages."

I'm sure the analogy was confusing. It was a very rough intuitive comparison. Yes it's true that material progress is real, as you state. But what matters to people is their status relative to others. So the rough analogy was something like saying that a nominal increase in income doesn't have the value that numbers might imply because the real worth of money is determined by the value relative to the whole economy, i.e. inflation adjusted purchasing power matters and number of dollars does not. From this I abstracted a core concept that comparing one's own progress between now and the past has less importance than comparing one's position now relative to others now. In this step it goes wrong from an economic standpoint because increased consumption is real and increased numbers of diluted dollars is not real.

But there is still relevence to people's sense of well being, which hopefully remains an important value to economists, though it is not easily quantifiable, if at all. How important is it to a middle income earner of the 21st century that his consumption of energy and technological convenience would be the envy of many medieval kings? How important is it to the poor of North America that their comfort level seems like unreachable opulence to the residents of a small rural Central African village? The answer, from a subjective psychological standpoint is, not a lot. Which is why the argument that the middle class is doing fine because they have smart phones, larger capacity energy efficient refrigerators stocked with food, and widescreen flat panel HD televisions, strikes me as somewhat bogus and missing the point. The point, as rsj mentioned, is declining share of total output.

It strikes me as profoundly unjust that people who own patents (but do none of the creative work), people with control over corporate purse strings, people close to the gushing channels of investment and finance, people who can employ the law to their advantage, and people who own stuff, are experiencing skyrocketing wealth and income (while paying a relatively small share in taxes compared to wage earners), while those with the actual skill and knowledge to discover, design, create, build, produce, transport, store, package, and sell stuff are earning a declining share of output. It seems there is a whole lot of rent seeking going on.

While the term "hollowing out of the middle class" may not be very sensical in a literal analysis, what it seems to mean in practice is that there is a growing anger among a majority of society who perceive, rightly or wrongly, that they are being short changed and gradually squeezed out of a fair share by those who don't always appear to deserve or earn their advantage by real effort or merit. There seems to be a moral component to this discontent, which I realize is outside the domain of economics, but remains important. The collapse of the housing bubble, subsequent financial crisis and bailout, and apparent lack of serious consequences for the highly rewarded people who made the errors or committed the intentional frauds most directly responsible for causing all the economic misery of recent history, seems to have brought the gradually growing spread of inequality over the last three or four decades into very sharp focus for a very large number of people.

[J.V. Dubois]

Pretty much what RSJ said plus some more

1. You have to compare median income to GDP growth or average income or some other measure to even start talking about "hollowing out" the middle class (rsj's point)

2. We did not even start talking about changing demographics. With larger share of population dependent on pension income + larger share of population being longer at school/university one would expect that median income actually shrunk even though the lifetime income is larger. This is what Scott Sumner talks about a lot. His example of university city being the second poorest in the USA (with students consisting around 70% of "households") illustrates this very well.

3. The same goes for other social changes. For instance now it is more common that people may be working part time. For instance mothers/father with small kids etc. - where in the past they would opted not to work for varisous reasons (cultural, overall infrastructure such as pre-school availability or taxreasons).

In general I have yet to see any meaningful study that takes all this into account. As of now I have similar feelings to those of Scoot Sumner - the data are so horrendous that it almost does not make sense to talk about them in a serious manner.

[Stephen Gordon]

Why are these things unique to the middle class? It certainly looks to me as households in lower-income groups are facing the same problems, if not worse. So why represent them as 'middle-class' problems?

[J.V. Dubois]

Stephen: If it was meant as a reply to me it is actually easy to explain. Compare these two hypotethical 15-65 year old population

1. 10% students or part-time workers, 80% full time workers, 10% pensions
2. 30% students or part-time workers, 40% full time workers, 30% pensions

There may be no change whatsoever in an actual income of these particular socioeconomic groups but nevertheless you will see sudden drop in median income that may easily be interpreted as "hollowing out of the middle class" in second example relative to first example using data similar to what you use in this post.

In a way it is true - given a snapshot of the society it they may look poorer on average compared to the first one. But from lifetime perspective of any particular person within society there is no change. They have the same individual lifetime income. So in a way it is a very misleading interpretation of data.

[Stephen Gordon]

If that was meant as an reply to me, then you didn't actually reply. And I'm getting not a little annoyed at the lazy accusations of 'misleading'. You're not addressing the point I'm making.

[hosertohoosier]

The reason for the "hollowing out of the middle class" mantra is sectoral, rather than really about class. It's primarily a lament about the loss of good-paying blue collar jobs as a result of de-industrialization. My guess is that a similar trend can explain the different wage group across income levels and across genders.

The premium on experience will is lower in blue collar jobs that require physical strength, and higher in service sector jobs that require knowledge. As you shift from a manufacturing-heavy economy to a service-based one, you shift to an economy where entry level jobs pay little (or nothing in the case of internships). At the same time, the highest salaries - generally those going to the most experienced workers - will rise as service sector jobs that reward experience replace blue collar jobs that do not.

This also explains why we see this phenomenon is less pronounced among females - the shift from manufacturing to services was less pronounced there.

AND it explains why people feel poor. The structure of our lifetime income has become back-loaded. However, many of the major costs people incur - starting a family and buying a house - occur early in life (when we are still poor). Our lifetime income is not synced well with our needs. This disjuncture is aggravated by government policy which, catering to more politically influential geezers, transfers money to the old by various mechanisms.

The solution is to develop better mechanisms for young people to cash in on future earnings (e.g. lifetime income taxation).

[Jacques René Giguère]

Conspicuous consumption has a goal, that of its owner telling you: "Oh ye suckers of the middle class,the engineers who designed that car could have worked on a better MRI. But they didn't and you will die of some horrible disease while that undrivable contraption sit in my garage. I am enjoying that thought so much! Stuffit."

[Patrick]

"The premium on experience will is lower in blue collar jobs that require physical strength, and higher in service sector jobs that require knowledge."

Tell that to any unemployed IT worker over 40.

[Determinant]

Why are these things unique to the middle class? It certainly looks to me as households in lower-income groups are facing the same problems, if not worse. So why represent them as 'middle-class' problems?

Politican-speak. Nobody wants to be "Poor" or "Lower Class" so the NDP calls them "Ordinary Canadians" and the Liberals call them the "Middle Class". Middle Class is anyone who does not self-identify as "Rich" (as a crude approximation). That it may encompass anyone below the 80th Percentile in income just goes to show how very useful it is as a political term.

You also have the loss of job security/tenure and the rise of precarious work (temps, contract labour, etc.) which lowers effective income through risk discounting. You also have the loss of pensions and health benefits which also lowers income available for discretionary consumption.

[The Keystone Garter]

rsj is correct. You have to view it in terms of the potential tax base. Policies and tax shifts that may potential increase income or employment, aren't funded out of the median income. That being said, the main middle class shrinkage arguments refer to the USA.
I've pretty much figured out why Victoria's Cross recipients came from Vancouver and Wpg. The rail industry right now employs about 1.2 persons for every million dollars of market cap. Oil, is five times this. Tar, somewhere in between. Aerospace and banking are employment intensive. Potash, not. Winnipeg expanded during 1900 via rails. The wealth/power was basically tied to education during the Scottish Enlightenment. In France, their precursor Enlightenment probably was founded by gvmt military figures; the antithesis of trade, all else equal (sometimes you should destroy capital and in a Depression sometimes war is the only way to get rich people to pay for stimulus).
Education, certain courses, are very important in qualifying for a future GAI, and in selecting wise leaders. People are monkeys that don't want to forage for bananas when some aren't doing there part. When imperial oil is only employing one person for every $7M in market cap...
Enough funds lets you fund employment intensive health and education. There weren't too many silver spoons in Wpg or GD era Vancouver; merit based economy where people had experienced unemployment in another city, made people proud to sacrifice. This could form the basis of selecting future leaders. Could we get the 2011 stats?! When will Trudeau be able to restore Census?

[Henry]

It would appear Canadians have borrowed from the US the notion that we have had a hollowing out of our middle class. The following report says no, but shows a more nuanced picture that goes beyond the median income metric: http://www.td.com/document/PDF/economics/special/ff1212_income.pdf, See p.3, at "Keeping Up With The Joneses..." in particular.