Higher Education Strategy Associates

Tag Archives: Statistics Canada

March 05

The Long-Term Benefits of Higher Education

A very good Statscan report came out last week, and didn’t get nearly enough attention.  Authored by the excellent Marc Frenette, it’s called, An Investment of a Lifetime? The Long-term Labour Market Outcomes Associated with a Post-Secondary Education, and it deserves a wide readership.

What Frenette did was link the 1991 census file to the Longitudinal Worker File (LWF), which integrates data from Records of Employment, annual T1 and T4 files, and some data on employers as well, for a 10% random sample of all Canadian workers.  From this, he created a sample of about 8,000 people who were born in Canada between 1955 and 1957 (i.e. who were about 35 years old at the time of the census), and who held jobs in 18 out of 20 years since then. From this, he worked out what the added value of university and college credentials were over that period.

Figure 1 shows earnings by education level.  For men between the ages of 35 and 55, the added benefit of a college education (vs. high school) was $153,000; for a university education it was $445,000 (for women, the figures were $115,000 and $280,000).  In addition to higher salaries, higher levels of education were also associated with higher levels of union membership, and lower frequency of layoffs.

Figure 1: Net Present Value of 20-Year Earnings of Canadian-Born Workers, by Level of Education














Figure 1 isn’t exactly ground-breaking; more education = more money, and more so for men than women.  Where it gets interesting is when the results are disaggregated by gender, and attention is paid not just to means and medians but also at the distributional tails.  Figure 2 compares the wage premiums at various percentiles for female college and university graduates, over high school graduates in the equivalent percentiles.

Figure 2: Cumulative Additional 20-Year Earnings for Female College and University Graduates, at Selected Percentiles














Figure 2 shows three things.  First, women with university degrees make more money than those with college or high school across all percentiles.  Second, that said, down around the 5th-10th percentile, the premiums are so low that it’s really not clear that women are better off with higher education.  And third, the premium for higher education really flattens out above the median – which, as Figure 3 shows, is not even vaguely the case for men.

Figure 3: Cumulative Additional 20-Year Earnings for Male College and University Graduates, at Selected Percentiles














Crazy stuff.  Recall from Figure 1 that average gains for men were significantly higher than for women.  But Figure 2 and Figure 3 show that the median gains – those at the 50th percentile – are about the same.  The difference is that among males, the top ten percent – and especially the top five percent – are reaping astronomical rewards from higher education.

The last amazing thing in the paper has to do with how men and women with bachelor’s degrees fare comparatively in the public and private sectors.  And the numbers there are astonishing: in the bottom ten percentiles in the private sector, women are making less money, cumulatively, than their counterparts with just high-school education.  But what’s really interesting here is the fact that in the public sector, at least, women actually reap higher gains than men.

Figure 4: Median Cumulative Additional Earnings for Male and Female University Graduates in Public and Private Sectors














Nitpickers will likely snub this study – it deals with a cohort that finished school 35 years ago, it doesn’t disaggregate by field of study, etc.  But methodologically, it points to ways to conduct future studies (we could do the same with a shorter period for 25 year-olds in the 2001 census, for instance), and substantively it gives us a lot to chew on, not only in terms of average earnings, but also with the distribution of those earnings.  Kudos to Marc for this work.

February 27

New Student Debt Numbers

So, the more stat-minded among you may have noted the release, this past Tuesday, of Statistics Canada’s 2012 Survey of Financial Security (SFS).  Though the main talking points were largely about mortgage debt, it also contained some interesting statistics on student debt.

Now, remember that these are figures on outstanding student debt.  Some of it will be in repayment (i.e. held by graduates now in the labour force), and some of it will not (i.e. held by current students).  The way to think of these debt figures is as a collective portrait of people who borrowed in the decade or so prior to the snapshot, and who had not yet fully repaid their debt (because those who had successfully completed repayment would be out of the sample).  So the 2012 figure for student debt is actually a collective picture of the outstanding debt of everyone who borrowed in the period 2002-2012, and who had not yet repaid, the 2005 figure covers the period 1995-2005 or so, etc., etc.

Anyways, the headline that the usual suspects would like you to focus on is the one about aggregate debt outstanding: $28 billion, up by $5.5 billion (23%) in real dollars since the last time the study was conducted, in 2005.  Why is that a big deal?  Because!  $28 Billion!  Big Number!  But a slightly more intelligent look at the data shows a different story.

Figure 1 shows that the average outstanding student loan was about $15,000.  That’s up about 6% from 2005, and 13% from 1999 (again, all figures are inflation-adjusted).  Why is this figure so much smaller than the one for total debt?  Simple: more people have outstanding student debt than in 2005, so it’s divided among a larger population.  That might be because people are taking longer to repay their loans – more likely, though, it’s a reflection of the fact that student numbers as a whole rose substantially over the 00s.

Figure 1 – Average Student Debt Among Holders of Outstanding Student Loans, in $2012














Intriguingly, the data for median student debt (that is, the mid-point value, rather than the mean) tells a slightly different story, in that it fell 2% between 2005 and 2012 (though it has still risen a bit since 1999).

Figure 2 – Median Student Debt Among Holders of Outstanding Student Loans, in $2012














How should we interpret this?  This isn’t the easiest data to unpack.  It probably means, as I pointed out back here, that student debt hasn’t been increasing.  But it also might mean that debt repayment rates have been increasing along with indebtedness, or (less likely) that a greater fraction of student loans are held by individuals who graduated from shorter programs.

Whatever the truth, what we do know for sure is that young people aren’t drowning in student loan debt.  Among family units headed by people under-35, only a quarter hold a student loan, and the loan debt constitutes just 5.3% of their total debt, down from 6.7% in 2005.  Whatever the effects of student borrowing is, it would appear that deterring graduates from taking on ever-larger mortgages isn’t one of them.

October 22

Faculty Salary Data You Should Probably Ignore

Recently, the Ontario Confederation of University Faculty Associations (OCUFA) published a comparison of American and Canadian academics’ salaries.  Using Canada’s National Household Survey (NHS) and the US Occupational Employment Statistics (OES) survey (which they described as being not quite apples-to-apples, but at least Macintosh-to-Granny Smith), they noted that average salaries for the combined college-and-university instructor population (the OES cannot disaggregate below that level) were $76,000.  In Canada, the figure was $65,000.  Hence, according to them, with the dollar at par, there is a 17% gap in academic pay in favour of the Americans… and much more of a gap if PPP is taken into account.

There are three reasons why this conclusion is deeply suspect.

First, OES and NHS are not even vaguely comparable.  One is a world-class instrument, based on administrative data collected at over 200,000 places of employment; the other: a self-report from a nonrandom sample of Canadians which has been widely panned as a steaming pile of horse manure.

Second, the actual numbers seem to be slightly off.  When I go to the OES, the category for 2- and 4-year post-secondary teachers (25-1000), I get $77,600.  The Canadian NHS files show that “university professors and lecturers” (category 4011) earn $87,978 and “college and other vocational instructors” (category 4021) earn $57,275.  Together, weighted, that’s an average of $70,033.  So, a 10% gap, not a 17% one.

Third, since the two countries don’t have identical proportions of instructors in the 2- and 4-year sectors, it’s hard to tell how well these numbers reflect differences among university professors.  Neither do we have any sense of the proportion of part-timers and sessionals in the count, on either side of the border.  In other words, this comparison is based on a hodgepodge of non-comparable data, and proves absolutely nothing with respect to relative salaries of professors on either side of the 49th parallel.

More direct comparisons are possible.  Oklahoma State University has been doing an annual survey of salaries at Public and Land-grant Universities – the grouping of US institutions that look most similar to Canadian universities – for 40 years.  The figure below compares the 2012-13 OSU data with that of Canadian profs from Statistics Canada’s last UCASS study (2010-11), as published by CAUT.

Canada vs US Professors’ Salaries













One can quibble with this graph, of course.  The Canadian numbers have probably gone up another 6-7% in the intervening two years.  The US numbers don’t include the income professors get from summer research grants, which would probably add another 10% or so to their averages (see here for that calculation).  But effectively, there’s about a 15% pay gap in Canada’s favour, if dollars are counted at par, not a 17% gap the other way.

Naturally, one could get into arguments about purchasing power parity, living standards, and the like – that’s all fair game.  What’s not fair game is using a set of bad statistics when better ones are available, just because the bad data happens to better serve your cause. You’d think an association representing academics, of all people, would know that.

October 02

A New Study on Postdocs

There’s an interesting study on postdocs out today, from the Canadian Association of Postdoctoral Scholars (CAPS) and MITACS.  The report provides a wealth of data on postdocs’ demographics, financial status, likes, dislikes, etc.  It’s all thoroughly interesting and well worth a read, but I’m going to restrict my comments to just two of the most interesting results.

The first has to do, specifically, with postdocs’ legal status.  In Quebec, they are considered students. Outside Quebec, it depends: if their funding comes from internal university funds, they are usually considered employees; but, if their funding is external, they are most often just “fellowship holders” – an indistinct category which could mean a wide variety of things in terms of access to campus services (are they students?  Employees?  Both?  Neither?).  Just taxonomically, the whole situation’s a bit of a nightmare, and one can certainly see the need for greater clarity and consistently if we ever want to make policy on postdocs above the institutional level.

The second – somewhat jaw-dropping – point of interest is the table on page 27, which examines postdocs’ training.

Level of Training Received or Available, in % (The 2013 Canadian Postdoc Survey, Table 3, pg. 27)














As the authors note, being trainees is what makes postdocs a distinct group – it’s basically the only thing that distinguishes them from research associates.  So what should we infer from the fact that only 18% of postdocs report receiving any formal training for career development, 15% for research ethics, and 11% on either presentation skills or grant/proposal writing?  If there’s a smoking gun on the charge that Canadian universities view postdocs as cheap academic labour, rather than as true academics-in-waiting, this table is it.

All of this information is, of course, important; however, this study’s value goes beyond its presentation of new data.  One of its most important lessons comes from the fact that a couple of organizations just decided to get together and collect data on their own.  Too often in this country, we turn our noses up at anything other than the highest-quality data, but since no one wants to pay for quality (how Canadian is that?), we just wring our hands hoping StatsCan will eventually sort it out it for us.

But to hell with that.  StatsCan’s broke, and even when it had money it couldn’t get its most important product (PSIS) to work properly.  It’s time the sector got serious about collecting, packaging, and – most importantly – publishing its own data, even if it’s not StatsCan quality.  This survey’s sample selection, for instance, is a bit on the dodgy side – but who cares?  Some data is better than none.  And too often, “none” is what we have.

CAPS/MITACS have done everyone a solid by spending their own time and money to improve our knowledge base about some key contributors to the country’s research effort.  They deserve both to be commended and widely imitated.

September 05


At some point in the next week or so, Statistics Canada will be releasing its annual statistics on tuition fees.  Hopefully it will be less of a fiasco than last year, when they released data a few days after the Quebec election, but didn’t bother to note that the planned tuition fee hike was being reversed.

What I want to do today is to put the inevitable “rising fees” stories that always accompany the Statscan release into some sort of context.  Students pay two types of fees – tuition and “ancillary fees”.  Statscan data on the latter is only marginally better than hopeless, so these fluctuating annual figures need to be treated with extreme caution; but they’re a non-negligible part of total tuition (15% or so), and so I include both in the graph below showing the evolution of total fees.

Figure 1 – Average Tuition, Canada, Nominal Dollars













Figure 1 is the graph that the zero-tuition crowd love to show: steady 5.1% annual tuition increases from 1995 to the present.  That’s actually a trick of scale – in fact, during the era of maximum government skintness (the 90s) tuition was going up about 9% per year to make up for cuts in government grants.  After 1999, the economy improved, public finances improved, and the rate of fee increase fell to just about 4%.

There is, however, a little thing called inflation.  It’s kind of important if you want to understand real prices over time.  Here’s what the tuition graph looks like if you take inflation into account.

Figure 2 – Average Nominal and Real ($2103) Tuition, Canada













This changes things a bit.  Those annual increases since 1999-2000?  Just two percent, after inflation.

But, as apparently nobody in the press or politics seems to understand, those increases in fees have been accompanied by increases in subsidies, too.  The most important of these are the increases of various forms of tax credits.  Say what you want about them – they reduce the actual cost of education by about a third.  Their value is eroding slightly at the moment due to inflation, but they are still worth $2,220 to the average Canadian student.

Figure 3 – Average Nominal and Real ($2013) Tuition plus Net Real Tuition Canada













Finally, if we’re looking at affordability, we also need to take into consideration a measure of ability-to-pay, because cost on its own is meaningless.  Televisions cost more than they did, say, 40 years ago, but no one thinks they’re “less affordable”, because incomes have risen even more quickly.  So to compare affordability across time, what we need to do is look at cost over time with respect to a measure of purchasing power, such as average family after-tax income.  Which I do, below.

Figure 4 – Real Net Tuition as a Percentage of Average After-Tax Family Income













So, is tuition less affordable than it was?  Well, a bit, yes.  Fifteen years ago, it took up 4.8% of average, after-tax income; now, it takes up 5.2%.  But calling it a crisis, the way the usual suspects routinely do, is a bit of a stretch.

And we haven’t even taken into account need-based student aid yet.  We’ll do that tomorrow.

July 15

More Money Than You Think

If there’s one thing everyone knows, it’s that Canadian universities have had a hard time of it during the recession during the last few years, yes?  Absolutely starved for income because of government cutbacks, etc etc.

Not so fast.  Check out this data on university operating budgets from the CAUBO/StatsCan financial survey:

Figure 1: Indexed growth in University Operating Budgets 2007-08 to 2011-12

That’s right – across the country, university budgets went up by 28% between 2007-08 and 2011-12.  That’s more than twice the rate of inflation.  (Note: if you’re wondering why Alberta skews high, its because MacEwan and MRU were re-classified as universities in 2009 – take them out and Alberta basically looks like BC).

How is this possible, you ask?  Haven’t governments been cutting back?  Well, the last two years haven’t been very good, but let’s not project that too far backwards.  In fact, during the heart of the recession years, the worst any major province fared (Ontario – big surprise) was to keep pace with inflation.  Across the four big provinces which make up 90% of our national system, spending was actually up nearly 20%.

Figure 2: Indexed Growth in Government Contributions to Operating Grants, 2007-08 to 2011-12

Those of you with heads for numbers may now be scratching your heads.  Government grants are a little over half of all institutional income.  So if overall income is up 28%, and this half of it is only up 20%, that means the other half – the student half  must be up by…

 Figure 3: Indexed Growth in Tuition Income, 2007-08 to 2011-12

Yeah, that’s right: tuition income is up 40%.  Four.  Zero.  How is this possible when Statscan says tuition fee increases are only about a third of that?  Because this is aggregate tuition and Statscan looks at average tuition.  One is larger than the other partly because of increases in domestic enrolments, but more importantly because of spectacularly increased international enrolments, which also carry much higher tuition fees.

Obviously, with extra students come extra costs, which is why it doesn’t necessarily feel like there’s 28% more money floating around these days.  Between enrolment increases and cost increases (mostly labour costs, including rise through the ranks (Link to: http://higheredstrategy.com/rise-through-the-ranks-rtr/)), Ontario is still slightly down on the deal in per-student terms, while other provinces are up, but only slightly. 

“Cutbacks” aside, governments are still spending far more than they were on PSE six years ago (even in Alberta) and institutions have been absolutely raking in cash from tuition.  We don’t have the 2012-13 numbers yet, but they’ll likely be in the 8-9% range everywhere except Quebec.  That means operating budgets overall likely expanded by about 3-4% last year, even as governments reduced funding.

Two final thoughts: One, if institutions still feel squeezed when income is rising twice as fast as inflation, it means there are some serious issues to work out on the cost side.  And two, God help us if those international students stop coming.



July 05

Today’s Statscan Youth Jobs Report

Hi there.  Just a slight deviation from the summer publication schedule to bring you some perspective on the youth employment numbers coming out of StatsCan today.

Unless something has gone seriously gaga in the youth labour market in the past few weeks, today’s Labour Force Survey release will say that slightly over 70% of students aged 20-24 are employed and that unemployment among these students is in the 7-9% range. That sounds pretty good; the problem is that StatsCan’s definition of unemployment doesn’t even vaguely correspond to how students see the issue.

The basic problem is that StatsCan defines someone as being “out of the labour force” if they are in full-time studies; as a result, students taking summer courses are excluded from the calculation.  But in fact, as our own 2012 survey of summer employment showed, over 70% of summer students are also either working or looking for a job; among this group, unemployment typically runs at between 20 and 30% (last year, the figure was 29%; this year, it is 23%).  Indeed, one reason many students take summer courses in the first place is precisely because their jobs search was unsuccessful!  

Although our full annual employment report won’t be out for a bit, I want to provide you with some statistics on one other labour issue currently generating a lot of attention: unpaid internships.  Our preliminary examination of the data suggests that 5.4% of students are in some kind of internship or practicum this summer.  Of these, roughly half are educationally-related (e.g. mandated practicums in teaching or social work), meaning that about 2.7% of all students (or about 27,000 across the country) are in unpaid internships this summer.  That’s a long way below the 100-300K estimates one sees in the press these days, but it’s not inconsistent with those numbers since a) those larger figures represent internship positions across an entire year rather than positions at any one time, and b) our survey looks only at current university students and does not include either college students or recent graduates. 

Lastly, a key point about these unpaid internships: they’re mostly part-time affairs.  The median unpaid internship is just a 14 hours per week commitment; as a result, fully half of the students with unpaid internships are able to gain an income by working either full- or part-time. 

Have a good weekend, and be wary of overly rosy LFS statistics.

June 04

Some Insights Into Medium-term Education Outcomes

As I noted yesterday, Canada is unnecessarily bad at looking at medium-term outcomes of education. The only place where we have data on university graduates even five years out is in BC, and they publish the data in such a weird format (seriously: check it out) that no one really explores them.

It could be worse. In 2005, Statscan, did a 5-year follow-up of the class of 2000 and elected not to publish any results relating to employment or income. *Facepalm*, as the kids say.

However, because I have nothing better to do, I have put together three interesting figures on how graduates fare between years 2 and 5, in select disciplines (chosen because of sample size). It’s all courtesy of that same BC data on the graduating class of 2004. I won’t bore you with employment both at 2 and 5 years, it’s uniformly quite low. Let’s start instead by looking at incomes five years out. It turns out that while some disciplines do have precarious earnings in the first two years after graduation, median incomes rise across all fields by 35% between years 2 and 5 (that’s more than 10% per year, if you’re counting). Just for comparison, the median earnings among all Canadian workers in 2009 was $46,500. So, even in the “soft” disciplines, the ones that allegedly leave people without valuable skills like English and History, graduates five years out show median incomes above the national average.

Figure 1 – Median incomes, 2 and 5 years out, BC class of 2004, selected disciplines

Ah, you say: but are they using their skills? Aren’t they, perhaps, underemployed? Well, not really. Figure 2 shows the percentage who are in jobs which have been classified by the National Occupation Classification system as either being managerial or requiring university education. In the three disciplines where that percentage is lowest after 2 years (Biology, English, and Business) the rates of employment in high-skilled jobs jumps by 50-65% in the following three years. Five years out, the difference between history grads and computer science grads is only five percentage points.

Figure 2 – Percentage of graduates in jobs classified as “Management” or “Skill Level A” by NOC, 2 and 5 years out, BC class of 2004, selected disciplines

What’s perhaps most interesting is how graduates feel about how their education changes over time (figure 3). Across the board, graduates five years out feel less satisfied with their education and are less likely to say they’d do the same program again that they did at two. But while there’s a generalized malaise among students, the regret factor is clearly a lot higher in arts and science programs than it is in professional ones.

Figure 3 – Percentage-point change in graduates indicating satisfaction with program and indicating they would take same program again, 2 and 5 years out, BC class of 2004, selected disciplines

Anyways, that’s just what one bored dude can do with available data on a crappy 12-hour flight. Imagine if governments actually wanted to improve data and analysis in this area! Possibilities: limitless.

June 03

A Better Way to Track Graduates

The real problem Canada has with respect to the whole “does-education-pay” debate is data. It’s not that we don’t have people collecting data – we do, lots of them. The problem is that they’re all collecting data over time frames so short as to be largely meaningless.

The gold standard used to be the National Graduate Survey, which surveyed every fifth graduating class two and five years out. Now the 2-year survey is a year behind schedule and the 5-year follow-up has been discontinued. That’s right, folks – at the start of the recession, when Statscan took a look at their suite of surveys and decided which ones to can and which ones to keep, they decided that the one on medium-term educational outcomes was among the least policy-relevant and canned it. You know, so they could keep funding their monthly poultry storage reports .

For about a decade now, a number of provinces (all except MB, SK and NL) have started collecting data too; indeed, they have been doing so on a biannual basis, which is much better than Statscan could ever manage. However, most only track them out to 24 months, so the issue of long-term outcomes is still unaddressed. BC is the only province which does 5-year reports, and they’re quite interesting (more about them tomorrow).

The long-term outcomes of degrees and programs clearly matter a great deal. So why can’t we measure them? Cost, mainly. Anything further out that about 24 months is expensive to do well (BC’s 5-year response rates are disappointing, for instance), and so – penny-wise pound-foolish nation that we are – we don’t do it.

But there actually is a very cost-effective way to do this; namely, to link student records to tax records. Virginia, Tennessee and Arkansas have already linked their grads’ data to unemployment records and others seem poised to follow. In Canada, we could quite easily do the same thing by having Statistics Canada link its Post-Secondary Student Information System (PSIS) to the T1 family file. Instantly, with no new data collection expenses, you’d have income data by institution, program of study – what have you – as many years out as you like. As always with Big Data, there are some privacy concerns, but frankly none of them are very convincing, certainly not compared with the major public policy gains available.

Linking administrative databases is cheaper, faster and more accurate than what we do now. Why we haven’t moved to this system already is one of the biggest mysteries in Canadian higher education policy.

May 02

Those Statscan Cutbacks

Many will have seen news yesterday about large cutbacks in the works at Statistics Canada. On the basis of the news that lots of PSAC members had received notices that their jobs may be “affected,” a number of pro-Statscan commentators rushed to say that the agency needed to be saved because it provided such fantastic, non-partisan analysis.

Well, yes. But yesterday’s notices appear not to have gone to any analysts, since they are not PSAC members.  The employees who got notices would appear to be the ones involved in data collection, field interviews, data processing, etc.  Not to put too fine a point on it, these units are the ones that make Statistics Canada a target, because they are unbelievably expensive.

I have had occasion in the past to compare prices between a private sector data collection agency and Statistics Canada. Using exactly the same methodology, Statistics Canada costs were between two and three times that of the private sector agency. On any given survey, that’s going to be a seven-figure difference.

Now, obviously, quality costs money. And Statistics Canada, just by dint of being itself, has some advantages over a private sector provider in terms of getting higher response rates (people are a lot more willing to stay on the phone with “Statistics Canada” rather than some outfit running out of RackNine). But it’s not that big a gap; in the project in question, the difference between our project and a comparable one Statistics Canada had done was about twelve percentage points (65% vs. 77% if memory serves). Every percentage point matters, of course, but that’s a lot of money per point.

Besides, it’s not as though private sector firms are incapable of dispassionate and highly professional data collection. South of the border, data collection and analysis of the entire suite of post-secondary education surveys (the National Post-Secondary Student Aid Survey, the Beginning Post-Secondary Survey and the Baccalaureate and Beyond Survey) conducted by the U.S. Department of Education has always been awarded via competitive tender. As anyone who has used these studies knows, the work is world-class.

And, who knows? If budget cuts for surveys are significant enough, then Statscan might have to get serious about pursuing more solutions based on administrative rather than survey data, a development which – as I’ve argued before – could actually improve the quality and timeliness of the information.

So while the Tories’ past behavior has given all Canadians cause to worry about the honorability of their intentions, let’s not jump to the conclusion that a budget cut is necessarily a disaster. It might actually open the door to a better statistics agency.

Page 3 of 41234