« Vocations and altar servers. | Main | Postsecondary education questions: 3 ways of being able, 3 ways of being unable.. »

29 August 2012


Feed You can follow this conversation by subscribing to the comment feed for this post.


The only negative I would point out is that this looks at college solely from the monetary benefits to be accrued -- which is definitely an important point to be made. That is, of course, relevant to the question you are discussing regarding what is owed to your children. However, have you considered looking at the intangible benefits of college? Though now that I think of that, I must confess that many of the intangibles that occur to me can be found in other ways than just college.

Which means that I just argued myself in a circle. Feel free to ignore.

Christy P.

I think I got more intangibles from grad school than college, but of course the first degree was the prerequisite.


Does my spouse count as an intangible? That certainly gives this post an appealing recursivity.

I think you "get" intangibles wherever you go, and whenever you do it -- the key is that you get different sets of intangibles from different paths. And some intangibles are more worthwhile than others. I mean, there are intangible benefits to college, certainly. And there are intangible costs. And there are intangible benefits to other postsecondary education choices, and also intangible costs.


BTW, ChristyP, I forget -- do you typically teach undergraduates?


Hmmm. Okay, a couple quibbles with the data analysis and then a more general comment.

1) I'm not sure it makes to assume 7% interest on the loans long term. As I recall, the loans we had for my wife started out that way, but within six months of graduation she was flooded with offers to consolidate her student loans are rates <3%. The difference between 3% and 7% would be pretty material over 20 years.

2) Taking the loan payments out of net income makes sense from a mathematical point of view, but I don't know if it makes as much sense from a behavioral point of view. People tend (willingly or not) to make financial decisions based on their current payment load. It's a moderately good bet that all other things being equal, the person without college loans will compensate by spending more on a car or some other loan payment. One could point out that at least the other loans would be chosen, but overall, since we can't control for other loan decisions, it seems cleaner to look at income pre loan payments rather than post.

General Comment:
One of the reasons I always struggle with this kind of argument is that according to it I did everything wrong, and yet things have worked out for me very well. I went to a private college and got a degree in Classics. (I didn't take out any loans myself, living off scholarships and savings, but I pay for my wife's student loan debt and she doesn't work outside the home, so I figure I'm not that far from the experience of having taken loans out.) However, I make a high income working jobs that pretty much absolutely require a college degree. (Indeed, these days I'm in "MBA strong preferred" territory, so even just having a BA I'm under-degreed.) Maybe I could have taught my self the same kind of work that I've taught myself without the actual college education, but it would be very, very hard for me to get these jobs without the piece of paper. (I put a fair amount of energy a while back into getting a guy without a college degree hired for a position on my team, and even though he had 20 years experience helping to run a family company, and was well qualified on experience, it was a fight to get HR to people to believe that someone without a college degree would be able to do the job.)

Now, as the averages show, a lot of people who have degrees like mine don't end up doing as I have. I know people who also have private college humanities degrees who struggle to keep any kind of job at all in this environment. But I'd have a very hard time advising a child of mine not to do what I did, when I know that the doors that I've managed to open would almost certainly have remained closed if I couldn't cite a BA in my background.


Darwin: I don't know very much about who gets offers to consolidate loans at lower interest rates. Do you think it makes sense to plan based on the assumption that "someone will probably offer me a lower interest rate later?"


Bearing and Mark,
Will you be encouraging your kids to attend college?



I think we're trying to figure out the answer to that question right now. (Mark may have a different comment -- I know he probably won't be able to appear in comments till after work today, though).

One thing I think we've already figured out: it's a mistake to frame the question as "should we encourage our kids to attend college?" It's far too general of a frame. (I think Mark tried to make that point in this post.) I expect each of our children to face choices among different sets of options.

I think we want to frame the question more like [now it's me talking to a kid]: What skills and strengths do you have to offer that people will want to buy? What skills and strengths do you think you could develop, and what different ways could you develop them? How much risk are you able to absorb?

I think we're still heavily biased towards encouraging our kids to obtain post-high-school education or training, but I remain unconvinced that 4-year college is automatically worth "encouraging" over other means of self-improvement.


To my understanding (at least unless things have changed in the last ten years) basically anyone who doesn't go into default within the first couple months gets offered the chance to consolidate -- consolidation itself is a highly regulated and government subsidized business.

That said, poking around a bit more, it looks like things have settled down a lot since 2006, because federal student loans now have fixed rather than variable rates. The big question is whether you get subsidized or unsubsidized loans (ours were subsidized). Subsizided are around 3.5% and unsubsidized around 7%.


I know you're kind of a success story when it comes to humanities majors, Darwin, and I love giving your story as an example, but I have to ask -- what was your plan for how you were going to support a family someday when you *started* college?

Why did it work for you, but not for other people?

Is this one of these things where you're really quite deviously smart and so from the very beginning you had it mapped out how you were going to leverage your BA into a sufficiently-paying career, or do you wonder how you got so lucky?

Are all those other humanities majors just doing it wrong?


That said, now I run the numbers, if you come out of undergraduate with 30-60k in loans, the difference per year between 3.5% and 7% is only one to two thousand per year, so it's not going to be the amount that makes or breaks your lifetime earnings.


I think your interest rate example is a good one to point out that people's intuitive grasp of the economics situation is not very good.

In your first comment: "The difference between 3% and 7% would be pretty material over 20 years." In your latest one: "Now I run the numbers... it's not going to be the amount that makes or breaks your lifetime earnings."

I'm not criticizing you in particular. I was nodding my head in agreement with you on the first comment. :-) I think people don't have much intuitive grasp of powers/logarithms.

And I think a lot of folks are basing their thinking on "intuitively sensible" guesses that turn out to be wildly wrong. Or just on bad information (cf. law schools outright lying about employment rates.)


Honestly, I didn't have much of any plan, other than that I was pretty sure I didn't want to go on in academia. (Or, alternatively, I had lots of vague plans, none of which I followed, which is almost the same thing.)

On the other hand, while there was some luck involved, a lot of it was just trying to figure out how to do well wherever I found myself and looking for more opportunities.

Now, I do think, at the risk of sounding like a jerk, that there's a minority percentage of people who are good at academics, and can enjoy and thrive going to college. I think in general those people should be encouraged to go to college (unless they don't want to), and that the same skills that allow them to excel at academics will generally allow them to excel at careers, unless they have other issues or insist on following unremunerative paths. People who don't particularly thrive on academics should not be pushed to go to college on the theory that "it will help you", because those people typically won't be helped much by college anyway.

I keep meaning to write up some lengthy thoughts on this series, but I haven't read all the posts yet, so I keep putting it off. (The novel thing has been sucking up most writing time.) This one I couldn't resist because it was relatively short and included graphs.


Yeah, and different estimates serve different purposes. For the purposes of doing your monthly budget, saving 1000-2000 per year is pretty significant. However, when you're trying to calculate "what is your annual income minus loan payments" that might be the different betwee, say 57k and 56k, and that's not going to be the big thing that determines whether it was a good idea to go to college.

But you're right, people often have economic instincts that don't play out all that well in the given scope and situation. I catch myself on that kind of thing all the time. The trick is just to think to catch yourself on it before someone else does.


A difference in $1,000 annual income might well be the final deciding factor. Suppose I really want to do X, but I am being told that the less desired path Y is much more lucrative. I run the numbers and discover that the expectation value of the difference is only $1,000 per year. I decide that the loss of $1,000 per year is a reasonable price to pay in order to choose my preferred path. I choose X with a clear conscience.


It seems to me that the difference between Darwin and many humanities majors is that he wasn't cemented to the idea of having a career exactly like his college degree. I don't want to put words in your mouth, Darwin, so correct me if I am wrong. He went to college, had the whole experience, expanded his academic horizons, and then went about creating a career with an eye to his interests, but not holding out for a specific job. Many in the liberal arts areas buy the college line so completely, they cannot imagine working outside of their highly specialized and limited field. And employers take advantage of this glut to limit their hires to very specific degree requirements.

Two cases:
1)My brother majored in sports journalism. Yes you read that correctly. He spent four and half years and thousands of dollars for a specialized degree that pays a starting salary of less than ten dollars an hour.

When he graduated, he and my parents were shocked that not only were there limited job opportunities (1-2 per town), but the pay would be under 20K and probably never reach 40K. They never investigated for themselves and trusted that the college would never offer a degree that had such miserable outcomes. The expected wage in this career was never mentioned while he was in school and my brother did not truly find out what his prospects were until he actively started looking for a job.

I think at his first job he made $8.50 an hour. There is blame to go around: the college for not disclosing, my brother for not investigating, and my parents for trusting blindly. But the truly perverse thing about his situation was the employers offering jobs would not hire non-journalism majors. So these newspapers require four year degrees for jobs that pay slightly above the minimum wage. Obscene, really. He worked for a few years, figured out he was going nowhere fast, and went back to school to get an MBA. He works in HR now.

2) My cousin majored in English education and then went to graduate school for library science. She wants to work in a library. Period. Other jobs are not considered. The financial rewards are somewhat better than journalism, but the job openings are extremely, extremely limited in a traditional library setting.

She cannot get hired. She lives with her parents. She is 30 years old. Her response to her unfortunate situation is more graduate school. But she is hard-headed enough (and coddled by her parents enough) to hold out for her library job for as long as it takes racking up graduate degrees along the way.

I don't know if she is paying for all this through loans or if her parents are picking up the tab. She has drunk the 'do what you love' kool-aid down to the last drop. She honestly believes she should be able to major in whatever she chooses and have a well-paying job doing that exact thing waiting on her. Just 'cause.

So I think college is a bad deal for a lot of people, but I don't how to transition out of a job market that requires it for most non-minimum wage jobs.


I think the fact that many college graduates have to settle for work outside their fields is obscured -- maybe sometimes deliberately.


Part of my point here is that although college is a good deal for a majority of students in terms of return on investment, it is a bad deal for the country because we spend so much money teaching the other 40% skills they are unlikely to use. Retail trade is still the largest employer of college grads. We ought to be able to train people to interact with customers, track inventory, and put together a work schedule for far less than a four year degree costs. Even for the students who get a good return on investment, they still paid too much. Many high paying employers (e.g. the consulting firms) don't care what you know they just want smart people they can train and who will work hard. Colleges are sorting machines that make it easier for these companies to find smart people. We ought to be able to sort students into top, middle and bottom for less than $88,000.


Mark, I just finished reading all of the paper that you cite, and I'm a bit curious about your use of it. You say that the real earnings boost from going to college is only about 20%, however, the paper cites a number of different returns on education, and it kind of looks like you picked the lowest one.

Ashenfelter and Zimmerman (1997) and Altonji and Dunn (1996) give results of roughly 5% increased wages per year of schooling based on studies of siblings (thus 20% for four years).

However, the identical twin studies they cite in the next sections give return on education percentages of 8-10% (which would be 32-40% for a four year college degree).

The instrumental variables regression method suggests 8-10% per year as well. (Thus, again 32-40% total.)

Analysis by ethnic group suggested roughly a 10% return on education (40% total.)

Analysis by parental education suggested anywhere from 6-14% return on education.

And analysis by family economic bracket suggested returns on education from 6-16% (with the highest return for those from middle income background.)

(Also, interestingly, this paper suggests the opposite of the Academically Adrift study: that intelligence does not have any impact on the effect of additional education on income. I'm not sure right off how to judge between those.)

Obviously, in calculating ROI, it makes a big difference whether we put the impact on income from a college education at 20% or 40% or even higher if we accept the middle income number of 16% per year: 64% total difference in income resulting from a four year college education.

What caused you to pick the 5% figure from the basic sibling study over the ones based on identical twins (thus controlling for genetics) or on various socio-economic or demographic characteristics?


Also, I'd like to understand a bit better your lifetime earnings calculation. What did you use as the per year earnings for each example (college and high school) and how did you arrive at that number.

Obviously, given your assumptions, we wouldn't want to use average earning for either high school graduates or college graduates. Did you get data in average income by age and then deviate up and down from that number in order to get the 20% spread between the college and high school grad?


The 20% comes from direct comparisons between the earnings of siblings who have varying levels of education. By comparing siblings we reduce the effects of the family's wealth on earnings. I admit to ignoring the data for identical twins on the grounds that the data sets are smaller, although these pairs might also eliminate genetic variation from the earnings results. For each year of post high school college completed, a sibling can expect an earnings increase of slightly over 4%. Compounded for 4 years we get a total increase in earnings of 20%. the comparisons are between siblings so if one makes $10 per hour we expect the other to make $12. If one makes $100 per hour we expect the other to make $120. Certainly there is wide variation, but the average difference is 20% more for the 4 year degree over the HS diploma. The key is that even wealthy college grads are still only making slightly more than their non college educated siblings. This is a fact the universities will not want to spread.

For the first part of the explanation of my simulation, I am talking about dollars that are always inflation adjusted to 2012 value. I discuss later how inflation shifts things to the better for borrowers. There are a number of things I did in the model. First, I used a wage of $36,000 for the non college educated sibling and 20% more for the 4 year degree. $36k is the median wage for individuals in the US. One might argue that I should use $47k and $56k instead since $56k is the median wage for workers with a college degree. Higher wage assumptions do slightly favor the college grad in this simulation, but $56k is still not enough to get the college grad ahead if they take 6 years to get the degree as most folks do.

Wages typically are higher for older workers that younger. So if the average wage is $36k you might expect to make only $30k when you start and $42k by the time you retire. I could not find perfect data for this but census data on household income appeared mostly to support an assumption of a linear rise in income starting at 66% of the average wage and rising to 133% of the average wage by age 65. The model is not sensitive to the the slope of earnings over time. Going from 50% earnings at 18 to 150% at retirement or going from 75% at 18 to 125% at retirement only shifts the crossing points for the lifetime earnings plots by 1 year. This is a minor difference.

I assume a fixed rate loan for the borrower. Since everyone's wages are assumed to rise with inflation, the net result is that over time the loan payments are a smaller fraction of the college grad's income. This is a case where inflation is good for the college grad.

So borrowing for college is a better deal if the following are true:
-you graduate in 4 years
-you pick a high paying degree
-you get a fixed rate loan and then inflation skyrockets

College looks pretty bad if:
-you take the typical 6 years to graduate
-your degree pays less than median
-inflation is low relative to you loan rate.

The thing that I conclude from the sibling pair data is that if you are the kind of person who would go to college, graduate on time and get a high paying degree, you are also the kind of person who would have done nicely as a small business owner, tradesman, or salesman without a college degree.


I agree that sibling pair data or other data that controls for differences in socioeconimic and family circumstances is a pretty decent way to try to look at the real impact of college on earnings. However, all the studies summarized in the paper you cite are controlled for those factors in one way or another, and you seem to have cherry-picked the very lowest figure. That's helpful in getting an attention-grabbing headline, but it's not necessarily a good way to do analysis.

Now, some of the control methods use in these studies are complex, and I can appreciate that the simple, uncorrected sibling data is probably the easiest to understand. However, if we want a simple, easily derived number it seems like we'd lean more towards the identical twin data (8-10% impact per year of education) or if we're concerned about the small data set, run an average between the two.

Another factor we need to consider is that these studies are about 15 years old. However, as the paper points out, the impact of education on earnings has been growing over time. So in addition to either using the identical twin number or averaging between the two studies, we'd probably also want to trend those numbers in order to bring them up to date.

By the time we finish dealing with all these factors, it's probably better to assume a 30-40% difference between college and high school grads for our analysis.

For wages by age, this seems to have a pretty good breakdown:


It looks like the high school grad could expect to make around 22,300 during the years his sibling is in college. This takes the lifetime earnings of those four years down from 144,000 to 89,200, a difference of 54,800, which means its going to be easier for the college grad to "catch up".

I actually don't think that lifetime earnings is a very good metric here. I think people care a lot more about their current earnings and assets than they do about their lifetime earnings. But that's an argument for a later time.


Okay, I think I've successfully built a model that mimics yours now, so that I can play with it. The differences are:

- I used 2% inflation (interest rates currently include a very low inflation assumption, so if inflation went up, we'd see interest rates go up)
- I assumed a round 100,000 in college debt (my alma mater currently charges $25k/yr all in)
- I used the age adjusted hourly wages from the BLS
- I ran both 20% and 30% college factor scenarios.

What I'm getting is that with 20% college factor, and the BLS age adjusted wages, the college grad exceeds the HS grad in lifetime earnings at age 47 (that's removing loan payments of 9300 per year from the college grad's wages for the first 20 years out of college). However, at the more experiential level, the college grad will exceed the HS grad in net annual earnings at age 29.

With a 30% college factor, the college grad will exceed the high school grad in net earnings at age 39, but he'll exceed the HS grad in net annual earnings at 25.

I'd argue that the age at which the college grad exceeds the high school grad in earnings net of loan payments is probably a much better measure of the quality of life improvement that comes from going to college, and it looks like even with the 20% factor, we see that before age 30.


So, Darwin and Mark, obviously there is some disagreement here -- but is there some consensus that you can come to about the windows in parametric space within which it is likely a financial net gain to go to college and within which it is likely a financial net loss to go to college?

I think everyone here agrees that college is not a panacea, but i am interested in delineating a little more precisely whom it benefits and whom it harms.


One question: Are you having your graduate get their degree in 4 years or 6? This makes a big difference.

I did throw out all twin data for small sample size and very large variation between studies. For all the sibling data I took the simple mean of the sibling data reported in the table. I think that was fair although the better route would be to weight all the studies by the number of data points in each study but I was too lazy to dig for that. Also it doesn't make sense to me that the twins could have a bigger boosts. The siblings should have genetic variance plus education variance effecting earnings. Since all the genetic variance would be confounded with the education difference, the earnings boost for college should look artificially high for siblings not the twins.

All that said, to Erin's general point on gathering consensus, perhaps we would agree that the models are highly sensitive to the estimate of earning boost and to the time to graduation. Also, would we agree that the within group variation in earnings is as big or bigger than the between group variation (even using the twin data figures for the earnings boost)? If we agreed on that, then I think it would follow that a prudent selection of degree or trade is as important as the college/ not-college decision.


Well, keep in mind, that the genetic differences being lumped in won't necessarily be ones that result in the college educated person making more relative to the other. Say you have a non-identical pair of siblings and one is very well suited for working with his hands, and the other is very well suited for a "knowledge worker" job. The one goes to college, the other doesn't. Both of them would actually be better off (and the gap between them might be less) than if either had been forced into the mold of the other. The nice thing about the identical twin data is that it zeros in on the effect of college all other things being equal. That's why identical twin studies are kind of gold when it comes to social studies, it eliminates so many other elements that it's really hard to control for otherwise.

I had been working with four years, but I've now re-done with six. The really brutal thing, of course, is taking total borrowing up to 150k and missing out on two more years worth of inflation. Even so, assuming the 20% college factor, the college grad "catches up" in net earning after loan payments at age 41 -- before his loans are paid off. Once his loans are paid off, (age 44 for both) the HS grad is making $75k (with inflation) and the college grad is making $90k. Even though the college grad won't catch up in lifetime earnings until age 55, I think just about anyone would go for that math.

That said, one point on which we could certainly agree is that taking more than four years to get a bachelors degree is a bad, bad thing. (Since big state schools are the prime offenders in regards to making it impossible to graduate in four years due to course scheduling, I'd strongly recommend people take that into account when looking at which college to go to. The more expensive private school may make more sense if you can get your degree on time.)

Another probably point of agreement is that having to borrow the money for college (i.e. no scholarships) is financially pretty devastating. You're looking at paying $750-$1000/mo for twenty years, depending on how long you spent in college.

I think the most basic thing to consider is: Do you want to go to college? (This is in part because I don't tend to see college just as a salary generator, though I do think the data is pretty strong that it's good at that.) If you don't really want to go to college or don't know what to study, just taking a job and seeing how you like the HS grad workforce experience is probably the best call. I don't support going to college just on spec. While the data in the cited paper seems to support the claim that on average just about anyone capable of finishing a degree is likely to be better off having gone to college, it such an expensive purchase these days that I think people are wise to think about whether or not they really want it and have the drive to do it well, get scholarships, etc. After all, though on average just about anyone capable of graduating is better off having gone to college -- the average is necessarily a distribution. There are going to be a lot of people at the lower margin for whom going to college turns out to result in a lot of debt and still not much earnings. Those same people might earn even less without the college degree, but at least they wouldn't be crushed under the debt.

Christy P.

I teach graduate students exclusively.


I thought so, CP - thanks for the reminder.

The comments to this entry are closed.

Screen Shot 2015-07-19 at 6.07.09 PM
My Photo

I think I read something somewhere about this

  • Google

    bearing blog


Become a Fan