Bryan Caplan  

Major Premium

PRINT
David Stockman Screeches... Carden and Horwitz on Market F...
Economists usually talk about the college premium, but the college premium heavily depends on your major.  At the same time, though, stronger students typically choose harder - and more lucrative - majors.  Thus, the college premium is doubly infected by ability bias: People who would have made more money anyway are more likely to go to college, and college graduates who would have made more money anyway are more likely to select demanding majors. 

To weigh whether a particular major is better than skipping college entirely, you need to adjust for all these factors, then see whether the adjusted college premium outweighs your full cost of college.  As far as I have found, however, no researcher has published such figures.  Don't get me wrong: Existing research provides all the numbers necessary to make this calculation.  But you have to snap all the pieces together yourself. 

Here's how in six tolerably difficult steps:

Step 1: Get the average college premium. (+83% on the latest Census data).

Step 2: Adjust the average college premium for ability bias. (leaving a +46% "reasonable estimate" based on my reading of existing literature that controls for measured ability).

Step 3: Get major-specific premia (usually calculated relative to the education major, roughly the lowest-earning major). (The best source I've found is Altonji, Blom, and Meghir 2012).

Step 4: Adjust these premia for ability bias. (50% ability bias is a standard estimate in this literature, which routinely controls for pre-existing ability).*

Step 5: Get the ability-adjusted premium of the average-earning major (probably business). (Just apply Step 4 to Step 3).

Step 6: The major-specific college premium relative to high school alone then equals: Step 2 * (1+Step 4)/(1+Step 5).

Here are the final results for common majors:

Earnings Compared to H.S. Grads

Major

Males

Females

Electrical engineering

+63%

+72%

Computer Science

+61%

+63%

Mechanical engineering

+61%

+72%

Finance

+61%

+55%

Economics

+60%

+59%

Accounting

+53%

+53%

Mathematics

+53%

+50%

Nursing

+52%

+59%

Chemistry

+48%

+47%

General business

+46%

+46%

Political science/gov't

+46%

+47%

Biology

+44%

+43%

Communications

+37%

+45%

History

+35%

+37%

Sociology

+35%

+36%

Liberal arts

+34%

+36%

English language/lit.

+34%

+37%

Anthro./archaeology

+32%

+36%

Fine Arts

+25%

+29%

General Education

+24%

+30%

Needless to say, these major premia lack the precision of Planck's constant.  Compared to what the typical 12th-grader vaguely thinks, though, this table is good as gold. 

STEM fans will revel in the top three rows.  To avoid confirmation bias, though, notice that the payoff for English, history, and communications is only modestly lower than the payoff for biology, chemistry, and math.

The real winner in this table, I do declare, is economics.  I've long told my undergrads, "Economics is the highest-paid of all the easy majors."  Still, the magnitudes surprised me.  After adjusting for pre-existing ability, male econ majors (+60%) are almost as well-paid as male electrical engineers (+63%).  Given engineers' workload, they definitely seem to get the short end of the stick. 

When I was an undergrad at UC Berkeley, a popular campus t-shirt read, "As the limit of GDP --> 0, GOTO poli sci."  Given current major premia, I'm tempted to make a t-shirt that says, "Do not stop in STEM.  Go directly to econ."

* 50% ability bias means a 50% decline in the coefficients measured in log-dollars.  Some major premia are so large that the log dollar=% change approximation breaks down, so be careful.


Comments and Sharing





COMMENTS (21 to date)
Bryan Willman writes:

I don't know about any other majors, and my assessment is dated, but inside computer science there is another issue. (I stopped having any contact with this about 2003)

In CS the ratio of skills between the worst and the best boggles the mind. So there are people with CS degrees who understand computers and software in powerful ways. There are also people with CS degrees who don't (I was always mystified interviewing them.)

And that spread can skew compensation - a CS person who is a "real winner" can end up working for some "hot" org where they literally win the lottery. Whereas those on the bottom, well, they end up doing something else.

So it may (!? may !?) be that the returns to a particular major are uneven.
(Whereas the disreturns to not finishing high school seem consistently high.)

Noah Yetter writes:

I am going to both endorse, and attack this proposal.

The endorsement is simple, I was an econ major and I have done quite well. I work as a software engineer and earn quite a nice living at it. I feel that an education in economics (and statistics!) is rather useful in this field, more than one might guess. This is particularly true for me since I do a lot of work with data analytics.

My decision to major in econ even followed similar lines to Bryan's argument. It was 1999 and the first internet bubble was peaking (though no one knew it just then), and it seemed anyone with a pulse and a modest grasp of building computer systems could land a good job. So instead of continuing with my initially-declared Information Systems major -- that program having turned out to be mediocre relative to the skills I already had -- and decided to spend my college years studying something interesting.

The attack is a bit subtler.

First, there's the obvious disadvantage you may have already guessed: I began my career at a disadvantage compared to Computer Science or Information Systems majors. After a few years in the industry one's education becomes largely irrelevant, but at the beginning it matters quite a bit. I would have had a hell of a time finding even an entry-level job on the open market. I lucked out and got a job through a friend. The irony of my decision to study something interesting in light of the hot tech market is that by the time I was looking, the tech job market was the weakest it's ever been. After a one-semester experiment with grad school (where I met Bryan, though I doubt he remembers), I took an entry level job at an embarrassing salary of $34k/yr.

And even though education becomes less relevant the more years you spend actually doing this stuff, there's still a space of skills I'm simply not familiar with having not studied CS. I could catch up now in my spare time, but frankly it's not just interesting enough to hold my attention.

Second, well, let me just link to this:
http://josephwalla.com/attrition-is-a-group-statistic-it-dont-apply-to-me

Statistics like what Bryan presents above are interesting, but not necessarily relevant to an individual in the position of choosing a major. Given sufficient introspection, you the student know what you find interesting, and perhaps whether you need/want the boost of a relevant major rather than just any old degree.

Continuing in the vein of my own experience, a bright student interested in writing software professionally could certainly choose a relevant major like CS or IS, or a mostly-irrelevant major like Econ or anything else on this list, and do well afterwards. What's likely though is that taking CS or IS classes and buddying with other CS or IS majors will give that student more opportunity to build and refine their skills, and acculturate to the norms of the field. It's unlikely that taking an "easier" major and building software in one's spare time would be as rewarding or engaging. If one were to survey entry-level software industry hires regarding their choice of major, I bet you would find very, very few who deliberately chose a non-relevant major while holding the intent of entering the industry after graduation. This is partially a belief that a relevant major is necessary or advantageous (a notion which Bryan is attempting to dispel), but also simple self-selection. I would further bet that in light of the above data very few aspiring software engineers would change majors.

James writes:

There's also the fact that coding jobs, even very high-paying software engineer positions, are much more likely to not require a degree than those in e.g. electrical engineering. It's pretty simple to actually demonstrate basic programming proficiency in the hiring process, and as Bryan Willman notes, having a CS degree is curiously only slightly correlated with knowing how to program computers anyway. I'm not sure how that should influence this analysis...

David Adsit writes:

Suddenly, I am curious what percentage of serious software developers are hobbyist economists.

FWIW, I left school instead of finishing a CS degree because the job market in the late 90s was so hot I couldn't afford to stay in school anymore.

Brian Moore writes:

I guess the comp sci majors are up early on April Fool's -- not too surprising, they have lots of content to check out/write. But I will add my 2 bits: if you truly want to examine the additional premium of the actual sheepskin, CS is an amazing place to start, and it's actually very easy to assess the actual competence of a programmer (though truly being a computer science person is slightly different) with easily designed, automated tasks.

And as they say above, there are CS majors with no clue whatsoever, and ones that are quite good, and then many, many others with no formal programming/CS education at all who are quite good as well. But that competence variable usually has little to do with their education, and should be easy to disentangle for new grads.

Bostonian writes:

The most prestigious universities offer degrees in economics, but many do not offer degrees in "general business". How does this affect the results? I think part of the reason business majors earn more more money than liberal arts majors is that the former group is more financially motivated. At Harvard the financially motivated cannot major in business, so they major in economics.

dave smith writes:

@Brian Williams, the same is true with Economics majors. And I'd bet with every other major as well.

Hazel Meade writes:

I work as a software engineer and earn quite a nice living at it. I feel that an education in economics (and statistics!) is rather useful in this field, more than one might guess. This is particularly true for me since I do a lot of work with data analytics.

In my experience, any major where you acquire a firm grasp of mathematics, such as calculus and/or statistics, prepares you to cross over into any number of other profitable occupations.
That's why you frequently hear of physicists crossing over into financial analysis, economists becoming software engineers, even people who do statistical analysis in the social sciences moving over into programming or data analysis. Math is universal.

I'd wager that you could divide up college grads between those who took a few calculus and statistics courses and those who didn't and find a sharp wage differential later in life.

It would be wonderful if we could teach calculus in high school and spare a lot of kids the need to get an expensive four year degree just to get the bulk of the profitable knowledge (minus the signalling aspect, albeit).

Someone from the other side writes:

@Bostonian: there is that but in my experience,the Econ crowd genuinely is smarter on average.

I did my MA in Econ/Business in a top 10 European department (it ended up being overweighted towards finance and micro, I could never stand macro and business always bored me out of my mind) that has business and econ largely intermixed in the undergrad space and the crowd that went for a econ (or finance, for that matter) genuinely was significantly smarter than the people who stuck with pure business stuff.

I later on added a top tier MBA to my CV (technically, I knew it was bad idea when I signed up but I had non diploma related reasons to do so plus I basically got, opportunity cost aside, a free ride) and now the top crowd was mostly former engineers with a few rare econ MA's thrown in. Business undergrads were decent but usually nowhere near the top.

Having said that, some of the top business undergrads I have come across (being in consulting, I see no small amount of those) are every bit as smart as the econ crowd (and probably also as smart as the top STEM albeit with less quant skills) but generally have a quite different personality - more extroverted, more aggressive in going out and getting things at the price of sometimes shooting from the hips.

As for computer science, that was my initial preference but after half a term of analysis I decided I don't like math nearly enough to be bothered to do that (plus that was at the through of the dotcom bust). Looking back, I never once truly regretted doing econ instead.

Brian Moore writes:

David Adsit :

Suddenly, I am curious what percentage of serious software developers are hobbyist economists.

Lots! :)

Skeptic writes:

Here is one important caveat for all you young people using this information to decide your future path: these studies are always retrospective. Using this data, your parents, your professors, and I are able to tell you exactly the right positions to take in 1980, 1990, 2000, and probably 2010. If you are expecting to live in 2020, 2030, and beyond, our power degrades the further one goes.
Is it possible that software coding will be done cheaply in India in a few years? Can most accounting be outsourced to overseas shops? Do you have any natural advantage over a Chinese engineers? Do you have a natural advantage over a Chinese person in English studies?
Past perforance is no guarantee of future results.

Dan Carroll writes:

Curious as to the dispersion of earnings within each discipline and the rate of decay over time.

Also, while I agree that econ is relatively easy, when I was in school, I found that I was somewhat unique in that assessment. Many of my classmates found econ to be quite alien to their thought process. Perhaps that could partially explain the higher than expected earning rate. Also, econ majors often go into finance. Indeed, econ PhD's often call themselves finance professors in order to get a raise.

Bryan Willman writes:

A note on the value of formal CS, and a revision to Hazel Meade's proposal:

1. It is indeed not too hard to weed out people who cannot code an algorithm. The value of a "deep CS background" is the ability to think about the structure of systems, how software interacts with hardware, how it interacts with users, etc. The ability to devise new algorithms or invent really new things is different yet again.

By the way - this same thinking seems (to me) to be helpful in projects that aren't about CS - somehow the processes that let you think about millions of lines of code help you think about factories, logistics, and so on.

2. @Hazel Meade - I think the actual list might be: formal logic, expression of ideas in formal languages (including math), calculus (even if poor at it), statistics (even basic), and at least some observed interest (passing a course) in observed or experimental science. Uh, maybe add some course where you "build something".

Bryan Willman writes:

Here's another question that might illuminate the returns question - and it probably involves natural ability, work habits, and human capital development all mixed together.

Does the major have some "horrible challenge course" and did the graduate survive it?

Often these courses have modest names. "Compiler construction" or "graphics seminar". How hard can they be? Hard. Very hard. I think for EE's it was some course called "individual device assembly" which at least at one school meant you had to build a computer and make it work. Hard.

Very few people write compilers, but having that class of ability comes in handy all sorts of places.

(But in this and the prior post I am arguing for transferability of learning. Others suggest that doesn't apply much in the real world. Am I that weird???)

Finch writes:

> Curious as to the dispersion of earnings within
> each discipline and the rate of decay over time.

+1

I have the feeling there's more variance in econ major outcomes than engineering. Also, the census data Bryan quoted is a bit deceptive since it uses averages and time value of money is important here.

Engineering starting salaries are quite high. Econ majors don't really begin to earn until they get their MBA. Getting an MBA is a risky bet - it may work out really well and it may saddle you with a lot of debt and a job not really that different from what you had before.

I'd encourage a child to pursue economics, particularly if the child didn't seem smart or driven enough to make it through engineering. But there's more to it than choosing a major, and econ is a little harder to turn into a career.

And to address some of the other comments, econ would be a lot better if it had a lab in the engineering sense. Some sort of requirement that people who got the degree had really beaten a substantial project into working. I'd preferentially hire those people who'd taken the econ equivalent of 6.111. I say this partially because most econ majors are really employed as a sort of subject-matter-trained programmer anyway. It's not like they spend their time theorizing.

Michael_M writes:

Great and thought-provoking data! The breakdown between females and males in the list defies easy explanation, so perhaps a "Step 7" is required: correct for time elapsed since graduation. The US Census data groups together all earners 18 years and older and a brief glance at the Altonji et al paper suggests it looks at lifetime earnings only. This will distort interpretations of the data, as men and women will likely spend different numbers of years in a particular profession after graduation, and seniority (years in grade) may inflate some of the numbers here.

This also may change the perception of the relative value of the majors by the students themselves, based on their discount on future earnings: if the poli. sci. premium, let's say, is 46/47%, but I need to spend 10 years to realize that, perhaps nursing is a better option, as I might be able to realize the maximum premium in my first year?

MingoV writes:
... the payoff for English, history, and communications is only modestly lower than the payoff for biology, chemistry, and math...

BS degrees in those three fields don't mean much. MS and PhD degrees in chemistry provide excellent opportunities for high-paying jobs. Mathematicians and general biologists with PhDs have good opportunities in academia and moderate opportunities in other fields. Those with specialist training (eg: cryptographic math or crop genetics) can do very well.

Terry Hulsey writes:

David Smith ends the discussion: As he correctly observes, there are those of greater ability and those of lesser ability in every field.

Here, as often happens in discussions of education, the topic wanders off into a discussion of who's got the smarts. But the stated theme of the article is who's got the bucks.

And on that score, my guess would be that the bucks are held by those middlin' folks having an acquaintance with business.

Glen S. McGhee writes:

The "through a friend" says it all -- where are the mediating social networks that make all the magic happen? Without the "through a friend," nothing necessarily happens, yet none of this is specified in the model.

steve writes:

Need to add doctor, lawyer, IT, and entrepreneur starting with a college degree's worth of capital to the list.

Gary S writes:

Econ majors are not subject to government immigration ('guest worker') programs that STEM majors are. This is a significant factor.

Comments for this entry have been closed
Return to top