Chris Blattman

PhD production as a process of self-discovery

There is an oversupply of PhDs. Although a doctorate is designed as training for a job in academia, the number of PhD positions is unrelated to the number of job openings.

Meanwhile, business leaders complain about shortages of high-level skills, suggesting PhDs are not teaching the right things. The fiercest critics compare research doctorates to Ponzi or pyramid schemes.

From The Economist.

Many of the arguments are valid. But make two plausible assumptions and you get a different answer.

Assumption 1: Innovation, including academic research, is the fundamental driver of long term health, wealth and happiness for the human race. (The “including academic research” bit is the biggest leap.)

Assumption 2: Unfortunately it’s very difficult to say beforehand who will and who will not produce great, or even good, research. (Even after five years departments have trouble predicting which of their crop will excel.)

In this world, each extra PhD raises the chances of one more brilliant, world-changing idea. While hardly comforting to the thousands who toil without job prospects, the collective benefits just might outweigh all the individual misery.

The decision might be individually rational as well, especially if students are no better at predicting their success than their advisors (they probably aren’t).

(A similar analogy comes from Lant Pritchett, who points out that you need a system that produces an enormous number of terrible dance recitals to get the handful of sublime performers. The same logic applies, he argues, to development projects and policies.)

One counterpoint: Here is where I would expect to see overconfidence bias lead to oversupply (and few of the collective benefits thereof). So maybe we need a system that gives the least promising an easier out that saves face.

20 Responses

  1. as someone who spent 6 years in their ph.d. program proving that they are smart enough and good enough but having to leave with a master’s instead of a ph.d. i’ll say that i agree with most parts of your assumptions. If I currently had my Ph.D. I would be unemployable but because I have a master’s I have a great job now. Lest we forget in many labs Ph.D.s are also consolation prizes! These individuals often have a much harder time. Not all theses are created equal and most theses have little to no application to a broad commercial sector with a glut of opportunity.

    I do have some problems with assumption 1. Most notably that to create real change for ordinary people, academic science must scale in order to translate. This is something that industry and government develop that academics have no interest in doing really; if they did they wouldn’t be an academic! Thus while academia can provide the fortuitous spark thanks to immense amounts of “disposable” income thrown at it, academia really doesn’t drive anything but self-perpetuation. It’s marketing and self-promotion at its best and worst with ungodly overinflated claims and goals from people who can’t function in normal society usually, let alone be drivers for change of that society. I would say most academics drive progress as much as plot drives the action in Family Guy. (read: very very loosely)

    Second, your second assumption is intellectually flawed. The reason why we can’t predict who will succeed and who will “fail” at academia is because the standards are not concrete. For every grad student there’s a constantly shifting landscape of success and failure that largely the grad student has no control over. This landscape is driven by departmental politics, advisors’ fluctuating egos, moods, and budgets (which is based on past decisions the grad student often has little to no knowledge of), and competition that you don’t know you’re competing against in many cases. Hence an inability to cherry pick which grad students will “succeed” at this perverse apprenticeship or “fail” is not surprising. It’s much easier to determine which professor is on the upswing or the downswing and what they need their minions to accomplish in order for their situation to improve or not accomplish for their position to worsen. While academic research in some fields is largely independent and isolated, that of the sciences is incredibly dependent on others while the system largely ignores this in its criteria for awarding degrees. Most successful science ph.d.s that i know did not work in isolation, most failures did. In most cases this isolation has little to do with the individual and most to do with the advisor’s priorities which are largely self-focused. Failure of a grad student should be viewed as a failure of a larger group rather than saying “oh this person couldn’t produce good work and can’t and won’t while this other person obviously can”. Just look around at “successful” phd students who can’t hack it in their post-doc for the proof that there’s no predictive ability because it’s not a judgment of the individual.

    As such the system that creates new ph.d.s and determines its own standards of success (which are meaningless outside of it). The least promising leave after 2 years as is while the promising stay until the very end. The only thing you can do to save more face is require departments to inform their students of life opportunities and value outside of academia.

  2. PhD programs are more likely to drain individuals of creativity than they are to inspire it. The higher education establishment has become incredibly top-heavy, relying on graduate students to do grunt work in classrooms and labs, while college administrations have ballooned into enormous white-collar job-creation programs.

    When all of these grad students (after losing years and years of potential earnings) walk onto the job market, there are simply no places for most them. Society pays a heavy price for this waste, but the burden is truly terrible for the individuals involved.

    Some of these issues are a addressed on the “100 reasons not to go to grad school” blog:

  3. I accept both assumptions, but not the conclusion.

    Science PhD students are, if I may over-generalize, mostly foreign students who need the PhD to immigrate here or get a good job back home. The American students are those who prefer making less money in Academia than toiling in the real world. These are not risk takers.

    Government grant applications are critiqued by committees, prioritized by political appointees, and managed by bureaucrats. Awards are mostly based on the Professor’s prestige, resume length, and the trendiness of the subject matter. These are backwards looking criteria. Individual academic research success is determined by tenure committees based on total grant money awarded. In sum these are not factors that tend to reward actual innovation.

    You are correct that professors can’t predict which students will excel at research. Neither can government bureaucrats predict which professor, which subject, which proposal, and which university will produce innovation. The private sector is better at this — because if they fail they go bankrupt.

    The Federal Government funds about ½ of all the USA’s Basic Research, ¼ of Applied Research, and 1/16 of Development ( So the more measurable and practical the research is, the more likely it is funded by a company. Conversely, the harder it is to predict the future value of research, the more likely it is that the government is funding it through an Academic institution. In the end research in Academic settings produces more research. Research in private settings produces things people can use. This argues that PhDs in academic settings produce less real-world innovation per $ than the private sector.

    Oh, and teamwork is not a core academic skill.

  4. “In this world, each extra PhD raises the chances of one more brilliant, world-changing idea.”

    Sounds like a sensible idea at first, except when you realize how many of the smartest students steer clear of PhD programs because they know about the low pay and poor job prospects. The Economist article mentions this when it says that the percentage of foreign-born PhD students in American programs has grown from 23% to 48%.

    Want to encourage the very most brilliant students to go into PhD programs? Reduce the oversupply, so that the wages and job prospects improve. Turning PhD programs into menial labor does not attract the very best candidates.

  5. Normally I am really impressed with the quality of comments, but here I feel that my argument has been misunderstood by most of the commenters. It may be my poor explanation, so let me clarify some points:

    – Mine is a simple logical argument with a premise following from two simple assumptions (that may or may not be true)

    – Essentially, if both assumptions are true, or partially true, then we need to accept the conclusion.

    – The argument may be more true for some disciplines than others. Disciplines with lower barriers to entry and high “consumption” or intrinsic value will fir more poorly.

    – None of the comments falsify the assumptions or undermine the logic. For instance, lots of innovation certainly comes from outside academia, but a lot of innovation does come from academia. In no way does innovation come from any one sector of the economy. Neither all nor even a majority of innovations need to come from academia for the logical argument to hold. The Bill Gates critique does not make any sense.

    – If I wanted to challenge this logical argument I would look for evidence that there is a collective opportunity cost of academics studying academic subjects rather than innovating in the real world.

    – My basic point is that the “oversupply” complaint is a pretty weak one, and doesn’t take into account the broader, collective manner in which innovation happens

    1. I’d argue that the biggest driver of human progress is not simply research or even innovation, but more importantly turning this into something that is practical, replicable and frequently commercial that will drive its widespread use. Yes, of course you need the research that this rests on but without the follow up from entrepreneurs, policy makers, governments, communicators citizens and consumers much of this research will lead to naught.

      You might therefore argue that supply of researchers is not the main bottleneck here but rather the ability of society to apply the knowledge gained for human progress. If you accept this “assumption” then maybe we need more entrepreneurs and others who strive to apply research, or maybe we need to reform the research establishment to encourage greater links between research an application in terms of the type of training we give researchers.

  6. “In this world, each extra PhD raises the chances of one more brilliant, world-changing idea”

    This statement is not backed by any actual data or research.

    In my experience every good idea has been thought of a 10 or 100 or a 1000 times before someone actually does the very hard work of making it reality. And in my world (Medical Technology) people with Bachelors and Masters Degrees produce just as many of them. But we don’t have to argue. This is a provable conjecture. I will wait eagerly for someone (A PhD candidate in the History of Science perhaps? Or would that be a conflict of interest?) count up and weigh the 20th century’s world changing ideas, and determine which ones would have never come about without a PhD or PhD program.

  7. PhDs in what? I mean, I’m a long-lapsed ABD who is almost certainly not going to complete her dissertation, but my own “brilliant, world-changing idea” had to do with how composers conceived their audience in a certain classical-music repertory, and I rather think that isn’t exactly what you meant. I think I influenced — very slightly — about half a dozen scholars in my own very small neck of the academic woods before I quit working on it, and I’d imagine that’s typical for liberal-arts PhDs, even the completed ones.

    There are only two real values to a liberal-arts PhD — I mean, a completed one, as opposed to doing the coursework and the study (which is always going to be valuable, even if only to the student). One is that if you want to teach at the collegiate level, you might have a chance of getting work. The other is that you’ve proved (to yourself, and to prospective employers outside academia) that you can take on a long-range project and see it through to its conclusion.

  8. Your argument rests way too much on what might be called the Lone Wolf theory of scientific discovery. Sometimes that happens, it used to happen a lot. But these days scientific advancement is the result of a team effort.

    And no, I don’t consider Facebook to be a scientific advancement.

  9. An apprenticeship system would be more effective. Departments could select doctoral candidates based on their teaching/research needs, and would only select new people when those needs are no longer being met. The job could pay a little more (say, $30K-$35K, instead of the $20-$25K that’s standard now), and doctoral candidates would face the same limitations (seven years from entrance to graduation) that exist today. Those candidates who proved to especially good teachers, or administrators, or research assistants could stay on in that capacity, which would allow the department to increase its overall productivity without having to hire expensive new tenure-track faculty.

    This kind of system should have two effects: 1.) reduce the number of people admitted into doctoral programs, and 2.) modestly increase their quality of life and create a stable, if not glamorous alternative to tenure-track.

  10. I think a three to five year related work experience requirement might make the PhDs a bit more useful to world (while also probably contributing to less frustrations during the research process itself?)

  11. “So maybe we need a system that gives the least promising an easier out that saves face.”

    True. I think that North America’s biggest problems right now are 10% unemployment and dismal finance possibilities for small business and startups (collapse of housing prices have cat back the value of collateral that anybody can offer on a bank loan, for example.) It’s interesting to see the strange things Research In Motion did before they came up with a killer product.

    There’s a consumerist angle too in the idea that a Ph.D. should be a product with a known price and known benefits — an Industrial idea rather than a Renaissance one. Life, unfortunately or not, is not a product.

  12. This seems highly dependent on the field you’re talking about. I’m a stats phd candidate. you can argue about the marginal (in economic terms, not stats terms) value of another dev econ phd, and though I’ve worked with dev economists I don’t have a super well informed view on that, but I know for a fact that there’s an undersupply of good statisticians.

    1. I have a degree in psychometrics (i.e., psychological statistics) from one of the best programs in the country, now teach PhD students and will attest to two things:

      (1) Amen brother to the fact that there’s a massive shortage of competent statisticians. This means job security for me in some respects, but is also hard to deal with as well—overwork, lack of understanding of what I do from potential colleagues, etc.

      (2) A master’s and a PhD are far from the same thing. For some students’ goals, a master’s would be much better, but many students don’t know that. A good advisor (I like to think I am one, but of course nobody’s perfect) will help a student figure out what her goals are.

      The real oversupply of PhDs tends to be in areas where there isn’t a private sector and universities have serious service course loads.

  13. yeah, this one has been doing the rounds – its possibly the worst article I’ve read in the economist. The bit I liked the least was this statement discussing a “study in the Journal of Higher Education Policy and Management by Bernard Casey”:

    “The earnings premium for a PhD is 26%. But the premium for a master’s degree, which can be accomplished in as little as one year, is almost as high, at 23%.”

    great, an interesting (partial) correlation between wages and dummies for different levels of education. but nothing more than a correlation, hence meaningless in trying to figure out the financial returns to the different degrees. what about the endogenous selection of people into PhDs? (come on the economist; apples with apples). and what about compensating wage differentials for the different type of jobs these different type of people select into after they finish? (this time the jobs are fruit)

    I think a prerequisite for writing a piece on economics in the economist should be an economics degree.

  14. There’s something about this post that strikes me as self-important, verging on elitist. Yes, innovation is critical, but the academic part is a HUGE leap. There is a distressing amount of academic work that 1) rarely sees the light of day and is merely a means to secure life time employment, and 2) is sub-standard. Similarly, there is an amazing number of brilliant folks that never set foot in these hallowed halls. I’m not sure the presence of more Ph.Ds really correlates (or causes, as seems to be the suggestion) to more advances. It could also be said that it merely leads to more navel-gazing. And then there is the tendency to produce safe research that fits the vision of the old guard of a tenure committee. I’ve seen a lot of academic work (particularly Ph.D. dissertations) that are just derivative. I personally learned far more from my Master’s program than my Ph.D. The more innovative development practitioners I’ve met rarely hold terminal degrees.

  15. I do wonder what an econ PhD might look like if it’s objective was to prepare for being useful outside academia. For instance, investors want to be predict what’s going to happen to variables like exchange rates, output growth, inflation across countries, but macro PhDs don’t usually train you to know how to monitor economic indicators across countries and forecast these things. But one can imagine some theoretical and empirical work that might help.

    Perhaps development economics research is more applicable in policy work?

    by the way, self discovery would be a fine thing, but speaking as a soon to obtain doctorate (touch wood) individual, I don’t feel like I have a good idea of my own prospects. There is a joke that goes something along the line of “how many PhD grads does it take to screw in a light bulb? none – once the bulb they screwed in with their supervisor’s help has gone out, they’re left in the dark” which feeling worryingly apt

  16. What about masters degrees? I thought PhD programs were giving them as “consolation prizes” in recent years for those who decide to leave early. In fact, I have heard of some people signing up for PhD programs with the hidden intent of pursuing a funded masters. Since few reputable PhD-granting departments continue to support traditional masters programs, prospective students often must jump directly from the bachelors to the PhD level. These factors probably contribute to increased levels of attrition in PhD programs.

Why We Fight - Book Cover
Subscribe to Blog