The Barriers to Entry to the Economics Profession
“We received applications from many more qualified applicants than we have space to enroll.”
This idea, or some variation, should be familiar to anyone who has applied to graduate school, and it is certainly true in the case of applications to most, if not all, Economics PhD programs. MIT’s Economics department lists on their website: “The Department receives, on average, about 800 applications each year. About 40 students are admitted, and 20–22 matriculate.” That’s around a 5% admissions rate. This is less than the undergraduate admission rate of every college in 2020, except for Harvard University which narrowly edges it out at 4.9%. Admission to Business School’s Economics PhD programs can be even more competitive; Columba GSB posts Fall 2020 enrollment in their Finance and Economics program of 4 students which likely represents about a 50% yield on their admitted students for an estimated 8 admitted students… out of 300 applicants, a 2.6% admissions rate.
If it is not already clear, these admission rates suggest very high levels of competition in the market to be admitted to one of these programs. The level of success should be considered even more dismal than these numbers suggest as most of the students admitted to these top programs are likely being offered admission from several schools, which will inflate these admission rates relative to the actually successful pool of applicants at a market level.
With such a high level of competition it may be worthwhile to ask, what are the defining competitive elements in this market? In other words, what are admissions committees trying to assess in their review and selection of applicants? Now, this has a number of potential answers including: the promise of impactful research, the ability to expand existing research areas in novel directions, or even something as simple as the ability to complete the program and do the work. I think the precise answer to this question may vary from discipline to discipline, e.g. an admissions committee for a Biology PhD program may have a different goal relative to a Psychology PhD program, and perhaps even vary between admissions committees within the same discipline. For Economics programs in particular, I think the underlying selection criteria is to maximize academic placements of graduating students, specifically with regard to department rankings.
Understanding the Selection Criteria for Economics Admissions Committees
Let’s first understand what the PhD degree is. A PhD program is a research apprenticeship. In short, this means that the program takes in an aspiring researcher, invests its time and money into training that individual, and then expects that training to generate some return to the program. An optimistic person might say that the return is the generation of knowledge and the advancement of the field, but an Economist knows better; private value is what motivates behaviors and decisions, not the public good (generally speaking).
Private value varies from discipline to discipline. In physical sciences like Chemistry that return is pretty tangible, usually in the form of very cheap high-skill labor for lab work (no offense to Chemistry). In Economics however, it seems to be in the form of influence. Economics has a unique feature in the academy, at least among mathematical and social sciences, in that it is very top heavy, by which I mean that the “top departments” have outsize influence on the profession.
As an introductory example, consider research by Marion Fourcade, Etienne Ollion, and Yann Algan (FOA) in their paper “The Superiority of Economists” which shows the composition of the Executive Councils of the three primary disciplinary organizations for Economics, Political Science, and Sociology; respectively the American Economic Association (AEA), the American Political Science Association (APSA), and the American Sociological Association (ASA). The Top 5 departments in Economics constitute over 70% of the Executive Council between 2010–2014, and departments ranked below 20 do not even make an appearance. This contrasts with both the APSA and the ASA which have much more even spreads of Executive influence. This means that the views of Economists at the Top 5 Departments have dramatically outsized influence on determining what is important in the economics profession, the direction of the field, and even what behavior is acceptable in the profession. For a numeric representation of that influence, the AEA notes that it “now attracts 20,000+ members from academe, business, government, and consulting groups within diverse disciplines from multi-cultural backgrounds.” 5 departments/schools represent the majority of interests of over 20,000 people. That sure sounds like a lot of influence to me. What’s more is that the top heavy characteristic of Economics even has influence on who advances through the profession.
“Publish or perish” is a common refrain in the academic world. For those who are unfamiliar, it is “an aphorism describing the pressure to publish academic work in order to succeed in an academic career.” This idea has a unique realization in Economics: publish in a top 5 journal or perish.
In “Publishing and Promotion in Economics: The Tyranny of the Top Five” James Heckman and Sidharth Moktan (H&M) explore the relationship between publishing in the Top 5 economics journals and progressing to tenure as an academic Economist. These journals are:
- The American Economic Review (AER) administered by the AEA
- Econometrica (ECMA) administered by the Econometric Society
- The Journal of Political Economy (JPE) administered by the University of Chicago
- The Quarterly Journal of Economics (QJE) administered by Harvard University
- The Review of Economic Studies (ReStud) administered by The Review of Economic Studies
The top 5 journals in economics are regarded as the de facto standard of research excellence in the discipline. If admissions committees are optimizing academic placements, academic economists are optimizing for placement into a top 5 journal and H&M explore the potential pitfalls of this focus in their paper. For this article I want to focus on the impact of the top 5 journals, specifically on tenure outcomes. A graphical representation of the impact of these journals on tenure is shown below.
Having 3 top 5 publications by tenure review is associated with a 0.62 probability of tenure success, ~2.5x more likely than the next best probability outside of the top 5 (2 TierA publications). But more than that, the number of top 5 publications is the only variable that sees a change in probability with number of publications. These estimates do not imply causality in the relationship, but I think it is fairly clear what the effect is. Next, we want to establish the private value to the program.
H&M provide an analysis of what they call “incest coefficients” between the top 5 journals and the top 10 US economics departments as well as New York University and University College London. The main point of the table is to show that in-house journals, JPE & QJE, have an outsized affinity for faculty submissions from the affiliated schools, whereas the professional associations’ journals, ECMA & ReStud, have much less deviation from the average incest coefficient for all schools.
As H&M note: “If the explanations in the literature hold for the T5 journals, tenure-track faculty with connections to T5 editorial boards gain an advantage over colleagues who lack such networks.”
Thus, the private value to an Economics PhD program in optimizing selection of applicants who are most likely to place into top programs should be clear now. Selection of students who will place well extends the influence of the department to the top 5 journals and the AEA, which the department can then leverage for influence that has positive outcomes for the department, e.g. for tenure (across the profession, not just at the department), and setting standards in the discipline via AEA Executive membership. It should be noted that while H&M focus specifically on top 5 journals, the effect of networks on publishing have much further reach as top faculty are often editors of TierA journals and also serve as referees for papers; having networks associated with top departments is useful for publishing outside of the top 5 journals as well.
In this section I have made an argument as to why admissions committees selection criteria might be built on maximizing academic placements above, but for those unconvinced I think you will find that any other reasonable objective function will lead to the same effect… because applicants value top placements.
Understanding the Value of Top Placements to Applicants
To understand the value of top placements to applicants we first need to understand what applicants might value after graduation. There are three things every graduating PhD with aspirations to be an academic economist is likely to care about:
- Salary
- Teaching Load
- Location
Salary
The value of salary should be clear. The European University Institute provides information on average salary of Associate Professors in economics by National Research Council (NRC) rankings of graduate programs.
The gap between in salary between Tiers 1–2 and Tiers 3–4 is significant at around $20K. The gap between Tier 4 and Tier 5 is another $25K.
Teaching Load
For non-academics, teaching load may seem like a strange thing to list but academics understand the value of a lower teaching load. Generally speaking, people become professors primarily to do research not to teach, and their career progression is determined by their research output not their course reviews. This means that there is an opportunity cost to spending more time teaching relative to doing research and this shows up in the assigned teaching load in a professor’s employment contract with their university. Generally speaking, higher ranked schools are expected to be more research focused and in accordance with this tend to have lower teaching loads relative to lower ranked schools. This is great for academics as it aligns their employment contract with their career progression, which means that academics at lower ranked schools are likely to find it more difficult to make career progress after placing into those schools.
We have data courtesy of the EUI, again, but for Assistant Professors this time.
Note also the salary differentiation. This grouping does not identify teaching loads by rank as in the Associate Professor salary breakdown above but note the variance on teaching load relative to the mean. A 3.5 mean load vs 2.4 variance suggests very high heterogeneity in teaching loads within the PhD institution category, the same category that will contain all the schools in the previous table. What’s more, the higher ranked schools are more likely to have funding to hire teaching assistants to do some of the heavy lifting for courses, such as grading, which further entrenches the difference in experience along the ranking distribution.
Location
The trend here is less rigid but still has a fairly strong realization. Top ranked schools, especially the Top 20, tend to be in more desirable places to live. Harvard and MIT are in Boston, UChicago is of course in Chicago, Stanford is in Palo Alto, UCLA in LA, UCSD in San Diego, etc. The outliers among top schools are few, primarily Princeton and Cornell which are well known to be fairly remote.
Consider Michigan State University, ranked 29th by USNews in their most recent ranking dated to 2017. The university is located in East Lansing, MI. To give a sense of the desirability of the location I quote an answer from Quora about what it’s like to live in East Lansing, MI: “Four years ago, I moved from East Lansing to the Washington DC area. I naively presumed that East Lansing, being a college town, had a lot of activities and a diverse population. Only after living somewhere else, does one come to realize what they’ve left behind. In this case, I am thankful that I now live in a much more diverse, thriving, and, frankly, much more open-minded environment.” This is also echoed in a tweet from an unnamed economist below.
Of course, I do not mean any offense to Michigan State and I am sure East Lansing has its strong points. I merely mean to suggest that higher ranked schools often correlate with more desirable locations. For a within state comparison, the University of Michigan-Ann Arbor is ranked 12th by USNews and Ann Arbor was labelled as one of the best place to live in America in 2018 by livability.com.
Having identified the value of top placements to applicants, we can explore the relationship between PhD programs and placements.
Academic Placements
In their working paper “Staying at the Top: The Ph.D. Origins of Economics Faculty”, Todd Jones and Arielle Sloan (J&S) examine the relationship between faculty across economics rankings and their PhD institution. In short, PhD students tend to place into departments at or below the ranking of their PhD institution. The most striking graph from their paper is shown below.
Note the displacement of the department axis relative to the PhD axis. A thread on the diagram going directly across is toward a lower ranked group, a thread has to go upward to meet its corresponding ranking group, and the thickness of the thread represents the relative number of placements. Most of the upward moving threads are relatively thin. The graph is based on current department membership, not based on first placement after graduate school and for non-economists it may be worthwhile to know that first placements cannot be to the same program a student is graduating from.
To help give a more precise idea of the placement information capture in the previous graph, consider the following graph which zooms in on the top 8 departments.
For any prospective PhD students who are reading this and ever thought, “it seems like everyone went to Harvard or MIT”, while going through Professors’ CVs, that’s because a sizable portion did.
Review of Discussion so Far
- PhD programs may be optimizing their selection criteria to maximize their placements.
- Applicants are likely to value top placements independently of the PhD program’s selection criteria.
- Departments typically place graduates into departments at or below their rank.
With the last two points established we can revisit my earlier statement: any other reasonable objective function in the selection criteria will lead to the same effect as optimizing for placements.
What is a “reasonable objective function”? They are the ones that our optimist friend mentioned earlier might hold, e.g. maximizing selection based on applicants’ potential to advance the field, or do novel research. Since applicants value top placements, in order for departments to have access to the best students (in our reasonable objective function world) they must be able to offer top placements.
In other words, even if my conclusion that Economics PhD programs are selecting to maximize placements is incorrect, programs must still be able to offer good placements in order to attract good students as any reasonable objective function on the program side must be positively correlated with placement outcomes. If they aren’t positively correlated, then there would be a serious market mismatch where we reward students (with good placements) for things that the profession does not value (our uncorrelated or inversely correlated objective function), which could not be an equilibrium. Departments seem to understand this as they all list their job market placements, not a standard practice across all of academia, and MIT’s Economics department even goes so far as to highlight its placement into “Top Departments”.
Thus, we can proceed with a discussion of the competitive elements of the Economics PhD admissions market having established the defining criteria that admissions committees are likely optimizing, but first a quick note about placements and departments.
Correlation or Causation: Placements
As established above, economics departments generally place students at or below the ranking of the department. It is very difficult to disentangle whether this is merely correlation or causation. To illustrate, in the story where it is causal we might think that these top ranked departments are good at educating and advising students such that they are able to produce research and job market papers which are highly valued in the academic job market allowing these students to place well. A more cynical story of “causation” could also be that networks strongly affect placement and having lots of affiliated faculty at similarly ranked schools effects a placement at a similarly ranked school. In the story where it is correlation, admission committees could simply be selecting the applicants available to them who are most likely to place well, independent of which program they go to. It’s difficult to know.
Competition in the PhD Application Market
For the remainder of the article we will assume that admissions committees are selecting to maximize placement outcomes.
In order for PhD students to place well, they will have to do two things:
- Successfully pass the qualification exams.
- Produce research and a job market paper that is likely to be valuable on the academic job market.
Programs seem to be well coordinated on the answer to the question “what should every economist know?” as evidenced by a pretty standard first year core in the form of sequences in Microeconomics, Macroeconomics, and Econometrics. They also appear to have fairly similar standards for these courses, which means to say they likely have similar valuations of applicants with regard to their ability to succeed in these courses. Almost all advice about coursework for Economics PhD applications will include the following:
- Multivariable Calculus
- Linear Algebra
- Probability and Statistics
- Differential Equations
- Real Analysis
In other words, the selection criteria based on the first requirement is vertically differentiated and can be used to rank candidates directly: An applicant with coursework in all 5 areas is strictly better than an applicant with coursework in only 2–3 areas, and an A grade in a given area is better than a B grade in the area.
Having done research work in economics I can confirm that the first three are the bread and butter of economics. My understanding is that differential equations are mostly used in macro fields, while the virtues of emphasizing Real Analysis has been a point of debate among economists in recent years as captured in the tweets below.
In an article for Quartz, Miles Kimball says much the same: “Linear algebra (matrices, vectors, and all that) is something that you’ll use all the time in econ, especially when doing work on a computer. Multivariable calculus also will be used a lot. And stats of course is absolutely key to almost everything economists do. Differential equations are something you will use once in a while. And real analysis — by far the hardest subject of the five — is something that you will probably never use in real econ research, but which the economics field has decided to use as a sort of general intelligence signaling device.”
The quantitative portion of the GRE is also meant to help evaluate candidates on this requirement but I suspect it is used more as a screening tool to filter the application pool rather than a confirmatory tool for satisfying the requirement.
While we are on the topic, here is a preview of how programs are integrating the importance of this requirement into their applications. They directly ask for some number of the most relevant/difficult courses, and often times the textbook, usually ranging around 5–6 courses but Stanford Econ is an outlier asking for not only the most information (they ask the applicant to determine the most comparable Stanford course!) but also the most at 15.
The second requirement is much less specific and is likely the requirement that admissions committees spend most of their time discussing. I once spoke with a faculty member at a high ranking Economics program and s/he viewed the application market as being primarily horizontally differentiated rather than primarily vertically differentiated, the latter of which is my position. I think this view came about as a result of the time s/he spends in the committee focusing on this requirement and its horizontally differentiated character.
To provide a more concrete example, how might a committee trade off the potential of an applicant to introduce new methods to the field, like Machine Learning, against another applicant who has more traditional characteristics associated with good placement such as having undergraduate research experience in a well established economic field, like Development Economics? It’s not clear which one of these is “better” or “worse” in the context of producing good research and a strong job market paper. A large part of it might even uniquely depend on what the specific members of the admissions committee forecast will be important in 3–5 years and their risk appetite. Ultimately, admissions committees will likely hedge their admission offers by selecting some “traditional” candidates and some higher variance candidates.
Given the difficulty of measuring the potential of an applicant to satisfy the second requirement, admissions committees have often placed very high emphasis on the recommendation letters attached to the application. Consider the following information from John Bound found in the AEA CSWEP Summer 2014 issue:
“While applicants must be adequately prepared in mathematics and economics, we also look for individuals capable of doing independent research. In fact, applicants without evidence of such capacity — for example, through their undergraduate or master’s thesis, work for a professor or employment research — are unlikely to gain admission to our program. In all cases, we are looking for evidence of both creativity and seriousness, and in this regard we pay a good deal of attention to letters of recommendation (especially from economics professors) and to applicants’ statement of purpose.”
To add emphasis, the faculty member I referred to earlier once commented: “There’s nothing a strong letter of recommendation can’t fix”. Letters seriously help solve the information asymmetry related to requirement #2!
But even better than a letter describing an applicant’s potential to do good economics research is a letter describing an applicant’s existing good economics research; the strongest predictor that someone can do something in the future is generally going to be whether they did it in the past. Combined with the demand side changes that led to the rise of pre-doctoral fellowships, full-time research experience before applying has come into vogue and it provides a much stronger signal that an applicant can satisfy requirement #2. Assuming that the signal is positive, it helps admissions committees look more favorably on an application as a benefit of the reduced noise.
As a quick aside: There is one potential caveat to the benefit of full-time research in the application market. Applicants who clearly identify that they wish to do theory will likely be evaluated more heavily on their mathematical ability and coursework relative to applicants who might be interested in more empirical work. Of course, prior theory focused research will help in this case as well, but my impression is that theory research before the PhD program tends to be very uncommon.
Before we close out the discussion of competitive market structure, it may be worthwhile to review some evidence about what type of research places well in the academic job market. In his paper “Young ‘Stars’ in Economics: What they do and where they go”, Kevin Bryan constructs a dataset of academic job market “stars” based on candidate flyout lists during the academic job markets between spring 2013 and spring 2018. Bryan justifies this definition by quoting from the FOA paper referenced earlier: “Although there are of course differences in tastes across schools, economists famously have ‘more homogeneous standards of evaluation within, greater confidence in their judgment about research excellence even in other fields, and a higher likelihood to stick together as a group than panelists from other disciplines.’”
Using his definition, Bryan finds 226 such stars for his dataset. He produces data on the field associated with each star and the general style of the job market paper. Most notably, most stars are broadly in Applied micro, and over 85% of stars produce a theory based or guided job market paper.
I would caution against relying on these as definitive of what is most important for a job market paper, and over time these results quite possibly may change, but at the very least it is suggestive that the job market does value elements of theory based research on the job market. It may be that Real Analysis is used in the admissions market as an indicator of potential to use theory in the job market paper rather than merely a general intelligence signaling device as Kimball suggests.
Finally, with the infrastructure developed we can move onto the main topic of this article.
Barriers to Entry into the Economics Phd Program
The barriers to enter economics seem to have become a great point of concern among some, and with good reason. Economists have a large amount of influence on policy, or at least we would hope they do, and the PhD program serves as the de facto entry point into the profession; bachelors and masters degrees are generally not sufficient for employment as an economist in academic, industry, or other professional, settings.
The reason I spent the the better part of this article developing a framework for thinking about competition in the PhD application market is to identify what are “barriers to entry” and what are simply artifacts of competition in the market. As a point of illustration, in the first tweet above there is a suggestion that doing “multiple pre-docs” is a barrier of sorts. Even adjusting for what is likely some degree of hyperbole, the assertion that a pre-doctoral fellowship is a requirement or a formal barrier is not true. If an applicant had the foresight to pursue and gain access to research opportunities during their undergraduate program sufficient to provide a similar signal as a full-time research position, then that applicant is likely to be as competitive as someone who has full-time research experience. The response to the second tweet above is similar: pre-doctoral fellowships aren’t a conspiracy by faculty to delay entry to the PhD program, they’re an artifact of increasing competition within the market in response to an increased share of empirical work in economics and requirement #2.
This is not to say that the lengthening of the time to enter and complete a PhD program is a good thing, but we need to address those items at a system level by thinking about how we might enable people to demonstrate sufficiency for requirements #1 and #2 in more efficient ways. Even then however, there are many more qualified applicants than there is space to enroll; we cannot get away from competition. With this in mind, while it is noble to advocate for a better work-life balance as in the tweet below, an applicant who produced more by spending their nights and weekends working more is going to have a better chance on the market than an applicant who produced less by only working 40-hour weeks.
The incentives are probably well captured in a prisoner’s dilemma framework; if one applicant works more than another, that applicant is better off, but this is symmetric so both applicants will simply just work really hard, even though they might both be better off if they instead worked a normal amount and had their weekends to themselves.
With all this said, what do I think are the real “barriers to entry”? They group largely into two categories: (1) Lack of ability to effectively compete or unfair competitive advantages (2) Money. Time plays a factor in both.
Barriers to Competing
Information and Opportunity
In their paper “Political Dynasties”, Ernesto Dal Bó, Pedro Dal Bó, and Jason Snyder (DDS) explore the hereditary persistence of power in political office, i.e. how often does political power pass from parents to children, specifically from father to son. They examine legislative power in their paper but also provide a bit of data on the heredity of being an economist. In the table they report, reproduced below, economists are second in “dynastic bias” only to legislators, which was the focus of their paper.
For comparison on these numbers, let’s look at data from a New York Times piece also based on the General Social Survey.
The NYT piece groups all professors and lecturers together so it is difficult to compare the effect for economists in specific but we can compare the numbers for some other professions. We see that lawyers, doctors, and plumbers have around similar numbers. The only main deviation is for dentists. Given the generally good fit between the two pieces of data, I am inclined to believe DDS’ assertion that 37 times is the right figure for economists. This is very high. What benefit might having an economist (or related) parent bestow? Well, one is likely money, which we will discuss later, but the other is likely information.
Having an economist parent, or more generally a parent with prior exposure to a graduate academic environment in a STEM field, provides a wealth of information and strategic planning for setting up oneself for academic success the first time around. Consider the tweet below from an Economics professor.
This isn’t very uncommon for high-talent children of well educated parents. I happen to know other graduates from the same school as that professor who have also communicated to me that their undergraduate curriculum was much more accessible to them because they took some of the difficult foundational mathematics courses in high school. The benefits of taking a difficult course multiple times (especially if lower stakes the first time) cannot be understated. This is not to suggest the professor is any less capable than anyone else, or that s/he does not deserve their position; the tweet is used only as motivation.
I’ll use myself as another anecdotal example. I took Real Analysis 3 times. The first time I ended up dropping it. The second time I got a B-. The third time was this past fall (Fall 2020) and having built the intuition for the material the first 2 times, I got an A with a final grade of 99%. I happen to have had some advantages that allowed me to spend this additional time patching up weak points in my application; in particular I won a government scholarship from my home country that allowed me to graduate completely debt free from both my undergraduate program and my masters program. If I could have simply gotten this right the first time, or rather if the first time had been lower stakes, having a scholarship might not have been a necessary condition in my case.
Parents really do play an important part in supplying this information. Here is another anecdote to demonstrate this. When I was in high school I qualified for the national mathematics olympiad in my country and won first place. The group supervising the local International Mathematics Olympiad (IMO) team invited me to join them for IMO training on the weekends. I went to a few sessions but as I was joining in the middle of training, I was having difficulty following all the material, I didn’t know anyone, and there wasn’t much effort to help integrate me so I was having a hard time staying on. From my perspective at the time, this training thing was just something to do on a weekend, so I told my parents I wasn’t making any friends and I didn’t really want to go. Looking back I understand the loss of that opportunity, but at the time neither me nor my non-college educated parents could have had any idea about the academic value of that type of opportunity. Having parents who can at least understand the value of opportunities, and even better if they can seek out opportunities, makes a big difference.
Why do economists have such a high dynastic bias? I think it is in part because economists (and other highly educated parents) can set their children up for success by understanding the currency of success indicators and other necessary preparation, and then steering their children to obtain those indicators and preparation. The lack of access to this information at a similar time for everyone, during high school and undergrad, and the subsequent time investment (along with accompanying opportunity costs) needed to make up for not having this information is a barrier to competing.
Stata
Every economist reading this may have just perked up in their seat after reading the header. Yes, Stata is a barrier to entry but probably not for the reasons one might first think. Stata is the de facto language of computational economics; a statement I built up in my article on Pre-Doctoral Fellowships.
As shown in this graph from the AEA, Stata code appears in almost 80% of economics research, with the context that ~10% of research uses no code at all, so conditional on using software, Stata is used ~90% of the time. This reinforces the point I made in my pre-doc article: “if the predoctoral position is to be spread across more than one faculty member, it is very likely that Stata will be a silent, or explicit, requirement for the position.” This by itself is not a problem, but when we explore the context around Stata I think the barrier should become clearer.
Who uses Stata? Basically, just economists. Students outside of economics never touch the software; outside of economics people are trained to use Python, R, Julia, Matlab, etc.
Who has access to Stata? The software is not open-source/free so generally speaking the only people who have access to Stata are economists, or students at a university with a subscription to Stata. There is one industry group that I know of who also uses Stata and therefore has access, and that is the Economic Consulting world (again, because all the economists working at these firms know Stata).
Why is this a problem? Conditional on the importance of pre-doctoral research experience, there are two problems:
- The need to be fluent in Stata to get a position limits would-be applicants who were trained in a different field like Applied Mathematics or Computer Science.
Even if the pre-doctoral applicant has strong programming skills in R or Python that can accomplish the same tasks (and more) as can be done with Stata, applicants who know Stata are favored due to the use of the same code base as the hiring professor. By itself, not a problem but it seems that the economics field values applicants from these economics-adjacent fields as they often list on the requirements for a pre-doctoral position: “Bachelor’s degree in finance, economics, statistics, computer science, mathematics or a related field”, but then this requirement unfairly disadvantages applicants from these adjacent fields. - Assuming the candidate is fluent in Stata then, in order to do the skills test associated with the pre-doctoral application, they have to be at a university either as a student or as a researcher, or working in economic consulting, to get access to Stata at no cost to themselves. Without being in one of those groups, a single license Stata/SE license is USD$765.
In my opinion the pervasiveness of Stata is also a bigger problem the profession needs to deal with, as being limited in computational methods available for research to whatever the Stata Corporation thinks is important to include in their software and the accompanying delays associated with updates to that decision function has contributed to economics falling behind in the ongoing data revolution happening in other fields.
But I get it, after investing in training oneself in one statistical software we all want to get as much ROI on that investment as possible and there is a real opportunity cost to every hour spent learning new software that we could otherwise spend on research as illustrated by the tweet below.
Anyways, independent of opportunity costs, Stata’s status as the computational language of economics is a barrier to competing.
Money
Budget constraints are nothing new in economics. If a person has more money they can afford to do more things, but in an environment where we might value merit and diversity I’d advocate for viewing budget constraints as a problem to be solved rather than simply a feature of the environment.
There are a number of places money shows up as what I would consider a barrier to entry.
Applications
Applicants are generally required to pay an application fee between USD$90–140 for each application they submit. Sure, this is an opportunity beneficial to the applicant and a price mechanism is useful for its signaling value (applicants really want to be in the program, at least up to the cost of applying) and its effect of reducing the number of applications requiring review, conceivably allowing admissions committees to invest more time in reviewing the applications that are submitted. One of the problems is that applicants do not know their market value when applying and this leads to applications acting as a search mechanism rather than just a matching process. Let me unpack that a bit more.
The admissions process is very opaque from the outside. We have a general sense of requirements (some listed here) but there is a lot of noise in how those suggested requirements transition into likelihood of acceptance. Consider this advice from a premier economist, Susan Athey: “Real analysis is an especially important class because it tends to be demanding everywhere, and forces you to do logical and formal proofs. Get a good grade in this class.” Then consider this statement from an economist who did their PhD at MIT, one of the top economics programs in the world (see above for rankings):
Is a B a “good grade”? I don’t know, and likely many other applicants don’t either. In fact, there are many economists who have never even taken Real Analysis, but some of these economists were applying many years ago and perhaps the level of competition in the market was lower as compare to today. What information is accurate and what isn’t? You can hopefully see why I describe applications more as a search process than a matching process. What is the strategic response to this noisy application market?
Applicants are increasingly applying to more and more schools. Applicants do not understand their general market value when they are applying and so part of what they are doing in the application process is paying to determine what that market value is. It is very common to see an applicant apply to 19 programs or more, I know of at least one student who applied to 34 and got into 3, one of which was a top program in the form of Columbia University’s Econ Department.
Now, this particular candidate had a very nontraditional background so this is not meant to represent a typical scenario. That said, one might ask how s/he could have gotten into a top program but then only into 3 total, even assuming the 34 schools applied to were just rank ordered top 34 schools. The answer to this question resides in part in the horizontally differentiated aspect of applications and in part in the phenomenon known as “yield protection”.
Yield protection, in short, is rejecting qualified candidates if the admissions committee believes the candidate will get a strictly better offer of admission elsewhere. Now, in some cases this is just for trivial things like reporting high yield or reducing acceptance rates, but in economics PhD admissions there may be something more noteworthy happening, specifically funding.
Based on my discussion with anonymous economics professor at a top school from earlier, funding budgets are outlined by a central graduate administrative body before admission decisions are made, for example the central graduate admissions office at the Faculty of Arts and Sciences. These budgets are assigned to each department based on planned offers of admission. What’s interesting is that if an offer of admission is rejected, the department cannot convert that released funding into an offer of admission to the waitlist but rather it goes back to the central office for reassignment to any department. In other words, if the department makes offers of admission to candidates unlikely to accept it, they are effectively squandering the limited resource of funding for getting students into their department for the next academic year, which is strictly a bad thing for the department. Consequently, admission to the #2 ranked program does not imply admission to the #15 ranked program, because #15 may view such a high market-value candidate as too likely to have a strictly better offer (such as from the #2 ranked program).
This makes search hard for most candidates. They cannot just apply to a certain tier of schools or lower, they may instead have to apply very broadly to find their correct tier. This leads to total application fees of $2000+, probably with an average around $2500. For a 22–24 year old that is a lot of money, often prohibitively a lot of money. Consider the tweet below from a professor at Wharton, clearly someone who would have been a high-talent applicant during their admissions cycle.
Here is a short thread about this for a PhD student in statistics. Without family money to help offset these costs, it is very difficult to apply sufficiently to satisfy the search process, and remember, applicants are also discounting the cost-benefit analysis based on a low expected probability of acceptance anywhere.
This is a barrier to entry.
Stipends and Pre-Doctoral salaries
One of the great things about doing a PhD in economics is that it is almost guaranteed to be funded by the department/university with free tuition and a stipend. Similarly, pre-doctoral fellowships are paid research positions so they come with a salary as well, and often times some facility to take courses for free or at a significantly reduced price. However, while the salary for pre-doctoral fellowships are generally consistent at around USD$50K per year, PhD stipends can vary greatly.
As a quick aside: Why do schools fund PhDs anyway? They don’t fund MDs or JDs, or any other degree programs really. Here’s my conjecture, the academy wants to propagate itself and without training PhDs to continue the academic institution, it would whittle away over time. Academic positions notoriously do not pay as well as non-academic options especially for high-talent students who can find lucrative employment elsewhere. Combine this lower payout with the 5-year investment and we can see that in order to make the academic path appealing schools must reduce the loss relative to other options in order to attract high-talent students, hence funded PhDs. The macro-dynamics behind this have been changing in recent years though, so I do wonder about how this will evolve in the future.
Back to PhD stipends.
Consider two schools which are within a few minutes of each other in Boston: Harvard Business School (HBS) and Boston University (BU). HBS offers a stipend of $43,860 per year while BU offers a stipend of $23,922 per year.
According to the MIT living wage calculator, the living wage needed for Boston-Cambridge-Newton as a single adult with no children is $34,824 per year.
As someone living in Boston currently, I think $50K per year is a more realistic living wage estimate in order to be within a reasonable distance to HBS or BU campus but this is just my opinion according to my lifestyle. Regardless, the point is that stipends can vary wildly and while I commend BU for putting up the financial data on their website, it allows for only one of two types of people to attend their program: people who are comfortable taking on tens of thousands of dollars of graduate student debt over 5-years or people with family money to cushion their time in the program.
The pre-doctoral fellowship salary is only marginally better. Most of these predoctoral positions are at top programs, and as noted before top schools tend to be located in more desirable locations which tend to result in higher costs of living. Not only that, but predoctoral positions are becoming increasingly competitive with some even asking outright for a masters degree.
Apart from living costs, would-be pre-docs have two other fees they may have to consider: (1) Loans for their undergraduate (or masters) degrees (2) Application fees during their PhD applications. We have seen the cost of applications can range around USD$2500 and student loans are likely to be much higher. For potential applicants who have to contend with these costs, can they realistically consider going through 2 years of a predoctoral program and then 5 years of a PhD program? Perhaps, but they are going to come out on the other side in a lot of debt. Consider the tweet below; it is about lower ranked programs, but the idea applies more generally.
This problem extends in weird ways beyond just the number of applications and paying off loans. My prior is that the situation described below is not at all very common, but I could be wrong. Regardless, it should be considered an incredibly sad state of affairs for the economics profession.
Aggregating the financial environment associated with the economics PhD program, I can think of only one group who can realistically choose to pursue it, people with the family money available to cushion 11 years of schooling (including undergrad), and this is made even more salient when I think about the high levels of competition to get into a program to begin with. I’ll assert that an applicant who can successfully overcome the competitive hurdles to get into a PhD program is likely to be just as competitive for a much better paying entry level position in industry. This opportunity cost only reinforces the idea that the current system is designed to select those coming from some level of comfortable family wealth.
I can say with absolute confidence that had I not gotten a government scholarship to support my bachelors and masters degrees, my family would not be able to support me while I pursue the PhD path; both things that are generally not true for most people regardless of how high-talent they are.
This is a barrier to entry.
Conclusion
Competition in the market for Economics doctoral programs is increasing at a somewhat alarming clip. Competition is good; it helps us funnel the best and brightest to the top and they should then be able to give us large societal ROI through their knowledge generation and policy influence. However, the idea of efficiency vs. equity is not lost on us economists. There are some elements of efficiency in the current system that is fundamentally flawed both from an optimal efficiency standpoint and an equity standpoint, these are what I consider the “barriers to entry” I list here.
Loretta Mester discussed her thoughts on why economics isn’t diverse in a piece for the Federal Reserve Bank of Cleveland. Her main reasons were:
- “lack of role models in the field”
- “different preferences at the time students enroll”
- “Other hypotheses have to do with the way economics is taught or with the content of the courses”
These ideas seem to be shared among other members of the economics field.
I’m sure these ideas have some influence on the diversity problem as well; my hypothesis is that it has the most to do with gender diversity. But as an aspiring economist myself, it brings me great turmoil to almost never see the issue of the financial incentives around entry into the profession brought up in the larger discussion. We make models of how people make financial decisions all the time, except on this one particular issue. Why? I have no idea, so I decided to introduce the argument myself. Hopefully this has convinced some of the profession to extend the discussion of diversity and inclusion to ideas about financial and informational barriers to entry into the profession.
To close, I will leave a story shared on Twitter by an economics PhD student at a top institution that captures the barriers I described here.
