My Journey into Econ PhD

Why exchange at Harvard?

Coming to HK to study ‘business’ wasn’t what I dreamed of when I was 18— I had wanted to go to Peking U to study history or sociology, because I have always wondered how the world came to be the way it is + why society works the way it does. But (1) my dad laughed it off and demanded that I study something useful and (2) I was 7 points short from the admission cut-off for PKU in the National College Entrance Exam (高考).

How to exchange at Harvard?

The normal route would be to use HKUST’s exchange program, but it’s too competitive: the whole business school, which has probably 1000 students per cohort, have 2 spots for NYU, and that’s already the best institution partner HKUST B-school has, and every year it’s some ‘Global Business’ (GBUS) student getting it (GBUS is HKUST b-school’s ‘flagship program’, which is not what I was in). In fact, this situation is broadly true for universities in Asia: it’s very hard to get a very good undergrad exchange via the university, because resource is extremely scarce.

Why Econ PhD?

But it never came true — 2 weeks into my exchange at Harvard, I changed my mind.

How to Econ PhD?

I again got it the easy way — YG has already figured it out, and he basically told me the ‘playbook’ — work for Raj Chetty as an RA, and you’ll get into Harvard/MIT.

  • It’s extremely hard to become top 1/2 in your cohort at Tsinghua, because everyone is *extremely* good
  • Even if you’re top one or two, you may still not get in, because US top econ programs may not know how to read foreign undergrad’s transcripts
  • Even if they know — because, to be fair, Tsinghua often place people into top US econ grad school — it’s not enough, because any US school applications nowadays, be it undergrad, visiting, or grad school, it’s all about the rec letters.
  • If you are a IMO gold medalist, you can ignore the above, because your extreme math ability is enough to convince the admission committee
  • Some top European masters program consistently place their top students into top US econ grad schools, e.g. LSE, PSE, TSE, etc. There are some in the US too: Duke has a research-oriented master program in econ. But you basically need to be top 1 or 2 in those to directly place into, say, Harvard/MIT
  • If you’re a Harvard/MIT undergrad who has started working as a research assistant part-time for the faculty since, say, your sophomore year, and you did well, and took grad classes and Real Analysis, and got As, you can also ignore the above

How to Econ RA?

Now that I know I need a full-time RA job with one of the faculty here at Harvard (or MIT, or NBER), the question is how to get there.

What classes to take?

At the same time, I also extended my visiting at Harvard from 1 semester to 1 year, so I can keep working for her, and take some grad level classes and Real Analysis to strengthen my profile. During my year at Harvard, I took and got A’s from 6 courses: an undergrad economic history class from James Robinson, an undergrad level econometrics, a master level class at Harvard Kennedy School ‘Designing Government Policy’, Real Analysis, a 2nd year grad level class at MIT on Development Economics, and Senior Thesis Seminar, so that I can write a paper.


During my second semester as a Visiting Undergrad Student at Harvard, MA asked me if I want to work for her as a full-time RA for a year, which I of course said yes to. 3 months into my full-time RAship, in Sept, she asked me if I want to extend my RAship to 2 years.

  • GRE: I spent non-trivial amount of time prepping for the Verbal and Writing, and took the test multiple times to get the Verbal above 160, which in hindsight was a waste of time. As MA said, no one cares about the GRE as long as you have a 170 on the math section — which I didn’t even get… which didn’t stop me from getting into Stanford
  • Statement of Purpose: I again spent time revising it multiple times + asking for help from then-Harvard econ phd students, which again turned out to be a waste — ‘nobody reads SoPs’ says more than 1 faculty in econ. Usually they only read your SoP *after* you got in to figure out which field you are and which faculty they should assign to send emails to you to convince you to come. Also, you should make sure your SoP is aligned with the letters — for example, I said I wanted to do economic history / development, because MA is in this field and she is my main letter writer and my RA work was in this field. You should send your SoP to all your letter writers for feedback — this is a chance for you and them to get on the same page and align your stories. You might say, ‘but didn’t you already decide to not do economic history / development?’ — you can always switch field after you get into grad school, which means in the application, you should focus on how to get into the best program you can get into, and once you’re in, worry about picking fields/professors to work with
  • Letters: Of course there’s the letter from MA. I also asked for a letter from SA, a Northwestern faculty who was visiting MIT at that time, when I took the 2nd year grad Development. Why did I not ask for a letter from BN, the MIT faculty who co-taught the class? It’s a huge mistake. At that time, I thought I can only ask for 1 letter from faculty teaching the devo class, because otherwise their content will be overlapping. In fact, it’s totally wrong —First, they taught different parts(!) so the content will be different. Second, they have *different connections* — the MIT dept will of course take more seriously an MIT faculty’s words than a visiting faculty, and vice versa for Northwestern. Again — grad school application is fundamentally about the letter writers’ distribution network. Different people have different distribution networks, so even if two people taught the same class, it’s still ok to ask for letter from each of them!
  • Coursework/Transcript: I had A from Real Analysis, which is pretty much the only thing that mattered. If you’re from a US undergrad who often place people into top econ phd, your undergrad transcript will matter because they know how to read it. In my case, HKUST is a place that just doesn’t place students into top US econ phd programs that much, so as a result, my transcript from HKUST actually didn’t matter at all — it worked well for me, because my grades weren’t good at all (3.7 out of 4.3). Instead, because I spent a year visiting Harvard, all that mattered was my Harvard transcript. But, most of the classes I took had no signal value at all, as I said before. So the A in Real Analysis was essentially the only piece of info that’s useful on my transcript. (I was dumb enough to take a probability class during my last semester at HKUST because I thought I didn’t have enough math — the truth is, as long as you have an A in Real Analysis from Harvard, it doesn’t matter anymore if you didn’t have math classes easier than that. Of course having A’s in harder / grad level math or stats classes is a plus but it’s unnecessary given my strategy, i.e. being sold as an empiricist instead of theorist.

How to do well in Econ PhD

There are lots of articles on ‘how to present’ or ‘how to structure a paper’ for job market candidates, but there seem to be fewer articles in general on ‘how to do well in grad school’.

  • Always go to the strongest department. For example, if you have MIT/Harvard don’t go to Stanford. If you have Stanford don’t go to Northwestern, etc. unless you have a very very good reason, e.g. you’ve already started coauthoring with a big shot at the school who clearly will place you well when you’re on the market
  • Once you get into your program, figure out which professors place well / which field is the department strongest at. For example, if you ended up at Stanford, please don’t do econometrics (unless you go across the street to the business school to join SA/GI’s causal inference /machine learning crowd). I once heard the students complaining in a Town Hall meeting ‘why isn’t there an econometrics seminar? why isn’t there more econometrics faculty?’ and was thinking ‘if you had wanted to do econometrics why would you have come to Stanford?’
  • Professors who publish well and professors who place well are not the same! The biggest common mistake 1st years make is ‘oh I want to work with RC’ —have you heard of anyone who’s main advisor is him and placed well? He has a group of full-time RAs who churn papers out for him; he doesn’t work with grad students. Again, at the professor level, it’s another game: some keep their influence at the department by placing students well, some do so by lots of QJE/ARE, and some decided to not play this game anymore and do consulting all day on the side. You really should figure out who is doing what — who is important in the department; whose strategy was writing lots of papers, and who has a track record of always placing students well.
  • Of course on the job market there’s a ranking: every department ranks its job market candidate into several tiers: push as stars (at places like Stanford, maybe 1/2 each year, but MIT/Harvard can have a lot more) i.e. top 5, top 10–20, top 20–50, not good for academia. Of course it’s easy to get into a competitive mindset, but the best thing you can do is to (1) pick a good advisor who will and can vouch for her/his student because she/he has always placed well and have the credibility to say someone is good (2) write the best job market paper you can so that they can credibly say you’re good. Sometimes the strongest advisor’s 3rd best student place better than a median advisor’s best student — for example, in 2016, DF’s 3rd best student placed into UPenn. That’s already an extremely good placement by Stanford’s standard — yes, the gap between MIT/Harvard and even Stanford is huge. MIT can consistently place almost all of its students into good academic placements, while maybe half of Harvard can, whereas maybe 1/5 of Stanford can
  • But after all, it’s really on you having a good job market paper, and good presentation. If you don’t have a good JMP, no amount of ‘following the playbook’ will help you. Also, once you get the interview, the rest is on you: no matter how strong your rec is, if you did really badly in the interview, you still won’t get a flyout. (Although there are cases where someone failed all his interviews/flyout and miraculously got a job because his advisor made a call, but it’s very rare.)
  • So how to write a good JMP? Unfortunately, it’s not ‘writing whatever you want to write the most’ — there’s pretty much a certain way a good JMP should look like nowadays — read recent JMC’s JMP to get a sense. If you’ve picked a good advisor, then listen to their advice — again, similar to my interactions with MA when I was applying for grad school, I pretty much followed all her advice when deciding what classes to take, etc. because it’s her job to sell me and she should have control over how my packet look like. — If I had insisted writing in my SoP that my real dream is to become a theorist, then her letter would be void, because as a good empiricist, there’s no way she can credibly talk about my research potential as a theorist.
  • Should you listen to *all* your advisor’s advice? Not necessarily. Advisors appreciate people who can think independently, and convince them that they are wrong. It’s really an art without a ‘right answer’ — there are cases where a JMC later said ‘when I’m on the market everyone asked me that question which my advisor has told me to fix that I ignored’, but there are also JMC who said ‘my advisor thought the idea was shit, but after cleaning it up and presenting it in a better way, they were finally convinced that it’s a good idea, and it eventually turned out to be a good paper’.
  • In general, it’s good to have *ideas* that are come up independently and novel — I’ve heard many times that ‘if you work on an idea clearly derivative of your advisor’s work, it’s a very negative signal on the job market’ or ‘AS placed well precisely because the topic of her JMP has nothing to do with her advisor’s work’, etc. But on the execution, you should listen carefully to the advisor’s (conditional on you having an advisor who has a track record of placing people well in recent years) because they know what the market is looking for in a JMP in *your field*: as I mentioned before, there are impressionism, realism, etc. in art and you need to draw in a certain way to be recognized as part of the gang — same for academia. There’s a certain way to write a paper to indicate that ‘you’re an IO person’, or ‘you’re a development person’, etc.

Q&A + Miscellaneous

Q: Wait, so did you work for Raj Chetty?

A: No — when working as a full-time RA at Harvard I was funded by the LEAP Pre-Doctoral Fellowship, which is something Raj created, but I worked for MA, not Raj Chetty.

Q: I heard these RA programs only hire US citizens

A: Not true — nowadays many of these programs provide J1/H1B sponsorship. I worked under J1 visa. One important thing you should know is that, if you used a J1 visa, you’re very likely under a ‘two year residency requirement’, meaning that you need to stay at your home country for 2 years (cumulatively) before you can apply for H1B or green card in the US, unless you apply for a ‘J1 Waiver’ from the Department of State + USCIS. If you’re in the same situation, check DoS and USCIS website and your home country’s relevant website for detail.

Q: I’m a master student outside of the US but really want to go to a top US econ PhD. Is it too late? What should I do?

A: As mentioned before, rec letters from ‘within the circle’ is key. ‘within the circle’ doesn’t really mean ‘within the country’ — for example, there are European (or even Asian) universities with faculty who are very connected with top US academia.

  • there are people with a previous math or engineering phd (or dropped out of such a phd) and got in top US econ phd
  • there are people who went to do banking/trading/hedge fund for a few years and got into top US econ phd
  • Did a master in China, then another master in the US, then got in
  • Did a master in the US, then 3 years of RA, then got in
  • Did 2 years of RA at the Fed, then 2 years of RA at a top econ dept., then got in
  • If you’re an undergrad somewhere outside the US in a non-target school, transfer / do an exchange / apply for a visiting undergrad student program, to come to the US, then work as part-time RA, do well, get a summer full-time, then do well, get a full-time RA job for a year (in this process, you don’t need to stay at the same institution/work for the same advisor; you can use their letter to bootstrap upwards: from a top 50 to a top 20, then to a top 10, then to top 5, then to top 2 institution or RA program)
  • If you don’t have the money/time to transfer or do an exchange/visiting year, send cold emails to all top 50 US econ dept’s faculty asking to work for them part-time remotely for free. To successfully get attention, you should already know Stata/R and econometrics, i.e. know how to plot graphs and run regressions — in other words, you need to be useful. You’ll find something — probably not from Harvard/MIT, but it’s a starting point. Then, again, you can bootstrap upwards: if you did well in your part-time RAship, ask to come to the US over the summer to ‘work’ non-paid full-time, then use that letter to get in US top master programs or top RA programs, etc.
  • If you’re already a master somewhere, things are not *so* different — the above should apply, i.e. either break into the US by education (i.e. another master in the US) or RAship (cold email to start + bootstrap upwards)
  • If you’re already a PhD, you can try transferring / doing visiting semesters or years. There are people in Stanford econ (more than one) who started as a visiting PhD student. But to be fair, not all universities have such connections with Stanford econ. Again, every institution/program’s connection, or every faculty’s connection is different — check with your advisor or your department chair and see what they have
  • Yet another way is to apply for a program that’s related to econ but less popular and thus easier to get in (because many people don’t know and didn’t apply). For example, at Harvard, other than econ, there’s HBS’s program and HKS’s program, and there’s also the public health program, if you’re already set on doing health econ. At Stanford, there’s not only econ, but also GSB’s econ, GSB’s OIT, and MS&E (‘management science and engineering’ program) whose PhD works with econ and GSB faculty all the time, and some of them even transferred into Stanford econ. At MIT, there’s not only econ, but the Media Lab, and dept of Urban Planning (I’ve met people who are PhD there and work with Ed Glaeser, for example.)
  • This might sound unintuitive but location matters: Disproportionately many Harvard/MIT econ phds eventually place into *geographically* adjacent departments such as BU/BC (Why? Because then they can keep working with people at Harvard!). As a result, if you’re a BU student, it’s very likely that your advisor came out from Harvard, and still work with people ‘in the circle’/at Harvard, and it’s very likely for you to go to Harvard as a visiting PhD for a year (again I know someone like this). I don’t think this piece of info is already priced into the dept ranking. So when choosing programs, consider places like BU that seem not in the top but actually has ‘special connections’ with the top
  • Other than geographic adjacency, you should consider field adjacency — finance, marketing, accounting, operations research, information, urban studies, health, law and economics… these are all programs that are related to econ, and consistently see econ phds placing into. Econ is very crowded nowadays — this can be seen from the increasing number of RA programs and RAs — many of them have profiles that really should be going to a PhD program, but they are still doing RAs because there are too many people who want to get in. However, these adjacent fields are a lot less crowded — many people wouldn’t even give it a thought, making it easy to get in; placement is also easier, for example, NB at Stanford keep telling his student to consider the accounting job market, because how much easier it is than econ/finance.

Q: Did you enjoy/like/have a good time at your PhD at Stanford?

A: Yes I did. But that’s not the right question to ask.

  • At Harvard/MIT there are a lot more people who are depressed
  • But Harvard/MIT placement is also much better than Stanford

Q: So what are some good VS bad reasons to get an Econ PhD?

A: great question. Let me start with bad reasons:

  • ‘Undergrad econ classes are fun/I like thinking about these interesting questions’: Undergrad econ classes are a complete mis-representation of what econ grad school is like and what econ research is about. In chemistry or physics, the undergrad classes teach you the basic version; in econ they teach you the wrong version. Econ departments are smart — they know that, in order to keep econ faculty employed, and keep the whole econ academia boat afloat, they need more young people entering it, including attracting undergrads into majoring economics. To do that, they make the undergrad classes easy (e.g. IS-LM curves shifting around) and fun. As a result, what undergrad econ teaches has nothing to say about grad school or research, which is most of the time math-y, unintuitive, and disconnected from the real world.
  • ‘I’ve been working and felt I haven’t been learning things. I want to learn things by getting a phd’: As I mentioned before, consumption and production of knowledge are two completely different things. In undergrad or a master program, you pay the school money and consume knowledge. In grad school, the department pays you money and you produce knowledge aka papers. When you’re paying, you are a customer, and of course it’s their job to make you happy. But when you’re in grad school, you’re a worker/employee, not so different from who you are right now at the firm. It is your job to make sure you’re learning things. You’ll most likely find the first year grad sequence un-interesting/irrelevant, and 2nd year onwards it’s all up to you to learn what you need to produce research, and you’ll be focused on a very specific topic — if you thought coming to econ grad school means you’ll be able to talk about the stock market, Fed policy, bitcoins, macro trends, trade policy, at a dinner table with your family, you’re completely mis-informed. Econ grad school teaches you nothing about comprehending our economy or economic policies in general. Instead it asks you to dive extremely deep into one tiny little topic that most people at a party won’t be interested in
  • ‘I’m really good at school. I don’t like talking to people. So I don’t want to work.’: being good at school doesn’t mean you’re good at research. As I said, your job in undergrad or masters is to take classes and consume knowledge. That’s completely different from grad school where you have to come up with new ideas and execute research projects. Also, increasingly, being able to present is extremely important on the job market. If you ‘can’t talk to people’ that’s a very good reason not to come to econ grad school… There are disciplines where ‘salesmanship’ is less important, like hard sciences or lab sciences, where the ‘facts’ are still more important. Econ is too ‘soft’ to be objective, and it’s all about opinions, the story, whether you can sell it, whether you’re in the network, etc.etc.etc. If you really hate ‘playing’ the politics / schmoozing / networking / talking to people / sell yourself, it might not be a great fit.
  • ‘I’m in the Fed/European Central Bank and to rise up the ladder I need a PhD’: go ahead! But notice that in Europe there are PhDs that you can get part-time, so you don’t need to waste 5–6 full years of your life just to get the degree for you to rise up the ladder
  • ‘My whole family is in academia/econ academia. I have a lot of insider info, know how the system works, have connections, know the playbook, can get advice on how to succeed in academia from my dad.’ — despite how sad it is, still, in the US nowadays, many of these ‘upper middle class professions’ rely heavily on info/connections. For example, when I was an undergrad, trying to break into investment banking, I realized that you need a summer internship to get a full-time, and need to already have had previous summer internships to get a summer internships at the ‘Bulge Bracket’ banks in your sophomore year. But where to get the ‘first’ internship? — from your family connections. This is true for econ as well: a large part of doing well is knowing the system and knowing how to ‘play’. If you already have that through family connection, you essentially already have a head-start.
  • ‘I want to become an upper-middle class in the US, and considering the options, academia is the one where I’m most likely to succeed, and I’m happy to play the game’: Like it or not, for a first-gen immigrant with good grades to remain in the US, and join the ‘upper middle class club’, academia is still one of the best options. Other such ‘paths’ include: get a CS or related degree and become a programmer and get a job in FAANG, get H1B/green card sponsorship. Or do finance (investment banking, etc.) after undergrad and get similar work visa sponsorship. But there are always people who fail the H1B lottery and has to go back. So if you absolutely want to stay in the US, and set your kids for a good start, academia is really not a bad choice, especially if you’re already good at math and don’t hate research and don’t mind ‘playing’ the game


-- Senior Data Scientist@Shopify. PhD in Economics@Stanford. Ex-google.

Love podcasts or audiobooks? Learn on the go with our new app.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Scarlet Chen

Scarlet Chen Senior Data Scientist@Shopify. PhD in Economics@Stanford. Ex-google.