Are Ideas Getting Harder to Find?

Interviewer: Do you like Stony Brook? 

Ekerdt: I like Stony Brook University. I mean, it's my first job as a professor, so I'm still getting used to teaching, as you probably can attest to. But no, I like it. I mean, the department's really nice. 

Interviewer: Nice, nice. Good to hear that you like Stony Brook. You came from the University of Rochester? 

Ekerdt: Yeah, well, I came from Washington initially because I was working with the census, but before that, Rochester. Rochester is like an alumni conference, uniquely to PhD programs. And I've noticed that like, it seems like everyone's arc at Rochester is like, they come first here and they're like, oh my god, this is the worst place in the world. You know, it's a small American city and then it's really cold, snows a lot. And then by the end, they're like, oh my god, Rochester is so nice. Forests, lakes, it's not too expensive. There's not much traffic. Yeah, I like Rochester. 

Interviewer: I got that kind of Stockholm syndrome with Stony Brook. So into the meaty stuff, why do you choose to study economics? 

Ekerdt: Yeah, okay, so let me think about it for a moment. So my parents are both researchers, they're both in pharmaceutical research, or were, I should say. My dad's retired by now. So I always felt like I wanted to do some kind of research and then I was never sure between biology and economics. The reason I got interested in economics is because I took a high school economics class during the financial crisis, which I thought was really exciting. And it seemed like there was a lot to do there. So I did a biology econ double major in college. I even considered pre-med. But so at some point I decided I was totally not cut out to be a doctor A, just like personally. And I'm married to a doctor. And B, that I didn't really want to work in a lab. So then, yeah, economics I think is fun because it's like you get to be really curious in what topics you choose and what methods you employ and so on, which I think is fun. 

Interviewer: You were an economist at the Census Bureau. What was that like? 

Ekerdt: Yeah, so I was a postdoc and then I was in their research department, which basically is like they have tons of data, right? Including a lot of like microdata, so like things they never publish and which are actually restricted legally. But the goal of the research economists is to put those things together into new ways to make new data sets and get new insights out of that. So that's what I did there. And my project there was to attach demographics to business owners. A lot of people talk about who owns a business, etcetera. And it turns out it's quite hard to say what is the gender or the race or the age of this person. Usually we only have survey data for that. But in principle, you can do it by linking tax records and things like the census. So that's what we were doing there. And it's pretty far away from my own research, but I think it's really fun to do things like that.

Interviewer: What makes stuff like that important? Why do we still need it? 

Ekerdt: I mean, I think it's always good to know what's going on in the economy, right? If you want to make decisions about the economy, the best thing to do is to first figure out what's going on in the economy, right? And so, if you were concerned about representation among business owners. I mean, step one would be to just figure out what representation is among business owners, right? And how's it changing over time? And the survey is okay with that. But this, what we're doing, I think it's a lot better. 

Interviewer: Do the Census and the Fed work together at all? 

Ekerdt: A little bit, yeah. There's, like, one division at the Federal Reserve Board of Governors, for instance, that is staffed with former Census employees which is their industry output section to make statistics, basically. There's definitely some back and forth. They tend to be a lot more focused on the day-to-day of the economy, whereas the Census moves very slowly. In most cases, they're trying to change that. 

Interviewer: And so more of a process question. How did the process of doing that differ from your current research? 

Ekerdt: Yeah, yeah. So I mean, I split my time there, like I did my own research and this thing. So usually they say, 50-50. And maybe mine, I did a little more of my own research, because I was a postdoc. But I think it's a little different, because it's like, first, the audience is much different. So like, mostly we're presenting to higher ups in the census. And then my boss, for instance, was thinking a lot about l the quote, unquote, stakeholders. So this would be people who use the data. And a lot of companies or researchers, but not necessarily in economics. And so then when you think about things like how do we structure the data to be useful and not confusing to them? You need to start thinking about things like who's going to use the data and how are they going to use it, and what would they find most useful, and what would be most misleading to do, things like that. And then process-wise, I mean, it's a ton of data work, and a lot of, like, things that would only ever end up in the appendix of a research paper. If we change this definition slightly, are things going to change a lot, because they shouldn't, things like that. And then it's a little more, like, team-oriented. I mean, so, my papers are mostly with one other person, one of my classmates, and we work in a team, but not as actively as this. So, like, my team was, like, you know, seven or eight people, and we met once a week, and split up tasks, and passed things back and forth quite a lot, which is a little different. I think it's a fun compliment to research. Research, the progress can sometimes be very slow, which is frustrating. Whereas here, you can, like, you know, kind of always make a little progress, at least. 

Interviewer: What is a working paper? 

Ekerdt: A working paper is a paper that you usually haven't submitted to a journal yet. You're working on it actively and you're presenting to people. In economics at least, it's very usual to travel around to universities or conferences and present something that you're working on. And then for other people to see it and have ideas and for them to give you comments about it. Usually working papers, like, once you have a full draft, you post it online so people can go read it when they want. 

Interviewer: At what point do you get to submit it? 

Ekerdt : I mean, I think at some point you just decide, like, you've reached a point for, like, putting a lot more work into it. But it wouldn't change much, it's complete in some sense. It's a little hard to say what that is. You submit, right, and then there's something called a desk reject, which is that the editor looked at just the manuscript, usually it's the abstract and the intro and said, “no thank you.” You know, if you get desk rejected everywhere, then, of course, you would have to conclude that maybe I need to work on it a little more. 

Interviewer: Two of your current working papers are about the economics of research, the role of R&D factors in economic growth, and self-selection and the diminishing returns of research. What drew you to that topic? 

Ekerdt: Yeah, so, there's a quote by an economist who said, once you start thinking about growth, you can't think about anything else, which I think applies. I don't know, like, growth to me is a fundamental economic topic because it's about living standards, right? And, how do living standards grow, and what determines how they grow? I think modern economics kind of focuses on research as the thing of how we continue to grow. So it used to be about, you know, capital accumulation and things like that, but now it seems like it's mostly about changing technologies, and so then about research. So that's, like, broadly how I got to the topic. And then I think, idiosyncratically, it's that my co-author and I were in a reading group in the second year of PhD, and we did a paper about whether ideas are getting harder to find. And we were the only two people that were super interested in the topic of the people in our PhD program. So we decided we should work together on something like this. 

Interviewer: Is that commonly how people pick their research topics? What are other things that drive people to a research topic? 

Ekerdt: I hope that everyone picks something that is personal to them in some way or they really enjoy doing. I think partially it's about who's going to be your advisor? Like, who are you personally compatible with? Because, you know, you're never personally compatible with everyone, even if you really like their topic. And then sometimes, like, you pick your advisor and then you work with them, and then you start basically doing, you know, whatever their topic is. I think that's a really common path too. 

Interviewer: So what is generally the role of an advisor when you're completing a PhD or doing your own research? 

Ekerdt: Yeah, so at least in Rochester, I think it was, like, that we would meet with our advisor, like, you know, every three or four weeks or every two weeks, depending on the time of year and how close we were to the job market, and talk about what we've been doing. And he would give us advice, he would tell us, “you guys should check this and this and this,” or “I'm concerned about this”. So, like, very specific topics about what we're doing. And then also, like, I think where the advisor's most helpful is, like, framing. Because, you need to present your paper to an audience, and they need to think it's interesting, right? Given whatever results you have, there's different ways to frame it, right? And so figuring out that framing is really hard as a PhD student just because you haven't seen that many presentations. So you don't quite know how to do it. So I think the advisor's especially helpful for that. Then in the end your advisor usually is, like, really key about getting a job because they're writing you a recommendation letter. And you need a solid recommendation to get a good job. 

Interviewer: Economics is obviously, like, this ultra-quantitative field. Do you guys ever work with qualitative data? 

Ekerdt: Do you mean, like, interviews and things like that, for instance? 

Interviewer: Interviews, things that literally are not numerical. When you're working with qualitative data, you have to code it and make it into something that can be used in a software. But interviews, text, stuff like that. 

Ekerdt: I see, I see. I personally haven't used it. And at least in Rochester, I don't remember that there were any people using things like that. But I think it is becoming, like, more common, especially with AI tools, to, like, try and process text into data. One thing I saw is sometimes on surveys they have a write-in section where someone can just do free form and so on. And then usually it's completely impossible to use that because you might have to go through thousands of responses to try and figure out some commonalities. Or if it's not possible it gives you too many degrees of freedom as a researcher to decide that they're saying what you wanted to say. But if you have an algorithm to do, like, what's the sentiment of this, then you could do it. So I think more and more that will be used. 

Interviewer: Follow-up question from that specifically. Could you describe your process of data collection for your research? 

Ekerdt: I don't collect my own data generally. I'm using a survey from the NSF that they give to college graduates. And then also things using, like, the census. So basically public use data, which sometimes are gigantic data sets, like the decennial census. And then my other project mostly uses restricted use census data. So this is the micro data that underlies, like, the statistics they publish, which you can get access to. It’s just kind of a pain in the ass. You need to apply. You need to do some kind of background check and things like that to make sure that you're not stealing it. And then usually you need to access it through a secure platform. But I would say most people in economics don't have sufficient funding to do their own data collection. And I'd fall in that vein as well. Data collection is so much work, so much labor, so much manpower. 

Interviewer: One of your claims slash findings is that there are diminishing returns of knowledge production. And that researchers are becoming literally less able. Why do you think this is so? 

Ekerdt: Let me first say that the first thing is not our finding. That's the paper I was telling you about “are ideas getting harder to find”. Which, roughly speaking, if you take productivity growth in the U.S. or most developed countries it's flat or declining. On average 2% every year. But then you take investment into research, which we would think should make productivity grow right? So it looks like we get less and less out of our research unless we're mismeasuring something dramatically. That’s not our claim. What we say is, look at the share of researchers among the labor force. It's tripled since 1960. Much more of the labor force is doing research. And we think it's implausible that you haven't changed the composition of researchers. So we're not saying a given researcher would be less able or more able in 1960 versus now. I mean, that one would even say it's probably that they're more able now, right? Because they have all these technologies that they can use to help them do research. But aside from that, what we're saying is that people select into different occupations, right? And they select based on their returns. So not just financially but they like doing it, for instance. And if you have that view, then, like, when the sector is really small, the only people in the sector are the ones who have really high returns from being in that sector. So they're really good at research. That would be the argument. And then if you make it bigger and bigger as a share, then you need to be drawing in some people who, you know, they're not quite as good as the first people. Because the first people have the very highest returns. So they still have returns that are higher than in other occupations for them. But they're a little lower than the people who started there. And so that would be self-selection. That would mean the average ability of researchers is declining over time. That’s something we're applying to this context. It's not a mechanism we came up with ourselves. Any sector you could argue that the average ability of those workers should be declining in the sector. Because you're drawing in people who are a little worse on average. 

Interviewer: So returns meaning ability to produce research? 

Ekerdt: Yeah, ability is a little nebulously defined, usually. In our paper a component of your earnings. So it would be if you have higher ability in research then on average, we would say that you should be having higher earnings. So it could be because you end up at a particularly good university, they tend to pay more, things like that. 

Interviewer: So there are more researchers, but the amount of research produced stagnates. The explanation that is being given is that there are more researchers, and they're all slightly worse than the one above them. 

Ekerdt: On average, our researchers are slightly worse now. Well, it's not slightly, it's a lot, according to us, worse than in 1960. And it's because the average researcher in 1960 was a part of an extremely select group, and now they're still part of a very select group, but less than in 1960. 

Interviewer: What do you, in relation to the current changing landscape of academia, where do these things connect? Is there a connection? 

Ekerdt: There would be a connection in the sense that if you think, like, we're going to throttle the share of researchers for whatever reason, then what we're saying would stop happening as much, right? Because it's about the fact that we're adding a ton more people to the research sector. So if we stop doing that, then what we're saying, like, no longer applies. And it will still apply historically but from here onward, it wouldn't be like that anymore. My impression is that even though maybe it's harder to get into academia in certain fields, like, being a researcher at any place, not just a university, is still getting easier, I would say. But research as a whole, I would expect it to keep growing. 

Interviewer: Would you say that's just part of how higher education is generally getting more accessible? 

Ekerdt: Yeah. So, actually we showed the paper that it seems like most of the change is just because there's way more college graduates. So the share of researchers among people with college degrees is actually declining. The average college graduate is less likely to do research. Just like from 1960 to 2021, we've gone from the share of college graduates being 10 percent to like 38 or something like that. So a massive change, right? If you're drawing in all these people a lot of them will discover, oh, you know, what I'd like to do best once I have my college degree is to do research. So then you're going to think if the share of people with college degrees stagnates, I think it's slowed down. Then I guess you would also say that maybe your research will stagnate a little bit. 

Interviewer: Do most of the people who you're seeing becoming researchers, do they stay in that career their whole lifetime? 

Ekerdt: Some people, we can see them across multiple surveys. So we can say, like, in this survey, you were a researcher. What's the chance that you're a researcher in the next survey? That chance is actually, like, quite low. It seems like between two or three years something like 20 percent of people move out of research, which makes a lot of sense if you think about graduate students or untenured faculty, things like that where you're in research for a bit, and then you slowly figure out whether you can stick with it or equivalently, whether it makes sense for you to stick with it. And then eventually you're out. I think there's a lot of, I guess, like, discovery, basically, people need to discover if they're good at research or not. 

Interviewer: If it was still that only the best of the best that became researchers, would we have more new findings, new research?

Ekerdt: No, definitely not. We're not saying, the people you're adding don't contribute. We're saying they contribute a little less than the people you had initially, but they still contribute. If we had the same share of researchers that we did in 1960 now, we'd have way less accumulated knowledge, right? Because all these people are accumulating knowledge. I'm definitely not advocating for any kind of exclusiveness. It's like an accounting thing, right? It’s like, if you calculate R&D productivity by saying, like, here's average productivity growth, and here's how many people I added, then you're making the mistake that the people are not all the same, so you can't just add them up. 

Interviewer: Do you have any ideas on how or if this will be affected by the advent of AI? 

Ekerdt: I just have speculations. Because it's so new. You could think, first, that, like, AI should make a lot of productivity growth, and that's still, I think a little unclear. How important it will be in terms of productivity, right? Clearly, in terms of people using it, it's very important, but it doesn't make you that much more productive. We could say at least in the short term, maybe we'll have a lot of growth. And then some people will say maybe it'll make researchers more productive, right? Because they can use AI as an aid, for instance, like an unpaid research assistant, you know? And I think there's a little bit of that right now.Definitely for coding, it's really useful. But I think it's not quite there yet. For instance if you ask it for references, it often makes up references. Or if you ask it to code, like, a whole thing, and not just, like, ask it about a snippet of code, it's going to make a ton of mistakes. So it's kind of like if you had an unpaid RA, but you had to triple check everything the RA did, then I don't know if the RA would be worth using, right? But, yeah, I don't know. I'm definitely not an expert in that field. 

Interviewer: In another one of your working papers, you write about the rise of specialized firms. One of your findings is that specialized firms are more resistant to replacement by foreign producers. What would this mean for the consumer? 

Ekerdt: The consumer as worker or the consumer as buying something?

 Interviewer: Both. 

Ekerdt: The argument we make in the paper is that, like, that people like to buy, quote-unquote, high-quality goods. Imagine you're buying a washing machine. Probably you're only buying one washing machine, right? If you were really rich, you wouldn't go out and buy ten washing machines. You'd buy one really expensive washing machine, right? That's usually what we call a higher-quality washing machine. Maybe it lasts longer, it has more features, etcetera. And it tends to be true that more wealthy, higher-income people will buy higher-quality goods. And then we make the argument in the paper that specialized firms have a comparative advantage in making higher-quality goods. The idea is if you're really good at producing a high-quality good your being really good at it that doesn't transfer very well across sectors. If you're really good at making vacuum cleaners, there's, like, no reason that you'll be good at making wind turbines or something like that. The technological differences between those two sectors expand as you get higher up the quality ladder. In terms of the consumer, I think we don't explore it too much. 

Interviewer: If there is one piece of advice you could give to econ students, what would it be? 

Ekerdt: Stay open-minded in the sense that I think one part of economics that's really fun is that very little of it is settled. So you go into biology and you say “this is the cell, you know, these are the pieces of the cell, and this is mitosis, and blah, blah, blah.” Whereas in economics, like, you know, most things are up for debate. Like, someone can always come along and change 

conventional thinking, which I think is really fun. But then, of course, you've got to be receptive to changing how you think about things. 

Interviewer: Do you have an economics text or book you would like to recommend? 

Ekerdt: I think one book that's really fun is called The Lords of Finance, which I think won a Pulitzer. And it sounds really boring when I give you the topic. It's about central bankers, like, before, during, and after the crash in 1929. And how they were trying to shore up the monetary system. It's kind of written as a thriller. But then it won't make sense if you don't know a little bit about exchange rates and things like that. I think that's a really fun one. 

Interviewer: Is there anything that I should have asked that you would like to add?

 Ekerdt: Don't go to grad school unless you like doing research.