Wisarut Suwanprasert (@invariance) profile picture
Murfreesboro, TN
Joined September 2009
Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)Photo by Wisarut Suwanprasert (@invariance)
Wisarut Suwanprasert retweeted
Is all (referee) information revealed in the long run? Let’s see if someone will claim credit.
Econ history mystery 🕵️: who refereed Ayse’s 1989 JPE paper? Her computational heterogeneous-agent model preceded Huggett and Aiyagari, but gets far less recognition. The referee saw the methodological breakthrough. We have the report—but no name. Does anyone have a lead?
1
2
30
7,836
I was fortunate to graduate in 2016 🫣
Update from the front lines of the econ job market: it's not looking great out here.
2
29
5,000
Wisarut Suwanprasert retweeted
In 2020, Fed researchers graded six economics textbooks on how the Fed actually implements monetary policy. Half scored below zero. The best got 6/12. federalreserve.gov/econres/f…
10
90
27
385
80,696
A very sad day for macro and Penn economics. "Equilibrium Price Dispersion" is one of these papers every macroeconomist should read and re-read.
Gracias totales, Ken.
6
59
490
32,485
Wisarut Suwanprasert retweeted
My friend Laura has 56 days to find a job or leave the US. Last Friday her company NextSense furloughed the whole team. Laura is on an H1-B, which requires her to be paid while employed so her role was terminated and the 60 day clock started. She's a biomedical engineer with a PhD in neuroscience. She builds ML for sleep and time-series signals like EEG/EMG and she has built wearable AI models in production. She's looking for research scientist roles in the Bay Area. If you're hiring in health tech or wearables, DM me and I'll intro you. @ycombinator founders: if you need a founding ML or research hire, she's it. Her H-1B transfers so no lottery needed. And if you're not hiring, a repost might reach the person who is 🙏 linkedin.com/in/laurarose95/
1,800
1,368
147
5,158
1,911,712
I am sharing a new paper today with Dario Caldara: “Can Peer Review Survive AI-Generated Text?” sas.upenn.edu/~jesusfv/Peer_… Dario and I, as editors, are increasingly getting papers submitted to our journals that are 99 percent AI-generated. Even worse, we are also receiving many AI-generated referee reports, both as editors and, more frustratingly, as authors. And we cannot emphasize this enough: the real problem is not that the papers and reports were AI-generated; it is that no one had done the thinking behind them, and you can so easily tell. The papers were eminently forgettable, and the referee reports were disappointingly uninsightful. We have seen too many reports of the form: “The paper is well-written and addresses an important problem. However, the authors should consider additional baselines.” This deluge of AI-generated text puts peer review in jeopardy: journals will be flooded with low-quality submissions, economists will decline requests to referee papers or routinely use AI to referee them, and which papers get published will become less and less linked to their quality. We do not want to defend the status quo entirely. Peer review is slow, many reports are poor, and editors (including us!) make bad calls all the time. Still, well-executed peer review remains the best system we have for assessing each other’s work and advancing the profession. AI could improve peer review if used deliberately for that purpose, for instance, to check proofs and code. That is not what we are seeing. The difficulty is that readers, including editors and referees, cannot see how AI was used. As a result, in practice, the burden of verification shifts from the author to the reader. Since we reject this shift, Dario and I did what economists always do when something bothers us: we wrote a paper. So, what do we do in it? First, we check whether AI detectors such as Pangram get this right: is a referee report flagged as AI-generated indeed AI-generated? Or is Pangram just flagging proofreading and editing? No. We run an experiment with 25 referee reports from the pre-AI era. Proofreading and editing do not trigger AI-generated flags. Only when we ask the LLMs to substantially rewrite the referee report or to generate a new report do we get the AI-generated flags. Pretty compelling evidence, if I may say so. Second, we build a theoretical model to explore who should bear the burden of verification: the author or the reader? (Yours truly always enjoyed the courses on the rules of evidence in law school; yes, as Billy Wilder would say, nobody is perfect.) The answer is that it is much more efficient for the author to do it, because the author can verify at a much lower cost and only needs to do it once (rather than each reader repeatedly). Before AI, the authors more or less had to verify the results, even if only because they had to type them in. This was not perfect (and all papers ever published in economics probably have at least one mistake). But the problem is that, with AI, the author does not even need to read the text. Or can just skim it. Writing is thinking, and without writing there is usually not much thinking. So, a paper flagged as AI-generated is prima facie evidence that the author has not verified it. Should this lead to an automatic rejection? No. But there should be a zero-tolerance policy for serious mistakes. The first serious mistake we find in the paper, and it is out! In comparison, for a non-flagged paper, a serious mistake is something that the editor might determine it can be fixed in a revision. We are stricter with referee reports. There is absolutely no point in the referee generating a report using an LLM. As an editor, I can do that directly and save myself the trouble of asking you for it. Now, of course, you can use the LLM to help you with the report (if the journal allows it! Many journals do not allow you to upload the paper to an LLM, and you need to follow the rules). For example, you can use the LLM to search for existing results in the literature. But peer review is not about finding one sentence that is not entirely precise or a slightly sloppy mathematical proof. Referees are asked: “Is this idea important despite its flaws?” And, relatedly: “Is the paper a good fit for the journal in terms of quality and focus?” That judgment is the task for which you are selected as a subject expert. We also propose that journals require that, before a paper is submitted, the author submit it to an LLM using an approved prompt and include the LLM report and a discussion of what was and was not fixed. That way, the referee would not need to check (or at least not check so carefully) every technical detail, and they could focus on the big picture: Is this a good paper? Is this a good fit for this journal? What is fundamental for this paper to become publishable at this journal? Anyway. The paper has many other arguments, and we try to be nuanced. So, before ferociously typing the first complaint that crosses your mind, would you consider reading it? Chances are that we have already discussed your objection. We have been thinking about this for a long time. But by all means, leave good comments!
65
242
35
1,032
102,696
Wisarut Suwanprasert retweeted
The more home ownership or marriage seem out of reach for young adults, the more they turn to financial gambles such as crypto currencies. nber.org/papers/w36010
5
27
6
116
9,508
I'm happy to do for the price of zero
Economics job market candidates whose job market papers were promoted by established big accounts received ~20% more flyouts and ~30% more job offers.
3
331
Wisarut Suwanprasert retweeted
California professor who teaches at nine colleges rakes in $800,000-a-year by holding 30 classes with 1,100 students | Victor Heller, Daily Mail A new investigation found that professors are playing California's education system by teaching dozens of classes with pre-recorded online lectures, allowing them to cash in big time. Thien-Huong Ninh Villarreal, who teaches sociology, ethnic studies and business, has managed to hold two full-time and seven part-time jobs at different community colleges in the Golden State and made nearly $800,000 in 2024. Ninh Villarreal is currently teaching 30 classes to as many as 1,100 students, an investigation by CalMatters revealed. And she is not alone. There are no rules limiting how many colleges or districts an instructor can work within and no tracking systems to monitor a professor's workload as each of California's 73 districts is governed independently. This has led to a widespread phenomenon, where more than 70 teachers are leveraging these conditions to their advantage and giving lectures across at least five different community college districts, Cal State or University of California campuses at the same time, CalMatters found. Sarah Thompson, president of the Faculty Association of California Community Colleges, told the outlet: 'Obviously no one has enough time in a week to teach 30 classes and do a good job, so that’s very problematic.' Another instructor, Andrew Vanden Heuvel, taught around 50 online classes and labs in astronomy to students in various California cities and even at schools in New Mexico, Michigan, Washington and Florida - all from an office above his garage in Michigan. He instructed as many as 1,000 students in a single semester, which earned him a salary of $758,485 total in 2024, according to the report. Vanden Heuvel insisted his students still get excellent education from his material, and that he is the only astronomy professor available at some rural colleges. He posts much of his material on Youtube, where he has garnered millions of views. He told CalMatters: 'Now, did I push it further than some people think is morally conscionable? I don't think so.' He also explained that 'in its current form, I am not grading anything' but instead all of his quizzes are multiple choice and graded by computer. Bobby Seals also teaches 14 extra courses at four districts aside from his full-time job as ethnic studies professor at Evergreen Valley College in San Jose. He has about 930 students this semester. CalMatters listed four more professors teaching 10 or more classes at different campuses, including disciplines such as computer science, psychology, business and art history. Community colleges evaluate the quality of all courses, but have no insight into the actual workload of their professors. The Cal State and University of California systems can track employment and limit the number of courses each professor can teach. But they might not know whether a professor is simultaneously teaching courses at a community college. Read more: dailymail.com/news/article-1…
754
1,552
316
5,867
814,945
Wisarut Suwanprasert retweeted
My favorite thing about presenting at an academic conference is getting feedback from the audience.
104
171
55
7,212
261,708
Wisarut Suwanprasert retweeted
It's essentially what Ed Prescott called a chicken paper: 1. People like chickens. 2. The private sector won't produce chickens. 3. Therefore the government helps by producing chickens. (I think that's from Prescott, but I got it from a presentation @HaraldUhlig gave several years ago on DSGE models). SIDE NOTE: We have good reasons to think that competitive pressures probably limit the scope for discrimination. Why? At its core, a claim of discrimination is a claim about mispricing -- i.e., about the existence of arbitrage. You have to explain why different prices for the same services persist, not least of all because of how this interacts with other government policies. A great motivation for getting into lobbying is rent seeking. Maybe some of what we see *is* discrimination, but it persists partly *because* the government is busy doing stupid things. Another issue is that the implementation of DEI almost guarantees misallocation of workers and capital across sectors. Why? A good example is that racial categories such white, black, hispanics, etc. correlate with cultural dispositions towards leisure, wealth, social status, etc. Basically, different people have different views about what constitutes a good life -- i.e., their utility functions are different -- and it would affect their choices even absent any discrimination. If all cultures are not perfectly evenly spread across racial groups, the problem of undoing discrimination is not trivial. A similar comment applies to composition in terms of age (old and young people don't have the same priorities and consequently don't strike the same trade-offs). But, in practice, DEI policies are basically enforced as quotas. They would therefore ask the government to undo choices people make freely even in the absence of any discrimination because those choices don't avarage out to be identical... That's a cost. Another related issue is that if you have a really small pool of, say, men who want to be nurses relative to women who want to be nurses, but you need it to be a 50/50 split, you're almost certainly going to hire worse nurses. That's another cost. And shifting opportunities across sectors conditional on things like your skin color and what kind of genitals you happen to have means that you will likely misallocate talent just as you misallocate dispositions. That's a cost. This might also change capital allocation across sectors by affecting productivity and labor costs. That's another cost. If you're going to tackle these problems, don't just posit the existence of discrimination and conclude that undoing the mispricing you imposed from above is beneficial. That's silly.
'DEI is good because I simulated a world in which DEI is good.' We need a term for this type of argument where you simulate something unlike the real world to try to argue something about the real world.
3
12
128
17,617
There are too many dimensions to this discussion. I think we should distinguish four separate economic questions: 1. Is $51K enough to live on in Boston? Perhaps for a single graduate student sharing an apartment, but probably not for someone who is married with children.
7
1
39
11,605
4. Is working toward a PhD the same as working as a university employee? Which activities count as employment, and how should working hours be measured when the same activity may contribute to both academic training and university research?
1
1
6
1,865
These are four distinct economic questions, and the answers need not be the same. We should be careful not to conflate the cost of living, the market value of labor, the value of education, and the distinction between studying and employment.
1
12
1,689