←
Henrik Sigstad61 posts
Henrik Sigstad avatar

Henrik Sigstad

@hsigstad

Economist studying legal institutions

◷ Joined September 2009▧ 1 Photos and videos
Ingar Haaland avatar
⟳ Henrik Sigstad retweeted
Ingar Haaland@Ingar30·Nov 11

How should PhD students leverage AI in their research? First, you should write papers where you leverage AI to analyze something at scale. There are many low-hanging fruit because gen AI is essentially still a new thing. Exploit it!🧵

Fernando Meireles avatar
⟳ Henrik Sigstad retweeted
Fernando Meireles@meirelesff·Nov 6

Eleição após eleição há discussão sobre erros nas pesquisas. Eu e o Henrik temos um novo working paper mostrando que, sim, há erros, mas onde eles ocorrem não é tão óbvio. Explico 🧵 papers.ssrn.com/sol3/papers.…

Post attachment
Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Nov 6

The bias can't be explained by the poll's declared design or methodology, so compliance with formal rules doesn't solve the problem. But we do find suggestive evidence that pollsters' reputational concerns can limit the bias. Full paper: papers.ssrn.com/sol3/papers.…

Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Nov 6

Do candidates tilt the polls measuring their own support? Yes. In a new WP with @meirelesff, we find that in Brazil's 2024 mayoral elections at least 1 in 8 polls are paid for by a candidate or party in the race. Those polls overstate their sponsor by ~8 points.

Post attachment
Ingar Haaland avatar
⟳ Henrik Sigstad retweeted
Ingar Haaland@Ingar30·Sep 6

HOW TO REFORM ECON REFEREEING WITH AI Frontier AI is clearly extremely useful in the review process, but it should be done right. The right model is not that all referees ask ChatGPT to point out strengths and weaknesses of the paper. (The point of asking human referees is to get their human judgement.) The right model is something along these lines (helpfully illustrated by Astra in the flowchart): 1. If the editor plans to send the paper out for review, they first generate a frontier AI report (e.g. using a modified version of my open source paper reviewer. I think open source is nice here because the journal can modify it to their purposes.) 2. The authors then get the AI report back from the journal and are given the opportunity to provide comments, explain how they plan to address the points in the report (if they agree with them), and push back on any points they disagree with. 3. The journal then sends out the initial manuscript, the AI-generated frontier report, and the response from the authors. This makes everything constructive and avoids cheap duplication of AI efforts as well as removing the incentives of referees to suggest comments that most good AI systems are likely to flag in any case. The next extension should be to require authors to submit a full replication package with code and data at the submission stage to check consistency between the paper, code, data, and reported results.

Post attachment
Jonathan Roth avatar
⟳ Henrik Sigstad retweeted
Jonathan Roth@jondr44·Apr 23

I’m really excited about this paper! Some of my work has pointed out problems in empirical work, but this one is all about new 🔧s. If you (or your referees) want to know about the mechanisms by which a treatment affects an outcome, you may be interested. A 🧵.

Want to know about the mechanisms by which a treatment affects an outcome? This paper develops tools for testing hypotheses about mechanisms under weak assumptions. Check it out! New paper by @jondr44 and Kwon: restud.com/testing-mechanism… #REStud #EconX #EconTwitter

Post attachment
Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Apr 17

Includes pointers to where Brazilian institutional data lives. Feel free to fork and ask Claude Code to extend the repo to your area.

Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Apr 17

On top of dense notes on each institution, a resolver tool (`cite.py`) fetches verbatim statute text by compact ID (`CF.37`, `LIA.10`, `L14133.75.II`) and resolves STF cases and súmulas. CF articles come with STF annotations; CE articles with TSE.

Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Apr 17

Built with Claude Code and meant to be used with it. Point the agent at the repo and it will have a grounded understanding of Brazilian institutions when helping you design empirical work, instead of making up statute numbers or misreading legal terms.

Henrik Sigstad avatar
Henrik Sigstad@hsigstad·Apr 17

A possibly useful resource for people working on empirical papers in Brazil: github.com/hsigstad/brazil-i… A reference on the legal and political institutions — courts, procedure, MP, elections, procurement, corruption statutes — and how they work in practice.

Megan Stevenson avatar
⟳ Henrik Sigstad retweeted
Megan Stevenson@MeganTStevenson·Apr 5

If you're feeling a bit down about social science after that Nature article, consider the new approach to econometrics we outline in this paper. There's lots of valuable work to be done to reduce the problem.

Excited to share a new paper with @jfischman, just accepted at JEL. We argue that empirical research tends to be biased and overconfident due to a weakness in the dominant econometric framework: insufficient attention paid to humans “in the loop” with the research process. 1/

Post attachment
𝗡𝘂𝗻𝗼 𝗣𝗮𝗹𝗺𝗮 avatar
⟳ Henrik Sigstad retweeted
𝗡𝘂𝗻𝗼 𝗣𝗮𝗹𝗺𝗮@nunopgpalma·Mar 30

📢New paper: 𝐇𝐨𝐰 𝐚 𝐧𝐚𝐭𝐢𝐨𝐧 𝐰𝐚𝐬 𝐛𝐨𝐫𝐧: 𝐁𝐫𝐚𝐳𝐢𝐥𝐢𝐚𝐧 𝐞𝐜𝐨𝐧𝐨𝐦𝐢𝐜 𝐠𝐫𝐨𝐰𝐭𝐡, 𝟏𝟓𝟕𝟒–𝟏𝟗𝟐𝟎. Joint with @glambais. We built the first long-run GDP per capita series for Brazil, using 30,000+ archival price and wage observations across major regions🧵

Post attachment
Peter Hull avatar
⟳ Henrik Sigstad retweeted
Peter Hull@instrumenthull·6 Nov 2025

Paul G-P (@paulgp), Michal Kolesár, and I have a new guide to leniency designs (aka judge/examiner IVs), prepared for the JEP The bottom line? Keep calm and UJIVE on!

Post attachment
Henrik Sigstad avatar
Henrik Sigstad@hsigstad·20 Mar 2025

Devemos restringir o acesso aos tribunais eleitorais para evitar abusos? Provavelmente não—os tribunais são vitais para a integridade eleitoral. Mas regras como sucumbência, multas por litigância de má-fé e a limitação de ações pré-eleitorais podem conter processos infundados.