Gábor Békés

@gaborbekes.bsky.social

6.5K followers 2.6K following 2.3K posts

Economics Associate Prof, CEU Vienna.
Trade, applied IO, organisations. sites.google.com/site/bekesg
Author, "Data Analysis for Business, Economics, and Policy" (Cambridge UP) https://gabors-data-analysis.com/
@arsenalfc fan. .. more

Videos

Economics 56%

Business 25%

Posts Replies Media Videos

Pinned

Gábor Békés @gaborbekes.bsky.social · Jan 22

Cultural homophily is persistent, pervasive, and consequential, even in superstar multinational teams of high-skill people with common objectives, doing well defined, interactive, not particularly culture intensive tasks.
pubsonline.informs.org/doi/abs/10.1...
Oh, we use football data. #EconSky

Reposted by Wolfgang Huber, Gábor Békés

useR! - The R User Conference @user-conf.bsky.social · 1d

We are honored to announce Dariia Mykhailyshyna (@dariia.bsky.social) as a Keynote for useR! 2026 in Warsaw!

A postdoctoral researcher at the Kyiv School of Economics and a recent PhD graduate from the University of Bologna, Dariia has been an avid R user for seven years.
user2026.r-project.org

$A poster for the "USER! 2026 CONFERENCE WARSAW," which is scheduled for "6th-9th JULY 2026" at "WARSAW SCHOOL OF ECONOMICS & ONLINE." The poster announces "Keynote Speaker Dariia Mykhailyshyna." On the right is a photo of Dariia Mykhailyshyna, a woman with brown hair wearing glasses and a white t-shirt with "KYIV" text and an illustration of a building. Below the conference details are four colored squares (blue, yellow, orange, purple) showing a progression of a fractal geometric pattern, starting from a single square.$

Reposted by Gábor Békés

Gergely Orosz @gergely.pragmaticengineer.com · 7d

One thing every tech company is doing but few share with anyone:

evaluating AI tools for devs - for coding, for infra (eg gateway), for code review etc. Tons of vendors, unclear which one to buy.

Share what you / your team found, and I'll share what others are seeing and what they measure (cont'd)

Gábor Békés @gaborbekes.bsky.social · 7d

Yes clear event study time design is super key.

I recall @koren.mk found some issues for stata re base time
koren.dev/software/

Software

Lorem ipsum dolorem

koren.dev

Reposted by Gábor Békés

Alex @s3alfisc.bsky.social · 7d

I've been wondering if pyfixest syntax is ideal for the new DiD estimators - what moderndid does seems to make very good sense to me - first there is a model class call, and then users can specify aggregation levels of ATTs. This is more or less the same API that the R did package uses:

Gábor Békés @gaborbekes.bsky.social · 7d

I just learned the term #FAFO.
It's legit in international context, ok. But can I use it in college classrooms?

Gábor Békés @gaborbekes.bsky.social · 7d

What as european i checked today.
polymarket.com/event/will-t...

Will Trump acquire Greenland before 2027? Predictions & O... | Polymarket

Real-time odds on "Will Trump acquire Greenland before 2027?" as of January 3, 2026.

polymarket.com

Reposted by Gábor Békés, Maksym Polyakov

Gábor Békés @gaborbekes.bsky.social · 8d

@s3alfisc.bsky.social Alex can you not combine? Id rather keep pyfixest syntax as gold standard.

Reposted by Valerie Mueller, Gábor Békés

Alex @s3alfisc.bsky.social · 8d

There is a new-ish Python package, moderndid, that implements almost all of the "modern" DiD estimators:
github.com/jordandekler...

GitHub - jordandeklerk/moderndid: Python package implementing modern DiD estimators with diagnostic tools and sensitivity analysis.

Python package implementing modern DiD estimators with diagnostic tools and sensitivity analysis. - jordandeklerk/moderndid

github.com

Reposted by Gábor Békés

grand theft eigenvalue 🔆 @akhilrao.bsky.social · 11d

yeah the CLI is wolframscript, this is the MCP tool i used: github.com/paraporoco/W...

GitHub - paraporoco/Wolfram-MCP: MCP server for Wolfram Language mathematical computations

MCP server for Wolfram Language mathematical computations - paraporoco/Wolfram-MCP

github.com

Gábor Békés @gaborbekes.bsky.social · 13d

Gábor is human and can make mistakes.
Please double-check responses.

Gábor Békés @gaborbekes.bsky.social · 13d

Massive coolness factor.

Reposted by Gábor Békés

Paul Goldsmith-Pinkham @paulgp.com · 13d

[Trying this again with the right pictures!]
Really pleased with this paper -- posting a new version with a better discussion of the previous literature, some new results on calendar-time portfolios, and some fixed errors!

paulgp.com/papers/finan...

Gábor Békés @gaborbekes.bsky.social · 16d

I used to think like this.
Then I saw musk posts on destroying the eu and said I dont wanna be part of it.
Bc BlueSky is pretty weak and i no longer see a bold future as is, this means accepting that what used to be EconTwitter is gone.
Sad. But so it goes.

Reposted by Joe R. Feagin, Marc Goergen, Michael A. Clemens , and 2 more Joe R. Feagin, Marc Goergen, Michael A. Clemens, Olivier Mannoni, Gábor Békés

Laura Ballarín 🍒 @lauballarin.bsky.social · 18d

We condemn the scandalous US visa ban on former EU Commisioner Breton and digital activists. Protecting democracy isn’t censorship. ✋🏼

Disinformation isn’t free speech. Europe won’t take lessons from those undermining democracy and free speech.

Here in Europe we set our own laws. 🇪🇺✊🏼

Reposted by Gábor Békés

Economics LMU Munich @econmunich.bsky.social · 20d

🎁 ECONM Advent Calendar – Day 22! 🎁

Congratulations to the research team of @gaborbekes.bsky.social & 𝗖𝗹𝗮𝘂𝗱𝗶𝗮 𝗦𝘁𝗲𝗶𝗻𝘄𝗲𝗻𝗱𝗲𝗿 on their outstanding European research collaboration in the project “Globalization and Industrial Policy: Evidence from the Austro-Hungarian Monarchy (IP-KuK)”!👏

Gábor Békés @gaborbekes.bsky.social · 21d

I think the proactive way is to promote use cases that LLMs can scrape and learn

github.com/gabors-data-...

Release v. 0.9.0. "Frank Exchange of Views" · gabors-data-analysis/da_case_studies

Changelog: version 0.9.0. (2025-08-14) Evolution of code by language since v0.8.3 Changelog summary -- how the content of the repository evolved for each programming language (Python, R and Stata) ...

github.com

Gábor Békés @gaborbekes.bsky.social · 21d

Pyfixest forever.

Alex @s3alfisc.bsky.social · 21d

As everyone is querying LLMs for research these days, sometimes I check PyFixest's SEO game by asking "How can I estimate fixed effects regression problems in Python?". GTP: a) linearmodels b) statsmodels c) hand rolled d) call Julia / R from Python. Pf not (yet) recommended for arcane reasons.

Reposted by Gábor Békés

Alex @s3alfisc.bsky.social · 21d

Reposted by Gábor Békés

CEU Economics @ceu-economics.bsky.social · 22d

2026 january. Vienna.
#AI #education @gaborbekes.bsky.social at @weareceu.bsky.social
events.ceu.edu/2026-01-19/g...

GenAI in the CEU Curriculum: A Faculty Roundtable

events.ceu.edu

Gábor Békés @gaborbekes.bsky.social · 26d

AI policy, winter update (minor)
gabors-data-analysis.com/aipolicy/

Comments welcome

Gábor Békés @gaborbekes.bsky.social · 26d

One idea for Christmas or Hanuka.
@leleka.care
givebutter.com/Leleka25YE
"This winter, your impact is doubled. Thanks to a generous matching commitment by the Fooksman Family Foundation, all gifts made by December 31, 2025 will be matched dollar-for-dollar, up to $100,000. Your Gift Saves Lives"

Ukrainian medics are saving lives every day. You can help them save more.

By Leleka Foundation

givebutter.com

Gábor Békés @gaborbekes.bsky.social · 29d

While selection is indeed a major issue, maybe filtering on demographic observables could help. This would mean that maybe comparison over time only work for some sub populations where coverage is OK at every period.

super interesting problem, and I'm stunned by sharp decline.

Gábor Békés @gaborbekes.bsky.social · Dec 12

Cost of free speech: evidence from social media review program for h1b visa (job market paper title from 2027)

travel.state.gov/content/trav...

Announcement of Expanded Screening and Vetting for H-1B and Dependent H-4 Visa Applicants

travel.state.gov

Gábor Békés @gaborbekes.bsky.social · Dec 12

Tried. Does not really work.

Gábor Békés @gaborbekes.bsky.social · Dec 12

One reason I delay this is that where should I share stuff like this to reach out of echo chamber?
www.youtube.com/shorts/NvsDU...

Gábor Békés @gaborbekes.bsky.social · Dec 9

Loved @andrew.heiss.phd post.
This was mine.
gabors-data-analysis.com/aipolicy/
Ideas for january review welcome

Gábor Békés @gaborbekes.bsky.social · Dec 9

Yes it did.
But building a first can be organic.
Building up a competitor to a network is hard

Gábor Békés @gaborbekes.bsky.social · Dec 9

I mean, anti X, anti Trump stuff not enough.
I can only speak re #EconSky. It started off but eventually most ppl, r still back at x or linkedin.

Imho it needs to be 10x /100x bigger, more diverse. Vast majority of ppl not political.
I think building a network is hard.

Reposted by Michael A. Clemens, Kathleen J. Mullen, Paolo Crosetto , and 8 more Michael A. Clemens, Kathleen J. Mullen, Paolo Crosetto, Aaron Sojourner, Paul Goldsmith-Pinkham, Juan Rocha, Winston T. Lin, Simon Munzert, Samuel Coavoux, Gábor Békés, Jesper Wulff

Andrew Heiss @andrew.heiss.phd · Dec 9

Some closing thoughts for my students this semester on LLMs and learning #rstats datavizf25.classes.andrewheiss.com/news/2025-12...

Will you incorporate LLMs and AI prompting into the course in the future?
No.

Why won’t you incorporate LLMs and AI prompting into the course?
These tools are useful for coding (see this for my personal take on this).

However, they’re only useful if you know what you’re doing first. If you skip the learning-the-process-of-writing-code step and just copy/paste output from ChatGPT, you will not learn. You cannot learn. You cannot improve. You will not understand the code.

In that post, it warns that you cannot use it as a beginner:

…to use Databot effectively and safely, you still need the skills of a data scientist: background and domain knowledge, data analysis expertise, and coding ability.

There is no LLM-based shortcut to those skills. You cannot LLM your way into domain knowledge, data analysis expertise, or coding ability.

The only way to gain domain knowledge, data analysis expertise, and coding ability is to struggle. To get errors. To google those errors. To look over the documentation. To copy/paste your own code and adapt it for different purposes. To explore messy datasets. To struggle to clean those datasets. To spend an hour looking for a missing comma.

This isn’t a form of programming hazing, like “I had to walk to school uphill both ways in the snow and now you must too.” It’s the actual process of learning and growing and developing and improving. You’ve gotta struggle.

This Tumblr post puts it well (it’s about art specifically, but it applies to coding and data analysis too):

Contrary to popular belief the biggest beginner’s roadblock to art isn’t even technical skill it’s frustration tolerance, especially in the age of social media. It hurts and the frustration is endless but you must build the frustration tolerance equivalent to a roach’s capacity to survive a nuclear explosion. That’s how you build on the technical skill. Throw that “won’t even start because I’m afraid it won’t be perfect” shit out the window. Just do it. Just start. Good luck. (The original post has disappeared, but here’s a reblog.)

It’s hard, but struggling is the only way to learn anything.

You might not enjoy code as much as Williams does (or I do), but there’s still value in maintaining codings skills as you improve and learn more. You don’t want your skills to atrophy.

As I discuss here, when I do use LLMs for coding-related tasks, I purposely throw as much friction into the process as possible:

To avoid falling into over-reliance on LLM-assisted code help, I add as much friction into my workflow as possible. I only use GitHub Copilot and Claude in the browser, not through the chat sidebar in Positron or Visual Studio Code. I treat the code it generates like random answers from StackOverflow or blog posts and generally rewrite it completely. I disable the inline LLM-based auto complete in text editors. For routine tasks like generating {roxygen2} documentation scaffolding for functions, I use the {chores} package, which requires a bunch of pointing and clicking to use.

Even though I use Positron, I purposely do not use either Positron Assistant or Databot. I have them disabled.

So in the end, for pedagogical reasons, I don’t foresee me incorporating LLMs into this class. I’m pedagogically opposed to it. I’m facing all sorts of external pressure to do it, but I’m resisting.

You’ve got to learn first.

Add to Home Screen

Light up
your news

Add to Home Screen

Light upyour news

Sign in to Lightnews

Sign up to start reading

Connect Bluesky

Connect with Bluesky

Light up
your news