(With @ulojkine.bsky.social)

More details: github.com/oliviergodec...

Special thanks to @cdechaisemartin.bsky.social @dgirardi.bsky.social @pedrosantanna.bsky.social for clarifying chats!
Package still in debug mode—feedback welcome!

Execution time on 4M simulated obs?
Pretty fast.

Our lpdidcsa package does this:
• Simple LP-DiD
• Reweighted LP-DiD
• Adjusted regression LP-DiD
• IPW LP-DiD
• Callaway & @pedrosantanna.bsky.social cohort treatment effects in a local projection framework with IPW.

Tiny tweak to LP-DiD: inverse probability weighting (IPW).
Goal? Make control group resemble treated group.
Result: Estimate equally weighted ATTs faster than “adjusted regression.”

In our paper, we use @arindube.bsky.social @dgirardi.bsky.social et al. (2025) LP-DiD method.
Example: The horizontal 0 line = “control” group; blue bars = “treated” group.
Treatment: outsourcing (strange medicine).
Before (left): earnings evolved similarly.
After 0 (right): 12% drop.

- So? Second goodies?
- github.com/oliviergodec...
- What’s that?
- A staggered event study DiD R package.
- Again?!
- Yep—last event in a series of staggered DiD solutions.
- Explain?
- Unroll the thread…

Cutting hours through outsourcing - WID - World Inequality Database Cutting hours through outsourcing Olivier Godechot and Ulysse Lojkine study the effects of on-site outsourcing of low-skilled service jobs (cleaning, catering, security) in France.

Outsourced workers in low-skilled service jobs (cleaning, catering, security) in 🇫🇷 face a substantial earnings penalty—about 10 log points—persisting for up to 7 years after outsourcing, find @ulojkine.bsky.social & @oliviergodechot.bsky.social in new study.

More findings▶️ wid.world/news-article...

Enjoy!

And the SAS and R scripts for chaining BTS / DADS postes are available here : olivier.godechot.free.fr/hopfichiers/...

Here's the improvement in the number of individuals chained :

1/ We handle better workers' ages, which allows to do a perfect match on birth years.
2/ We leverage information on unemployment benefits between 2003 and 2015. This enables to continue chaining workers who are unemployed during a full calendar year. Sadly, this information disappears after 2015

Beyond the content on outsourcing, there are two additional goodies of potential interest for many social scientists.
Here's the first for CASD fans (as us) : we improved the Babet-Godechot-Palladino chaining of the French BTS (a.k.a. DADS).

And also all labour social scientists !

Cutting hours through outsourcing

New @wid.world working paper with @oliviergodechot.bsky.social

PS: our data and code is available here:
olivier.godechot.free.fr/hopfichiers/...

Le localisme universitaire, nouvelles évaluations En analysant près de 8 000 recrutements à la maîtrise de conférences en France entre 2017 et 2024, Olivier Godechot, Rachel Issiakou, Yann Renisio et Adrien Rougier reviennent sur la question ancienne...

This study invites debate on how to (self-)regulate recruitment practices to address this invisible but significant inequality. [n/n]
(And, the French version, published in November, is here:
laviedesidees.fr/Le-localisme...)

Academic inbreeding creates invisible inequality: Over half of applicants were on the market during years when their former department had no openings. As a result, their chance of landing a job was just 6.6%, compared to 10.2% for those with the inbred option. [9/n]

Part of this phenomenon stems from departments protecting their own PhDs against external competition—especially when their graduates struggle to find positions elsewhere. [8/n]

Of course, geographical distance and candidate preferences play a role. Yet even after controlling for these factors, the preference for inbred candidates remains strong. [7/n]

Academic inbreeding declined during the 2000s, thanks to national and local initiatives—such as the 2008 reform of recruitment committees and mathematicians’ deliberate policy to ban the practice. [6/n]

Parisian universities, which produce many PhD graduates, generally show lower inbreeding rates on our odds ratio scale than smaller provincial ones. However, there are exceptions: UTC Compiègne and Lyon I score very low, while Paris 2 (Panthéon-Assas) and CNAM score very high. [5/n]

Inbreeding varies by discipline—but not along the lines of natural vs. social sciences. Instead, fundamental sciences tend to be less inbred, while applied sciences are more so. [4/n]

Our key contribution is reconstructing the competitive landscape. Using the simplifying assumption that every candidate applies to every open position in their discipline, we find that inbred applicants are 9.5 times more likely to be recruited than outbred potential candidates. [3/n]

We combined three datasets: 1) 8,000 “maîtres de conférences” (assistant professors) recruited between 2017–2024, 2) 108,000 qualifications (the first national screening step authorizing candidates to apply) from 2012–2024, and 3) 312,000 PhDs defended between 1999–2024. [2/n]

Academic inbreeding: new estimates By analyzing nearly 8,000 recruitments for assistant professor positions in France between 2017 and 2024, Olivier Godechot, Rachel Issiakou, Yann Renisio, and Adrien Rougier revisit the long-standing ...

Academic inbreeding—when a university hires its own former PhD students as assistant professors—is widespread in many countries, yet it often sparks heated debate. In this paper, we assess its impact on the fairness of assistant professor recruitment in France.[1/n]
booksandideas.net/Academic-inb...

Fabian Muniesa - 'Paranoid Finance' @ Sciences Po Paris (AxPo / CEE) 19.02.2026 www.sciencespo.fr/centre-etude...

In Minneapolis the icy
Good was pretti
And Pretti was good
But shot in cold blood
For against what they stood

Your new year party is over. It is time to complete.

This paper sets a new empirical agenda for exploitation theory, through the notion of chains. Exploitation generally offers three attractive properties compared to more commonly used concepts—inequality and domination—in that it is simultaneously distributive, relational, and openly counterfactual, yet it remains an underexplored notion. While both Marxist and neoclassical traditions focus on dyadic relations—either worker-employer or through market exchange—most exploitative situations bear several relational components, where agents can simultaneously stand as exploited and exploiters. Building on a sociological-relational tradition, we identify four chains—I (connected), L (hinged), V (dual) and C (complicit)—which we argue represent the elementary structures of exploitation. We then contend that this meso-level approach, complementary to individual-transactional and structural accounts, bears potential for sociological analysis and then explore how these chains materialize in various economic sites—within the production unit, on the market, in the domestic sphere, and by the state.

Publication, with Simon Bittmann:

Chains of Exploitation: A Theoretical and Empirical Exploration

in Theory & Social Inquiry