Christian Vedel
banner
christianvedel.bsky.social
Christian Vedel
@christianvedel.bsky.social
Assistant Professor at @sdueconhist.bsky.social, University of Southern Denmark. Interested in Economic Geography, Machine Learning and Causal Inference in Economic History.
https://sites.google.com/view/christianvedel
So, as was recently announced, I'll be dealing with all of the great working papers flowing into this series in the future. Keep them coming! ehes.org/working-pape...
September 23, 2025 at 2:56 PM
Not only did I get to share numerous holiday pictures (which where somehow scientifically relevant) I also enjoyed insightful comments and questions (2/3)
June 19, 2025 at 3:42 PM
A true pleasure visiting the @camunicampop.bsky.social group to discuss my research on the effect of first nature Geography on Economic Development. (1/3)
June 19, 2025 at 3:42 PM
9/ Did Irish co-operatives fail because they were less efficient? The answer is “no.” Based on reports from 1898-1903 (541 dairies in Denmark and 171 in Ireland), we see that Danish dairies were actually less efficient.
April 28, 2025 at 10:02 AM
8/ But we don’t just rely on interpretation – we have the microlevel data. We have digitized reports from the creameries themselves, allowing us to test these claims with modern statistical methods.
April 28, 2025 at 10:02 AM
4/ This guy, Lars Emil Bruun, a butter trader, accumulated enough wealth to establish a famous coin collection, recently sold after 100 years (read about it here: news.artnet.com/art-world/br...). But it wasn’t just the rich who benefitted.
April 28, 2025 at 10:02 AM
Did you eat butter this Easter? Was it Lurpak? 🧈 New publication in Business History. It naturally leads to a little reflection I would like to share with you. And it also leads to a small corner of economic history that needs reinterpretation🧵
April 28, 2025 at 10:02 AM
We marked the occasion in two related events: An academic workshop on everything related to the theme, and a policy event, where we had invited in the public, heard from leading voices in the debate on gender inequality, and finally had a reception, where an exhibition on the theme was opened. (4/x)
March 10, 2025 at 9:47 AM
For most of history, women were excluded from knowledge creation. There has been a university in DK since 1479. But half of the population was excluded from knowledge creation and higher education until Nielsine Nielsen decided to try to change that, and was able to sit the entry exam in 1875 (2/x)
March 10, 2025 at 9:47 AM
Thursday and Friday was the 14th annual workshop on Growth, History and Development. But this year was different. This year marks the 150 year anniversary of women's access to university in Denmark. And what an obvious theme for a workshop on economic history. (1/x)
@sdueconhist.bsky.social
March 10, 2025 at 9:47 AM
Tweeting is the generic term for posts on twitter-like platforms. You can tweet on Threads, Mastodon, X and BlueSky - and a number of other platforms yet to be invented.
Currently my favorite Twitter is BlueSky.
November 25, 2024 at 7:04 PM
Was this just permanent? No, the lion's share of the effect is realized in the years after the crisis. The firms with an advantage in adaptation, were able to accrue an initial benefit, which then just grew: Adaptability is crucial for long term success. (6/7)
November 17, 2024 at 8:06 PM
We estimate that this gain in productivity alone corresponds to an output gain equivalent to 10.000 2010-USD every year, for each of these many and relatively small units. (5/7)
November 17, 2024 at 8:06 PM
Some firms, by accident of nature, were better equipped to deal with this crisis, simply because the main alternative fuel, peat, was readily available. Peat occurs in so-called freshwater deposits. We use this as a natural experiment to explore the effects of adaptability. (3/7)
November 17, 2024 at 8:06 PM
Energy crises are not new. During WWI, Danish industries, heavily dependent on imported coal, faced severe shortages. This wasn’t just a supply chain issue—it threatened the survival of businesses. Here we plot various components of Danish Energy consumption (1900-1935) (2/7)
November 17, 2024 at 8:06 PM
🚨New publication alert!🚨
Blue Sky has been growing fast lately—a testament to how platforms adapt and thrive in dynamic environments. Adaptability for SoMe users, but even more for industries facing challenges like energy crises. We explore adaptation to the WWI energy crisis in a new paper🧵👇 (1/7)
November 17, 2024 at 8:06 PM
The underlying training data was either publicly available or shared with us. All these research projects will now power endless new HISCO codes for other researchers. For this, we are immensely grateful. (10/12)
April 4, 2024 at 4:14 PM
The result is that OccCANINE, as we call our model, has picked up an intrinsic understanding of occupations. It recognizes that ‘farmer’ (Eng.) and ‘bonde’ (Danish) are similar, and place them in a similar location, when we visualise the embedding space. (8/12)
April 4, 2024 at 4:13 PM
The result is that OccCANINE, as we call our model, has picked up an intrinsic understanding of occupations. It recognizes that ‘farmer’ (Eng.) and ‘bonde’ (Danish) are similar, and place them in a similar location, when we visualise the embedding space. (8/12)
April 4, 2024 at 4:11 PM
The model has Accuracy, Precision, Recall and F1 all above 90%. This is on par with human labelers. It requires no string cleaning. Moreover, OccCANINE will always deliver the same results from the same descriptions --> HISCO coding replicates (4/12)
April 4, 2024 at 4:08 PM
We solve this problem. Our model can turn 10.000 such descriptions into HISCO codes in around 30 seconds on a laptop. This will free resources towards higher quality research with fewer hours and days of tedious labelling. (3/12)
April 4, 2024 at 4:07 PM
Censuses, parish records, marriage certificates, etc. are all sources filled with descriptions of what people did for a living. It is trivial to recognize that “he is a farmer” is a farmer (HISCO: 61110). But it is not trivial in thousands or millions of cases. (2/12)
April 4, 2024 at 4:05 PM
🚨New language model for historical occupations🚨
OccCANINE is a new tool, which turns occupational descriptions (in all their pesky variety) into standard HISCO codes. Accurate, multilingual and fast. Developed by
Christian M. Dahl, Torben Johansen, and myself (1/12)
April 4, 2024 at 4:04 PM
21 different research projects have already contributed more than 14 million observations as training data. Quality improves with more data. Any amount of data in any language helps. In return, I owe you HISCO codes. Please reach out via christian-vs@sam.sdu.dk (3/3)
December 8, 2023 at 12:35 PM
It transforms the task of occupational coding into something which can be done in an afternoon - rather than weeks, moths or years. It is still a work in progress. But that progress also relies on the community. (Yes, that's you reading this!) (2/3)
December 8, 2023 at 12:34 PM