In this hybrid research seminar on Thursday December 11th, 2025 at 12:00 pm Prof. Bruno Crépon will present his research titled " Designing Job Recommender Systems: Job Search Model and Evidence from Experiments."
Abstract:
In this paper, we address the question of how to design recommender systems (RS) for job seekers, and in particular whether such systems should prioritize jobs that match job seekers’ preferences or those for which they have the highest chances of being recruited.
We develop a job search model with an application stage, in which vacancies are modeled as lotteries. This framework allows us to derive the optimal RS from the job seeker’s perspective and to define a metric for comparing alternative algorithms. Using RS trained on rich administrative data from our partner, we run a series of randomized experiments with three objectives: (i) validate the theoretical model by estimating application behavior, (ii) construct and test an improved version of the algorithm, and (iii) build a ranking metric that compares the full set of algorithms from the job seekers’ perspective.
The zoom link to join the seminar online can be found here. This event is open to the public. All who are interested to this topic are very welcome to join the seminar. After the presentation there will be time for questions and discussion.
For any queries related to our UNU-MERIT Seminar Series, please contact Karthika Baby Sujatha and Lizeth Melissa Molina Alvarez by sending an email to seminars@merit.unu.edu.
