ALFRED GALICHON'S
MASTERCLASSES
'math+econ+code' masterclass
on discrete choice models and applications in economics
May 19–20, 2026
This intensive course, part of the 'math+econ+code' series, focuses on the theory, econometrics, and computation of discrete choice models, together with applications across economics. We will move from random utility and welfare and inversion results to logit and generalized extreme-value models, logistic regression, simulation methods, characteristics-based demand, endogeneity, instrumental variables, GMM, and the Berry-Levinsohn-Pakes framework.
Students will write their own Python code throughout the course, in the spirit of cooking lessons. The only prerequisite is the equivalent of a first-year graduate sequence in economics, applied mathematics, or another quantitative discipline.
This course is very demanding, but the learning rewards are high.
Practical information
-
The course will be taught online over two consecutive days, Tuesday May 19 and Wednesday May 20, 2026.
-
The instructors are Alfred Galichon (professor of economics and of mathematics at NYU and principal investigator of the ERC-funded project 'equiprice' at Sciences Po) and Antoine Jacquet (post-doctoral researcher at Sciences Po, Paris).
-
There will be four blocks of two hours each:
-
Tuesday May 19, 11:00am–1:00pm Paris time / 5:00am–7:00am New York time
-
Tuesday May 19, 2:00pm–4:00pm Paris time / 8:00am–10:00am New York time
-
Wednesday May 20, 11:00am–1:00pm Paris time / 5:00am–7:00am New York time
-
Wednesday May 20, 2:00pm–4:00pm Paris time / 8:00am–10:00am New York time
Course outline
Block 1:
Random utility, welfare, and inversion
-
Random utility models, choice probabilities, and the social-surplus representation.
-
Welfare analysis, convexity, and the Williams-Daly-Zachary theorem.
-
Identification and inversion of demand systems.
Block 2:
Logit, generalized extreme value, and regression
-
Logit choice probabilities, inclusive values, and max-stability.
-
Multivariate extreme-value models and nested logit.
-
Logistic regression, the GLM connection, identification, and regularization.
Blocks 3 and 4:
Simulation, demand estimation, and applications
-
Simulation, importance sampling, quadrature, and GHK methods.
-
Random-coefficients logit and characteristics-based demand.
-
Endogeneity, instrumental variables, GMM, and the Berry-Levinsohn-Pakes framework.
-
Applications in industrial organization, matching, international trade, and dynamic choice.