README

CupidPython contains Python code to compute the equilibrium and to estimate the parameters in separable one-to-one, bipartite matching models with perfectly transferable utility—see Galichon and Salanié 2021 for a general study.

It contains:

• in ipfp_solvers.py: implementations of the Iterative Projection Fitting Procedure (IPFP) algorithm for several variants and extensions of the Choo and Siow 2006 model and for a class of nested logit models.

• in poisson_glm.py: a function that estimates the original version of the Choo and Siow model (homoskedastic, with singles) for a semilinear surplus, using Poisson GLM.

• in min_distance.py: a function that applies a minimum distance estimator to separable, semilinear models with a user-supplied entropy function.

I also created a Streamlit app that demonstrates the basic Choo and Siow model. You can find a (hopefully) working version of the app here.

This package is released under the MIT license. I hope it is useful to you.

Bernard Salanié — Nov 7, 2021

bsalanie at columbia.edu