BonhommeLamadonManresa2021

Replication code for: "Discretizing Unobserved Heterogeneity", by Bonhomme, Lamadon and Manresa

Download the latest zip file with all results and source code:

This repository contains all the code to replicate the results presented in the paper. Reproducing the results should be close to be as simple as typing make all in your terminal. See however the required dependencies below.

As an alternative we provide a separate repository with a pip package and notebooks written in python to reproduce the results from the first model of the paper. You can launch the notebook either on google colab or using binder:

• 
• 

Generated figures:

• Figure 1
• Figure 2
• Table S1
• Table S2
• Table S3

Overview

• The matlab folder contains the code to generate the simulations used in the paper
• The Makefile can be used to regenerate all the results. Each matlab file can also be used to generate individual results where parameters can be changed easily
• The results folder contains the results that we generate for the paper using the random seed defined in the makefile

Dependencies

• To generate the mat files you only need access to matlab. You can use the makefile directly with make sims
• To generate the table and plots from the mat files you will need a few python dependencies. You can either:
  • install then by using the provided conda environment file: conda env create --file conda-env.yml and then activating blm2-env
  • install it through pip with pip install numpy pandas matplotlib tqdm seaborn scipy
• To compile the tables you need a working copy of latex.

Code content

The matlab folder contains 6 matlab files:

• Code_Earnings_Time_Invariant.m replicates Figure 1 in the paper and Table S1 in the Supplemental Material

• Code_Probit_Time_Varying.m replicates Figure 2 in the paper and Table S2 in the Supplemental Material

• Code_Probit_Time_Invariant_BinaryCov.m replicates Table S3 in the Supplemental Material

• lik.m, lik_bb.m, and lik_IFE2.m are functions to compute the likelihood function and scores & hessians of probit models.

Final notes

Thank you for using our codes.

For any feeback, please contact:

• Stephane at [email protected]
• Thibaut at [email protected]
• Elena at [email protected]