Replication Materials for "Employment-based Health Insurance and Misallocation: Implications for the Macroeconomy"
David Chivers, Zhigang Feng and Anne Villamil
2016

This archive contains all code and data used in the construction of tables and figures for this paper.

General notes:
All codes require relevant paths to the base Replication Materials to be specified. The following instructions assumes that all paths are set correctly.
C++ complier, Matlab and Numerical Receipes head files are all required.


STEP 1: RUN THE FULL DYNAMIC MODEL
The c++ files used to solve the models are in /C++files.
The results for baseline model, shown in Table 2, 3, 4, are solved using the c++ code: cfv_red_final.cpp.
The results for high risk aversion, shown in Table 5, are solved using the c++ code: cfv_red_final_sigmah.cpp.
The results (for robustness check) shown in Table 6 are solved using the c++ code: cfv_red_final_gnl.cpp, cfv_red_final_gnh.cpp, cfv_red_final_no_rwedge.cpp.


STEP 2: CREATE FIGURES
The matlab files used to create the figures are in /FigureFiles. 
Dont forget to set the base directory paths appropriately, and to specify a directory to save the figures if these are desired to be saved.

ENVIRONMENT:
IBM server 1350 cluster
Xeon 2.3 GHz processors
32 GB 1333 MHz DDR3
g++ compiler
Matlab R2015A

RANDOM NUMBER:
We use Matlab command "rand" to general uniformly distributed pseudorandom numbers for our simulations.

TIME TO RUN:
Solving the model takes 60-80 hours to run in above environment