Wuyi Wang, Peter C.B. Phillips, and Liangjun Su, "Homogeneity Pursuit
in Panel Data Models: Theory and Application", Journal of Applied
Econometrics, Vol. 33, No. 6, 2018, pp. 797-815.

All data and programs are included in the zip file "hp_panel.zip". It
contains Matlab programs to produce the simulation and application
results of this paper. It also includes a file named "app1.mat", which
is the cleaned "income and democracy" data used in the application. In
this panel data, there are 74 countries (N=74) and 7 periods (T=7). So
the number of observations is 518. In detail, "y" is the democracy
index"; the first column of "x" is log-income level; the second column
of "x" is the lag of the democracy index (the individual fixed effects
are concentrated out for all of them). The corresponding .csv file to
"app1.mat" is also included("app1.csv").

The original file for the dataset is available for downloading at
https://economics.mit.edu/files/5000.

Our Matlab codes utilize the Cplex solver by calling AMPL. AMPL/Cplex
is a commercial software. In the model where both time and individual
fixed effects are considered, we also use CVX Matlab, a free convex
optimization toolkit in Matlab.

Files description:

There are two sub-folders, viz, "model_fe" and "model_fe_and_te". As
the names suggest: the "model_fe" folder contains codes for the panel
structure model with individual fixed effects; the "model_fe_and_te"
folder contains codes for the panel structure model with both
individual and time fixed effects. 

1. In the "model_fe" sub-folder:
  "cardsest_panel.mod" and "cardest_panel.run" use AMPL to call Cplex to
  solve the optimization problem;
  "simu_dgp_1_and_2.m" is the script for generating DGPs 1 and 2;
  "simu_dgp3.m" is the script for generating DGP 3;
  "simu_dgp4.m" is the script for generating DGP 4;
  "codes_panel" folder contains functions used in the scripts.

2. In the "model_fe_and_te" sub-folder:
  "cardsest_panel.mod" and "cardest_panel.run" use AMPL to call Cplex to
  solve the optimization problem, as before (but the files here are
  different from these in the folder "model_fe");
  "app1.mat": cleaned data used in the application;
  "application1.m": script for the Panel-CARDS classification and estimation
  for the application;
  "codes_panel" folder contains functions used in the scripts "application1.m".

Please address any questions to:

Wang, Wuyi
wuyi.wang.2013 [AT] phdecons.smu.edu.sg