Chih-Sheng Hsieh and Xu Lin, "Social Interactions and Social
Preferences in Social Networks", Journal of Applied Econometrics,
Vol. 36, No. 2, 2021, pp. 165-189.

All files are ASCII files in DOS format. They are zipped in the file
hl-files.zip. Unix/Linux users should use "unzip -l".


PROGRAM FILES:

The program files are provided for implementing the Monte Carlo
simulation and the empirical study of the paper.

Note that you also need the INTEL Math Kernel Library (MKL) to compile
with the FORTRAN codes. 

An example command of compiling these files is

ifort -mkl -nologo -openmp -O2 -o EST -I /opt/intel/composer_xe_2013_sp1.3.174/mkl/include/em64t/lp64 random.f90 tool.f90 model_I.f90

Under the folder "simulation", there are FORTRAN codes to generate
Monte Carlo simulation results in Table 1 of the paper.  

- In the sub-folder "DGP", there are three FORTRAN 90 files and they
  should be complied together to form an executive file to generate
  artificial network and activity variable data. 

- In the sub-folder "estimation", compiling FORTRAN 90 file
  "model_I" with "random" and "tool" generates an executive file to
  estimate the Model (I). 

- In the sub-folder "estimation", compiling FORTRAN 90 file "model_II"
  with "random" and "tool" generates an executive file to estimate the
  Model (II). 

- In the sub-folder "estimation", compiling FORTRAN 90 file
  "model_III" with "random" and "tool" generates an executive file to
  estimate the Model (III). 

- In the sub-folder "estimation", compiling FORTRAN 90 file "model_IV"
  with "random" and "tool" generates an executive file to estimate the
  Model (IV). 

- In the sub-folder "estimation", compiling FORTRAN 90 file "model_V"
  with "random" and "tool" generates an executive file to estimate the
  Model (V). 

- "random" and "tool" are two modules which contain useful
  subrountines and functions

Under the folder "empirical", there are STATA, and FORTRAN codes to
prepare data and estimate models in Tables 3, 4 and Supplementary
Appendix Tables C.3-C.6 of the paper.

- In the sub-folder "data_preparation", the STATA file
  "read_AddHealth" reads the raw Add Health Wave-I in-school database
  and extracts the sample of 24 schools for the empirical analysis.  

- In the sub-folder "estimation/conventional_social_interactions_model",
  compiling FORTRAN 90 files "estimation_conventional_SAR" and "random"
  and "tool" generates an executive file to estimate Model (I) in
  Tables 3　&　４.

- In the sub-folder "estimation/conventional_social_interactions_model",
  compiling FORTRAN 90 files "estimation_conventional_SAR_with_externality"
  and "random" and "tool" generates an executive file to estimate Model
  (III) in Tables 3　&　４.

- In the sub-folder "estimation/altruistic_social_interactions_model",
  compiling FORTRAN 90 files "estimation_altruistic_SAR_exogenous_no_externality"
  with "random" and "tool" generates an executive file to estimate
  Model (II) in Tables 3 & 4.

- In the sub-folder "estimation/altruistic_social_interactions_model",
  compiling FORTRAN 90 files "estimation_altruistic_SAR_exogenous" with
  "random" and "tool" generates an executive file to estimate　Model
  (IV) in Tables 3 & 4.

- In the sub-folder "estimation/altruistic_social_interactions_model",
  compiling FORTRAN 90 files "estimation_altruistic_SAR_endogenous" with
  "random" and "tool" generates an executive file to estimate　Model (V)
  in Tables 3 & 4.

- In the sub-folder "estimation/directed_altruistic_social_interactions_model",
  compiling FORTRAN 90 files "estimation_directed_altruistic_SAR_exogenous"
  with "random" and "tool" generates an executive file to estimate
  Directed Altruistic Model in Table C5.

- In the sub-folder "estimation/directed_altruistic_social_interactions_model",
  compiling FORTRAN 90 files
  "estimation_directed_altruistic_SAR_exogenous_with_externality" with
  "random" and "tool" generates an executive file to estimate Directed
  Altruistic Model w/ externality in Appendix Table C5.

- In the sub-folder
  "estimation/heterogeneous_altruistic_social_interactions_model",
  compiling FORTRAN 90 files "estimation_heterogeneous_altruistic_SAR"
  with "random" and "tool" generates an executive file to estimate the
  model in Appendix Table C6.

- "random" and "tool" are two modules which contain useful subroutines
  and functions.

- The results in Appendix Tables C.3-C.4 can be obtained by changing
  the latent variable dimensions in the FORTRAN file
  "estimation_altruistic_SAR_endogenous". 


DATA:

The data used in the article are from the National Longitudinal Study
of Adolescent to Adult Health (Add Health) WAVE I in-school survey
data. This is a confidential data source administered by

     Add Health
     Carolina Population Center
     University of North Carolina at Chapel Hill 
     CB# 8120, University Square
     206 West Franklin Street
     Chapel Hill, NC 27516
     
Researchers should submit a research proposal to the above address to
get permission to use these data.


Please address any questions about the codes to:

Chih-Sheng Hsieh
cshshieh [AT] ntu.edu.tw
Department of Economics
National Taiwan University
No.1, Sec 4, Roosevelt Road, 
Taipei, 10617 Taiwan