Martin Becker and Stefan Kloessner, "Estimating the Economic Costs of
Organized Crime By Synthetic Control Methods", Journal of Applied
Econometrics, Vol. 32, No. 7, 2017, pp. 1367-1369.

The data used in this replication study is not included in this
archive due to copyright reasons, but a zip file containing the file
"dataset.dta" as well as a complete description of the dataset
("replication_README.pdf") can be downloaded at

  http://onlinelibrary.wiley.com/doi/10.1111/ecoj.12235/abstract
  
in the section "Supporting Information" (file
"ecoj12235-sup-0002-DataS1.zip").

Once the file "dataset.dta" has been extracted and put in the current
working directory, the following steps are sufficient to reproduce the
results of the paper:

1. Install R packages "Synth", "MSCMT", "readstata13", "ggplot2", "gridExtra",
   and "xtable" (all available from CRAN), usually by issuing the command
   
   install.packages(c("Synth","MSCMT","readstata13","ggplot2","gridExtra",
                        "xtable"))
   in R.

2. (Optional) Reproduce the STATA/EJ results by running "Replication.do"  
   after installing the Stata Synth package via
   
     ssc install synth, replace all
   
   in Stata.  

3. (Optional) Prepare the data for the Matlab code by running 
   "PrepareDataForMatlab.r" in R (output: files "main_X0.txt", "main_X1.txt",
   "main_Z0.txt", "main_Z1.txt" needed for step 4).
   
4. (Optional) Reproduce the Matlab results by running "Replication.m" in
   Matlab.
   
   Notice: Replication.m is a modification of the Matlab code contained in
   https://web.stanford.edu/~jhain/Synth_Matlab/Synth_MATLAB.zip 
   (loss_function.m is also part of this Matlab code).
   
   The corresponding readme file contains the following author information:
   
     Synth MATLAB Code (11/07/2006) written for MATLAB 7.0 
     by Alberto Abadie, Alexis Diamond, and Jens Hainmueller (all Harvard 
     University)
     Contact: jhainm@harvard.edu

5. Reproduce R/Synth and R/MSCMT results and import results from
   Matlab and Stata by running "Replication.r" in R (output: file
   "Replication.RData" needed for step 6). The results from Matlab and
   Stata are already contained in "Replication.r", the results
   obtained in the optional steps 2-4 should be exactly equal to these
   results.
   
6. Reproduce Table 1 and Figure 1 by running "ReplicationOutput.r" in R.

NOTE: To obtain more decimal places for the Stata results, the two lines 
         mata: roundmat("`wsol'")
         mat `wsol' = matout
      (line number about 590 in synth.ado) have to be removed/commented out in 
      Stata's Synth package.

The program files needed for the above steps are all zipped in the
file bk-files.zip. They are ASCII files in DOS format. Unix/Linux
users should use "unzip -a".