Jaerim Choi, Jiaying Gu, and Shu Shen, "Weak-instrument Robust
Inference for Two-sample Instrumental Variables Regression", Journal
of Applied Econometrics, Vol. 33, No. 1, 2018, pp. 109-125.

The micro data used for the analysis were obtained from two sources --
Olivetti and Paserman (2015); Currie and Yellowitz (2000)) -- and can
be downloaded from the following websites.

  https://www.aeaweb.org/articles?id=10.1257/aer.20130821
  http://works.bepress.com/yelowitz/7/

All Tables and Figures (except Figure 5) of the paper were produced in
R. Figure 5 is created using Stata version 14.  The main R code,
TSIV_function.R, carries out a) two-sample two-stage least squares, b)
proposed two-sample Anderson-Rubin, c) proposed two-sample CLR, and d)
proposed two-sample Kleibergen inference procedures. 

To replicate the Monte Carlo Simulations, please use the codes in the
folder "simulations".  The simulation tables will be saved to the
subfolder "mktable" and the simulation figures will be saved to the
subfolder "graph".

To replicate the first empirical application, please use the codes in
the folder "intergeneration".  First, you have to download Olivetti
and Paserman (2015)'s dataset available at
https://www.aeaweb.org/articles?id=10.1257/aer.20130821. In their
folder Olivetti_Paserman_AER/Olivetti_Paserman_Replication/final_data,
you have to download original data in "fathers_collapsed" and
"individual" and save those data in our folders.  Then, you have to
first run Table2_raw1.do, Table2_raw2.do, Table2_raw3.do,
Table2_raw4.do, Table2_raw5.do, and Table2_raw6.do to convert raw data
files of Olivetti and Paserman (2015) in the folder
"intergeneration/raw_data" to cleaned csv format.

To replicate the second empirical application, please use the codes in
the folder "public".  In this folder, you have to first run
Table3_row.do and Table4_row.do to convert raw data files of Currie
and Yellowitz (2000) in the folder "public/raw_data" to cleaned csv
format.

All data files are stored in ASCII (csv format). They are zipped in
the file cgs-file.zip. Unix/Linux users should use "unzip -a".

Please contact Shu Shen (shushen [AT] ucdavis.edu) if you have any
questions regarding the codes.

Tables:

"simulations"
Table1.R   Creates Table 1: Properties of 95% Benchmark Confidence Sets

"intergeneration"
Table2.R   Creates Table 2: Intergenerational Elasticities, 1870 - 1930

"public"
Table3.R   Creates Table 3: The Effects of Project Participation:
Baseline Specification

"public"
Table4.R   Creates Table 4: The Effects of Project Participation:
Robustness Checks

"simulations"
TableA1.R   Creates Table A.1: Coverage of 95% Robust Confidence Sets

Figures:

"simulations"
Figure1.R   Creates Figure 1: Size Comparison

Figure2.R   Creates Figure 2: Power Comparison (beta=0)

Figure3.R   Creates Figure 3: Power Comparison (beta=2)

Figure4.R   Creates Figure 4: Power Comparison Between AR and CLR

FigureA1.R   Creates Figure A.1: Size Comparison

FigureA2.R   Creates Figure A.2: Power Comparison (beta=0)

FigureA3.R   Creates Figure A.3: Power Comparison (beta=2)

"intergeneration"
Figure5.do, Figure5.csv  Creates Figure 5: Father-Son-in-law Elasticities 

Data:

intergeneration/raw_data
  fathers_collapsed folder	  Olivetti and Paserman (2015) raw data
  individual folder	          Olivetti and Paserman (2015) raw data

public/raw_data
reg1.csv	                  Currie and Yellowitz (2000) raw data
reg2.csv	                  Currie and Yellowitz (2000) raw data