Nathan Canen and Kyungchul Song, "Counterfactual Analysis under
Partial Identification Using Locally Robust Refinement", Journal of
Applied Econometrics, Vol. 36, No. 4, 2021, pp. 416-436.

Thank you for your interest in our work. The replication material is
composed of two separate folders: one for replicating our Monte Carlo
simulations (Sections 4.2-4.3) and the other containing the materials
to replicate the Empirical Application (Section 5).  We describe each
subfolder in turn in detail.

All files are zipped in the file readme.cs.txt. ASCII (text) files are
in DOS format, but .mat files are binary.

Monte Carlo Simulations:

-- The main file for replicating Section 4.2 is "maincode_sims.m". To
replicate the different specifications in the paper, set the value of
"Specification" in line 9 to either 1 or 2. Then, run the code within
the same folder as the other files.  The results will be output in
your screen after they are done and will be automatically saved to the
same folder.  We recommend running the code making use of the
parallelization.

-- The maincode_sims.m file will call three other files:

   -- "find_crit_LF_20Dec2020.m", which finds the critical value based on
      the Least Favorable configuration;

   -- "find_crit_RSW_20Dec2020.m", which finds the critical value based
      on the Romano, Shaikh and Wolf (RSW) based correction;

   -- "find_kalpha_20Dec2020.m" which finds the value of kappa_{\alpha_1}
      used in the RSW correction.

-- To replicate the results in Section 4.3, you simply need to run
   "stability_simulations.m". 

   -- To choose the specification for Figure 2, please set the variable
      Specification1 in line 12 to 1 (if Specification 1 is desired), and to
      another value if Specification 2 is desired. The results in Figure 2, 
      including those in the Notes, will be output automatically (see lines
      71-91). 
 
   -- To choose the specification for Figure 3, set the value for "pi"
      in line 94 accordingly. The results in Figure 3 are automatically
      output by lines 122-133.

The results for Figure 3 rely on a computation of the identified set.
This is already saved in the folder as "StabilityExercise2_sets.mat". 
It can be replicated by running the file "find_idset.m", although that
is not necessary.


Empirical Application:

-- The main file to replicate the results in Section 5 is
   "FinalEmpirical.m". To run the different specifications in the paper, 
   set the desired amount of top coding in line 13 to either 0.05 or 0.1.
   Then, run the code within the same folder as the other files.

-- The FinalEmpirical.m file will call the four other files: 

   -- "empirical_CPS_20Dec2020_short.m", the dataset for the empirical
      application.

   -- "find_crit_LF_20Dec2020.m", which finds the critical value based
      on the Least Favorable configuration;

   -- "find_crit_RSW_20Dec2020.m", which finds the critical value
      based on the Romano, Shaikh and Wolf (RSW) based correction;

   -- "find_kalpha_20Dec2020.m" which finds the value of
      kappa_{\alpha_1} used in the RSW correction.

-- The dataset is drawn from the 2000 Annual Demographic Supplement of
the Current Population Survey (CPS). Further information about the
data is described in the paper and in its supplementary material. We
use the data formatted by the Integrated Public Use Microdata Series
(IPUMS). If you choose to use the same data, please cite them
accordingly:

Sarah Flood, Miriam King, Renae Rodgers, Steven Ruggles and J. Robert
Warren. Integrated Public Use Microdata Series, Current Population
Survey: Version 8.0 [dataset]. Minneapolis, MN: IPUMS, 2020.
https://doi.org/10.18128/D030.V8.0

See https://cps.ipums.org/cps/citation.shtml for further details

-- Within this folder, we include two other subfolders which are not
   necessary for replication, but could be helpful to other researchers.

   -- "Code that Cleans Raw Data" includes two files: a raw dataset
      from the IPUMS website, and our .do file code which transforms it
      into the dataset used in the exercise.

-- "Table 2 Results (Output)": provides the saved output from running
   the codes following the instructions above. This could be useful
   because it might be time consuming to run the full code. To see the
   results in the format of Table 2, please load the appropriate
   dataset and then run lines 267 - 288 of FinalEmpirical.m.  This
   will output the results in Table 2 on your Matlab screen.


Nathan Canen, University of Houston
Kyungchul Song, University of British Columbia