Harry Kelejian and Gianfranco Piras, "An Extension of the J-test to a
Spatial Panel Data Framework", Journal of Applied Econometrics,
Vol. 31, No. 2, 2016, pp. 387-402.


EMPIRICAL APPLICATION 

The files to reproduce the empirical application are contained in the
zip files kp-applic.zip and kp-shape.zip. 

The file kp-applic.zip contains:

  - four R codes (try.R, predictors.R, datasetessential.R, and kpt.panel.R)
  - a dataset in format csv (cornwell1.csv)

These are all ASCII text files in DOS format, so Unix/Linux users
should use "unzip -a".

The file kp-shape.zip contains a shape file (north_carol.shp) to
create the spatial weights matrix, along with two related files,
north_carol.shx and north_carol.dbf. These are all binary files.
See http://en.wikipedia.org/wiki/Shapefile for information about shape
files.

The empirical application of the paper uses the well known economic
model of crime estimated by Cornwell and Trumbull (1994).  They use
panel data on 90 counties over the period 1981-1987.  The first two
columns of the dataset are the FIPS for the county and the year.  The
data are organized such that the fast index is year.  The dataset
cornwell1.csv only provides the final variables included in the model
along with the instruments. This is different from other version of
the dataset in that they include also pre-transformed variables. 

To reproduce the results In Kelejian and Piras (2014), first open the
file try.R. The first line set the working directory. This should be
the same directory in which the data file and codes are stored. This
is the only part of the code that the user should modify. The main
routine after performing some necessary calculation, uses the function
predictors_mod() (in the file predictors.R), and then the function
kptest() (in the file kpt.panel.R) to perform the J-test.

N.B. To reproduce the results, three R libraries are necessary (and all the 
libraries required by those three):

- maptools
- spdep 
- splm


MONTE CARLO

The files related to the Monte Carlo experiment are contained in the
zip file kp-monte.zip. These are all ASCII text files in DOS format,
so Unix/Linux users should use "unzip -a".

The folder contains four functions:

- run.R
- sim.Jtestpanel.R
- predictors.R
- kpt.panel.R

run.R is the function that can be used to run the entire experiment. 
For example, on a Linux machine, one could simply open the terminal
and type:

R CMD BATCH run.R out.R

The main function is named sim.Jtestpanel.R and performs all the
experiments for the various combinations of the simulation parameters
as described in the paper.  The tables of results are printed after
each of the experiments and saved on disk in the same directory. 

To replicate the Monte Carlo results one should have installed the
same libraries as in the empirical example.