Brent R. Hickman and Timothy P. Hubbard, "Replacing Sample Trimming with
Boundary Correction in Nonparametric Estimation of First-Price
Auctions", Journal of Applied Econometrics, Vol. 30, No. 5, 2015, pp.
739-762.
All data we used are already in the Journal of Applied Econometrics
Data Archive, because we revisited the application considered in
Sandra Campo, Isabelle M. Perrigne, and Quang H. Vuong, "Asymmetry in
First-Price Auctions With Affiliated Private Values", Journal of
Applied Econometrics, Vol. 18, No. 2, 2003, pp. 179-207.
See: http://qed.econ.queensu.ca/jae/2003-v18.2/campo-perrigne-vuong/
In addition, we have provided three sets of replication files coded in
Matlab. These are stored in three zip files. Below we describe the
files in each zip file. Briefly, the files in "MC-related.zip" can be
used to replicate the Monte Carlo experiments presented in our paper;
the files in "Application-related.zip" can be used to replicate the
application of the proposed estimator to the OCS data noted above; the
files in "Extensions.zip" are the core files we think a user
interested in applying boundary correction to a new environment and/or
application might want to start with.
Note that the programs can be used to execute the boundary-corrected
approach we propose as well as a trimming-based approach. We ask
researchers who use our code, in whatever capacity, to please cite
this paper.
All the Matlab files are ASCII files in DOS format. Unix/Linux users
should use "unzip -a".
Description of files in "MC-related.zip"
hh_mcstudy.m: main driver file to run and replicate Monte Carlo
experiments
computeise.m: compute integrated squared error between kernel-smoothed
density estimate and true density
computeise_stg1.m: compute integrated squared error between
kernel-smoothed density estimate and true density using a grid of
points for truth
computeise_stg1_postrun.m: variation of computeise.m called by
hh_mcstudy.m
computeise_stg2_postrun.m: variation of computeise.m called by
hh_mcstudy.m
evalkspdf.m: evaluate specified kernel at a grid of points
evalskpdf_num.m: evaluate function of a specified kernel at a grid of
points
evalkspdf_denom.m: evaluate function of a specified kernel at a grid
of points
kscdf.m: compute empirical distribution function
kspdf_bc.m: compute boundary-corrected kernel-smoothed estimate of
density function
kspdf_no_bc.m: compute kernel-smoothed estimate of a density function
mixbetarnd.m: function to help generate (pseudo) random valuations
from a mixture of Beta distributions
processoutput.m: takes as input the beta_mc_ex#.mat (where # =
{1,2,3,4}) files which are output from hh_mcstudy.m to replicate
results presented in the paper
risamint_mixbetas.m: function for computing the integral portion of
the symmetric IPVP bid function
trim.m: trim bids within a bandwidth of the minimum and maximum
observed bids
Description of files in "Application-related.zip"
Remember to download the OCS data from Campo, Perrigne, and Vuong as
described above before running these programs which read in a file
called Ocs702.dat.
bcappliedtoCPV.m: boundary correction applied to the CPV data
bcappliedtoCPVfcn: a function called by compare_figs.m which
implements boundary correction on the CPV data
compare_figs.m: a file which compares the boundary correction and
trimming-based approaches
evalkspdf.m: evaluate specified kernel at a grid of points
evalskpdf_num.m: evaluate function of a specified kernel at a grid of
points
evalkspdf_denom.m: evaluate function of a specified kernel at a grid
of points
jointDistbcfcn.m: compute boundary-corrected kernel-smoothed estimate
of joint density function
kscdf.m: compute empirical distribution function
kscdf2dCPV.m: compute two-dimensional empirical distribution function
using standard kernel
kscdf2dCPVbc.m: compute two-dimensional empirical distribution
function using boundary-corrected kernel
kspdf_bc.m: compute boundary-corrected kernel smoothed estimate of
density function
kspdf_no_bc_CPV.m: compute kernel-smoothed estimate of a density
function
kspdf2dCPV.m: compute two-dimensional kernel-smoothed estimate of a
density function
kspdf2dCPV.m: compute two-dimensional kernel-smoothed estimate of a
density function using boundary correction
kspdf2dCPVFAST_joint.m: compute two-dimensional kernel-smoothed
estimate of a density function using boundary correction in a
faster way
kzroutine.m: KZ routine for boundary corrected kernel
kzker_joint.m: another KZ routine for boundary corrected kernel
replicateCPV.m: a replication file of the trimming-based approach of
Campo, Perrigne, and Vuong
replicateCPVfcn.m: a function called by compare_figs.m which
implements trimming on the CPV data
trim_comparison_scatter.m: a program to compare the different ways of
trimming
trimasymmetricCPV.m: trim auctions for asymmetric APV case
trimsymmetricCPV.m: trim auctions for symmetric APV case
Description of files in "Extensions.zip"
bcgpv_driver.m: generates uniformly distributed data and applies
boundary correction
evalkspdf.m: evaluate specified kernel at a grid of points
evalskpdf_num.m: evaluate function of a specified kernel at a grid of
points
evalkspdf_denom.m: evaluate function of a specified kernel at a grid
of points
kscdf.m: compute empirical distribution function
kspdf_bc.m: compute boundary-corrected kernel-smoothed estimate of
density function