BDS Matlab distribution

The file bdsc.dll is a matlab (rev 5.3) mex file for estimating
the bds test for independence.  The file bds.m is a friendly
m-file front end, and contains full documentation on how to use
it.

Description of the BDS statistic and its asymptotic properties can
be found in: Brock, Dechert, Scheinkman, and LeBaron, A test for
 independence based on the correlation dimension,
 Econometric Reviews, 15, 1996: 197-235.

A description of the algorithm can be found in: LeBaron, A Fast
Algorithm for the BDS Statistic, Studies in
 Nonlinear Dynamics and Econometrics, 2, 1997: 53-59

Many examples of its use and monte-carlo experiments are in:
 Brock Hsieh and LeBaron, Nonlinear Dynamics, Chaos, and
 Instability: Statistical Theory and Economic Evidence,
 MIT Press, Cambridge, MA, 1991.

Note:  This program has been modified slightly from the earlier
c-versions.  It following the following conventions:

1.) The denominator is estimated using v-statistics for the entire
sample.

2.) Values in the numerator use the maximum amount of data
possible.  In other words when estimating

c(2) - c(1)^2

the sample is reduced by 1 for both c(2) and c(1), when estimating

c(3) - c(1)^3

the sample is reduced by 2 for both c(3) and c(1).

When estimating several dimensions, say 2-m, the sample size is
changed accordingly for each dimension.


The matlab code has not be tested extensively, so it should be considered beta.
Please send comments to:

blebaron@brandeis.edu
Blake LeBaron, GSIEF
Brandeis University


I would like to acknowledge the help of
Ludwig Kanzler
who provided many suggestions on this implementation of BDS.