This is a brief description of matlab files used in Simar and Zelenyuk
(2003) (note, some of the files are the same as those of Simar (2002), and
some others are modified, some are new created by Simar and Zelenyuk)

In the spirit of academia, promotion of science, etc, this code is free!,
but we hope that authors' work will be properly acknowledged.

Disclaimer: We do not take any responsibility for results and conclusions
anyone can obtain from using these methods.  We also ussume that fully legal
version of MatLab will be used for running this code.

	The MAIN program file is

DEA_SubSmpl2smplGWHet_Sh_J_CndX.m:    this file calls all other files 
				      and produces final results

In particular, it calls for 

	eff = deasel.m
	
for estimation of DEA individual efficiencies and for bootstrapping
(sub-sampling) of aggregate efficiencies and of RD statistic, it calls

  	DEA_SubSampl_Ag_2_Smpl.m
      
which in turn calls

		resample.m
		
for sub-sampling

		deaXNewYNew.m
		
for DEA with and at Bootstrapped data 
			
		P_indep_weights.m 
		
for computation of price-independent weights (can be modified to compute
other weights) using the bootstrapped data


NOTE: For solving linear programming problem, we use the Matlab function:
lp.m (Copyright (c) 1990-98 by The MathWorks, Inc.). There is a newer
version (linprog.m) but it is slower, and the speed is important 

lp.m uses another function :	qpsubold.m
(Copyright (c) 1990-98 by The MathWorks, Inc.)

Densities are estimated using bandwidth due to Sheather and Jones (1991), by
calling

       Sh_J_Run.m
       
which in turn calls
       
       Sheth_Jones_h.m	(can be made more efficient).


The first version of the paper has 3 simulated examples, all generated by:

	Sim_Ex_1.m
	Sim_Ex_2.m
	Sim_Ex_3.m

where the # of example corresponds to # cited in the original version of the
paper


which gave 3 data sets that were used for the paper:  

	Data_Ex1_Ag_Boot_Het.mat
	Data_Ex2_Ag_Boot_Het.mat
	Data_Ex3_Ag_Boot_Het.mat

(which were loaded, one at a time, by DEA_SubSmpl2smplGWHet_Sh_J_CndX.m)

All examples involved generation from truncated normal distribution using

	TruncNormGenerator.m

The main file also produced figures that are saved as:

	Figure1.fig	(this was actually produced by Sim_Ex_1.m)
	Figure2.fig
	Figure3.fig
	Figure4.fig

where the # of figure corresponds to # cited in the original version of the
paper


All the FINAL results of bootsrapping (subsampling) from the main file for
each example, respectively, are saved in 

	Ex1_BS_Agg_ppr.mat
	Ex2_BS_Agg_ppr.mat
	Ex3_BS_Agg_ppr.mat


where the # of example corresponds to # cited in the original version of the
paper.
