Package‘FEAR’
November9,2010
Version1.15
Licenfile LICENSE.
Description Routines for analyzing and estimating the productive efficiency offirms.
Title Frontier Efficiency Analysis with R
超过的英文
Author Paul W.wilson<pww@clemson.edu>
Maintainer Paul W.wilson<pww@clemson.edu>
Depends R(>=2.12.0)
Date11Nov2010油麦菜怎么炒好吃
Archs i386
R topics documented:
ap (2)
ap.plot (3)
boot.sw98 (4)
bootstrap.ci (7)
ccr (8)
cost.min (9)
cquan (10)
dea (12)
eff.bw (14)
fdh (15)
gen.unif (17)
genxy (18)
hquan (19)
malmquist (20)
orderm (27)
profit.max (28)
revenue.max (29)
xton (32)
show.dens (33)
treg (34)
wood (35)
1
2ap
Index37 ap Outlier Detection for Non-parametric Frontier Models
Description
This routine implements the Wilson(1993)outlier detection method.
Usage
ap(X,Y,NDEL=3,err.check=TRUE)
Arguments
X p×n matrix of obrvations on p inputs of nfirms who efficiency is to be estimated;
Y q×n matrix of obrvations on q outputs of nfirms who efficiency is to be estimated;
NDEL total number of obrvaitons to be deleted;i in Wilson(1993);
err.check equals TRUE(default)for error-checking,or FALSE for no error-checking.
Details
In terms of the notation in Wilson(1993),this routine computes R(i)min for i=1,...,NDEL.The QR decomposition is computed using Houholder transformations.See Wilson(1993)for furter details,but note that results in Table2and Figure2of Wilson(1993)are incorrect;e Wilson (2010)for details.
Value
r0a vector of length NDEL containing values R(i)
;thefirst element of r0contains
min
the value of R(1)min,the cond element contains the value of R(2)min,etc.
imat an NDEL by NDEL matrix containing indices of obrvations;indices in the i-th row of imat determine R(i)
.
min
ratio an n by NDEL matrix containing data that can be ud to produce log-ratio plots as in Wilson(1993);e the command ap.plot.
Author(s)
Paul W.Wilson
References
Wilson,P.W.(1993),Detecing outliers in deterministic nonparametric frontier models with multiple outp
uts,Journal of Business and Economic Statistics11,319-323.
Wilson,P.W.(2010),Detecing outliers in deterministic nonparametric frontier models with mul-tiple outputs:Correction,unpublished working paper,Department of Economics,Clemson Uni-versity,Clemson,South Carolina29634(available on-line at www.clemson.edu/ economics/faculty/wilson/Papers/ap-corrected.pdf.
ap.plot3
See Also
ap.plot
Examples
x=matrix(rnorm(25),nrow=1)
y=2+2*x+1.5*rnorm(25)
ap(X=x,Y=y,NDEL=4)
ap.plot Produce log-ratio plot for outlier analysis
Description
This routine produces log-ratio plots as in Wilson(1993)using data returned by the command ap.等闲的近义词
Usage
ap.plot(RATIO,NLEN=25,plot.options=list(pch=19),
ylim=NULL,ylab="log-ratio",xlab="i",main="",
err.check=TRUE)
Arguments
RATIO n by NDEL matrix returned by ap;
NLEN number of rows in RATIO to be ud in plot;default value is25,the number ud in Wilson(1993);
plot.options list of plotting options to be ud;
ylim vector of length2specifying minimum and maximum values on the vertical axis;default values are obtained when ylim=NULL;
ylab label for vertical axis;
xlab label for horizontal axis;
main an overall title for the plot;
err.check equals TRUE(default)for error-checking,or FALSE for no error-checking.
Author(s)
Paul W.Wilson
References
Wilson,P.W.(1993),Detecing outliers in deterministic nonparametric frontier models with multiple outputs,Journal of Business and Economic Statistics11,319-323.
See Also
吃药喝牛奶
ap,par,plot.
Examples
x=matrix(rnorm(25),nrow=1)
y=2+2*x+1.5*rnorm(25)
tmp=ap(X=x,Y=y,NDEL=4)
ap.plot(RATIO=tmp$ratio)
boot.sw98Homogeneous Bootstrap for Shephard(1970)Distance Functions
Description
This routine implements the bootstrap method of Simar and Wilson(1998)for estimating confi-dence intervals for Shephard(1970)input and output distance functions as well as for hyperbolic graph distance functions.
Usage
o2o模式是什么意思啊boot.sw98(XOBS,YOBS,NREP=2000,DHAT=NULL,
RTS=1,ORIENTATION=1,alpha=0.05,CI.TYPE=2, XREF=NULL,YREF=NULL,DREF=NULL,
OUTPUT.FARRELL=FALSE,NOPRINT=FALSE,errchk=TRUE)
Arguments
XOBS p×n matrix of obrvations on p inputs of nfirms for which confidence intervals are to be estimated.
YOBS q×n matrix of obrvations on q outputs of nfirms for which confidence inter-vals are to be estimated.
NREP number of bootstrap replications to be performed.
DHAT(optional)vector of n estimates of Shephard distance functions corresponding to(XOBS,YOBS)for which confidence intervals are to be estimated;the
efficiency estimates are relative to the technology supported by obrvations in
(XREF,YREF)if different from the technology supported by(XOBS,YOBS).
蓝鲸的图片RTS indicator for returns to scale(equal1for variable returns to scale,2for non-increasing returns to scale,or3for constant returns to scale).
ORIENTATION indicator input/output orientation(equals1for input orientation,2for output orientation,or3for hyperbolic graph orientation).
alpha either a scalar or a vector length n a of values giving statistical sizes of the con-fidence intervals to be estimated.
CI.TYPE logical variable indicating method to be ud to construct bootstrap confidence interval estimates;CI.TYPE can be:1-Efron percentile intervals;2-Hall
percentile intervals bad on differences;3-Efron’s bias-corrected intervals;or
4-percentile intervals bad on ratios.
XREF(optional)p×n r matrix of obrvations on p inputs of n rfirms that rve to define the technology.
YREF(optional)q×n r matrix of obrvations on q outputs of n rfirms that rve to define the technology.
DREF(optional)vector of n r Shephard distance function estimates corresponding to
(XREF,YREF)giving estimates of technical efficiency of each obrvation in
(XREF,YREF)relative to the technology defined by(XREF,YREF).
OUTPUT.FARRELL
logicalflag;if TRUE and ORIENTATON=2,Farrell output efficiency estimates
are returned(the will be greater than or equal to1)along with corresponding
bias estimates,bias-corrected efficiency estimates,and confidence interval es-绝句绝句
timates.If FALSE and ORIENTATION=2,Shephard output distance function
estimates are returned(the will be less than or equal to1)along with cor-
responding bias estimates,bias-corrected efficiency estimates,and confidence
interval estimates.If ORIENTATION=1then the value of this argument has no
effect.
NOPRINT logicalflag;if TRUE,warning messages are suppresd.If FALSE,warning
messages are printed on the console as appropriate.
errchk equals TRUE(default)for error-checking,or FALSE for no error-checking.
Details
The homogeneous bootstrap method described by Simar and Wilson(1998)is ud to estimate con-
fidence intervals for Shephard(1970)input or output distance functions,or for hyperbolic graph dis-
tance functions,corresponding to each obrvation in(XOBS,YOBS).Distance function estimates
from previous computations may be supplied in DHAT(or DREF);if the are not supplied,they
are computed automatically.Efficiency estimates are computed using the routine dea.Note that
distance function estimates are biad;conquently,confidence intervals obtained using Efron’s
percentile method(CI.TYPE=1)should be corrected by subtracting2times the estimated bias
from both the lower and upper bounds of confidence interval estimates when this method is ud;
e Simar and Wilson(1998)for details.This problem can be avoided by using CI.TYPE=2,3,
or4.The method described in Simar and Wilson(2000b)corresponds to CI.TYPE=2.
春节窗花剪纸教程
Value
A list containing the following items is returned:
bias vector of bootstrap bias estimates corresponding to obrvations in(XOBS,YOBS).
var vector of bootstrap variance estimates corresponding to obrvations in(XOBS,YOBS).
conf.int An(n×(2n a))matrix containing confidence inteval estimates corresponding
to obrvations in(XOBS,YOBS).
dhat efficiency estimates for obrvations in(XOBS,YOBS)relative to the tech-
nology supported by(XREF,YREF),or the technology supported by(XOBS,
YOBS)if XREF and YREF are not supplied or are not different from(XOBS,
YOBS).
dhat.bc bias-corrected efficiency estimates for obrvations in(XOBS,YOBS)relative
to the technology supported by(XREF,YREF),or the technology supported by
(XOBS,YOBS)if XREF and YREF are not supplied or are not different from
(XOBS,YOBS);equals dhat-bias.
dref efficiency estimates for obrvations in(XREF,YREF)relative to the technol-
ogy supported by(XREF,YREF).
bias.flag logical variable;equals TRUE if any bias-corrected distance function estimates
are less than or equal to zero,or FALSE otherwi.
boot an n×NREP matrix containing bootstrap estimates corresponding to the n
obrvations in(XOBS,YOBS).
