FEAR1.15Ur’s Guide
Paul W.Wilson
Department of Economics
222Sirrine Hall
Clemson University Clemson,South Carolina29634USA
pww@clemson.edu
9November2010
This manual is for FEAR version1.15,a library for estimating productive efficiency,etc.using R. Copyright c 2010Paul W.Wilson.All rights rerved.
阅读倡议书Contents
1Introduction1 2Licen Issues2 3Downloading and Installing R4 4Adding FEAR to R5
4.1Where to get FEAR (5)
4.2Installing FEAR into R (5)
5Estimation with FEAR7
5.1Getting help (7)
肉书海5.2Example#1:DEA estimates of technical efficiency (8)
5.3Example#2:Outlier detection for frontier models (12)
5.4Example#3:Other estimators of technical efficiency: (15)
5.5Example#4:Farrell-Debreu efficiencies (15)
5.6Estimating other things (17)
1Introduction
An extensive literature concerning the measurement of efficiency in production has developed since Debreu (1951)and Farrell(1957)provided basic definitions for technical and allocative efficiency in production.One large ction of this literature focus on linear-programming bad measures of effi
ciency along the lines of Charnes et al.(1978)and F¨a re et al.(1985).In addition,the free disposal hull(FDH)method of Deprins et al.(1984)is sometimes ud;this estimator can be written as a linear program,algthough it is easier to compute estimates using numerical methods rather than linear programming.Within this literature,the that rely on convexity assumptions are known as Data Envelopment Analysis(DEA).
DEA estimators have been applied in more than1,800articles published in more than490refereed journals (Gattoufiet al.,2004).DEA and similar non-parametric estimators offer numerous advantages,the most obvious being that one need not specify a(potentially erroneous)functional relationship between production inputs and outputs.Although much of the nonparametric efficiency literature has ignored statistical issues such as inference,hypothesis testing,etc.,the statistical properties of DEA estimators have recently been established;e Simar and Wilson(2000b)for a survey of the results,and Kneip et al.(2008)for more recent results.
Standard software ,LIMDEP,STATA,TSP)ud by econometricians do not include procudures for DEA or other nonparametric efficiency estimators.Several specialized,commercial soft-ware packages reviewed by Hollingsworth(1999)and Barr(2004)are available,and to varying degrees,the are good at what they were designed to do.Each includes facilities for reading data int
o the program,in some cas in a variety of formats,and procedures for estimating models that the authors have programmed into their software.A common complaint heard among practitioners,however,runs along the lines of“package X will not let me estimate the model I want!”The existing packages are designed for ea of u(again, with varying degrees of success),but the cost of this is often inflexibility,limiting the ur to procedures the authors have explicitly made available.Moreover,none of the existing packages include procedures for statistical inference.Although the asymptotic distribution of DEA estimators is now known(e Kneip et al.,2008,for details)for the general ca with p inputs and q outputs,bootstrap methods remain the only uful approach for inference.None of the existing packages incorporate the bootstrap methods propod by Simar and Wilson(1998,2000a).
FEAR1.15consists of a software library that can be linked to the general-purpo statistical package R. The routines included in FEAR1.15allow the ur to compute DEA estimates of technical,allocative,and overall efficiency while assuming either variable,non-increasing,or constant returns to scale.The routines are highlyflexible,allowing measurement of efficiency of one group of obrvations relative to a technology defined by a cond,reference group of obrvations.Conquently,the routines can be ud to compute Malmquist indices,scale efficiency
measures,super-efficiency scores along the lines of Andern and Petern (1993),and other measures that might be of interest.Routines are also included to facilitate implementation
of the bootstrap methods described by Simar and Wilson(1998,2000a).The features can be further ud to implement methods of inference for Malmquist indices as in Simar and Wilson(1999),statistical tests for irrelevant inputs and outputs or aggregation possibilities as described in Simar and Wilson(2001b).as well as statistical tests of constant returns to scale versus non-increasing or varying returns to scale as described in Simar and Wilson(2001a).A routine for maximum likelihood estimation of a truncated regression model is included for regressing DEA efficiency estimates on environmental variables as described in Simar and Wilson(2007).In addition,FEAR1.15includes commands that can be ud to perform outlier analysis using the methods of Wilson(1993),and to compute FDH efficiency estimates along the lines of Deprins et al.(1984),the robust,root-n consistent order-m efficiency estimators described by Cazals et al.(2002), and the robust,root-n consistent order-αefficiency estimators described by Daouia and Simar(2007)and Wheelock and Wilson(2008).Most of the features are unavailable in existing software packages.
2Licen Issues会计报表附注模板
R is distributed under the terms of the GNU General Public Licen Version2,June1991.This licen can be found on the internet u/copyleft/gpl.html.Further details on licensing of R can be found by typing licen()in the R console window described below.
FEAR1.15is provided under the licen that appears in thefile LICENSE included with the software. This licen states the following:
Licen for the u of the R package"FEAR:Frontier Efficiency Analysis in R,"
copyright2010,Paul W.Wilson
DEFINTIONS:
"OWNER"refers to the author,Paul W.Wilson,who address is given below.
"Software"means the frontier efficiency analysis software
developed by OWNER known as the"FEAR software,""FEAR package,"
骂人表情包
"FEAR library,"or any other designation referring to the R package
"FEAR:Frontier Efficiency Analysis in R,"including any source code,
binary code,or ur documentation.
"You"means you,the ur and licene of the Software.If you are
失望造句employed and intend using the Software in connection with your
努力学习的英语
employment duties,then you warrant that your employer has authorid
you to accept this Licen on behalf of your employer.
"Academic u"includes and is limited to u of the software for
scientific,academic purpos intended to result in publication of scientific
papers in academic journals,in your role as a faculty member or student
at a university,college,or condary school.
石参"Commercial u"is any u of the software that is not specifically
academic u as defined above.This includes any u by anyone working服务期
as an employee,contractor,paid consultant,or in any other capacity
