450 likes | 650 Views
Performance Evaluation: Network Data Envelopment Analysis. 高 強 國立成功大學工業與資訊管理學系 於 中山大學企業管理系 100 年 11 月 5 日. Contents 1. Efficiency 2. Data Envelopment Analysis 3. Mathematical Models 4. Network Models 5. Research Areas. 1. Efficiency. Definition Output point of view
E N D
Performance Evaluation: Network Data Envelopment Analysis 高 強國立成功大學工業與資訊管理學系 於 中山大學企業管理系 100年11月5日
Contents 1. Efficiency 2. Data Envelopment Analysis 3. Mathematical Models 4. Network Models 5. Research Areas
Definition Output point of view (Actual output produced)/(Maximal output can be produced) Input point of view (Minimal input required)/(Actual input used) Technically Efficient Production T. Koopmans:A feasible input/output vector where it is technologically impossible to increase any output (and/or reduce any input) without simultaneously reducing another output (and/or increasing any other input). => Pareto optimality
MeasurementParametric approachRegression analysis (Aigner-Chu)Nonparametric approachData envelopment analysis (Charnes-Cooper-Rhodes)
output Max. production Ave. production Output Eff.=A/A* Input Eff.=I*/I input Parametric approach Production function
X 2 o Dominated region o ● o Input Eff.=OA*/OA ● O X 1 Input efficiency Isoquant (Y0)
Y 2 ● Output Eff.=OA/OA* * A * O ● 2 A o O o ● ● 2 o Product transformation curve Dominated region O * O O Y 1 1 1 Single-input two-output (X0)
Production function: unrestricted in sign. Example:
Y Production function E* D ● C ● E ● B ● X A ● O Non-parametric approach
X2 Isoquant X1
Production transformation curve
Emrouznejad et al. (2008) Socio-economic Planning Science 42, 151-157
Ratio form Input i and output r of DMU j: (Xij , Yrj) DMU k chooses most favorable multipliers ur ,vi to calculate Ek
Envelopment form (Dual of the ratio form) ● is the target on the frontier.
Constant RTS Variable RTS Variable returns-to-scale Technical Eff.=A/A*, Scale Eff.= A*/A0, Aggregate Eff.=A/A0=(A/A*)×(A*/A0)
DMU k X1k Y1k X2k Y2k . . . . . . Xmk Ysk Conventional black box concept
Envelopment model θ unrestricted in sign
Zqk X1k X2k Xmk Z2k Z1k System DMU k Y2k Ysk Y1k Process 1 Process 2 . . . . . . . . . Two-stage series systemZpj:Intermediate product p of DMU j
Xi Yr Zp(l) Zp(t) h t l … r=1,…,s i=1,…,m p=1,…,q p=1,…,q … General case System efficiency is the product of the h process efficiencies.
1 X1, X2 Y1, Y2, Y3 3 2 A network system
Models I: Increasing marginal product II: Decreasing marginal product III: Negative marginal product- Congestion
Multipliers Strictly positive :non-Archimedean number,10-5 Absolute range Relative range (Assurance region, cone ratio)
Data type Traditional data Undesirable data Ordinal data Qualitative data Interval data Stochastic data Fuzzy data
Applications Novel application A new area A new journal Implications Special data type Derivation of multiplier restrictions
References • Chiang Kao and Shiuh-Nan Hwang, 2008, Efficiency decomposition in two-stage data envelopment analysis: An application to non-life insurance companies in Taiwan. European J. Operational Research 185, 418-429. • Chiang Kao, 2009, Efficiency decomposition in network data envelopment analysis: A relational model. European J. Operational Research 192, 949-962. • Chiang Kao, 2009, Efficiency measurement for parallel production systems. European J. Operational Research 196, 1107-1112. • Chiang Kao and Shiuh-Nan Hwang, 2010, Efficiency measurement for network systems: IT impact on firm performance. Decision Support Systems 48, 437-446. • Chiang Kao and Shiuh-Nan Hwang, 2011, Decomposition of technical and scale efficien-cies in two-stage production systems. European J. Operational Research 211, 515-519. • Chiang Kao, 2011, Efficiency decomposition for parallel production systems. J. Operational Research Society (accepted) (SCI) doi:10.1057/jors.2011.16. • Chiang Kao, 2008, A linear formulation of the two-level DEA model. Omega, Int. J. Management Science 36, 958-962. • Chiang Kao and Shiang-Tai Liu, 2004, Predicting bank performance with financial forecasts: A case of Taiwan commercial banks. J. Banking & Finance 28, 2353-2368.