Estimating output-specific efficiencies / by Dieter Gstach.

Format:

Book

Author/Creator:

Gstach, Dieter.

Series:

Applied optimization ; v. 61.

Applied optimization ; v. 61

Language:

English

Subjects (All):

Stochastic processes.

Physical Description:

xiii, 204 pages : illustrations ; 25 cm.

Place of Publication:

Dordrecht ; Boston : Kluwer Academic, [2002]

Summary:

This book presents a novel technique for comparative analysis of multi-output firms. In particular it provides a statistical foundation for output-specific efficiency estimation. This is necessary to find out whether firms of an economic sector systematically produce some good A more efficiently than some good B and to quantify eventual differences. In the presence of such differences the application of existing estimation techniques is either inappropriate or does not lead to statistically meaningful results. In this sense the book fills a remaining gap in the relevant literature. The key idea behind the approach is to use posterior distributions of target output-ratios of firms to determine reference points for performance comparison purposes. Monte Carlo simulations and an application to a real world data set indicate expectable small sample performance and the practical usefulness of the presented approach. Audience: Economists, MS/ OR researchers, and their libraries.

Contents:

Part I Motivating the concept

3 Motivation 27

4 Geometrical illustration 30

5 Interpreting the difference 34

Part II Operationalizing the concept

2. Technology Estimation 41

1 Statistical structures underlying DEA 42

2 Output-ratios to characterize technology 44

3 DEA bias correction 54

4 Estimator consistency 60

3. Relation to Radial Measures 63

1 Ouput-specific vs. radial efficiencies 64

2 An example that works 68

3 So why not use simple regression analysis? 71

4 A counterexample 72

4. Markov Chain Monte Carlo Analysis 77

1 The Metropolis-Hastings algorithm 79

2 Single-component updates 80

3 Sampling from conjugate distributions 81

5. Data Generating Process 83

1 Target output ratios 83

2 Output specific efficiencies 84

3 Distribution of output vectors 86

6. Identification 89

1 The basic tradeoff in an expectational perspective 90

2 The role of domain observations 92

3 Likelihood surface 98

7. Posterior Distributions 109

1 The prior assumptions 109

2 Sampling 110

3 Scale Invariance 115

Part III Evaluating the concept

8. Estimator Performance 127

1 Sample generation 127

2 Case of DEA-estimated frontier 132

3 Case of known frontier 141

Part IV Putting the concept to work

9. An Application 153

2 Estimating technology 157

3 The statistical model 158

4 Constructing the Markov chains 163

5 Data 167

6 Results 172

7 Conclusions from the application 183

2 Routes for future research 194.

Notes:

Includes bibliographical references.

ISBN:

1402004834

OCLC:

49739712