Data Envelopment Analysis (DEA) and Its Applications

Download Report

Transcript Data Envelopment Analysis (DEA) and Its Applications

Performance Evaluation
and Benchmarking Using
DEA
Joe Zhu
Department of Management
Worcester Polytechnic Institute
Worcester, MA 01609
[email protected]
www.deafrontier.com
Outline
• What is DEA?
• New Models/Uses
• Two-Stage Model
• Context-dependent DEA
• Benchmarking
• Books
Data Envelopment Analysis
Joe Zhu
2
DEA & Banking

The Banking industry has been the subject of DEA
analysis by researchers in various areas and
probably is the most heavily studied business


Branches
Banks across countries
Inputs
FTE in dollars
Premise/IT expenses
Other Expenses
Bank
Branch
Outputs
Loan Balances
Deposit Balances
Securities Balances
Gross Revenue
Source: Paradi et al. 2004
Data Envelopment Analysis Joe Zhu
3
DEA

Deals with multiple performance measures
(inputs and outputs) in a single integrated
model

Includes any necessary measures related to the
characterization of banking performance
Identifies a “base-line” for comparisons in
continuous improvement program
 Provides specific targets for improvement
(over time)

Data Envelopment Analysis Joe Zhu
4
Why DEA?
Output
DEA Best-Practice
Frontier
6
6 6
6
6
6 6
6
6
6
- Regression can accommodate
Multiple inputs or
outputs but not both
6
6
- Regression requires a
functional relationship
between in/outputs
Data Envelopment Analysis Joe Zhu
predicted
average
behavior
6
Input
- Regression provides only
average relationships
not best practice
5
Basic DEA Benchmarking Information

DEA gives




Efficiency rating, or score, for each DMU
Efficiency reference set: peer group
Target for the inefficient DMU
Information on how much inputs can be
decreased or outputs increased to make the
unit efficient – improving productivity &
performance
Data Envelopment Analysis Joe Zhu
6
DEA & Performance Improvement
Output
DEA Best-Practice
Frontier
D
6
D
6
6
6
6
Output
augmentation
K
D
Input reduction
Input
Data Envelopment Analysis Joe Zhu
7
Benefits

The establishment of the efficient frontier
consisting of the best performing DMUs
 A projection to the efficient frontier - a guide
to “what to do” for the DMU managers
 The identification of the peer group, a
reasonable argument why it is a FAIR
comparison
 An indication of how important a particular
DMU is as a role model
Data Envelopment Analysis Joe Zhu
8
How DEA works?
Supply Dollars
450
400
350
B1
B4
300
T1
250
200
B2
B5
150
100
50
B3
0
0
20
40
60
Teller Hours




5 branches
Three (B1, B2 & B3 are efficient – best practice frontier)
B4 and B5 are inefficient
Target for B4 is T1 (decrease inputs)
Data Envelopment Analysis Joe Zhu
9
How DEA works?
450
H1
400
Market Share
350
H2
H5
300
250
T2
200
H4
150
H3
100
50
0
0
100
200
300
400
500
Sales




5 branches
Three (H1, H2 & H3 are efficient – best practice frontier)
H4 and H5 are inefficient
Target for B4 is T2 (increase outputs)
Data Envelopment Analysis Joe Zhu
10
More Information on DEA

Web



Books




www.deafrontier.com
…
Cooper, W.W., Lawrence M. Seiford, and K. Tone. 2000. Data
Envelopment Analysis: A Comprehensive Reference Text with
Models, Applications, References, and DEA-Solver Software.
Kluwer Academic Publishers, Boston
Zhu, J. 2002. Quantitative Models for Performance Evaluation
and Benchmarking: Data Envelopment Analysis with
Spreadsheets. Kluwer Academic Publishers, Boston
…
Softwares


DEA Excel Solver (DEAFrontier)
…
Data Envelopment Analysis Joe Zhu
11
DEA & IT
• Indirect impact of IT on productivity
• Two Stage DEA Model
• Chen, Y. and Zhu, J., Measuring information technology’s indirect
impact on firm performance, Information Technology &
Management Journal, Vol. 5, Issue 1-2 (2004), 9-22.
Data Envelopment Analysis
Joe Zhu
12
What is Benchmarking?
... a process of defining valid
measures of performance
comparison among peer units,
using them to determine the
relative positions of the peer
units and, ultimately, establishing
a standard of excellence.
Data Envelopment Analysis Joe Zhu
13
Acceptance System Decision Rule


Trout et al. (1996, COR, Vol 23, 405-408)
– acceptance/rejection of credit risks
Seiford & Zhu (1998, COR, Vol. 25, 329332)
Benchmarking
Data Envelopment Analysis Joe Zhu
14
Approach
Output
DEA Best-Practice
Frontier/Benchmarks

6
6
6
6T
T
66
new
activities
T
T
Data Envelopment Analysis Joe Zhu
Input
15
Business Process Reengineering
• Compare new bank branches to the
traditional best-practice frontier.
performance
ss
s
traditional
best practice
ss
s
s
Data Envelopment Analysis Joe Zhu
time
16
Benchmarking results
New branch best-practice

Overall, new
branches’
performance is
improving
traditional branch bestpractice
Cook, W.D., Seiford, L.M. and Zhu, Joe, Models for
performance benchmarking: Measuring the effect of ecommerce activities on banking performance, OMEGA,
Vol.32, Issue 4 (2004), 313-322.
Data Envelopment Analysis Joe Zhu
17
Context-dependent DEA
• Context-dependent
DEA
• Consumer’s choice
is influenced by the
context
• The performance of
DMUs should also
reflect “context”
Data Envelopment Analysis
Joe Zhu
18
Journal of Marketing
Research
• Book Review
– contextdependent DEA
(identifying
possible
moderating
results) intriguing
and, conceivably,
breathtaking.
Data Envelopment Analysis
Joe Zhu
19
Service Productivity
• D. Sherman and J. Zhu, Service
Productivity Management: Improving
Service Performance Using Data
Envelopment Analysis (DEA)
Springer, Boston, 2006, ISBN 0-38733211-1.
Data Envelopment Analysis Joe Zhu
20
DEA Handbook
W.W. Cooper, L.M.
Seiford and J. Zhu
Handbook on Data
Envelopment
Analysis, Springer,
Boston, 2004, ISBN
1-4020-7797-1
Data Envelopment Analysis
Joe Zhu
21
Modeling Issues
 W.D. Cook and Joe
Zhu, Modeling
Performance
Measurement:
Applications and
Implementation
Issues in DEA,
Springer, Boston,
2005, ISBN 0387-24137-X.
Data Envelopment Analysis
Zhu
Joe
22
DEA & Finance
• Mutual funds
• CTAs
• Hedge Funds
G. Gregoriou and Joe Zhu, Evaluating
Hegde Funds and CTA Performance:
Data Envelopment Analysis Approach,
John Wiley & Sons, New York, 2005,
ISBN 0-471-68185-7 .
Data Envelopment Analysis
Joe Zhu
23
Data Envelopment Analysis
Joe Zhu
24