Transcript Measuring Embodied Technical Change in the US Local Telephone

Information Technology, Efficiency
and Productivity: Evidence From
Korean Local Governments
Nakil Sung
University of Seoul, [email protected]
International Telecommunications Society
15th Biennial Conference 2004
Berlin, Germany
Contents
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and TFP
Growth Estimation
4. Result 2: Regression Results
5. Conclusion
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and TFP Growth
Estimation
4. Result 2: Regression Results
5. Conclusion
Research Motivation
Yes, Too Many Studies on IT
Productivity Effects

The first generation of studies often provided
mixed empirical results on the Solow’s productivity
paradox until the late 1990’s
 The productivity paradox was partly resolved by
observing faster productivity growth in developed
countries.

The second generation of studies focuses on the
performance of IT-using sectors.
 Many studies agree that rapid productivity growth in
IT-producing sectors led to better performance of
national economy.
Research Motivation
The Second Generation of ‘IT
Productivity’ Literature

Recent studies are fairly successful in confirming
positive effects of IT.
 For example, Jorgenson (2001), Brynjolfsson and Hitt
(1996, 2000), Stiroh (2001), Mun and Nadiri (2002).

These studies mainly use micro data such as
industry or firm data.
 The use of micro data is a good way of identifying ‘IT
productivity effects’ because it provides researchers
with a chance of distinguishing IT-heavy users from
IT-light users.
Research Motivation
But, More Studies Are Still
Needed In Some Areas

As usual, the current literature does not
distinguish (technical) efficiency from productivity.
 Only Milana and Zeli (2002) examine the relationship
between IT investments and technical efficiency.

Is there any better measure of IT-using activities?
 Many studies use the purchase costs of IT-related
equipment as a proxy for the state of IT.
 On the other hand, the performance of IT users must
be affected by effective use and applications of IT.
Research Motivation
Korean Case Provides a Good
Research Opportunity Because…

The Korean government has reported an index of
IT-using activities (called Informatization Index)
for all local governments.
 This index measures a wide range of IT-related
activities.

Also, like other countries, good and reliable data
on local public services are publically available in
Korea.
Research Motivation
Informatization Index
Components
Support
Investment and
Equipments
Human and
Organizational
Factors
Usage and
Applications
Measures
• Number of IT related meetings and plans per year
•
•
•
•
•
•
Ratio of IT related to total budget
Number of servers and PC’s
Purchase costs of software
Diffusion rate of e-mail ID’s
Computer and information security activities
Efficiency of network management etc
• Ratio of IT related to all staffs
• IT related education activities
• Number of IT related license holders etc
• Usage degree and pattern of bulletin board and
homepage
• Application of IT to administrative process,
Development degree of e-government (including
electronic handling of public services)
• Degree of electronic approvals etc
Research Motivation
Then, the Study Has Two
Objectives

Measuring (technical) efficiency and productivity
growth for all Korean local governments
 By applying conventional methods

Examining the effects of IT on (technical)
efficiency and productivity growth
 By using the Information Indexes
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and TFP Growth
Estimation
4. Result 2: Regression Results
5. Conclusion
Research Methods
Research Strategy: Two Stage
Approach

First Stage: Measurement of (technical) efficiency
and TFP growth by using distance functions.
 Both efficiency and productivity growth are defined
and measured by using distance function
 The distance function is estimated by applying data
envelopment analysis (DEA).

Second Stage: Efficiency and productivity
regressions
 Efficiency scores and productivity growth rates are
regressed on some regional characteristic variables
and the Informatization Index.
Research Methods
Technical Efficiency: OutputOriented Measure
Y1
OA
TE 
OB
B
Production Possibility Curve
A
O
Distance Function :
Y2
d o ( x, y )  min {  :
y

 P ( x) }
Research Methods
Malmquist Productivity Index

Period-s (output-oriented) Malmquist productivity
index
s
d
s
o ( xt , y t )
mo ( x s , x t , y s , y t )  s
d o ( xs , y s )

Malmquist productivity index between period-s
and period-t
mo ( x s , xt , y s , y t )  [m ( x s , xt , y s , y t )  m ( x s , xt , y s , y t )]
s
o
s
o
s
o
t
o
t
o
t
o
d ( xt , y t ) d ( xt , y t ) 1 / 2
[

]
d ( xs , y s ) d ( xs , y s )
1/ 2
Research Methods
Decomposition of Malmquist
Productivity Index
d ot ( xt , y t ) d os ( xt , y t ) d os ( x s , y s ) 1 / 2
mo ( x s , x t , y s , y t )  [ s
][ t
 t
]
d o ( x s , y s ) d o ( xt , y t ) d o ( x s , y s )
Efficiency
Change
Technical
Change
Research Methods
Data Envelopment Analysis

Charnes-Cooper-Rhodes (CCR) Model: constant
returns-to-scale (CRS) assumption
Min ,
s.t.


x k  X  0
 y k  Y  0
  0
The optimal solution
to this LP problem is
the output distance
function.
Bankers-Charnes-Cooper (BCC) Model: variable
returns-to-scale (VRS) assumption
 convexity condition:

j
1
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and
TFP Growth Estimation
4. Result 2: Regression Results
5. Conclusion
Efficiency and TFP Growth
Two Levels of Local Governments
in Korea
KOREA
Metropolitan
Cities: 7
Districts (Gu):
69
Provinces: 9
Cities (Shi): 70
Samples
Counties (Gun):
83
Efficiency and TFP Growth
Input and Output Variables
Variables
Input
Variables
Output
Variables
Definition
NLSP
Number of local servants per 100 persons
CEXP
Annual constant expenditures per capita
PRWS
Penetration rate of water supply
AUPP
Area of urban parks per person
RRLA
Ratio of road length to area
NMVP
Number of registered motor vehicles per person
PRWR
Penetration rate of sewage and refuse disposal
CSWP
A seating capacity of social welfare institutions
per 100 persons
NSRP
Number of Basic Livelihood Security recipients
per 100 persons
NCPP
Number of building construction permits per 100
households
NCAP
Number of civil affairs and petition cases per
person
Efficiency and TFP Growth
Application of DEA Models

Both CCR (CRS) model and BCC (VRS) model are
applied to input and output data over the period
1999-2001. Then the estimates are averaged.

Operation environment of local governments
should be taken into account.
 Method 1: First, evaluate local governments under
handicaps and second, use this information to evaluate
local governments in better environments.
 Method 2: Evaluate local governments only within the
group.
Efficiency and TFP Growth
Average Technical Efficiency
Scores (1999-2001)
Method 1
Method 2
CRS Model VRS Model CRS Model VRS Model
District
(Gu)
Mean
0.850
0.999
0.851
0.999
STD
0.145
0.005
0.144
0.005
City
(Shi)
Mean
0.772
0.984
0.820
0.991
STD
0.156
0.027
0.142
0.019
County
(Gun)
Mean
0.657
0.976
0.657
0.976
STD
0.182
0.051
0.182
0.051
Mean
0.753
0.986
0.769
0.988
STD
0.181
0.036
0.180
0.035
Total
Note: STD implies standard deviation
Efficiency and TFP Growth
Average TFP Growth Rates
(1999-2001)
Method 1
Method 2
Efficiency
Change
TFP
Change
Efficiency
Change
TFP
Change
District
(Gu)
Mean
2.7%
4.2%
2.7%
4.8%
STD
9.2%
10.8%
9.1%
11.2%
City
(Shi)
Mean
3.5%
4.5%
0.6%
4.0%
STD
10.2%
15.4%
9.0%
14.3%
County
(Gun)
Mean
5.2%
-8.6%
5.2%
-8.6%
STD
10.8%
11.0%
10.8%
11.0%
Total
Mean
3.9%
-0.5%
3.0%
-0.5%
STD
10.1%
13.9%
9.9%
13.7%
Note: STD implies standard deviation
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and TFP Growth
Estimation
4. Result 2: Regression
Results
5. Conclusion
Regression Results
Efficiency and Productivity
Regressions
TIE or dTFP   0    j RC j   z ZSCORE  

Definition of variables
 TIE: technical inefficiency score,
1
TIE 
1
TE
 dTFP: TFP growth rate (Malmquist productivity index)
 RC: regional characteristic variables
 ZSCORE: Informatization Index

Estimation technique: censored Tobit method
 The TIE takes a value between 0 and infinity.
Regression Results
Regional Characteristic Variables
Variables
Definition
SIZE1
Dummy for regions with population of more
than 100,000 and less than 300,000
SIZE2
Dummy for regions with population of more
than 300,000
DISTRICT
Dummy for districts
CITY
Dummy for cities
Regional
POPD
Population density
Variables
APLS
Area per 100 local servants
NETP
Number of establishments, including
individuals and corporation, per person
NSEP
Number of service related establishments,
including hotels and restaurants, per person
RWTR
Number of workers per person
CLTP
Amount of collected local tax per person
Characteristic
Regression Results
Technical Efficiency Regressions
Equation 1
Equation 2
Equation 3
Equation 4
SIZE1
-0.237***
-0.271***
SIZE2
-0.374***
-0.457***
DISTRICT
-0.194**
CITY
-0.090
POPD
-0.015***
APLS
0.001***
0.002***
0.002***
NETP
1.347
2.453
2.618
NSEP
9.891*
13.721**
21.925***
RWTR
-0.257
-0.453
-0.822***
CLTP
ZSCORE
0.104*
-0.128*
-0.162**
-0.189**
-0.179**
Note: *,**,** implies statistical significance at 10%, 5%, and 1% level,
respectively
Regression Results
TFP Growth Rate Regressions
Equation 1
Equation 2
Equation 3
Equation 4
SIZE1
0.060**
0.063***
SIZE2
0.061**
0.057**
DISTRICT
0.111***
CITY
0.108
POPD
0.002
APLS
-0.001***
-0.000
-0.000*
NETP
-1.051
-1.105
-1.029
NSEP
4.061*
3.263
0.022
RWTR
0.126
0.143
0.264**
CLTP
0.073***
ZSCORE
0.063**
0.055**
0.077*
0.039
Adjusted R2
0.174
0.223
0.086
0.203
Note: *,**,** implies statistical significance at 10%, 5%, and 1% level,
respectively
1. Research Motivation
2. Research Methods
3. Result 1: Efficiency and TFP Growth
Estimation
4. Result 2: Regression Results
5. Conclusion
Conclusions
Summary

Local governments in more populous regions tend
to be more technical efficient and to experience
higher TFP growth.

Local governments in more business- or industrycentered regions may operate closer to production
frontier and enjoy higher TFP growth.

There exists a negative (positive) relationship
between gross regional product and technical
efficiency (TFP growth).

Local governments with higher level of
informatization operate closer to production
frontier and experience higher TFP growth rate.
Conclusions
Contribution

The study successfully confirms a positive role of IT
in improving technical efficiency and accelerating
productivity growth.

The study provides strong cases on the
development of e-Government projects in many
countries.
Thank You For Your
Attention!