84
ASSESSING GREENING ACTIVITIES OF SMALL AND MEDIUM ENTERPRISES IN THE AUTOMOBILE INDUSTRY IN THAILAND BY DEA APPROACH BY NIYAT TESFAI A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING IN LOGISTICS AND SUPPLY CHAIN SYSTEMS ENGINEERING SIRINDHORN INTERNATIONAL INSTITUTE OF TECHNOLOGY THAMMASAT UNIVERSITY ACADEMIC YEAR 2014

Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

  • Upload
    others

  • View
    0

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

ASSESSING GREENING ACTIVITIES OF SMALL AND

MEDIUM ENTERPRISES IN THE AUTOMOBILE

INDUSTRY IN THAILAND BY DEA APPROACH

BY

NIYAT TESFAI

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE

REQUIREMENTS FOR THE DEGREE OF MASTER OF

ENGINEERING IN LOGISTICS AND SUPPLY CHAIN SYSTEMS

ENGINEERING

SIRINDHORN INTERNATIONAL INSTITUTE OF TECHNOLOGY

THAMMASAT UNIVERSITY

ACADEMIC YEAR 2014

Page 2: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

ASSESSING GREENING ACTIVITIES OF SMALL AND

MEDIUM ENTERPRISES IN THE AUTOMOBILE

INDUSTRY IN THAILAND BY DEA APPROACH

BY

NIYAT TESFAI

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF MASTER OF

ENGINEERING (LOGISTICS AND SUPPLY CHAIN SYSTEMS

ENGINEERING)

SIRINDHORN INTERNATIONAL INSTITUTE OF TECHNOLOGY

THAMMASAT UNIVERSITY

ACADEMIC YEAR 2014

Page 3: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative
Page 4: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

ii

ABSTRACT

ASSESSING GREENING ACTIVITIES OF SMALL AND MEDIUM ENTERPRISES IN THE AUTOMOBILE INDUSTRY IN THAILAND BY DEA

APPROACH

By

NIYAT TESFAI

B.Sc. (International Management), University of Applied Sciences Karlsruhe, 2014

Small and medium enterprises (SMEs) are contributing more and more to Thailand’s economy. As a result, the Thai industry is increasing in energy intensity, mainly in the manufacturing sector due to outdated machines. Furthermore, studies have revealed that non-OECD countries will continue to increase its energy consumption, as a consequence of their growing numbers of inhabitants as well as economy. The purpose of this study is to assess greening activities and benchmark the energy consumption of SMEs in the Thai automotive sector. The study is fulfilled in corporation with GIZ Bangkok, which are implementing a Green Auto Project in which they consult SMEs regarding greening activities. With respect to the project, data has been collected for the evaluation, including the energy consumption as well as amount of goods produced. The contribution of this proposed study is a method of grouping several SMEs from the automotive sector according to production processes, so that the Data Envelopment Analysis (DEA) can be done amongst homogenous Thai SMEs in the automotive sector and benchmarking them is more representative. Furthermore, it is the first study to apply DEA to benchmark energy efficiency in the Thai automotive sector for SMEs. The DEA model is input-oriented, i.e. that given the same amount of outputs; inputs (energy consumption) shall be reduced. Efficient SMEs will be reported and results will reveal a slack analysis, indicating the amount of input that individual inefficient DMUs must reduce in order to reach the efficiency frontier.

Keywords: Data Envelopment Analysis, Energy Efficiency, Small and Medium Enterprises

Page 5: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

iii

Table of Contents

Chapter Title Page

Signature Page i

Abstract ii

Table of Contents iii

List of Tables v

List of Figures vi

List of Abbreviations vii

1 Introduction 1

1.1 Statement of the Problem and Motivation 2

2 Literature Review 5

2.1 Environmental Issues 5

2.2 Green Manufacturing 6

2.3 Measuring Green Production 9

2.4 Measuring Performance of Small and Medium Enterprises in

Thailand 14

3 Data Envelopment Analysis (DEA) 18

3.1 Theoretical Framework 18

3.2 Benchmarking with Data Envelopment Analysis 21

3.2.1 Parametric and Non-Parametric Methods 22

3.3 Advantages and Disadvantages of Data Envelopment Analysis 24

Page 6: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

iv

4 The German International Corporation 27

4.1 Company Profile 27

4.2 Introduction of the Green Auto Project 28

4.3 Approach for Working Packages of Project 31

5 Methodology and Procedure 37

5.1 Data Collection 37

5.2 Data Standardization Approach 38

5.3 Problem Formulation 42

5.4 Methodology and Modelling Undesirable Measures 47

6 Energy Efficiency Analysis 50

6.1 Profitable Environmental Management Approach 50

6.2 Best Available Technique Approach 55

6.3 Summary 58

6.4 Policy Recommendations 60

6.5 Conclusion 62

7 Limitations and Delimitations 64

7.1 Limitations of Study 64

7.2 Limitations of Instrument 64

References 66

Appendices 74

BAT Raw Data 74

Page 7: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

v

List of Tables

Tables Page

Table 2.3.1: Metrics and examples of SSCM .......................................... 10

Table 2.3.2: Major themes used by identified metrics ............................. 11

Table 5.2.1: Raw data of BAT peer group 1 ............................................ 40

Table 5.2.2: Standardized BAT data ........................................................ 41

Table 5.3.1: Supply Chain Operations ..................................................... 42

Table 6.1.1: PREMA Data ....................................................................... 51

Table 6.1.2: DEA Results of PREMA ..................................................... 52

Table 6.1.3: Slacks for PREMA Data ...................................................... 54

Table 6.2.1: DEA Results of BAT ........................................................... 55

Table 6.2.2: Standardized BAT Data ....................................................... 57

Table A.1: BAT Raw Data ...................................................................... 74

Page 8: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

vi

List of Figures

Figures Page

Figure 4.2.1: Project Structure ................................................................. 30

Figure 4.2.2: GIZ’s project target group .................................................. 30

Figure 5.2.1: Z-Standardisation ............................................................... 39

Figure 5.3.1: Efficient Frontier of 5 DMUs ............................................. 44

Figure 5.3.2: Output-oriented Efficient Frontier ..................................... 46

Page 9: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

vii

List of Abbreviations

BAT: Best Available Technique BMZ: Bundesministerium für wirtschaftliche Zusammenarbeit und

Entwicklung BREF: Best Available Techniques Reference Documents IC: Intellectual Capital CO2: Carbon dioxide CP: Clean Production CRS: Constant Return to Scale CSR: Corporate Social Responsibility DEA: Data Envelopment Analysis DEDE: Department of Alternative Energy Development and

Efficiency DFA: Distribution Free Approach DIP: Department of Industrial Promotion DMU: Decision Making Unit e.g.: exempli gratia (for example) EE: Energy Efficiency EENet: Energy Efficiency Network ENCON: Energy Conservation ESCO: Energy Service Companies etc.: et cetera EU: European Union FDH: Free Disposal Hull GDP: Gross domestic product GIZ: Gesellschaft für internationale Zusammenarbeit GSCM: Green Supply Chain Management i.e.: id est (that is) IEO: International Energy Outlook IQA: Intelligent Quality Assurance ISO: International Organization for Standardization KPI: Key Perfomance Index LRP: Loss Reduction Process ME: Medium Enterprises MEPS: Minimum Energy Performance Standards n.d. No date NGO: Non Governmental Organisation OECD: Organization for Economic Cooperation and Development OSMEP: Office of Small and Medium Enterprises Promotion RE: Renewable Energy SCP: Sustainable and Clean Production SE: Small Medium Enterprises SEA: Southeast Asia SFA: Stochastic Frontier Approach SME: Small and Medium Enterprises

Page 10: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

viii

SOP: Standard Offer Program SP: Sustainable Production SSCM: Sustainable Supply Chain Management TAI: Thai Automotive Institute TFA: Thick Frontier Approach VRS: Variable Return To Scale

Page 11: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

1

Chapter 1

Introduction

The importance of green manufacturing is raising awareness in today’s global

economy. Businesses are taking sustainable production and environmental activities

as part of their corporate social responsibilities (CSR), as they have a huge impact on

environmental changes (Tikul, 2014). Not only are companies modifying single

processes with respect to their CSR activities, but are involving the whole supply

chain as well. The social and environmental aspects can be integrated into a

companies’ supply chain through inter-organizational management (Kogg & Mont,

2012).

This study focuses on small and medium enterprises (SMEs) in Thailand and aims to

support the Green Auto Project of GIZ1 (Deutsche Gesellschaft für internationale

Zusammenarbeit) in Bangkok. Thailand’s SMEs have contributed 36.6% of the

overall gross domestic product (GDP) in 2011 and thus, have a great impact on the

economy. When looking at the size of enterprises it should be indicated that small

enterprises (SE) account more to the countries GDP than medium enterprises (ME).

While the value of SEs was 24.5% of Thailand’s total GDP in 2011, MEs only

contributed 12.1%. Another important finding from the Office of Small and Medium

Enterprises Promotion (OSMEP) in Thailand is that when considering the GDP

structure, the agricultural sector contributed 13.3% while non-agricultural sector

constituted 86.7% of Thailand’s total GDP in 2011. When breaking down the non-

agricultural sector the manufacturing sector stands out with the most influential

contribution to the economy with 34% of Thailand’s total GDP (OSMEP, 2011). The

economy in Thailand and Asia in general is growing continuously. The end of 2014

predicts that Thailand’s automotive production capacity reaches three million

vehicles. The automotive sector in Thailand has a large impact on the economy. In

fact domestic and exported sales accounted 10% of the GDP in 2012. Looking at the

largest automobile manufacturers of the world, Thailand ranked 15th largest in 2011

1  Engl: German International Corporation  

Page 12: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

2

producing 1.5 million vehicles, wherein the total world production accounted 80

million (Thailand Automotive Institute, 2012).

As a result, GIZ in Thailand started a project named Greening Supply Chains in

the Thai Auto and Automotive Parts Industries – in short, Green Auto Project - which

aims to improve the evaluation and optimization of outdated processes of SMEs

concerning a better environmental performance. To be specific, the project helps

SMEs to improve productivity and increase the quality of products, reduce waste and

decrease energy consumption. This can be done by using improved technologies,

sharing relevant information and developing policies (GIZ, n.d.). With respect to the

high influential number of SMEs in Thailand and GIZ’s project, this research focuses

on the energy consumption of SMEs in Thailand, which belong to GIZ’s target group.

1.1 Statement of the Problem and Motivation

SMEs are gaining a lot of thoughtfulness in CSR literature. CSR activities

include social, environmental and sustainable aspects. The rising relevance of SMEs

can be explained through the big contribution they make to the economy and their

uniqueness compared to large companies. The main characteristic is that families or

owners conduct SMEs, which are obliged to follow policies of business partners and

the community. Furthermore, SMEs have a shortage of resources and support that will

help them to execute greening activities. Hence, the characteristics of SMEs bring

obstacles along that hinder the implementation of CSR in terms of greening activities.

It should be noted that the community and non-governmental organizations (NGOs)

are demanding SMEs to perform more sustainable activities along their supply chain.

The goal is to influence business partners, so that a basis of pro-environmental and

sustainable behaviors can be developed (Ciliberti et al., 2008).

As mentioned above, the NGO GIZ in Thailand is currently implementing a

Green Auto Project targeting 250 out of 500 manufacturing SMEs in Thailand.

Activities that are done with respect to the project are advising SMEs on investing in

new technologies, so that waste and energy consumption can be reduced. In order to

be able to invest, GIZ supports in gaining access to green processing, i.e. the capacity

of the bank, giving trainings to bank officers about the importance sustainability,

Page 13: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

3

sharing information between banks and the target groups as well as guide the SME

bank in Thailand to develop a concept of green loan. Furthermore, giving seminars

about green concepts strengthens existing clubs. At last but not least, policy

recommendation on green production and sustainability are given through marketing

consultants so that more companies can be attracted (see chapter 4).

The target groups, in this case the SMEs that have actually implemented the

concept tend to have different results. Some companies may have invested in new

technologies, however, were not able to achieve the same amount of savings as other

SMEs. Also the diversity of SMEs leads to different results. Some organizations are

smaller, others bigger and some may have enough resources while others don’t. The

complexity lies in comparing the SMEs and finding commonalities, so that they can

be benchmarked to one another. This research will focus on energy efficiency that the

SMEs were able to accomplish through the participation of the project and benchmark

the enterprises respectively. The overall goal is to identify a best practice enterprise,

whose strategic approach can be implemented by those who did not do well. It aims to

motivate SMEs to participate in the Green Auto Project and inform them about the

energy savings they can possibly achieve.

The Thai consumption of electricity constitutes 169.4 billion kWh (2012) and

ranks number 23 compared to the rest of the world (The World Factbook, 2012).

According to the U.S. Energy Information Administration's International Energy

Outlook 2013 (IEO2013) countries that do not belong to the Organization for

Economic Cooperation and Development (OECD), will have a continuous increase in

energy consumption due to the growing economies and number of inhabitants. In fact,

the energy consumption share in non-OECD countries is said to increase from 54% of

the overall world energy consumption in 2010 to 65% in 2040 (EIA, 2013). As a

result, Thailand as one of the most influential manufacturing countries needs to adapt

new strategies and regulations, in order to be able to reduce its energy consumption.

GIZ’s Green Auto Project has strongly contributed on the achievement of reducing

energy consumption for the targeted SMEs in Thailand. The indicated growing energy

consumption in non-OECD countries shows the relevance of investing in new

technologies and green processing, as environmental changes have and will affect the

whole world. Thus, the purpose of this study is to benchmark the energy efficiency for

Page 14: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

4

the targeted SMEs in order to benchmark the enterprises using data envelopment

analysis (DEA) and identify the best-practice SME.

DEA functions as a performance tool that analyzes and compares multiple units

to one another at any given point of time. Most importantly, “DEA is a performance

assessment tool since it estimates relative efficiency of a set of decision making units

(DMUs)”. DMUs can be described as units that use multiple inputs in order to

produce multiple outputs. The overall goal of DEA is to optimize the efficiency of a

DMU by modifying input and output. Traditionally, this can be done in two ways.

One may either maintain the number of output while decreasing the number of input

or increase output while maintaining input (Azadi et al., 2014). Chapter 3 will give

further details on DEA.

Upon completion of the thesis, the following questions shall be answered:

a) What are the reasons of different achieved results for the SMEs?

b) Which SMEs have reached the efficient frontier?

c) What can be done to make the energy consumption more efficient?

After results have been obtained, the above-mentioned questions will be answered as

a summary.

Page 15: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

5

Chapter 2

Literature Review

2.1 Environmental Issues

The importance and awareness of green manufacturing is rising from year to year

and has reached a point where companies are considering strategies to implement it.

Within the last decades, manufacturing companies have caused tremendous harm to

the environment, as through production a lot of energy and water is consumed as well

as waste generated, which has lead to environmental changes that will affect all living

things. However, although manufacturing firms are aware of the harms caused, a lot

of firms do not consider to change their manufacturing process into a sustainable one,

as it is perceived that the benefits do not outweigh the risks involved (Lee et al.,

2014). Deif (2011) claims that because of the rising global awareness of

environmental changes and risks, customer preferences have changed and are in favor

of sustainable manufacturing. Tseng et al. (2013) highlight the importance of green

production by stressing out that the current used technologies are outdated, which is a

big factor adding pollution and harm to the environment. As a result, several changes

must be done in terms of policies, education and processes to eventuate in greater

awareness of environmental changes, especially in Asia. For Severo et al. (2014),

companies can reduce their contribution to environmental harm by adapting

environmental-friendly practices for e.g. cleaner production approaches.

One must bear in mind the increasing energy intensity of the Thai industry

mainly caused by the rising manufacturing sector, which are still using inefficient

industrial plants. The increasing energy consumption has led to environmental

changes, causing global warming. The transport sector is said to be the most energy

intensive, as Thailand has an increased level of motorization and not to forget the high

dependence on the road transport. The Thai Government has been trying to reduce

energy consumption by introducing laws and regulations. The under the Thai Energy

Conservation and Promotion Act 1992 developed Energy Conservation Fund

(ENCON Fund) has been a supportive financial act for developing energy efficiency

(EE) and renewable energy (RE). This aid is controlled by the baseline of the

Page 16: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

6

governments conservation plan (Wang et al., 2013). However, although the Thai

government has established several EE and RE projects, the energy consumption has

not changed much from the year 2000 onwards. The main causes for the stagnated

development are the small enhancements in EE and the fact that the economy requires

energy-intensive industries. As a result, the Thai government has developed a 20-Year

National Energy Efficiency Development Plan that ensures to reduce energy intensity

by 25% likened with levels of 2005 by 2030. Alongside, another 15-year Renewable

Energy Development Plan has been established, which commits to increase the usage

of alternative energy by 6.4% in 2008 to 20% by 2022 (Wang et al., 2013). Some

obstacles however still hinder the implementation to enterprises. For example, the

energy managers programs and measures which help to ensure a development of EE

did not manage to reach the promised results, as managers are not motivated in

implementing EE investments due to the high transaction costs. Additionally, EE

investments are avoided because of the risks involved. Managers often doubt a

successful realization. Furthermore, financial institutions such as banks do not have

the required knowledge and lack expertise that is necessary for establishing EE

business portfolios. This is the reason why Thai banks cannot provide corresponding

EE loans, especially to SMEs and Energy Service Companies (ESCOs) (Wang et al.,

2013).

2.2 Green Manufacturing

The term green manufacturing was introduced to indicate the current

manufacturing trend that aims to be more sustainable by using green

strategies/objectives and principles as well as green technology and innovation (Deif,

2011). However, existing barriers are making the implementation of green

manufacturing difficult. According to Voon-Hsien Lee et al. (2014) the barriers that

hinder manufacturing firms from implementing a green manufacturing process can be

classified into external forces in the industry and internal forces within the company.

One aspect that stems the implementation is the fact that not only single

manufacturing processes must be modified into a sustainable one, but the whole

supply chain instead. This includes all suppliers and partners, which makes it very

Page 17: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

7

difficult to control (Lee et al., 2014). Additionally, many enterprises are not aware of

the clear definition of “green”, which leads to confusion and restricts the modification

of green manufacturing. Not to neglect are the costs and regulations that result in

barriers too (Tseng et al., 2013). The term “green” relates to every activity that is

environmental-friendly. Thus, “green manuacturing” stands for the realisation of

production that intends to minimize the harm it causes to the environment (Yusuff and

Panjehfouladgaran, n.d.).

The number of researches concerning green manufacturing are getting more and

more, as the awareness and importance of it arises. The existing studies mainly focus

on the relevance and concept of green manufacturing as well as proposed models,

tools and practices to actually implement the green manufacturing paradigm.

Voon-Hsien Lee et al. (2014) have revealed that involving sustainability into the

whole supply chain will result in increased monitoring of environmental-friendly

activities and the implementation of the so-called “Re-“ activities, which are known

as remanufacturing, recycling, reclamation and reverse logistics. The adoption of

green supply chain management (GSCM) takes all environmental issues into account

and considers these aspects when doing strategic decisions. Tseng et al. (2013) agree

that the incorporation of partners throughout the whole supply chain will establish a

strong customer-supplier relationship, while additionally cutting costs. A good

structured GSCM requires the integration of material and information flow within the

circle of suppliers, manufacturers and customers (Tseng et al., 2013).

The main drivers that encourage manufacturing firms to implement green

production are on the one hand side ethical reasons, i.e. in terms of internal values of

the company and on the other hand side business-related reasons, as green supply

chain management (GSCM) is said to increase the competitiveness of a firm. Findings

have proofed that companies dealing with GSCM do generally add value to their

businesses. As a matter of fact, a study has found that the leading ISO14001-certified

companies in South-East Asia (SEA), which do have a GSCM have optimized their

competitiveness and their economic output (Lee et al., 2014). If companies change

their environmental-friendly activities into an internal strategy, they will be able to

enhance their overall environmental performance while at the same time earn better

economic results (Deif, 2011). For Severo et al. (2014), green manufacturing practices

Page 18: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

8

such as the cleaner production (CP) programs must be incorporated to their

organization and continuously be managed by the leading forces. Otherwise, a green-

manufacturing program cannot secure long-term environmental process if the

implementation of such practices takes place only during the duration of the

project/program. Therefore, organizations must be persuaded and satisfied with the

improvements that come along with cleaner production projects.

Deif (2011) approves that green manufacturing will indeed increase competitiveness

as well as their productivity and efficiency. Stakeholders, public attention,

communities and the media, are all keeping companies under pressure to be more eco-

efficient. Additionally, the author states that the three motivational aspects to invest in

green manufacturing are efficiency, market share and governmental support and

regulations. A more efficient manufacturing process leads to saving energy and

materials. Hence, the benefits of green manufacturing outweigh the investment costs,

which have a positive effect on the revenue. As already mentioned, customer

preferences have changed due to the higher awareness of environmental changes.

Thus, green manufacturing helps to gain more market share. The reduction of material

waste itself for instance, will increase productivity and lower production costs. Lastly,

global governmental organizations are developing many regulations, policies,

obligations, taxes and penalties, which is yet another point that puts pressure on

companies and motivates them to be eco-efficient (Deif, 2011). Tseng et al. (2013)

too, do guarantee that green innovation, as a practice for strategic development, will

result in improved productivity and competitiveness. Similar to the Green Auto

Project by GIZ, the United Nations Development Program has established the Cleaner

Production (CP) program, which aims to prevent further harm as a basic tool and has

been implemented by several developing countries (Severo et al., 2014).

Voon-Hsien Lee et al. (2014) developed five practices that can be done in order

to be more sustainable. These are internal environmental management, eco-design,

investment recovery, green procurement as well as customer-cooperation. In order to

achieve a certain level of sustainability, the above-mentioned practices must be

incorporated to one another, implying the collaboration of cross-functions. One way

to facilitate the complexity of GSCM is green procurement. Green procurement

requires design specifications that automatically integrates environmental aspects for

Page 19: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

9

the procurements from suppliers and supports the monitoring-process of GSCM as

well as ensures the cooperation with ISO 14001-certified suppliers (Lee et al., 2014).

According to Ahmed Deif (2011) green manufacturing requires products and

processes that consume less energy and materials. Other than that, it also involves the

substitution of input materials, such as renewable for non-renewable and the reduction

of waste as well as the transformation from output into input, also known as recycling.

The author developed a system model for green manufacturing, which aims to support

companies in green manufacturing with respect to the following:

• Acquire previous activities to comprehend the existent level of green

manufacturing

• Indicate the proposed green transformation plan and the required tools and

metrics in order to achieve the transformation

• Methods that support to sustain the accomplished development and maintain

the eco-efficient system

The proposed model by Deif (2011) is divided into two sections; first, the design and

planning process of the green manufacturing system is demonstrated and second,

control tools that measure the design and planning process at each level are suggested,

which helps to implement and evaluate green manufacturing. Tseng et al. (2013)

outlines the disassembly of products as another important green manufacturing

practice, as it contributes the highest effect on sustainability.

2.3 Measuring Green Production

As new strategies and processes are applied when doing sustainable projects,

organizations would like to measure and evaluate their progress, in order to see the

improvements achieved. In principle, “measuring performance can be described as the

process of enumerating the efficiency and effectiveness of an action quantitatively

and/or qualitatively”. The main reason to measure progress, is so that organizations

can control and assess their development, to signalize their improvements, strengthen

their knowledge on key processes, recognizing possible complexities as well as being

able to prepare cognition for future actions (Ahi and Searcy, 2014). Herani et al.

(2005) demonstrate that the level of performance measures is derived from what the

Page 20: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

10

organization wants to achieve, i.e. their goal. For instance, the available financial

measures must be regarded by organizations, such as ROI, market share, profitability

as well as the increased revenue in terms of a competitive and strategic level. Olugu et

al. (2011) state that the green process cannot be reached within a day and highlight the

necessity of measurement by demonstrating that it discloses the scope a company has

reached through investing in its GSCM at any given point of time.

Ahi and Searcy (2014) have analyzed 445 published articles after 2012 and have

identified 2555 unique metrics to measure green and sustainable supply chains.

However, most of the identified metrics have only been used once, which shows that

there is no consistent common metric that can be used to measure green supply

chains. Nonetheless, five identified metrics have been used for more than 20 times,

which indicates that they are appropriate and might also serve as potential metrics in

the long run. The most common identified metrics with respect to the research of Ahi

and Searcy (2014) are quality (31 times), air emission (28), greenhouse gas emissions

(24), energy use (24) and energy consumption (21). Subsequently, the environmental

characteristic is the most used metric within the analyzed publications in this study.

Table 2.3.1 indicates the importance of the economic and environmental focus in

sustainable supply chain management, as the study revealed that they were the most

frequently used metrics. The numbers in parenthesis in the table imply the frequency

rates of the used examples throughout the reviewed study’s. Furthermore, Table 2.3.2

reveals an overview of the topics used by at least 10 metrics. A total of 113 metrics

concentrated on energy, wherein most of the topics focused on energy efficiency and

energy consumption.

Table 2.3.1: Metrics and examples of Sustainable Supply Chain Management (SSCM)

characteristics (Ahi and Searcy, 2014)

SSCM Characteristics

Economic focus Environmental focus No. of Metrics

Examples of Metrics No. of Metrics

Examples of Metrics

Environmental focus

168 Environmental costs (11), Buying environmental-friendly materials (7)

Social focus 70 Customers’ satisfaction (14), Customer returns

78 Environmental Social Concerns (4), Cooperation

Page 21: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

11

Table 2.3.2: Major themes used by the identified metrics (Ahi and Searcy, 2014)

Major Themes No. of Metrics Example of metrics (frequency rates) Product(s) 261 Product characteristics (11), Product design for

remanufacturing (6) Cost(s) 176 Cost (12), Environmental cost (11) Waste 148 Solid waste (19), Reduction of solid waste (11) Recycle/Reuse 140 Recycling (19), Recycling revenues (7) Material(s) 131 Decrease of consumption for Hazardous/Harmful/Toxic

materials (8), Buying env. -friendly materials (7) Energy 113 Energy use (24), Energy consumption (21) Emission(s) 91 Air emissions (28), Greenhouse gas emissions (24)

The authors came to the conclusion that there is an urgent need for identifying

metrics that demonstrate a wider area of sustainability within a supply chain. Olugu et

(6) with customers for green packaging (2)

Flow focus 2 Cash flow (1), Cash flow provided by operating activities (1)

1 Annual mass-flow of different materials used (excluding energy carriers and water) (in tons) (1)

Coordination focus

1 Cooperation degree (1) 3 Collaborating with other companies and organizations for environmental initiatives (1), Improving opportunities for reducing waste through cooperation with other actors (1)

Relationship focus

1 Networks (2) 1 Interaction and harmony co-exist with natural systems on production and consumption systems (1)

Value focus 59 Profit (12), Market share (11), Sales (4)

3 Energy requirement per unit of net value added (1), Global warming contribution per unit of net value added (1)

Efficiency focus 5 Existing efficiency vs. cost of upgrading (2), Increased cost efficiency (1)

14 Energy efficiency (11), Recycling efficiency (3)

Performance focus

43 Cost savings (8), Operational performance (4)

39 Process optimization for waste reduction (8), Optimization of process to reduce air emissions (4)

Page 22: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

12

al. (2011) on the other hand have developed measures of the green supply chain in the

automotive sector. The focus lies on the automotive sector as it has a very complex

supply chain - as an automobile is made of many components from different locations

- and the rising importance of measuring such complex supply chains in terms of

greening. It should be noted that the authors divided the supply chain into forward and

backward chain. Forward chain refers to ensuring green manufacturing and the

termination of the product at the customer, whereas backward chain assures the

recycling process when the product returns to the chain. Aspects observed to measure

sustainability included environmental certification of supplier, supplier’s performance

on sustainability, greening costs, customer perspective and many more. In order to

examine the key performance indicators, a survey had been carried out to the industry

in which customers’ perspectives have resulted as the most important measure with

93.4% votes from organizations and academics. Hervani et al. (2005) suggest the

usage of ISO 14031, as it contains various environmental performance indicators. An

OECD green-framework has been used by Kim et al. (2014), in order to select 12

indicators that have been developed for measuring cross-country comparison of green

growth strategies. Selected indicators were compared to the value of a given country

and evaluated on a scale from 1-10. Respectively, five categories have been observed

which were:

(a) Environmental efficiency of production and changes in production patterns,

(b) Environmental efficiency of consumption and changes in consumption patterns,

(c) Natural capital stocks and environmental quality,

(d) Objective as well as subjective environmental quality of life and

(e) Economic actor responses.

As mentioned before, this study will use Data Envelopment Analysis (DEA) as a

performance and efficiency measurement tool, which has been applied for several

studies regarding sustainability and greening before. Azadi et al. (2014) used DEA as

a target-setting model for the application of GSCM of public transportation providers.

Developing a feasible and realistic plan with the target-setting model was the main

objective of the authors. Furthermore, Dobos and Vörösmarty (2014) also applied

DEA with the Common Weights Analysis (CWA) to identify the environmental

Page 23: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

13

factors as the influential decision factors when selecting green suppliers. However,

the applied method did not serve as an optimal solution but rather supported decision-

makers in evaluating the selection criteria’s. It ranks the suppliers according to the

most efficient DMUs (Dobos & Vörösmarty, 2014). Sueyoshi and Wang (2014) used

DEA to achieve a higher level of corporate sustainability, so that organizations can

assess their progress in terms of environmental performances. More importantly, the

proposed DEA tool provides information concerning the degree of investment for

technology innovation in order to reduce unwanted output, such as CO2. Another

research examines environmental assessment for corporate sustainability by resource

utilization and technology innovation using DEA in the Japanese industrial sector

(Sueyoshi & Goto, 2014). Houshyar et al. (2012) applied DEA for measuring the

sustainable and efficient energy of the corn production in Iran combining it with

multi-fuzzy modeling. DEA identified efficient and inefficient farmers, so that the

inefficient ones could apply some techniques from the more efficient farmers, in order

to improve their own efficiency.

Also, because a single outlier can influence all the efficiency score results of a

peer group, Lu et al. (2014) have worked on establishing a peer-wise energy

benchmarking model with DEA. The objective was to develop a model that is able of

identifying significant factors, discovering the influential outliers and to segment the

efficiency scores into either self-efficiency change or peer-efficiency change. This

aims to determine the causes of efficiency changes, i.e. either due to changes of

energy efficiency within itself or the peer-group. Thus, with the removal of outliers

the efficiency scores of the models cannot be influenced. This model has been applied

to multifamily properties for the purpose of benchmarking energy efficiency.

Furthermore, a large number of researches are using multi-stage DEA models.

Multi-stage DEA models are models wherein the DEA method is applied on every

stage and the results of one stage are the basis for the next stage’s analysis (Zhu,

2009). This can be applied for multiple process problems. This approach is favorable,

because generally it is a more precise analysis of DMUs analyzing the efficiency step

by step and it has been proofed that it obtains more trustworthy results (Wu et al.,

n.d.). An example of this is a research on economic and environmental efficiency of

Danish pig farms by Asmild and Hougaard (2006). In their paper they analyse the

Page 24: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

14

economic and environmental efficiency by using DEA with a 2-stage analysis. The

environmental efficiency concerns the reduction of undesirable outputs, i.e. nutrients,

while the economic efficiency concerns the revenue. In this case, the two efficiency

scores are computed seperately and compiled on a second step.

2.4 Measuring Performance of Small and Medium Enterprises in Thailand

The Office of SME Promotion (OSMEP) founded in 2001 aims to establish a

national SME policy strategy in coorporation with the Thai government. For this

matter, OSMEP introduces SME master plans every 5 years (OECD, 2011). Up to

now, two SME master plans have been completed. The first plan lasted from 2002-

2006 and targeted the following achievements: to increase the efficiency and capacity

of SMEs by building up an environment in which SMEs can grow, strengthen their

competitiveness and contribute into the regional economy. The second plan from

2007-2011 adressed the capability of growth as well as sustainability with respect to

skills and knowledge (Charoenrat & Harvie, 2014). The third master plan is ongoing

(2012-2016) and intends to make the SMEs the most influencial factor to the Thai

economy. This shall be achieved by increasing their capacity development and

improving the manager and sectoral standards in terms of innovation, know-how,

creativity and cultural uniqueness (Goverment Public Relations Department, 2014).

According to the OECD, the barriers apprehended by SMEs are political

instability, corruption, lack of financial assets, lack of railway transport system etc.

For this matter, it is important that barriers that hinder the development of SME

efficiency and effectiveness must be investigated, so that they can be erased. There

are several methodologies which can be applied to evaluate policies, programmes or

instruments including case studies, surveys, cost-benefit analysis as well as a number

of quantitative and qualitative approaches (OECD, 2011). Thus, the following section

aims to indicate research-based reports that have been applied to measure

performances of SMEs in Thailand.

Similar to this work, Charoenrat and Harvie (2014) have analysed the technical

efficiency of SMEs for the Thai manufacturing sector by applying the Stochastic

Page 25: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

15

Frontier Analysis (SFA). The study solely focused on SMEs due to its growing

relevance in the country as also pointed out in chapter 1. The data obtained for the

purpose of the analysis are from the years 1997 and 2007 as the objective was to

benchmark the level of technical efficiency performance in the pre and post Asian

Fincancial Crisis period. Furthermore, the authors examined firm-specific factors that

cause technical inefficiency as well as gave policy recommendations with the

foundation of the obtained results. As for the measurement tool, SFA has been

prefered over DEA because it considers random shocks and considers that these,

especially if not regulated by the firm , can possibly have an impact on the amount of

output. The study revealed that researches for Thailand that applied SFA or DEA

mainly focus on firm size, location and governmental support. The key findings were

that many SMEs were rather inefficient during the post crisis, except for medium-

sized enterprises that showed an improvement in efficiency. Furthermore, the SMEs

in the manufacturing sector have been found to be very labor-intensive with mainly

unskilled workers. This can be explained with the lack of capital investment.

Consequently, the authors suggest that governmental policy recommendations should

adress the improvement of technical efficiency and competitiveness of SMEs. The

policies should focus on capacity and capability through enhanced labor forces and

managerial quality as well as investing in technology (Charoenrat & Harvie, 2014).

Because several Thai SMEs went bankrupt after the Thai economic crisis in 1997

the authors (n.d.) Srivihok and Intrapairote measured the status of intellectual capital

(IC) of Thai SMEs enterpreneurs in order to strengthen the managerial understanding

of IC within their organization. The findings aim to be a supportive guide for firms

that are willing to enhance their managerial skills as well as increase their IC assets.

For this matter, the authors have adapted an intellectual capital measurement model

from a Scandinavian Intellectual Model. The proposed model is made up of three

clusters, namely human capital, structural capital and relational capital. To obtain

results the authors analyzed E-commerce websites of Thai SMEs in June 2013. The

surveys were collected for a total number of 55 websites and the sectors considered

were manufacturing, food-service, software house and agriculture. The results

revealed that most of the SMEs do not invest in the development of their human

resources and are not aware of the IT relevance within their organization.

Page 26: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

16

Furthermore, the SME websites have the intention of providing information than to

enable transaction function, which also leads to the lack of customer relationships

through their websites. At last, findings implied that SMEs rather have horizontal

relationships with partners than vertical relationships with suppliers. Subsequently,

the authors suggest the organizations should increase awareness of the importance of

intellectual capital as well as the integration of IT-based management to maximize

efficiency and competitiveness (Srivihok & Intrapairote, n.d.).

To point out the importance of applying both financial and non-financial

attributes when evaluation a companies’ performance, Sawan et al., (2007) have

published a report, in which they investigated the impact of a balanced scorecard

approach by measuring innovation effectiveness of 144 Australian and Thai SMEs.

Innovation is a major factor that improves performance of an organization. When

assessing the benefits an innovation has brought to an organization it might not be

sufficient to only focus on the financial benefits, as it does not reflect the overall

performance of a firm. Hence, a balance between financial and non-financial

measures should be applied. For this purpose, the authors quantitatively analysed the

degree of performance measure utilization of SMEs with respect to their innovation

effectiveness. The comparison of Thai and Australian SMEs is due to the

demonstration of metrics that are used across different cultures. The data was

obtained by using a questionnaire. Findings state that 79.5% Australian and 75,5%

Thai SMEs apply both financial and non-financial metrics to assess the innovation

effectiveness. Additionally, 17.7% of Australian and 12.2% of Thai SMEs reported

that they solely use financial metrics. Only a number of 2.9% Australian and 12.2%

Thai SMEs apply non-financial metrics. The most often metrics used by Australian

SMEs were; customer satisfaction (non-financial), sales and ROI (financial), while the

most often metrics used by Thai SMEs were the product/ service quality (non-

financial) as well as sales and profit margin (financial). No differences due to cultural

reasons could be found in the applied metrics by Australian and Thai SMEs. The

authors strongly recommend that both financial and non-financial metrics shall be

used for performance evaluation in terms of innovation as more positive percceptions

across several attributes could be found (Sawang et al., 2007).

Page 27: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

17

Research-based papers on the performance of Thai SMEs by applying

methodological approach are still minor but are gaining importance within the years.

SMEs are aiming to become Thailands most influential contributor into the economy,

especially now that the Asean Economic Community has been established (OECD,

2011). The next chapter aims to give a detailed discussion on the tool that will be used

for this work, namely DEA.

Page 28: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

18

Chapter 3

Data Envelopment Analysis (DEA)

3.1 Theoretical Framework

DEA is a data-oriented analysis tool that assesses the performance of a

particular number of peer entities, also known as Decision Making Units (DMUs),

which transfer multiple inputs into multiple outputs. The DMUs used for DEA are

diverse and flexible. Examples for DMUs are hospitals, banks, and organizations or

simply processes, regions etc. One crucial function of DEA is that it does not require

much internal information and assumptions; it simply considers internal information

as a “black box”. Solely the multiple inputs and multiple outputs of DMUs are of

relevance for the analysis tool (Zhu, 2009)

With respect to the DEA-results “full (100%) efficiency is attained by any DMU

if and only if none of its inputs or outputs can be improved without worsening some

of its other inputs or outputs”. Moreover, as the focus is on benchmarking DMUs to

one another, one has to consider the relative efficiency also. It states that “a DMU is

to be rated fully (100%) efficient on the basis of available evidence if and only if the

performances of other DMUs does not show that some of its inputs and outputs can be

improved without worsening some of its other inputs and outputs” (Cooper et al.,

2004). In other words, the efficiency of a DMU is assessed relatively to all the other

DMUs to be examined, with the only limitation that all DMUs are located on or under

an efficient frontier (Atici & Podinovski, n.d.).

Farrell first developed DEA in 1957 and has established the efficient frontier.

Subsequently, Charnes, Cooper and Rhodes enhanced this method by developing a

nonparametric analysis in using efficiency assessment in 1978. DEA functions by

using a linear programming methodology that converts inputs into outputs produced.

DEA identifies a frontier (best practice) and compares the relative performance of

organizations or units to the best practice. The range of efficiency is between 0 and 1

for the input-oriented model, in which 1 is the most efficient unit/organization and 0

the least. Furthermore, the DEA method helps to identify the reason of inefficiency

(Liu et al., n.d.). The output-oriented model on the other hand computes the target

Page 29: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

19

value as a number, to which the output variables must be increased. The inefficient

DMUs have an efficiency score of >1, for example 1.2 indicating that the output

values must be increased to 120%. The efficient DMUs have equivalent to the input-

oriented model, an efficiency score of 1 (Bielecki, 2012).

The two major types of DEA; can be as indicated above either input- or output-

oriented. The input-oriented model refers to the minimization of the objective

function, i.e. the objective is to minimize the amount of input, while maintaining the

same amount of output. Similarly, the output-oriented model considers the

maximization of the problem. The objective is to maximize the amount of output with

the same level of input (Atici & Podinovski, n.d.). However, it should be mentioned

that output could be classified into desirable (e.g. products) and undesirable (e.g.

waste) outputs. It is often the case that one wants to maintain the amount of input,

while reducing the amount of output (undesirable), in order to improve its

performance. The norm usually assumes that input needs to be decreased and output

should be increased, however, in this case we may treat the undesirable output as an

input, as it must be decreased. However, this procedure would not represent a proper

production process. There are also cases where the amount of input must increase, in

order to improve the performance. An example for this case is when enhancing the

performance of waste treatment process; the amount of waste (undesirable input) to be

considered must be increased (Zhu, 2009). Generally, we must acknowledge the

existence of interdependencies in between desirable and undesirable outputs. This

shall be demonstrated by considering a simple production process. Given the

assumptions that producing one product simultaneously produces one waste product,

it is obvious that the more products produced the more waste is generated (Pasupathy,

2002).

The methodology of how to solve these out of norm problems with undesirable

measures will be discussed later on.

Besides the difference in input- and output-orientation, DEA can also be

differentiated according to its condition, which can be a constant or variable-return-to-

scale, CRS and VRS. The CRS model implies that the output level is proportional to

the input level, while the VRS model supposes that the output level is higher or less

Page 30: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

20

than the increased input (Liu et al., n.d.). For example let us assume that there are 2

inputs and 2 outputs in a model. According to the CRS-Model, if both inputs are

increased by 10% then both outputs will increase by 10% accordingly. Under the

assumptions of a VRS-Model no proportionality is assumed (Atici & Podinovski,

n.d.). Lets take the CSR Model for instance and assume that Xij is the number of input

i that produces Yrj for output r. The objective is to maximize the efficiency score (Ej)

for all DMUs (j= 1,...,n). Thus,

𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒    𝐸! =!!"!!"!

!!!

!!"!!"!!!!

(3.1.1)

𝐸! =!!"!!"!

!!!

!!"!!"!!!!

≤ 1 (3.1.2)

where Vrj, Uij > 0; for all r, i and j (3.1.3)

The problem formulation of a CSR Model is the ratio of weighted outputs to

weighted inputs. Vrj and Uij are the weights for the multiple outputs and multiple

inputs. The maximum efficiency reached is 1, thus, the DMUs should score less than

or equal to 1 (Ozcan, 2008).

To aggregate the multiple input and multiple output factors, DEA combines

the individual factors through the weights to a single virtual input and a single virtual

output. Hereby, the weights of the aggregation are unknown and are a part of the

optimization process (Bielecki, 2012).

Also, there are three efficiency types that can be evaluated, which are known

as the technical, allocative and economic efficiencies. The technical efficiency can be

defined as “the degree to which a decision making unit produces the maximum

feasible output from a given bundle of inputs, or uses the minimum feasible amount

of inputs to produce a given level of output”. More precisely, it reflects the input- or

output-oriented models. Furthermore, the allocative efficiency examines “the ability

to produce a given level of output using cost-minimizing input ratios”. If we add up

both the technical and allocative efficiency, it will make up the economic efficiency

(Atici & Podinovski, n.d.).

Page 31: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

21

Another important criterion, which is crucial for the envelopment analysis that

needs to be considered, is the determination of the number of input and output data.

This is important because it can influence the discriminatory power of efficiency

scores. Assuming the number of DMUs is n and the amount is less than adding the

number of inputs and outputs (m + s) together, then the results will lead to a large

number of efficient DMUs. Hence, the efficiency discrimination of DMUs is rather

doubtful because of the inappropriate number of degrees of freedom. In principle, the

data set should be as few as possible. For this purpose, a rule of thumb has been

established that estimates the appropriate number of input and output:

n  (Number  of  DMUs)  ≥ max  𝑚×𝑠,  3  ×  (𝑚 + 𝑠)  

Thus, following this rule will help to avoid the problem of loss of discriminatory

power (Cooper et al., 2006).

3.2 Benchmarking with Data Envelopment Analysis

Benchmarking can be defined as “a measurement of the quality of an

organization’s policies, products, programs, strategies, etc., and their comparison with

standard measurements, or similar measurements of its peers” (Business Dictionary,

n.d.).

Especially managers are constantly seeking to enhance the performance of the

organizations they lead. In order to enhance their performance, organizations must

undergo evaluations, so that they know where they stand in comparison to their

competitors. As a result, benchmarking tools are widely used, as they identify best

practice organizations.

There are many benchmarking methods used to evaluate performances of

DMUs. Some practitioners use performance measurement tools while others use

performance indicators. Performance indicators are often used because they evaluate a

particular aspect of performance (Zhu, 2009). According to Stroobants and Bouckaert

(2014) indicators are more informative to managers as they are expressed in absolute

numbers and thus, are useful to evaluate a DMUs performance over time and

Page 32: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

22

benchmark them to their competitors. However, the disadvantages of performance

indicators often lead to misconceptions. On the one hand, they only have a limited

indication of performance. An example for this is that indicators only deal with

single-input and single-output data when considering efficiency performance. On the

other hand, they might as well reveal controversial results, as one DMU might be

efficient on one indicator, but inefficient on another (Stroobants & Bouckaert, 2014).

Generally, benchmarking methods can be classified as to parametric and non-

parametric methods. This shall be further discussed in the following chapter.

3.2.1 Parametric and Non-Parametric Methods

Parametric methods deal with the “description and examination of

relationships between different parameters, such as costs and schedules”

(BusinessDictionary, n.d.).

In principle, the parametric methods are classified into stochastic and deterministic

models. The deterministic approach encloses all the observations and computes the

distance amongst the maximum and observed production. Finally, it identifies the

technical inefficiency, whereas the stochastic method is capable of differentiating

between technical efficiency and statistical noise (Murillo-Zamorano & Vega-

Cervera, 2001). Three parametric methods have been widely used for frontier

efficiency approaches; these are the Stochastic Frontier Approach (SFA), the Thick

Frontier Approach (TFA) and the Distribution-Free Approach (DFA) (Paradi & Zhu,

2013).

According to Stroobants and Bouckaert (2014) “Non-parametric methods use

linear programming to construct a piecewise frontier that envelops all observations

(DMUs) of the sample used, against which each DMUs efficiency can be evaluated”.

The main difference between these methods and indicators is that unlike performance

indicators, it provides relative numbers as the DMUs are assessed relatively to one

another and thus, can be referred to benchmarking (Stroobants & Bouckaert, 2014). In

general, there are two mainly used non-parametric approaches for efficiency

evaluation known as Free Disposal Hull (FDH) and DEA.

Page 33: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

23

FDH considers the minimum number of restrictions on data, because it

supposes free disposability of resources and strong disposability of outputs. The first

refers to the capability of an entity to consume more inputs than technically required

for a particular amount of output, while the latter refers to the ability of an entity to

produce less output than technically feasible with the same amount of input. Similar

to DEA, FDH computes efficiency scores and identifies the best practice amongst a

peer group. Furthermore, it assesses the input- and output-efficiencies, i.e. identifies

whether the entities are input- or output-oriented. The difference between DEA and

FDH is that FDH only compares existing observations within a set, while DEA also

compares virtual DMUs. The efficient-frontier of FDH is stairway-shaped (Stroobants

& Bouckaert, 2014).

The main differences amongst the parametric and non-parametric frontier

efficiency approaches are the assumptions made considering the efficient frontier, the

presence of random errors as well as the allocation of inefficiencies and random

errors. Usually, econometric approaches include a random error and an error that

shows inefficiency. However, non-parametric methods do not include random errors

and the necessity of a few assumptions only occurs when determining the efficient

frontier (Paradi & Zhu, 2013).

According to Paradi and Zhu (2013) some studies have focused on comparing

the parametric and non-parametric methods in terms of accuracy of estimations.

Ferrier (1990) for instance has compared SFA and DEA with respect to measuring the

cost efficiency of U.S. banks. The outcome revealed that both approaches had similar

results, however, differ when considering the decomposition of cost inefficiencies

amongst the technical and allocative inefficiency. Moreover, there was a poor

correlation of rank-order between SFA and DEA. Furthermore, Bauer et al. (1989)

proposed a study where they examined DEA, SFA, TFA and DFA in order to

determine which frontier efficiency measure would be most precise for regulatory

analysis. The purpose was to assess the consistency in terms of efficiency levels,

rankings and identification of best practice. As a result, the outcome revealed that no

consistency was obtained amongst parametric and non-parametric approaches (Bauer

et al., 1998). Reinhard et al. (2000) also assessed environmental efficiency by using

both SFA and DEA approaches. Their conclusion was that different results were

Page 34: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

24

obtained, in this case mean technical efficiency scores of 89% SFA and 78% DEA

(output-oriented). The different results are caused because of the fact that with DEA

regularity is assumed, i.e. the environmental efficiency scores can be computed for all

specifications. SFA includes hypothesis testing and the rejection of particular

specifications such as bad inputs that do not satisfy the theoretical restrictions and

thus, cannot be estimated (Reinhard et al., 2000).

Consequently, DEA is a suitable benchmarking tool, because it involves

multiple performance measures. It involves mathematical programming techniques as

well as models, to assess the performance of peer units as mentioned above. In

particular, DEA monitors the existing resources to each unit and examines the

transformation of the inputs into outputs.

If we for instance consider a buyer-seller supply chain, then it would be of interest for

the buyer to benchmark the performance of the seller with regards to response time,

costs, flexibility, customer service, quality and customization. The poor operations

will be revealed and can thus be improved in efficiency. This will lead to decreased

costs of inputs as well as increased productivity. Benchmarking encourages

organizations or DMUs in general to get better, so that they can subsist in the global

business environment and stay competitive (Zhu, 2009).

3.3 Advantages and Disadvantages of Data Envelopment Analysis One of the main advantages of DEA is that it does not need any assumptions

in order to form the production function. Also, unlike other tools it has the ability of

determining best practices, which would be too difficult to identify via observations

(Bayraktar et al., 2012). Apart from that, one must emphasize on its ability of

handling multiple inputs and outputs that are expressed in different units as well as the

determination of the efficienct frontier, that identifies DMUs as best-practice and

leads to a comparison of DMUs to one another (Hababou, n.d.). In order to be able to

make efficiency ratios sensitive to the mixture of input and output, it is necessary to

weight the inputs and outputs according to their relative value. If the weights are not

considered, the ratio analysis is rather marginally helpful and may lead to a

misconception when analysing multiple-input and –output models. One factor of DEA

Page 35: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

25

is that it has the ability of determining relative weights that is crucial for identifying

the value of inputs and outputs. This especially enables managers to comprehend the

insights by using operating ratios, which is valuable information when improving

performance. DEA has the power of assessing relative performance even though such

weights are often unknown. DEA computes every possbile combination of weights in

order to make the DMU most efficient. This feature, i.e. the capability of assimilating

multiple-inputs and –outputs in their original value without the availability of relative

units makes DEA an appropriate tool for assessing organizations (Sherman & Zhu,

2006).

DEA also brings its limitations along. Because DEA generates individual

linear programs for every DMU, it may become difficult to compute large problems.

Furtermore, other disadvantages are the difficulties that occur when aggregating

several aspects of efficiency. This is often the case when DMUs perform several

activities. Another problem is the insensitivity towards intangible and categorical

components, such as the service quality in specific branches. This is also related to the

problem when analysing DMUs that operate on different dimensions. If for example

the analysis is based on a DMU performing in two functions, then it may result in

being efficient on one while being inefficient on the other. To give an example while

focusing on the bank branch sector, a branch might be efficient in sales but inefficient

in service. For this matter, it might be advisable to run two DEA models, however,

one must deal with comprising both models afterwards which might be difficult

(Hababou, n.d.). Another drawback of DEA that needs to be considered is that every

DMU that deviates from the efficient frontier is identified as inefficient. The existence

of random values, for example through measurement errors is not considered. This is

a problem for measurement errors that occur within efficient DMUs, as this has an

impact on the efficiency score of the other DMUs and possibly distort these. Hence, in

order to obtain robust results, it is necessary that the applied data is of high quality.

Furthermore, DEA identifies many DMUs as efficient when small samples or many

data sets are used. The higher the number of inputs and outputs, the higher the number

of efficient DMUs (Bielecki, 2012).

Page 36: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

26

At last, the lack of expressing the efficiency in absolute terms should be

mentioned. Especially for a managerial point of view, knowing the performance

compared to a theoretical maximum would be more useful, however, a frontier

expressed in absolute terms is difficult to obtain. In conclusion, it can be said that a

particular approach for testing the correctness of a given set of factors for the analysis

of an efficiency evaluation does not exist. DEA only reveals the efficiency scores of

every DMU of a given set, however, it does not give information whether the chosen

factors are the appropriate ones or not (Hababou, n.d.).

Page 37: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

27

Chapter 4

The German International Corporation

This chapter gives a brief introduction on the organization GIZ by giving a

company profile and will then give detailed explanation on the Green Auto Project.

The explanation of the project is of relevance and aims to understand the common

theme of this work.

4.1 Company Profile

GIZ is a German non-profit organization. Worldwide, GIZ employs 16,510

employees, operates in 130 countries and generated a business volume of 1.9 billion

Euros in the fiscal year 2013. As an institution, it engages on topics concerning

politics, economics, and social changes. Its target is to gain sustainable and effective

improvements worldwide. The majority of GIZ’s commissions are delegated by the

German Federal Ministry of Economic Corporation and Development (BMZ).

However, there are other organizations, which delegate projects or commissions to

GIZ, such as other ministry’s as well as private and public bodies in Germany and

foreign countries. The European Union institutions, the United Nations and the World

Bank are part of the commissioners (GIZ, 2013).

The German international corporation in Thailand began in 1959 when the

King Mongkut Institute of Technology has been established in North-Bangkok. First,

the main focus of the institution was to improve the agricultural as well as education

sector. As Thailand has managed to become an emerging and industrial country,

GIZ’s focus nowadays shifted to industrialization, renewable energy, climate security

and state modernization. Currently, GIZ Bangkok is implementing various regional

and international projects, in which approximately 100 international development

organizations are participating. The main focus of GIZ in Thailand is sustainable

consumption and production, environmental and climate security, agriculture and

nutrition security, energy efficiency and renewable energy, municipal sustainability

and transportations as well as regional integration (GIZ, 2015).

Page 38: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

28

4.2 Introduction of the Green Auto Project2

Project Scope and Structure

The goal of this project is to enhance sustainability in production within the Thai

automotive industry. For this project, only small and medium enterprises have been

targeted. The specific objectives and results are:

1. Enhancing productivity and environmental performance of Thai auto and

automotive parts production

- 250 SMEs have shown improvement in productivity and better product

quality, while at the same time, environmental harm per production unit

has been reduced

- More SMEs are now complying with international environmental

regulations

2. Improving networks, business and financial services for the greening of the

Thai auto and automotive parts industry

- Establishment of new financial packages for SMEs to invest in

improvement measures

- Improvement of alliances and networks for greening the supply chain of

Thai automotive industry

3. Disseminating good practices and promoting development and implementation

of related policy and economic instruments

- Promotion of good practices and related policies

In order to achieve the above-mentioned results, GIZ has established four

working packages, which was introduced to the participating SMEs and considered

project management and steering.

Working package 1 which is to “Improve productivity and environmental

performance of Thai Automotive Industry Clusters” is said to be the most extensive

and core element of the project. It addresses the objectives and results of number 1

indicated above. This working package includes the assistance in the enhancement of

2  Information from GIZ project proposal  

Page 39: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

29

technical machinery and optimization of operation procedures. This will ensure

productivity and competitiveness in the Thai automobile industry. Furthermore, the

improved technical equipment’s will lead to less energy and resource consumption as

well as reduced environmental harm. With regards to the new technical standards,

GIZ has planned to launch Best Available Techniques (BAT), which can be applied to

all targeted sectors and helps to implement the project by providing adequate

instruments and recommendations.

The objective and results of number 2 are addressed by working package 2

“Harmonize and implement SME specific financial support packages” and working

package 3 “Strengthen sustainable consumption and production (SCP) related

services”. One of the main drivers of this objective is investment. With respect to the

working packages, investments can be classified into financial support for technical

equipment and financial support for consultation and management services, audits and

evaluations. As it is difficult for SMEs to access Thai support programs such as the

Department of Alternative Energy Development and Efficiency (DEDE) or the

Department of Industrial Promotion (DIP), these packages aim to develop a concept

to increase SMEs’ access to financial support programs as well as direct support.

Finally, working package 4 “Dissemination of good-practice and policy

recommendations” addresses the objective and results of number 3. This package

aims to reach the final target groups and the dissemination of the improvements can

be used as input for the SCP Policy Support Component action in Thailand.

Moreover, working package 4 will correspond with the SWITCH-Asia3 philosophy

and will pursue to share the improvements with other countries through

communicating with the SWITCH-Asia Network Facility.

The following chart aims to illustrate the project structure described above:

3  With respect to the aspect of sustainable consumption and production (SCP) the European Commission has introduced a SWITCH-Asia program, which aims to support consumers, businesses and associations to switch to a better sustainable paradigm.  

Page 40: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

30

Furthermore, figure 4.2.2 demonstrates the significance of operations and processes as

well as the competitiveness of the targeted sector.

Figure 4.2.2: GIZ's project target group (GIZ Project Proposal, 2011)

Technical Standards &

Guidance (BAT)

Management Improvement

(PREMA)

Quality Improvement

(IQA)

Financial Support

Packages

Training and Consulting Program

Technical & Management

Support Services Financial Access Services

500 SME (Tier 2&3) group

+ individual services

Promoting Green Value, SC

& Green Procurement

Promoting SCP Service

Networks

Policy Recommendations Policy Support by Thai

Government

Policy Support by Thai

Government

Figure 4.2.1: Project Structure (GIZ Project Proposal, 2011)

Page 41: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

31

The following remarks aim to strengthen the understanding of the core

processes and give further explanation of the above-illustrated figure.

Tier 1 implies that the automakers are being supplied from Tier 2 and 3, which mainly

consists of multinational automotive manufacturers and have plants based in

Thailand. The next level of the automotive production is Tier 2 and 3, which are here

classified into groups according to their operation. These groups are stamping,

plastics, rubber, machining, casting, forging, functions, electric and trimming. The

target was to provide technical and environmental management guidelines to the 500

SMEs. Moreover, clustered networks were provided, wherein the chosen SMEs in the

network determined particular individual technical improvement measures. It should

also be noted that there were trained service providers that helped the networks to

reach the targeted improvement goals, by assisting the implementation of defined

measures. Another important point is that the financial support programmes, which

were established for SCP implementation, were harmonised and adjusted to present

programmes of diverse governmental agencies, for e.g. Energy Service Company

(ESCO). This too, was supported by trained service providers so that improvement

measures could be performed. Also, it should be mentioned that all data from the

networks and particular SME measures for further implementation would be collected

and transferred into a good practice database. This will allow data evaluation and will

reveal further development of achievable and applicable standards for industry

clusters. The data evaluation can then be shown to SME sectors, stakeholders and

administrators. At last, the determined good practices will be given as policy

recommendations for e.g. to the Thai Automotive Industry Master Plan.

4.3 Approach for Working Packages of Project

The project fulfills its objective by implementing 4 working packages that

include actions that support the advancement of greening the supply chains of Thai

SMEs in the automotive sector. This chapter aims to illustrate the actions taken within

the project and gives insights on each working package.

Page 42: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

32

Working Package 1 – Improve productivity and environmental performance of

Thai Automotive Industry Clusters

In order to be able to improve productivity and environmental performance in

the Thai automotive industry, three tools based on existing Sustainable Production

(SP)/ Cleaner Production-Technology (CT) have been established by the Thai

Automotive Institute (TAI). These are (1) Profitable Environmental Management

(PREMA), (2) Best Available Techniques Reference Documents (BREFs or BAT), as

well as initially (3) the Intelligent Quality Assurance (IQA), which has later been

replaced by the Loss Reduction Process (LRP) due to partner requests.

Profitable Environmental Management (PREMA) Approach

This tool has been established particularly for SMEs by GIZ, created to

identify, generate and implement measures, which aim to reduce production costs,

enhance environmental performance and optimize organizational competencies. Not

only has PREMA been used before in other countries, it has also been carried out in

other sectors. Previous experience and good practice recommendations will be given

to the training programs of the Green Auto Project. In particular, PREMA connects

technical and non-technical issues during workshops and advisory programs for

SMEs. It helps managers and operators in comprehending the importance of the

efficient consumption of raw materials and energy, as well with the connection

between product design, work planning and organizations and not to forget costs.

Hence, PREMA results in an optimized production process, internal organization and

product design, while continuously involving environmental aspects throughout the

production cycle. Furthermore, SMEs are obliged to reduce their undesirable output,

which are outputs that don’t end up being a part of the end product such as material

(waste), energy and water.

PREMA is applied by giving training systems to the SMEs and involves Resource

Management, Environment-oriented Cost Management, Chemical Management and

Good Housekeeping. The PREMA-trainers offer feasible tools and methods in order

to develop innovative solutions and enhance the potential of human resources. It is

based on international certifications such as quality standards (ISO 9001) and

environmental performance standards (ISO 14001).

Page 43: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

33

Best Available Technique (BAT) Approach

Initially developed by the European IPPC Bureau in Seville, Spain, this tool

involves detailed information on the best technical solutions. It gives recommendation

on the application of the identified solutions while considering its geographical

location, the technical characteristics of the installation as well as local

environmental/ economical conditions, costs and at last but not least, the general

availability of the technique.

In fact, horizontal BATs that can be applied to all industries have been developed.

These consist of BAT on Energy Efficiency, the BAT on Economics and Cross Media

Effects and the BAT on General Principles of Monitoring. As the Thai automotive

sector is very diverse, the Green Auto Project will use all horizontal BATs with the

main focus on energy efficiency. The focus on energy is mainly because of the high

costs and environmental harm it causes and a more efficient energy consumption will

lead to a higher profitability as well as improved competitiveness. Consequently,

SMEs learn about the BAT on Energy Efficiency in their sub-groups in workshops.

Managers might be more willing to implement this, as it has a positive environmental

and economical outcome. Energy efficiency is the best and most cost-effective

opportunity to reduce greenhouse gas emissions and comes along with the advantage

of saving costs. The EU plays a role model in this matter, as they were able to reduce

20% of its energy consumption solely through energy efficiency.

Intelligent Quality Assurance (IQA) Approach

The third tool known as the Intelligent Quality Assurance (IQA) is unlike the

other tools, an IT-based monitoring system, which was specified for the Thai

Automotive Sector. This approach involves (a) Hardware, (b) Software as well as (c)

Human Resources (professional staff). With the observation of input and output data,

machinery and equipment capacity the system evaluates the production and process

quality. However, in order to apply IQA the product quality, technology and the

production process must be computer-controlled.

The application of IQA with regard to the Green Auto Project has led to an

increased quality of data collection, which is very important for the evaluation of

processes. It also enabled SMEs to enhance their production efficiency in terms of

Page 44: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

34

decreasing input. However, within the years the IQA tool has been replaced by a tool

known as Loss Reduction Process (LRP). The latter approach identifies and

eliminates or replaces loss-causing processes.

The automotive sector in Thailand is very heterogeneous and consists of more

than 160 different fields of production and assembly. As mentioned before, tier 1

suppliers are mainly multinational manufacturing companies with their plants based in

Thailand. Companies worldwide are raising concern on environmental-friendly

production, as global and local communities are pushing them towards cleaner

production. Second and third tier suppliers are rather small and medium enterprises

and are highly diversified. As a result, SMEs are classified into clusters. The clusters

are formed according to the commonality on fields of operation and organizational

set-up size consisting of 15-20 SMEs. This way it makes it easier to apply particular

tools to each cluster accordingly. For example the cluster of the category rubber

involves SMEs that produce hoses, bushings, engine mounts, mudguards etc. In

addition to that, SMEs are classified into sub-groups under the clusters, as many of

the firms produce a large set of automotive parts. This enables better cross-linking of

SMEs that have similar production fields and require the same technical guidance and

support.

Working Package 2 – Harmonize and implement SME specific financial support

packages

With respect to this working package, the SME Bank aims to establish new

financial packages for Sustainable and Clean Production (SCP) improvement

measures. The establishment of the financial package is based on previous experience

and feedback from the training and consultancy sessions of working package 1.

Furthermore, other existing financial instruments will be taken under account such as

the Thai ESCO. The existing financial instruments in Thailand are difficult to access

for SMEs, as the financing sources are diversified and come from different

organizations, which use varied promotion agencies targeting different goals.

Financing support packages are relevant, as they provide SMEs a lot; starting from

debts financing to financing projects or provide loans, in order to go through training

Page 45: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

35

programs or consultancy that enriches the know-how and skills to reach green

production.

SMEs have to be advised on all possible financial schemes they could engage with, as

searching for the right one is considered costly. Through workshops and consultancy

the providers of financial packages promote and educate SMEs on their offer, so

SMEs can evaluate and make their decision accordingly. The new financial support

packages will be introduced to each cluster in a workshop. It should be noted that the

packages can be classified into funding for direct investments and improvement

measures, i.e. investing in and upgrading machines and tools and funding

improvement measures indirectly, for example through training programs and

management instruments. However, there is a possibility in participating in incentive

programs that will grant an excellence award and substitutes financial packages.

Furthermore, customized training programs for financial managers or providers will

be given, in order to increase awareness and enhance the comprehension of financial

officers who are responsible for approving loans to SMEs for fields related to green

production. This aims to ensure a better access to loans for SMEs required for SCP

measures.

Working Package 3 – Strengthen Sustainable and Clean Production (SCP)

related services and networking

The Green Auto project provides a number of 30 PREMA, 15 BAT and 15

LRP experts that will support 500 SMEs to implement the project for a duration of

three years. These experts are chosen by the project partners’ network, for example

Energy Efficiency Network (EENet), Technical and Engineering SP network or other

databases such as ESCO databases.

This working package is of relevance as the quality of service providers concerning

environmental auditing and technical or management consultancy in Thailand is said

to be insufficient. As a result, SMEs wish to have more insights in accessing

qualitative services and are willing to pay for it when proven quality can be expected.

The harmonization amongst the network of automotive parts manufacturers, service

providers, government agencies and financial institutes originated are based on good

practices of the EENet development. The EENet development consists of three

Page 46: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

36

platforms; (1) information exchange among key government agencies responsible for

improving energy efficiency in the industrial sector, (2) information platform and (3)

information exchange project implementation. The resulted successes, difficulties and

obstacles will then be transferred back to platform one, i.e. information exchange.

In addition to that, support packages and workshops will be provided with

regard to financial aspects such as investment options, turnover calculation and cost

analysis (e.g. waste and energy costs) under the consideration of working package

two. So that SMEs continue to use the service of service provided, the trustworthiness

must be given. Thus, service providers are required to update their certificates and

proof that they are undertaking additional training on a regular basis. Moreover, it is

anticipated that a database containing the qualification and experience of diverse

service providers will be established.

Working package 4 – Dissemination of good-practice examples and policy

recommendations

The goal of this working package is to collect all good practice examples,

which refer to the implementation of the project in which improvement has emerged.

This will then be made available to all other SME project participants, so they can

make use of it too. Useful policy recommendations extracted from the good-practices

will be submitted to the Thai Government.

The good practice examples can be classified into; (1) specific technical solutions for

individual SMEs and (2) cross-cutting technologies, instruments and procedures

applicable to all SMEs. The good practices that ended up having high results will be

forwarded to SMEs and stakeholders as well as be published on a website for

reference. Data will be collected through audits and evaluated for further analysis of

measures. The derived results will then be recommended as technical and operational

standards for particular production processes alongside the SMEs and TAI clusters.

The collected and analyzed data will reveal information on enhancement in green

production, reduction in greenhouse emissions as well as improved energy efficiency,

environmental performance and productivity. The noticeable potential can be

aggregated and determine the definition of emission limit values for specific

industries.

Page 47: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

37

Chapter 5

Methodology and Procedure

After chapter 3 focused on the detailed explanation on the methodology

applied for this work, namely Data Envelopment Analysis this chapter will discuss

how the author proceeded in order to obtain results. First, the data collection process

will be explained, and then the approach of standardizing the obtained data will be

elaborated, making the data ready for application. After that a mathematical

formulation of the DEA will be given, followed by a introduction of a methodology

that solves the problem with undesirable measures, which is the case for the analysis

obtained from the PREMA data.

5.1 Data Collection

As this study is written in corporation with GIZ Bangkok, available data of the

250 SMEs that participated in the Green Auto Project will be analyzed and evaluated.

GIZ Bangkok itself collected the data from the enterprises, in order to see the progress

that they are making. Data from the PREMA and BAT approach were made available

for the purpose of this study. GIZ has created data templates for the BAT tool, which

had to be filled out by the companies. Furthermore, consultants went to the companies

and have implemented audits. The collected data has then be sorted out according to

completeness and appropriateness for the application of DEA. The obtained data was

received in Thai language and has then been translated into English. The available

inputs, involve the consumption of gross energy and electrical energy. The output data

consists of the number of end products.

For the PREMA approach company reports have been collected, including flow

charts of the respective input and output data per process. Similar to the BAT data, the

input data for PREMA was the energy consumption. The output data mainly

represents scrap per process or gross energy consumption (heat), which are also

known as undesirable or non-product outputs. The selection of particular SMEs or

processes for the evaluation was decided upon the availability of data. As common

data and units of inputs and outputs are required for the application of DEA, the

author grouped the peer groups according to the commonality of data in order to make

Page 48: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

38

the evaluation feasible. Thus, the processes that were selected for the assessment of

the PREMA data were welding, piercing and cutting. As mentioned above, the BAT

data was collected with the help of a template and thus commonality of data was

already given. The only difference was that some products were expressed in

kilograms, others in tons or pieces. Hence, the author grouped the peer groups

according to the units of end products. However, because the SMEs in a BAT peer

group produce various and different types of automotive parts, one product might

require more energy than another, thus the data is rather heterogeneous. This is a

problem because DEA assumes homogeneity of data. As a result, the data must be

standardized for the purpose of applying DEA. The standardization of data will be

explained in the next chapter.

5.2 Data Standardization Approach

The availability of data for different SMEs within one sector, in this case the

automotive sector, does not imply equivalent interpretation of data. SMEs in the

automotive sector vary and have different backgrounds that cannot be considered by

simply observing data. One might wants to know whether the indicated value within a

data set is relatively high or low compared to the values in its distribution. This can be

reflected when using the standard deviation of a data set as an interpretation indicator.

It implies how many standard deviation units an indicated value is above or below the

mean of its distribution. Furthermore, if two indicated values of two samples

respectively should be compared to one another, then the deviation from the mean

should rather be considered for each sample. This transformation of data is known as

z-standardization, the transformed data are known as z-scores (Uni Hamburg, n.d.).

Thus, before applying the data for the DEA analysis, the data must first be

standardized for the purpose of comparison. Prior to the explanation of the

standardization process, the following variables shall be defined (Powers &

McMullen, 2000):

𝑋!": 𝑡ℎ𝑒  𝑣𝑎𝑙𝑢𝑒  𝑜𝑓  𝑎𝑛  𝑎𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑒  𝑗  𝑓𝑜𝑟  𝐷𝑀𝑈  𝑖

𝑋!: 𝑡ℎ𝑒  𝑚𝑒𝑎𝑛  𝑓𝑜𝑟  𝑎𝑛  𝑎𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑒  𝑗  𝑎𝑐𝑟𝑜𝑠𝑠  𝑎𝑙𝑙    𝐷𝑀𝑈𝑠

𝜎!: 𝑡ℎ𝑒  𝑠𝑡𝑎𝑛𝑑𝑎𝑟𝑑  𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛  𝑜𝑓𝑎𝑛  𝑖𝑛𝑑𝑖𝑐𝑎𝑡𝑒𝑑  𝑣𝑎𝑙𝑢𝑒  𝑓𝑜𝑟  𝑎𝑡𝑡𝑟𝑖𝑏𝑢𝑡𝑒  𝑗

𝑍!": 𝑡ℎ𝑒  𝑧 − 𝑠𝑐𝑜𝑟𝑒𝑠  

Page 49: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

39

The calculation of the mean and standard deviation is regular. The z-scores can be

computed by applying the following formula:

𝑍!" =(!!"!!!)  

!! (5.2.1)

A z-score is a standardized value and indicates how many standard deviation

units and in which direction (sign) a value deviates from the mean of its sample.

Through the transformation the values of a distribution with different means and

variations are made comparable with regard to its relative deviation from its mean.

Note that the z-standardization does not change the form of a distribution. The sum of

all z-scores of an attribute j equals 0, implying that there are negative z-scores as well.

The standard deviation of a sample is always 1 and the mean 0 (Uni Hamburg, n.d.).

The following graph indicates a normal distribution and illustrates a z-standardization:

Figure 5.2.1: Z-Standardization (Lucerne University of Applied Sciences and Arts, n.d.)

Page 50: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

40

The graph illustrates that the mean is 0 and the samples have been transformed

to a normal distribution. The distances on the x-axis represent the standard deviations,

while the y-axis indicates the amount or rate of an attribute. Through this

transformation the data is made comparable and can be applied for DEA.

However, one must bear in mind that DEA is not capable of assessing negative

data. Thus, the application of negative z-scores is not feasible. To circumvent this

problem, a random variable R will be added to the z-scores of an attribute j, so that all

the z-scores become positive. This procedure is based on the study of Powers and

McMullen (2002), where the absolute value of the minimum z-score is added to each

z-score of an attribute. The mean of an attribute is then relocated from 0 to the

random variable R. Graphically explained, the distribution is moved to the right side

until the mean reaches the random variable R. However, the standard deviation

remains 1.

The following indicates how this statistical approach has been applied to the

BAT data. A detailed example will be given by using the data of BAT peer group 1.

The following table illustrates the raw data of peer group 1, as well as the mean, the

variance and the standard deviation that are required to calculate the z-scores.

Table 5.2.1: Raw data of BAT peer group 1

DMUs Electrical Energy (kw/year)

Gross Energy (MJ/year)

Products in Kg

DMU1 5,207,000 18,745,200 1,272,231.92 DMU2 4,749,080 17,096,688 2,605,490.39 DMU3 42,652,553 159,727,480.8 20,841,472.38 Mean 17,536,211 65,189,790 8,239,732

Variance 3.1545E+14 4.46914E+15 7.96982E+13 Standard Deviation 17,760,920 66,851,630 8,927,385

When applying the above-mentioned formula for the z-scores, the z-score for

DMU1 in attribute 1, i.e. the electrical energy is calculated as follows:

𝑍!!:(5,207,000 − 17,536,211)

17,760,920=  −1

Page 51: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

41

As mentioned before, the value of the score Z11 = -1, reflects that the value lies

exactly one unit below the mean (which is 0 for the transformed data), meaning below

average compared to the rest of the distribution. The same calculation applies for the

rest of the z-scores. The table below illustrates the z-scores for all the peer groups. It

shall be noted that the indicated random x-values have been added to the computed z-

scores in order to overcome negativity. For example for Z11 = -1+2 (random X-value)

=1. Table 5.2.2: Standardized BAT Data

BAT Peer Group 1 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in Kg

DMU1 1 1 0,22 DMU2 1 1 0,37 DMU3 3 3 2,41 X-Value 2 2 1

BAT Peer Group 2 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in pieces

DMU1 3 1 2 DMU2 3 4 2 DMU3 2 2 1 DMU4 2 2 1 DMU5 1 1 5 DMU6 4 4 2 DMU7 2 2 1 DMU8 1 1 1 DMU9 1 1 2 X-Value 2 2 2

BAT Peer Group 3 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in tons

DMU1 1.064 0.752 1.223 DMU2 0.079 0.433 3.532 DMU3 2.245 2.981 1.223 DMU4 2.612 1.834 1.222 X-Value 1.5 1.5 1.8

At this point, a standardization of data has been reached and the above-

indicated data has been applied for the DEA application of this study for the BAT

approach. Now that the data has been standardized, homogeneity of data is assumed.

The raw data of the DMUs can be found in the appendix.

Page 52: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

42

5.3 Problem Formulation

In the following a basic DEA example by Zhu (2009) will be given, which is

input-oriented and variable-return to scale and aims to strengthen the understanding of

the analysis tool. The input-oriented envelopment VRS model will also be applied for

the analysis of this research. As the presented linear formulation is input-oriented, the

objective is to minimize input and maintain the amount of output.

Let 𝜃∗be the efficiency:

𝜃∗ = 𝑚𝑖𝑛𝜃 (5.3.1)

Subject to:

Σ!!!  ! 𝜆!𝑥!" ≤  𝜃𝑥!" i=1,2,...m; (5.3.2)

Σ!!!! 𝜆!𝑦!" ≥ 𝑦!" r=1,2,…s; (5.3.3)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j=1,2,…n. (5.3.4)

Here, DMUo stands for one out of n DMUs that will be assessed and 𝑥!" and

𝑦!" represent the ith input and rth output for DMUo. If 𝜃∗, which indicates the

efficiency score equals 1, then the input can no further be reduced, as DMUo is

already on the frontier. Thus, if 𝜃∗ < 1 then the efficiency frontier is above DMUo

and DMUo can be pushed towards the frontier.

For the purpose of strengthening the understanding an example will now be given. Let

us assume there are 5 DMUs, wherein each of them has the same profit every week of

$ 2000 (output) and different supply chain costs and response time (inputs). The

following table summarizes this: scenario Table 5.3.1: Supply Chain Operations (Zhu, 2009)

DMU Cost ($ 100) Response time in days Profit ($ 1000)

1 1 5 2

2 2 2 2

3 4 1 2

4 6 1 2

5 4 4 2

Page 53: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

43

The next step is to analyze which out of the 5 DMUs are located on the

efficient frontier. If we take the following modulation for DMU5 with respect to the

table above, then the formulation is as follows:

Objective: Min 𝜃

Subject to

1𝜆! + 2𝜆! + 4𝜆! + 6𝜆! + 4𝜆! ≤ 4𝜃 (5.3.5)

5𝜆! + 2𝜆! + 1𝜆! + 1𝜆! + 4𝜆! ≤ 4𝜃 (5.3.6)

2𝜆! + 2𝜆! + 2𝜆! + 2𝜆! + 2𝜆! ≥ 2 (5.3.7)

𝜆! + 𝜆! + 𝜆! + 𝜆! + 𝜆! = 1 (5.3.8)

𝜆!, 𝜆!, 𝜆!, 𝜆!, 𝜆! ≥ 0 (5.3.9)

Note that 𝜃 represents the efficiency and 𝜆 the weights assigned to each

DMU for the purpose of identifying the best practice. The weights (𝜆) will be

elaborated more detailed later on. The solution of the above formulated problem is

𝜃∗ = 0,5, 𝜆!∗ = 1, and 𝜆!∗ = 0 (𝑗 ≠ 2), which demonstrates that DMU5 is inefficient

and DMU2 is its benchmark. Furthermore, it reveals that DMU5 must reduce cost and

response time in order to reach the efficient frontier. The following graph indicates

the efficient frontier for the input-oriented model and shows that DMU1, 2, 3 and 4 lie

on the efficient frontier. Additionally, it shows that DMU2 is indeed the closest

benchmark for DMU5 and that it must reduce its cost by $200 and its response time

by 2 days to be exactly in DMU2’s position.

Page 54: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

44

Figure 5.3.1: Efficient Frontier of 5 DMUs (Zhu, 2009)

The individual reduction of input or output (depending on the model

orientation) in order to reach the efficient frontier is known as DEA slack. The above

modulation for DMU5 can be done for every DMU and in this case will be 𝜃∗ = 1,

for DMU1, 2, 3 and 4 as they already reached the frontier. However, through

observation one can see that DMU4 could reduce its total supply chain costs by 2

units and would be better off it would be in the same position as DMU3. Thus, DMU4

is only weakly efficient. The characteristics of weakly efficient will be explained

more detailed.

Apart from the basic DEA envelopment model, DEA is mostly used as a 2-

stage procedure as it first analyzes the efficiency scores and the slacks later on. The

slack models can be designed on the basis or extended to the efficiency model, which

is the following in particular:

𝑠!! = 𝜃∗𝑥!" − Σ!!!! 𝜆!𝑥!" i= 1, 2,…,m (5.3.10)

𝑠!! = Σ!!!! 𝜆!𝑦!" − 𝑦!" r= 1, 2,…,s (5.3.11)

Here, 𝑠!! and 𝑠!! stand for input and output slack, as input needs to be reduced (-) and

output be increased (+) to become more efficient. The efficient DMUs will have zero

slack values (Zhu, 2009). Generally, first the efficiency scores are computed by

Page 55: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

45

ignoring the slacks. In the second stage, the efficiency scores will be fixed and the

slacks optimized by applying the following linear programming problem (Sherman &

Zhu, 2006):

𝑚𝑎𝑥Σ!!!! 𝑠!! + Σ!!!! 𝑠!! (5.3.12)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝜃∗𝑥!" i= 1, 2,…,m (5.3.13)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝑦!" r= 1, 2,…,s (5.3.14)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j= 1, 2,…, n (5.3.15)

It should be noted that a DMU is fully efficient if and only if 𝜃∗ = 1  𝑎𝑛𝑑  𝑠!! =   𝑠!! =

0. Should the efficiency score be 1 and the slacks ≠ 0, the DMUs are considered as

weakly efficient. Weakly efficient units will still be considered as efficient, however,

when calculating the slacks, i.e. in stage two, weakly DMUs will not be included in

the reference set as a benchmarking unit to other DMUs (Sherman & Zhu, 2006).

As a result, the input-oriented VRS envelopment model including the slack

calculation can be expressed as follows:

𝑚𝑖𝑛𝜃 − 𝜀 Σ!!!! 𝑠!! + Σ!!!! 𝑠!! (5.3.16)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝜃𝑥!" i= 1, 2,…,m (5.3.17)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝑦!" r= 1, 2,…,s (5.3.18)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j= 1, 2,…, n

Wherein the variable 𝜀 enables the minimization of 𝜃 in order to avoid the

optimization involving 𝑠!!  𝑎𝑛𝑑  𝑠!!. Consequently, 5.3.16 runs the model in a two-

stage process. First, a maximal reduction of inputs possible is calculated and the

optimal value of 𝜃∗ from model 5.3.1 reached. Then, the inefficient DMUs are moved

towards the efficient frontier by optimizing the slack variables (Zhu, 2009).

The output-oriented VRS envelopment model is expressed as:

Page 56: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

46

𝑚𝑎𝑥𝜙 + 𝜀 Σ!!!! 𝑠!! + Σ!!!! 𝑠!! (5.3.19)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝑥!" i= 1, 2,…,m (5.3.20)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝜙𝑦!" r= 1, 2,…,s (5.3.21)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j= 1, 2,…, n (5.3.22)

Similarly, this model is also calculated in a two-stage process. After 𝜙∗

indicating the efficiency has been calculated, the slacks can be computed by fixing the

optimal efficiency score. Thus, the linear programming problem for calculating the

slacks is:

𝑚𝑎𝑥Σ!!!! 𝑠!! + Σ!!!! 𝑠!! (5.3.23)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝑥!" i= 1, 2,…,m (5.3.24)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝜙∗𝑦!" r= 1, 2,…,s (5.3.25)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j= 1, 2,…, n (5.3.26)

This makes the DMUs efficient if and only if 𝜙∗ = 1 and 𝑠!! = 𝑠!! = 0 for all

i and r. It shall be demonstrated that 𝜙∗ ≥ 1 and 𝜙∗ = 1 if and only if 𝜃 = 1.   This

reflects that the models 5.3.16 and 5.3.19 determine the same frontier. Furthermore,

𝜃∗ = 1/𝜙∗ (Zhu, 2009). The following figure illustrates in contrast to figure 5.3.1 an

output-oriented frontier in which inputs are fixed and outputs must be increased to

reach the efficient frontier. It shows a total number of 4 DMUs, wherein DMU 4 is the

only inefficient one. Only two outputs are considered and DMU 4 must reach point G,

in order to reach the efficient frontier.

Page 57: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

47

Figure 5.3.2: Output-Oriented Efficient Frontier (Zhu, 2009)

5.4 Methodology of Modelling Undesirable Measures

The objective of this thesis is to assess greening activities of SMEs in the Thai

automotive sector. As energy efficiency should be improved and for example waste

belong to undesirable outputs, this chapter aims to understand how to deal with

undesirable measures when using DEA.

There are several ways of modeling undesirable measures when using DEA.

As mentioned before, undesirable measures do occur and must be modeled

differently, as they do not correspond to the norm. For example in an output-oriented

model, the objective is to increase the amount of output while maintaining the amount

of input in order to improve the overall performance and to become more efficient.

However, if a model involves undesirable outputs then the objective is to reduce these

(see chapter 3.1).

For example, one way of dealing with undesirable outputs is to simply treat

them as inputs and compute the efficiency score with the input-oriented model.

However, this does not represent a real production process. The undesirable output

can also be ignored, but this approach will not lead to the achievement of the target,

which is to reduce the amount of undesirable output, thus, other approaches must be

considered. Another way to deal with undesirable output is to use a non-linear DEA

model, however, this procedure might be more difficult. Other possibilities are to treat

Page 58: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

48

the undesirable outputs as a regular one and later on, limit the expansion of the

undesirable output by modifying the distance measurement. Finally, one could also

use a monotone decreasing transformation by translating the undesirable output data

and use the adapted data as regular output (Seiford M. & Zhu, 2002). The latter

approach will be demonstrated in detail, as it will be used for the analysis of this

work.

Seiford and Zhu have developed a strategy to use undesirable measurements in

the DEA models. In principle, the approach “is the use of DEA classification

invariance under which classifications of efficiencies and inefficiencies are invariant

to the data transformation” (Zhu, 2009).

While considering the efficiency invariance let us assume that input and

output have been transformed into 𝑥!" = 𝑥!" + 𝑢! and 𝑦!" = 𝑦!" + 𝑣! respectively and

𝑢 and 𝑣 are ≥ 0. With respect to the transformed data, the input and output-oriented

VRS models are as follows:

𝑚𝑖𝑛𝜃 − 𝜀(Σ!!!! 𝑠!! +  Σ!!!! 𝑠!!) (5.4.1)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝜃𝑥!" i= 1,2,…, m; (5.4.2)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝑦!" r= 1,2,…, s; (5.4.3)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j=1, 2,…, n. (5.4.4)

𝑚𝑎𝑥𝜃 − 𝜀(Σ!!!! 𝑠!! + Σ!!!! 𝑠!!) (5.4.5)

Subject to:

Σ!!!! 𝜆!𝑥!" + 𝑠!! = 𝑥!" i= 1,2,…, m; (5.4.6)

Σ!!!! 𝜆!𝑦!" − 𝑠!! = 𝜃𝑦!" r= 1,2,…, s; (5.4.7)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0 j=1, 2,…, n. (5.4.8)

The attribute Σ!!!! 𝜆! = 1 enables to have negative values for input and output before

using the VRS-Model.

Page 59: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

49

The following example indicates a model with undesirable output referring to

the work of Seiford and Zhu (2002).

Let us assume 𝑦!"! represents the good and 𝑦!"! the bad output. Thus, the objective is to

increase 𝑦!"! and decrease 𝑦!"! . To achieve the multi-objective problem, the

undesirable output must be multiplied by “-1” and additionally a proper value 𝑣! must

be found which will be added and will turn the negative bad output into positive

numbers. This can also be expressed as 𝑦!"! =  −𝑦!"! +  𝑣! > 0 and 𝑣! = 𝑚𝑎𝑥! 𝑦!"! +

1. With respect to the above efficiency invariance, the objective function is:

Max h (5.4.9)

Subject to:

Σ!!!! 𝜆!𝑦!"! ≥  ℎ𝑦!"

! (5.4.10)

Σ!!!! 𝜆!𝑦!"! ≥  ℎ𝑦!"! (5.4.11)

Σ!!!! 𝜆!𝑥!" ≤  𝑥!" (5.4.12)

Σ!!!! 𝜆! = 1 and 𝜆! ≥ 0, j= 1,…,n (5.4.13)

Note that h represents the efficiency. The model for (5.4.9) makes sure that desirable

output is increased and undesirable decreased. Furthermore, having a translation

vector 𝑣, we assume ℎ∗ is the ideal value for (5.4.9) at ℎ∗𝑦!"! ≤  𝑣!. It provides non-

negativity for the optimized undesirable output of 𝑦!!! (= 𝑣! − ℎ∗𝑦!"! ) (Zhu, 2009).

Page 60: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

50

Chapter 6

Energy Efficiency Analysis

The following analysis refers to an input-oriented model with undesirable

input for the data obtained from the PREMA approach and a regular input-oriented

for the data obtained from the BAT approach. The first model is based on the

undesirable input-oriented methodology by Seiford and Zhu (2002) explained in

chapter 5.4. It shall be noted that because of the diversity of PREMA data, the SMEs

have been classified according to internal production processes. As for BAT, the

SMEs have been grouped according to the units of their end products, i.e. kilograms,

tons and pieces. The tool IQA will not be considered for the analysis, as it would

exceed the scope of this work.

6.1 Profitable Environmental Management Approach - Data and Analysis

Because the available data and the different measures available for this project

- or approach to be specific - are very diverse, different approaches for benchmarking

have been considered. First, the objective was to collect all the available data of

energy consumption, costs and goods produced. However, this approach did not seem

to be representative, as the companies differ in size, production of goods and variety

and number of production processes. Thus, the companies are heterogeneous even

though they all focus on the automotive sector. The complexity of this particular

model mainly lies on grouping the SMEs according to particular attributes for

benchmarking. The second approach considered focused on evaluating the

companies’ specific operating processes with the purpose of eliminating inefficiency.

For many reasons benchmarking the operating processes are more important than

benchmarking the institutions as a whole. First, information on specific production

processes helps to understand the basics of efficiency on a production level and might

reveal some measurement problems. Second, companies’ production processes are

responsible for the majority of operational expenses, which is of more interest from a

manager’s perspective. Looking into the available data and having selected processes

(i.e. cutting, piercing and welding), the next step leads to collecting data that would

Page 61: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

51

apply to all the SMEs within the particular production processes. This is necessary for

the DEA approach, as otherwise the performance evaluation would not be feasible. As

a result, particular data that could be found for every company within the processes

were energy consumption as an input and scrap as an output, while the welding

process has heat as output data. The following table indicates the PREMA data

selected for the benchmarking analysis.

Table 6.1.1: PREMA Data

Welding Process DMUs Input (Undesirable)

Electrical Consumption (kw/year) Output Heat (kw/year)

DMU1 256,471 256,471 DMU2 10,214 36,771 DMU3 9,800 204,873

Cutting Process DMUs Input (Undesirable)

Electrical Consumption (kw/Year) Output Cutting Scrap (kg/year)

DMU1 63,180 1,428 DMU2 4,708 236.7 DMU3 18,200 657.34 DMU4 876 1270.4 DMU5 12,286 2385 DMU6 189,447 7200

Piercing Process DMUs Input (Undesirable)

Electrical Consumption (kw/year) Output Piercing Scrap (kg/year)

DMU1 15,343 17,280 DMU2 760,500 7,740 DMU3 15,731 3,500 DMU4 2,520 75,860

The presented data expresses the input in kilowatt per year and the outputs in

kilogram and kilowatt per year, respectively. Obviously, the outputs heat energy and

scraps are bad outputs and shall be reduced.

For the purpose of the applied method, translation vectors of v= 260,000 for

welding, v= 7,300 for cutting and v= 76,000 for piercing have been selected,

respectively. The undesirable input data of each process is subtracted from the

translation vector, leading it to a reduction of input, instead of an increase of input,

which is the norm in an input-oriented model. Hence, the translated data is used for

Page 62: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

52

further evaluation. The data has been analysed by applying an input-oriented VRS

envelopment model by using Microsoft Excel Solver. The efficiency scores with the

applied method are indicated in the table below, as well as optimal benchmark

weights of the DMUs, which is also known as the reference set.

Table 6.1.2: DEA results of PREMA

Welding Process DMUs Efficiency Scores Reference Set DMU1 0.04 1 (DMU3) DMU2 1 - DMU3 1 -

Cutting Process DMUs Efficiency Scores Reference Set DMU1 0.02 1 (DMU4) DMU2 1 - DMU3 0.2 0,6 (DMU2); 0,4 (DMU4) DMU4 1 - DMU5 0.07 1 (DMU4) DMU6 0.004 1 (DMU4)

Piercing Process DMUs Efficiency Scores Reference Set DMU1 0.86 0.81 (DMU3); 0.19 (DMU4) DMU2 0.002 0.94 (DMU3); 0.06 (DMU4) DMU3 1 - DMU4 1 -

Within the peer group of the welding process DEA identified 2 efficient

DMUs amongst 3 DMUs. DMU1 has an efficiency score of <1, which implies that it

is inefficient. The scores can also be read as percentages and state that DMU2 and

DMU3 should keep 100% of its input and output; whereas DMU1 is advised to reduce

it to 4% from its currents level.

The two efficient DMUs, i.e. DMU2 and DMU3 encompass the best-practice

DMUs and are located on the efficient frontier. However, it is unknown which DMU

out of the best-practice is more or less efficient than the other one. The DMUs located

on the efficient frontier and which are amongst the best-practice DMUs are those that

are clearly not inefficient relative to the other DMUs in the evaluation. Also, it should

be demonstrated that these DMUs are fully efficient, as they have an efficiency score

of 1 and the slacks equal 0, which will be indicated later. The calculation of the slacks

was based on the formula 5.3.10 and 5.3.11.

Page 63: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

53

Furthermore, DEA reveals the information of a subgroup of the SMEs, which

are referred as the efficiency reference set. For example, DMU1 has found to be

inefficient and has operating inefficiencies in direct comparison to DMU3. The values

in the column of the reference set indicate the relative weight allocated to each DMU

included in the reference set in order to calculate the efficiency rating. In other words,

the input and output of DMU3 are multiplied by the weights determined by DEA, in

this case by 1. Afterwards, these are added together and result in a composite DMU

that uses a more efficient set of input and output than that of DMU1. By applying a

combination of the operating techniques used by DMU3 the new hypothetical

composite DMU can decrease its input or decrease the bad output. More specifically,

the new hypothetical composite DMU can move towards the efficienct frontier by

either reducing its energy consumption and maintaing its heat energy consumption or

maintaining its energy consumption while reducing its heat energy consumption. This

information should be an adequate evidence to cause the SMEs to question why they

were not able to be as efficient as other SMEs and be the trigger of some new plans

and implementations with respect to energy efficiency. DMUs 2 and 3 are their own

benchmarks, thus, they do not need to be compared to other DMUs, which explains

why they have no weights indicated in the column of the reference set. The same

interpretation of results applies for the remaining DMUs in the peer groups for cutting

and piercing process.

Additionally, DEA computes the slacks of the inefficient DMUs advising

them what input or output to reduce, in order to become more efficient. For the

welding process, DMU1 is recommended to reduce its heat energy consumption by

51,600 kw/year towards the output level of DMU3. The slacks for the cutting and

piercing processes are represented in the table below.

Page 64: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

54

Table 6.1.3: Slacks for PREMA data

Cutting Process

DMUs Reducing Electrical Energy (kw/year) Reducing Scrap (kg/year)

DMU1 - 157.6

DMU2 - -

DMU3 - 7.16

DMU4 - -

DMU5 - 1114.6

DMU6 - 5929.6

Piercing Process

DMUs Reducing Electrical Energy (kw/year) Reducing Scrap (kg/year)

DMU1 2122 32

DMU2 774 4240

DMU3 - -

DMU4 - -

The inefficient DMUs within the cutting process are advised to reduce their

scrap, while the inefficient DMUs within the piercing process can either reduce their

electrical energy or reduce their scrap. The efficient DMUs have zero slacks, which is

why no slack-reduction or recommendation is indicated. The implementation of the

electrical energy consumption or scrap can be done by updating outdated machinery

or revising the production processes. For more specific recommendations, further

energy saving opportunities will be discussed in chapter 6.4 as policy

recommendations.

At this point, it is also of relevance to emphasize that the efficient DMUs are

relatively efficient and not absolutely. Thus, amongst the DMUs in a peer group, for

example within the welding process no other DMU is more efficient than DMU 2 and

3. Consequently, they are the best available and not the best possible.

Page 65: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

55

6.2 Best Available Technique Approach - Analysis

For the BAT approach and data, 16 SMEs were available. These DMUs had

common inputs and outputs, i.e. electrical energy consumption, gross energy

consumption and goods produced. The inputs are expresses in kilowatt/year and mega

joule/ year while the products had different units. Thus, the DMUs had to be classified

according to their unit of products, i.e. products in pieces, kilogram and tons. In this

case, the outputs are desirable, which should be increased if possible. Each peer group

has a number of 3, 9 and 4 DMUs to be evaluated, respectively. The DMUs use 2

inputs (electrical and gross energy) and produce one output, expressed in different

units. The statement of the problem is the very same as the one applied for the

PREMA data, the only difference is that this model has a desirable output. Hence, the

regular input-oriented VRS envelopment model will be applied. At this point it shall

be recalled that the data illustrated in table 5.2.2 has been used for the computation of

DEA results with the DEAFrontier Software in Microsoft Excel.

The following table reports the results of DEA that have been applied to the

below-indicated DMUs. Table 6.2.1: DEA Results of BAT

Peer group 1 DMUs Efficiency Score Reference Set DMU1 1 weakly efficient 1 (DMU2) DMU2 1 efficient - DMU3 1 efficient -

Peer group 2 DMUs Efficiency Score Reference Set DMU1 1 weakly efficient 1 (DMU5) DMU2 0.33 inefficient 1 (DMU5) DMU3 0.5 inefficient 1 (DMU5) DMU4 0.5 inefficient 1 (DMU5) DMU5 1 efficient - DMU6 0.25 inefficient 1 (DMU5) DMU7 0.5 inefficient 1 (DMU5) DMU8 1 weakly efficient 1 (DMU5) DMU9 1 weakly efficient 1 (DMU5)

Peer group 3 DMUs Efficiency Score Reference Set DMU1 0.57 inefficient 1 (DMU2) DMU2 1 efficient - DMU3 0.14 inefficient 1 (DMU2) DMU4 0.23 inefficient 1 (DMU2)

Page 66: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

56

Out of the 3 DMUs in peer group 1, DEA identified 2 DMUs as efficient.

DMU1 has an efficiency score of 1 and is weakly efficient. Recalling from chapter 5.3

a DMU is fully efficient, only and only if the efficiency score equals 1 and slacks are

zero. Because DMU2 uses less inputs and produces more output than DMU1, DEA

indicates that DMU1 has the potential to either decrease its input by producing the

same output or remain its input level and produce more output. In other words, DMU1

has found to be weakly efficient in comparison to DMU2, which is why it is being

referred to in the reference set by 100%. The second peer group has also 1 efficient

DMU, namely DMU5. DMU1, 8 and 9 are only weakly efficient, due to the fact that

the slacks are non-zero. The remaining DMUs are relatively inefficient and are

referred to DMU5 for improvement. As DMU5 is identified as relatively efficient, it

will not be compared to any other DMU within the peer group and is its own

reference. The computed weights in the column of the reference set indicated in the

above table are assigned to each reference set member in order to calculate the

efficiency rating. In other words, it reveals information for weakly efficient or

inefficient DMUs about how much weight of which reference set member to apply in

order to move towards the efficient frontier. In this case, only one efficient DMU has

been identified per peer group, which is why a full weight of 100% is assigned to the

best practice in the reference set. If there would be more than one efficient DMU, then

DEA would compute every possible combination of weights to make the weakly

efficient or inefficient DMU move towards the efficient frontier. Hence, the weakly

and inefficient DMU should apply 100% of the input and output combination the

best-practice DMU is using.

In regards to the slacks, DEA provides some recommendation with respect to

the computed weights in the reference set. Looking into peer group 1, the

recommendation is that DMU1 should increase its products. DMU1 consumes more

electrical and gross energy than DMU2 and produces less kg of output. Through the

data information of GIZ it is known that DMU1 and DMU2 in this particular peer

group operate in the very same field and are branched SMEs. Thus, it is meaningful

that DMU1 is benchmarked with DMU2. Given the fact that DMU1 uses more input

to produce less output, it has the potential of increasing its products by 1,333,259 kg

Page 67: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

57

to 2,605,490 kg, which is the same amount DMU2 produces. This can be achieved

without increasing energy consumption to reach the efficient frontier. But because the

focus of this study is to improve energy efficiency, it is recommended to reduce the

energy level to the same amount of DMU2. The slack recommendations for peer

group 2 and 3 are illustrated in the table below:

Table 6.2.2: Slacks for BAT data

Peer group 2

DMUs Reducing Electrical Energy (kw/year)

Reducing Gross Energy (kg/year)

Increase products in pieces

DMU1 decrease - -

DMU2 - decrease -

DMU3 - - increase

DMU4 - - increase

DMU5 Best practice

DMU6 - - increase

DMU7 - - increase

DMU8 - - increase

DMU9 - - increase

Peer group 3

DMUs Reducing Electrical Energy (kw/year)

Reducing Gross Energy (kg/year)

Increase products in tons

DMU1 decrase - increase

DMU2 Best practice

DMU3 decrease - increase

DMU4 decrease - increase

For peer group 2, a recommendation to decrease the electrical energy

consumption goes only to DMU1. DMU2 is advised to decrease its gross energy

consumption, while the remaining DMUs have the potential to increase their amount

of output while using the same amount of input. DEA computes the slack by

analysing the data sets and revealing the slack recommendation that it thinks is most

feasible to achieve. Because for peer group 2, DMU5 uses relatively low energy

consumption compared to its peer members, DEA rather recommends to increase the

Page 68: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

58

output level by remaining the input level, due to the probability of feasibility. DMU5

is the best practice, thus, does not need to reduce input or increase output.

Within the third peer group, the DMUs are advised to either reduce their

inputs or increase their output. Unlike the previous cases, this peer group has the

option of choosing the slack-adaption, as two recommendations are provided.

However, this is an either-or option rather than achieving both ways simultaneously.

Because of the targeted energy efficiency input reduction is suggested. DMU2 has

been identified as best practice and thus, no slack recommendations are given.

Because the data for BAT has been standardized and transformed the author rather

gave a slack-suggestion for the energy saving areas, rather than giving exact numbers

to which the energy consumption should be reduced.

Note that the slack recommendation rather serves as an orientation to what

DEA thinks is feasible. It shall be emphasized that knowing which input to reduce or

which output to increase is more important, than the actual slack numbers. The SMEs

can use this advice and try to achieve the slack recommendations in every possible

way, especially the reduction of electrical and gross energy. It will lead to saving

costs and a greener way of producing the automotive parts.

6.3 Summary

For the purpose of concluding the analysis, the research questions shall be addressed.

a) What are the reasons of different achieved results for the SMEs?

Out of the large number of SMEs it was observable that the companies are very

diverse in size and internal processes. Especially when considering the BAT

approach, one can assume that the main problematic of achieving different results is

due to the fact that out-dated machines are still in use. This can be confuted by

referring to the BAT DEA results of peer group 1. It was mentioned that DMU1 and

DMU2 are branched companies. However, DMU2 is able to consume less energy

consumption and achieve a greater amount of output at the same time. This is just an

obvious example reflecting the internal processes as a problem for inefficiency.

Furthermore, aspects such as know-how or technical feasibility are possible reasons

for inefficiency.

Page 69: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

59

b) Which SME has reached the efficient frontier?

Upon analysis it was found that the efficiency score of both PREMA and BAT

data ranged from 0.2% to 100% out of which a total number of 10 SMEs were found

efficient with a score of 1.

With regard to the PREMA data, DEA identified a total number of 6 out of 13

efficient SMEs. These were DMU2 and 3 in the welding process, DMU2 and 4 in

the cutting process and DMU3 and 4 in the piercing process. The lowest score

obtained for PREMA was 0.2% ranging up to 100%. Whereas for the BAT data,

DEA identified 4 efficient SMEs out of 16, namely DMU2 and 3 for peer group 1,

DMU5 for peer group 2 and DMU2 for peer group 3. The lowest score within the

BAT data amounted 14%. Generally, when looking at the inefficient DMUs, higher

efficiency scores could be observed for BAT. Also BAT revealed 4 weakly efficient

DMUs, compared to the PREMA approach, which revealed none. The differences

within the ranges could be due to the standardization of data within the BAT peer

groups.

At this point it shall again be emphasized that the number of efficient DMUs

might also depend on the number of data sets and number of DMUs. Preferably, the

number of DMUs should be twice as much as the number of added inputs and

outputs together. Having too many inputs and outputs as well as few DMUs can lead

to a slight risk of increased loss of discriminatory power. In other words, more

DMUs are identified as efficient.

c) What can be done to make the energy consumption more efficient?

This highly depends on the internal processes of the SMEs. However, if no

investments in new technology have been done with respect to the Green Auto

Project and SMEs are found to be inefficient, then it is advised to upgrade

machinery, as it has the highest influence on green manufacturing or more

efficient energy consumption. Through this analysis, DEA has identified energy

saving opportunities by indicating which input to reduce in case of inefficiency

and measures energy consumption. More on the way to achieve green

manufacturing will be discussed in the following chapter.

Page 70: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

60

6.4 Policy Recommendations

Recommendation 1: Specific data collection

With respect to the analysis the development of a data template is

recommended to GIZ, so that they can collect common data for all SMEs and

benchmarking will be made easier, especially for the PREMA tool. The obtained data

was very valuable as such as it gives high-valued information on individual processes

of SMEs but lacked in commonality across all SMEs, which made benchmarking

difficult. As there was no template for the data collection, it was complex to group

and analyse the companies according to commonality of data and other attributes,

such as internal processes, so that the application of DEA or any other benchmarking

tool that will be applied for future projects could be made easier. The data also

included many data sets that were not energy-related. That is why the availability of

common energy-related data throughout the SMEs is the foundation of developing

policy recommendations that aim to improve energy consumption. Also, it is

advisable to continuously monitor the data and evaluate the improved results in a

given time period and consequently, develop energy efficiency reports. The

application of key performance indicators (KPI’s) is advisable as through indexes the

developments achieved can be easily illustrated. This shall support the achievement of

energy efficiency.

Recommendation 2: Promoting the Energy Efficiency Development Plan

In chapter 1 it has been mentioned that the Thai Ministry of Energy has

launched a National 20-year Energy Efficiency Development Plan (EEDP), which

targets particularly the transportation and industrial sector. The goal is to reduce

energy intensity by 25% in 2030, compared to the level of 2005. The EEDP will be

achieved by implementing rules and regulations as well as providing subsidies. The

rules and regulations involve the establishment of Minimum Energy Performance

Standards (MEPS) and the labelling’s of energy efficiency on appliances. The

subsidies include a Standard Offer Program (SOP) or the approval of fundings for the

achieved energy savings (Ministry of Energy). GIZ is already working with the

Ministry of Energy; however, it shall be emphasized that the integration of the EEDP

Page 71: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

61

to the Green Auto Project should be expanded, as GIZ has a great impact on

promoting the 20-year Energy Efficiency Development Plan to its target group and

the Thai Ministry of Energy is willing to reach out to SMEs nationwide. Former

Energy Conservation Programs also included regulations and financial incentives. The

previous worked out strategic approaches will remain but further measures that

accelerate the implementation and lead to energy efficiency will be added. The five

strategic approaches that aim to support the realisation of EEDP are following

mandatory requirements through regulations, rising public awareness, developing the

know-how of human resources, the development of an Energy Conservation

Promotion and Support as well as increasing innovations and consequently, apply

improved technology (Ministry of Energy).

Recommendation 3: Establish an Energy Management Team and Implement

Small Energy Saving Actions

For the SMEs that have been found to have high-energy consumption relative

to the rest of the peer group, it is suggested to update their energy-intensive

appliances. This can be done by for example the elimination of inefficient lighting

products or air conditioners and their substitution by improved sustainable lighting

and cooling systems. Additionally, as machinery is known to be the most energy

consuming factor consultants could introduce and promote mandatory minimum

energy performance standards (MEPS) and labels that might be more cost-effective.

These kinds of investments have a high impact on energy efficiency. The achievement

of higher energy efficiency is guaranteed if the individual SMEs start an action plan

that involves the above-mentioned energy saving opportunities and the continuous

audits on energy consumption by engineers (International Energy Ageny, 2011). If

financial measures allow it investment in photovoltaic systems are suggested. By

developing individual action plans, SMEs are able to undertake self-assessments. In

particular, this could include the nomination of an energy manager and building-up a

team that focuses on managing energy efficiency. The team should define policy

regulations, audit the energy consumption and implement energy saving activities.

Also, the team would be responsible for following up the actual implementation of the

regulations. Measures, such as the Energy Efficiency Index (Wajer et al., 2007) or

Page 72: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

62

other statistical approaches like DEA illustrate and document the actual improvements

that have been achieved. DEA provides slack-recommendation, so the slack reduction

for inputs are known and can be targeted.

Conclusively, recommendations to the Green Auto Project are to include

further workshops that address energy efficiency and educate SMEs on the impact of

energy efficiency and the implementation of action plans and policy recommendation

for their enterprise. Furthermore, with respect to the DEA results obtained for the

purpose of this work it is suggested to further investigate the reasons of inefficiency.

6.5 Conclusion

This paper assessed the energy efficiency of SMEs in the Thai automotive

sector. The relative performance of SMEs in terms of energy efficiency has been

analysed. For the obtained data from the PREMA tool that amounted 13 SMEs, 6

DMUs have been identified as energy efficient. Out of the 16 DMUs from the BAT

tool, a total number of 4 energy efficient DMUs have been identified. Furthermore, 4

DMUs are identified as weakly efficient and 8 as inefficient. The findings highlight

the value of benchmarking as a foundation for improvement opportunities in energy

savings and the demonstration of energy consumption of SMEs in the Thai

automotive sector. Additionally, results have indicated that some SMEs are able to

use less energy consumption to produce more goods than others. Hence, results

suggest that with respect to the GIZ project the reason of inefficiency should be

further investigated. The value added to the SMEs is the target-oriented slack

recommendation that gives flexibility on selecting actions to reduce energy

consumption. The SMEs identified as best-practice are not the absolute best SMEs,

but the relative best, meaning that they are not the best possible but rather the best

available.

The results are of high importance because Thailand as one of the

manufacturing countries in the world needs to further address the issue of energy

efficiency as it has a huge potential in improving it. It is important that the energy-

intensive sectors consider the high influence an efficient energy consumption can

Page 73: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

63

make and conduct arrangements that lead to a productive way of managing energy

efficiency. The information obtained from the benchmarking results can be used to

improve the relative performance of individual SMEs. Energy has a great impact on

the economical performance of an organization. Improving energy efficiency reflects

a great opportunity to save economic resources, improve production processes while

also having a positive impact on the environment. Especially SMEs in Thailand that

contribute a lot into the Thai GDP as pointed out in the introduction have a huge

potential in enhancing energy efficiency, as it is often the case that out-dated

machinery is used, awareness is missing or financial measures are unavailable. For

this to be implemented, awareness of the latest technology should be demonstrated.

Other actions include the substitution of inefficient lightings and air conditioners to

more efficient ones, the application of efficient appliances such as Minimum Energy

Performance Standards or with efficiency labelling’s. Lights and equipment’s should

be turned off when not in use and if financial resources allow it, investments in

photovoltaic solar systems are recommended. The National 20-year Energy Efficiency

Development Plan (EEDP) is an important project that will boost the achievement of

energy efficiency in Thailand. The study has provided information on the

performance and identification of a particular peer group within the Thai automotive

sector that need to adapt the above-mentioned policy recommendations for efficiency

maximization.

The contribution of this study is that it is the first empirical study to apply the

Data Envelopment Analysis methodology to measure energy efficiency in the Thai

automotive sector for SMEs to measure energy efficiency. It has benchmarked the

level of energy consumption of SMEs and used the obtained results to identify energy

saving potentials and suggests policy recommendations.

Page 74: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

64

Chapter 7

Limitations and Delimitations

As every work has some limitations that could be avoided in order to not

exceed the scope of the study or because the methodology applied restricts some

procedures, this chapter gives some limitations that appeared in this work and aims to

encourage future researchers to consider and avoid the following restrictions.

7.1 Limitation of Study

In order to narrow the scope of this study, only SMEs that belong to the

automotive sector in Thailand and only those that have participated in the Green Auto

Project by GIZ were examined. Thus, the limitation is associated with the available

data set that could be collected with respect to the Green Auto Project. Furthermore,

as this study focuses on Thailand only, one cannot compare the results to similar

SMEs which might be best practice in other countries.

When applying a data envelopment analysis, the data set for the performance

evaluation is mostly quantitative. However, many evaluation problems contain

qualitative aspects that should be taken into account when assessing the efficiency of

DMUs. For example, the manager’s competence or know-how are essential for the

efficiency of the unit to be evaluated. For the purpose of including this attribute to the

data set, a ranking of the DMUs scaled from best to worst can be established. A more

precise way of quantifying the attributes that reflects the value of the factor is not

possible. Thus, the best way of presenting the qualitative data sets is to rank them in

an ordinal scale rather than in a numerical sense. Each DMU must be assigned to only

one category, which might be high, medium or low in the case of the manager’s

competence (Zhu & Cook, 2007). Qualitative data has not been applied to this work,

therefore, future studies are encouraged to assess energy efficiency with DEA by

applying both quantitative and qualitative data.

7.2 Limitations of Instrument

Some limitations come along with the instrument used in this study. The scores

attained through DEA tend to be sensitive when specifiying input and output as well

Page 75: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

65

as the size. If the size of the organization/unit to be examined is large, DEA will have

a greater scope of determining comparison partner and thus, the average efficiency

score will be reduced. However, if only the amount of input and outputs is increased

without adding further organization, the average efficieny score increases (Steering

Committee for the Review of Commonwealth/State Service Provision, 1997).

Because of data availability, peer group could not have a great number of DMUs in

this study which restricted the selection of best-practice DMUs as for example BAT

results peer group 2.

Additionally, a total number of 7 inefficient DMUs have been identified for

PREMA and 8 for BAT. The inefficient SMEs have been referred to the best practice

within the reference set for achieving improvement. As mentioned in the summary,

the lowest score obtained was 0.2%. It is difficult to compare a DMU that scores 0.2%

to a DMU in the reference set with a score of 100%. For feasible improvement, an

inefficient DMU with a low efficiency score should be compared to a DMU with

identical characteristics and capacity, which makes benchmarking to one another

more justifiable. This is a drawback of DEA as it does not concern the internal

processes and only requires input and output for computation. Especially when it

comes to the slacks, it might not be feasible to reach the recommended slack amount

as the best practice and the inefficient DMU have a high score deviation.

Page 76: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

66

Reference

Ahi, P., & Searcy, C. (2015). An analysis of metrics used to measure

performance in green and sustainable supply chains. Journal of Cleaner Production ,

86, 360-377.

Asmild, M., & Hougaard, J. L. (n.d.). Economic and Environmental Efficiency

of Danish pig farms. Retrieved February 24, 2015, from

http://www.unioviedo.es/7ewepa/pdf/enveff.pdf

Atici, K. B., & Podinovski, V. V. (n.d.). A REVIEW OF DATA

ENVELOPMENT ANALYSIS STUDIES IN AGRICULTURE. University of Warwick,

Warwick Business School, Coventry, United Kingdom.

Azadi, M., Shabanib, A., Khodakaramic, M., & Farzipoor Saena, R. (2014).

Planning in feasible region by two-stage target-setting DEA methods: An application

in green supply chain management of public transportation service providers.

Transportation Research Part E: Logistics and Transportation Review , 70, 324–338.

Bauer, P. W., Berger, A. N., Ferrier, G. D., & Humphrey, D. B. (1998).

Consistency Conditions for Regulatory Analysis of Financial Institutions: A

Comparison of Frontier Efficiency Methods. Journal of Economics and Business , 50

(2), 85-114.

Bayraktar, E., Tatoglu, E., Turkyilmaz, A., & Delen, D. (2012). Measuring the

efficiency of customer satisfaction and loyalty for mobile phone brands with DEA.

Expert Systems with Applications , 39 (1), 99-106.

Bielecki, A. (2012). Methodik der empirischen Forschung - Efficient Frontier

Analysis. University of Kiel, Kiel.

Page 77: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

67

Business Dictionary. (n.d.). Benchmarking. Retrieved January 25, 2015, from

http://www.businessdictionary.com/definition/benchmarking.html

BusinessDictionary. (n.d.). parametric-analysis. Retrieved February 6, 2015,

from http://www.businessdictionary.com/definition/parametric-analysis.html

Charoenrat, T., & Harvie, C. (2014). The efficiency of SMEs in Thai

manufacturing: A stochastic frontier analysis. Economic Modelling , 43, 372-393.

Ciliberti, F., Pontrandolfo, P., & Scozzi, B. (2008). Investigating corporate

social responsibility in supply chains: a SME perspective. Journal of Cleaner

Production , 16 (15), 1579–1588.

Cooper, W. W., Seiford, L. M., & Tone, K. (2006). Introduction to Data

Envelopment Analysis and its uses with DEA-Solver Software and References. New

York: Springer.

Cooper, W. W., Seiford, L. M., & Zhu, J. (2004). Handbook on Data

Envelopment Analysis. Springer.

Deif, A. M. (2011). A system model for green manufacturing. Journal of

Cleaner Production , 19 (14), 1553–1559.

Dobos, I., & Vörösmarty, G. (2014). Green supplier selection and evaluation

using DEA-type composite indicators. International Journal of Production

Economics , 157, 273-278.

EIA. (2013, December 3). eia.gov. (U. E. Administration, Producer) Retrieved

October 12, 2014, from http://www.eia.gov/todayinenergy/detail.cfm?id=14011

Ferrier, G. D. (1990). Measuring cost efficiency in banking: Econometric and

linear programming evidence. Journal of Econometrics , 46 (1-2), 229-245.

Page 78: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

68

GIZ. (2013). GIZ Company Report 2013. Retrieved April 21, 2015, from

https://www.giz.de/de/downloads/giz2014-en-giz-companyreport.pdf

GIZ. (n.d.). Greening Supply Chains in the Thai Auto and Automotive Parts

Industries (EU Project). Retrieved October 10, 2014, from http://www.thai-german-

cooperation.info/eu-green-auto.html

GIZ. (2015). Thailand. Retrieved January 11, 2015, from

http://www.giz.de/de/weltweit/354.html

Goverment Public Relations Department. (2014). thailand.prd.go.th. Retrieved

April 18, 2015, from http://thailand.prd.go.th/mobile_detail.php?cid=4&nid=1373

Hababou, M. (n.d.). DEA Tutorial. Retrieved February 22, 2015, from

http://www.ryerson.ca/~mhababou/DEAtutorial/deatutorial.html

Hervani, A., & Helms, M. M. (2005). Performance measurement for green

supply chain management. Benchmarking: An international Journal , 12 (4), 330-353.

Houshyar, E., Azadi, H., Almassi, M., Javad Sheikh Davoodi, M., & Witlox,

F. (2012). Sustainable and efficient energy consumption of corn production in

Southwest Iran: Combination of multi-fuzzy and DEA modeling. Energy , 44 (1),

672-681.

International Energy Ageny. (2011). 25 Energy Efficiency Policy

Recommendations. Retrieved March 21, 2015, from

http://www.iea.org/publications/freepublications/publication/25recom_2011.pdf

Kim, S. E., Kima, H., & Chae, Y. (2014). A new approach to measuring green

growth: Application to the OECD and Korea. Futures , 63, 37–48.

Page 79: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

69

Kogg, B., & Mont, O. (2012). Environmental and social responsibility in

supply chains: The practise of choice and inter-organisational management.

Ecological Economics , 83, 154-163.

Lee, V.-H., Ooi, K.-B., Yee-Loong Chong, A., & Seow, C. (2014). Creating

technological innovation via green supply chain management: An empirical analysis.

Expert System with Applications , 41 (16), 6983–6994.

Liu, W.-B., Wongchai, A., & Peng, K.-C. (n.d.). Adopting Super‐efficiency

and Tobit Model on Analyzing the Efficiency of Teacher’s Colleges in Thailand.

National Pingtung University of Science and Technology, Taiwan R.O.C, Department

of Tropical Agriculture and International Cooperation.

Lu, J., Ashuri, B., & Shahandashti, M. (2014). A Data Envelopment Analysis

Model for Building Energy Efficiency Benchmarking. Construction Research

Congress 2014 , 1073-1082.

Lucerne University of Applied Sciences and Arts. (n.d.). hslu.ch. Retrieved

March 28, 2015, from http://www.empirical-methods.hslu.ch/h-einfache-

regression.htm

Ministry of Energy. (n.d.). Thailand 20-Year Energy Efficiency Development

Plan (2011 - 2030). Retrieved April 4, 2015, from enconfund.go.th:

http://www.enconfund.go.th/pdf/index/EEDP_Eng.pdf

Mohd Yusuff, R., & Panjehfouladgaran, H. (n.d.). INNOVATIONS IN

MANUFACTURING FOR SUSTAINABLE GROWTH. Universiti Putra Malaysia,,

Department of Mechanical and Manufacturing Engineering, Faculty of Engineering.

Serdang, Selangor, Malaysia: Proceeding, International Seminar on Industrial

Engineering and Management.

Page 80: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

70

Murillo-Zamorano, L. R., & Vega-Cervera, J. A. (2001). The use of

parametric and non-parametric frontier methods to measure the productive efficiency

in the industrial sector: A comparative study. International Journal of Production

Economics , 69 (3), 265-275.

Myers, M. B., Borghesi, A., & Russo, I. (2006). Assessing the global

environment. In T. Mentzner, M. B. Myers, & T. P. Stank, Handbook of Global

Supply Chain Management (pp. 39-49). Thousand Oaks: Sage Publications.

OECD. (2011). Thailand: Key Issues and Policies, OECD Studies on SMEs

and Entrepreneurship. http://dx.doi.org/10.1787/9789264121775-en.

OSMEP. (2011). Office of Small and Medium Enterprises Promotion.

Retrieved October 10, 2014, from http://eng.sme.go.th/Pages/home.aspx

Ozcan, Y. (2008). Health Care Benchmarking and Performance Evaluation -

An Assessment using Data Envelopment Analysis (DEA). Springer.

Paradi, J. C., & Zhu, H. (2013). A survey on bank branch efficiency and

performance research with data envelopment analysis. Elevier - Omega , 41 (1), 61-

79.

Pasupathy, S. K. (2002). Modeling Undesirable Outputs in Data Envelopment

Analysis: Various Approaches. Virginia Polytech Institute and State University.

Powers, J., & McMullen, P. R. (2000). USING DATA ENVELOPMENT

ANALYSIS TO SELECT EFFICIENT LARGE MARKET CAP SECURITIES.

Journal of Business and Management , 7 (2), 31-42.

Reinhard, S., Lovell, C. K., & Thijssen, G. J. (2000). Environmental efficiency

with multiple environmentally detrimental variables; estimated with SFA and DEA .

121 (2), 287-303.

Page 81: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

71

Sawang, S., Unsworth, K., & Sorbello, T. (2007). AN EXPLORATORY

STUDY OF INNOVATION EFFECTIVENESS MEASUREMENT IN

AUSTRALIAN AND THAI SMEs. International Journal of Organisational

Behaviour , 12 (1), 110-125.

Seiford M., L., & Zhu, J. (2002). Modeling undesirable factors in efficiency

evaluation. European Journal of Operational Research , 142, 16-20.

Severo, E. A., Cesar Ferro de Guimarãesa, J., Charles Henri Dorionb, E., &

Hermann Nodari, C. (2015). Cleaner production, environmental sustainability and

organizational performance: an empirical study in the Brazilian Metal-Mechanic

industry. Journal of Cleaner Production , 96, 118-125.

Sherman, H. D., & Zhu, J. (2006). Service Productivity Management:

Improving Service Performance using DATA ENVELOPMENT ANALYSIS (DEA).

New York: Springer.

Srivihok, A., & Intrapairote, A. (n.d.). MEASURING INTELLECTUAL

CAPITAL: WEB SITES ANALYSIS OF THAI SMEs. Retrieved April 18, 2015, from

http://www2.warwick.ac.uk/fac/soc/wbs/conf/olkc/archive/oklc5/papers/k-

4_srivihok.pdf

Steering Committee for the Review of Commonwealth/State Service

Provision. (1997). Data Envelopment Analysis: A technique for measuring the

efficiency of government service delivery. AGPS, Canberra.

Stroobants, J., & Bouckaert, G. (2014). Benchmarking local public libraries

using non-parametric frontier methods: A case study of Flanders. Elsevier - Library &

Information Science Research , 36 (3-4), 211-224.

Page 82: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

72

Sueyoshi, T., & Goto, M. (2014). Environmental assessment for corporate

sustainability by resource utilization and technology innovation: DEA radial

measurement on Japanese industrial sectors. Energy Economics , 46, 295–307.

Sueyoshi, T., & Wang, D. (2014). Sustainability Development for Supply

Chain Management in U.S. Petroleum Industry by DEA Environmental Assessment.

Energy Economics , 46, 360-374.

Thailand Automotive Institute. (2012, December). MASTER PLAN FOR

AUTOMOTIVE INDUSTRY 2012 - 2016. Retrieved October 31, 2014, from

http://www.thaiauto.or.th/2012/research/research-detail.asp?rsh_id=39

The World Factbook. (2012). cia.gov. Retrieved October 11, 2014, from

https://www.cia.gov/library/publications/the-world-factbook/geos/th.html

Tikul, N. (2014). Assessing environmental impact of small and medium

ceramic tile manufacturing enterprises in Thailand. Journal of Manufacturing Systems

, 33 (1), 1-6.

Tseng, M.-L., Fung Chiuc, S., R. Tane, R., & Siriban-Manalang, A. B. (2013).

Sustainable consumption and production for Asia: sustainability through green design

and practice. Journal of Cleaner Production , 40, 1-5.

Udoncy Olugu, E., Yew Wong, K., & Mohamed Shaharoun, A. (2011).

Development of key performance measures for the automobile green supply chain.

Resources, Conservation and Recycling , 55 (6), 567–579.

Uni Hamburg. (n.d.). z-Transformation von Messwerten. Retrieved March 28,

2015, from http://www2.jura.uni-

hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Lehre/StatIKrim/Z-

Werte.PDF

Page 83: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

73

Wajer, B. H., Helgerud, H. E., & Lackner, P. (2007). Benchmarking and

energy management schemes in SMEs. Retrieved April 4, 2015, from

http://www.eceee.org/library/conference_proceedings/eceee_Summer_Studies/2007/P

anel_7/7.370/paper

Wang, X., Stern, R., Limaye, D., Mostert, W., & Zhang, Y. (2013,

September). worldbank.org. Retrieved October 11, 2014, from

http://elibrary.worldbank.org/doi/book/10.1596/978-1-4648-0020-7

Wu, T., Fowler, J., Callarman, T., & Moorehead, A. (n.d.). Multi-stage DEA

as a Measurement of Progress in Environmentally Benign Manufacturing. Retrieved

February 24, 2015, from

http://www.ceibs.edu/knowledge/papers/images/20061023/4570.pdf

Zhu, J. (2009). Quantitative Models for Performance Evaluation and

Benchmarking. New York: Springer.

Zhu, J., & Cook, W. D. (2007). Modeling Data Irregularities and Structural

Complexities in Data Envelopment Analysis. New York: Springer.

Page 84: Assessing greening activities of small and medium ...ethesisarchive.library.tu.ac.th/thesis/2014/TU... · Thai SMEs in the automotive sector and benchmarking them is more representative

74

Appendix

Bat Raw Data

Table A.1: BAT Raw Data

Peer group 1 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in kg

DMU1 5,207,000 18,745,200 1,272,231.92 DMU2 4,749,080 17,096,688 2,605,490.39 DMU3 42,652,553 159,727,480.8 20,841,472.38

Peer group 2 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in pieces

DMU1 2,094,720 2,109,072 3,220,290 DMU2 3,085,000 15,498,914.88 6,735,169 DMU3 953,620 3,433,032 334,613 DMU4 1,233,000 4,438,800 180,700 DMU5 345,244 1,242,878.4 26,843,769 DMU6 3,694,720 13,300,992 1,615,290 DMU7 910,712 3,278,563 47,733 DMU8 124,292 447,451 220,481 DMU9 595,000 2,142,000 4,533,638

Peer group 3 DMUs Electrical Energy

(kw/year) Gross Energy

(MJ/year) Products in tons

DMU1 10,160,000 40,457,774 33,018.46 DMU2 852,900 5,984,742.45 38,585,072.56 DMU3 21,318,720 281,286,259 18,402.4 DMU4 24,782,058 157,335,545.55 8,268.6