Residential Location Choice Behavior in Mandalay

THESIS

RESIDENTIAL LOCATION CHOICE BEHAVIOR IN MANDALAY

THEINT HTET HTET AUNG

GRADUATE SCHOOL, KASETSART UNIVERSITY Academic Year 2019

THESIS APPROVAL GRADUATE SCHOOL, KASETSART UNIVERSITY

DEGREE: Master of Engineering (Civil Engineering) MAJOR FIELD: Civil Engineering DEPARTMENT: Civil Engineering

TITLE: Residential Location Choice Behavior in Mandalay

NAME: MISS THEINT HTET HTET AUNG

THIS THESIS HAS BEEN ACCEPTED BY

THESIS ADVISOR (Associate Professor Varameth Vichiensan, Ph.D.)

DEPARTMENT HEAD (Associate Professor Suphawut Malaikrisanachalee, Ph.D.)

APPROVED BY THE GRADUATE SCHOOL ON

DEAN (Associate Professor Srijidtra Charoenlarpnopparut, Ph.D.)

THESIS

RESIDENTIAL LOCATION CHOICE BEHAVIOR IN MANDALAY

THEINT HTET HTET AUNG

A Thesis Submitted in Partial Fulfillment of the Requirements for the Degree of Master of Engineering (Civil Engineering) Graduate School, Kasetsart University Academic Year 2019

ABSTRACT THEINT HTET HTET AUNG : Residential Location Choice Behavior in Mandalay. Master of Engineering (Civil Engineering), Major Field: Civil Engineering, Department of Civil Engineering. Thesis Advisor: Associate Professor Varameth Vichiensan, Ph.D. Academic Year 2019

Mandalay, as a major commercial and industrial hub of Myanmar is located in the middle of Myanmar, which is about 390 km north of Nay Pyi Taw and about 700 km north of Yangon. Based on 2013 population figures provided by the Department of Immigration, there are an estimated 240,000 households in Mandalay City with an average household size of 5.25 persons per household. The second biggest city in Myanmar and, besides being a religious and cultural centre, is now facing great challenges due to a substantial increase in automobile, traffic volume, air pollution, and urban sprawl. The rapid urban growth can be seen by the high-density housing development in many parts of the city. However, the current development of high-density housing that is going on in the city center may not best match with people’s preference, on the contrary it will even make the problem more severe resulting in traffic congestion and accelerating the urban sprawl.

The main purpose of this study is to examine the factors that have influence on residential preference, i.e., attractiveness of attributes with respect to various groups of people in Mandalay and this study was focused on the people who live in six townships of Mandalay. A multinomial logit model was developed based on the SP survey data. Although the discrete choice model has a long history of application in the economic, transportation, marketing and geography fields, it is not well developed in location analysis. The results supported that the existence of factors which differ in their housing choice preference and reveal that people are considering factors not only house size, house price, but also locational convenience in terms of commuting time, and neighborhood quality. It is also found that different socio-economic groups, i.e., ethnicity, income and number family members, exhibit different location preferences.

______/ ____ / ____ Student's signature Thesis Advisor's signature

ACKNOWLEDGEMENTS

I would like to express my deepest gratitude to my thesis advisor Associate Professor Varameth Vichiensan, Ph.D for his invaluable advice, suggestion and encouragement throughout my study. More importantly, I would like to thank him for his patience and willingness in spending his precious time to guide in doing research and writing of my thesis.

I am grateful to my thesis committee members for their useful suggestions and comments. I wish to thank Associate Professor Suphawut Malaikrisanachalee, Ph.D, the department head, for his always support and continuing to confirm that everything would be fine. I am also indebted to all of my teachers for contributing transportation knowledge and ideas. I would like to thank to all the office staffs.

I would like to express the special thank to Kasetsart University Scholarships for ASEAN for Commemoration of the 60th Birthday Anniversary of Professor Dr. Her Royal Highness Princess Chulabhorn Mahidol for providing scholarship and thesis grants to undertake the study. I would sincerely like to thank my country to give a chance to pursue my Master Degree study in Thailand. In addition, I would like to thank my friends that helped me for the data collection and gave me all the support I needed.

I also wish to acknowledge all of my colleagues in the Transport group at Kasetsart University for a pleasant working environment. My friends, Wan and Deelh, thank you for your advice and ideas and for always taking care of me.

Finally, I would like to thank my lovely parents for their unending support and encouragement toward my success. I also wish to thank my two younger sisters for supporting my decision to study at Kasetsart University.

THEINT HTET HTET AUNG

TABLE OF CONTENTS

Page

ABSTRACT ...... C

ACKNOWLEDGEMENTS ...... D

TABLE OF CONTENTS ...... E

LIST OF TABLES ...... H

LIST OF FIGURES ...... J

INTRODUCTION ...... 1

OBJECTIVES ...... 5

STUDY AREA ...... 6

LITERATURE REVIEW ...... 7

1. Mandalay ...... 7

1.1 Socio-economic Overview of Mandalay City ...... 7

1.2 Population, Employment and Ethnic groups ...... 8

1.3 History of the Mandalay City ...... 9

1.4 Urban and Housing Situation ...... 11

1.4.1 Central Urban Area ...... 12

1.4.2 Surrounding Area of the Old Downtown ...... 12

1.4.3 New Housing and Business Development Areas ...... 13

1.5 Land Use ...... 17

1.6 Transport in Mandalay ...... 22

1.6.1 Road and traffic problems ...... 23

1.7 Bus Service in Mandalay ...... 23

1.7.1 Use of Transport by Purpose ...... 24

1.7.2 Public Bus Transportation Supply...... 26

1.7.3 Public Transport Institutional and Regulatory Structure ...... 29

2. Residential Location Choice Model ...... 31

2.1 Factors Influencing in Residential Location Choice ...... 31

3. Discrete Choice Model ...... 39

3.1 Multinomial Logit Model ...... 39

3.2 Goodness of Fit ...... 41

RESEARCH METHODOLOGY ...... 45

1. Questionnaire Survey Design ...... 46

1.1 Stated Preference Experiment ...... 47

1.2 Orthogonal Design ...... 50

2. Data Collection ...... 54

2.1 Data codes for analysis ...... 55

2.2 Variables used in the model ...... 57

2.3 Relationship between each independent variables ...... 58

3. Data Preparation in the NLOGIT software ...... 62

RESULTS AND DISCUSSION ...... 63

1. Characteristics of the Respondents ...... 63

1.1 Face-to-face survey respondents’ characteristics ...... 63

1.1.1 Household income and Personal income ...... 65

1.1.2 Respondents residence and work/school location ...... 66

1.1.3 Respondents vehicle ownership and daily vehicle usage ...... 67

1.2 Online survey respondents’ characteristics ...... 68

1.2.1 Household income and Personal income ...... 70

1.2.2 Respondents residence and work/school location ...... 71

1.2.3 Respondents vehicle ownership and daily vehicle usage ...... 72

2. Attitudes of the Respondents in considering house location ...... 74

3. Results of all sample Face-to-face and online interview ...... 77

3.1 Burmese ethnic and Chinese or other ethnic groups result of Face-to-face interview ...... 79

3.2 High-income and Low-income groups result of Face-to-face interview.. 81

3.3 Big family and Small family groups result of Face-to-face interview ..... 84

CONCLUSION ...... 87

LITERATURE CITED ...... 89

APPENDIX ...... 92

Appendix Questionnaire Survey Design with Google Form ...... 93 CURRICULUM VITAE ...... 108

LIST OF TABLES

Page

Table 1 Population, percentage of urban population in the ASEAN countries in 2011...... 2

Table 2 Population of Mandalay District by Township in 2014...... 7

Table 3 Population in each township of Mandalay City...... 8

Table 4 Employment in each township of Mandalay City...... 8

Table 5 Current Land Use (General) of Whole Mandalay City...... 18

Table 6 Sub-categorized land use type of area and percentage...... 19

Table 7 Estimated Trip Rate of Mandalay and Actual Trip Rate of Yangon...... 22

Table 8 Vehicle Fleet and Modal Share...... 24

Table 9 Main Modes of Travel for Different Trip Purposes...... 24

Table 10 Comparison of past studies...... 36

Table 11 Summary of the experiment...... 48

Table 12 Data in questionnaire...... 55

Table 13 Variables used in study ...... 57

Table 14 Correlations in three alternatives of Face-To-Face survey...... 58

Table 15 Correlations in two alternatives of Face-To-Face survey...... 59

Table 16 Correlations in three alternatives of Online survey...... 60

Table 17 Correlations in two alternatives of Online survey...... 61

Table 18 Descriptive statistics of the Respondents...... 63

Table 19 Descriptive statistics of the Respondents...... 69 Table 20 Results comparison of all sample in Sub-game A and Sub-game B of Face- to-Face and Online survey...... 78 Table 21 Burmese ethnic and Chinese group result of Face-to-face interview in Sub- game A ...... 79 Table 22 Burmese ethnic and Chinese group result of Face-to-face interview in Sub- game B ...... 79

Table 23 High-income and Low-income groups result of Face-to-face interview in Sub-game A ...... 81 Table 24 High-income and Low-income groups result of Face-to-face interview in Sub-game B ...... 82 Table 25 Big family and Small family groups result of Face-to-face interview in Sub- game A ...... 84 Table 26 Big family and Small family groups result of Face-to-face interview in Sub- game B ...... 84

LIST OF FIGURES

Page

Figure 1 Six townships of Mandalay City...... 6

Figure 2 Central Business District (CBD) area, Old-town and New-town areas...... 16

Figure 3 Current Land Use (General) of Whole Mandalay City...... 17

Figure 4 Land use percentage of Mandalay...... 18 Figure 5 Current Land Use (Detailed) of Part of Mandalay City (Mainly Urban Area)...... 19

Figure 6 Current Land Use (Detailed) of Urban Area of Mandalay (Enlargement). .. 21

Figure 7 Current transport mode share...... 25

Figure 8 Current Public bus supply...... 28

Figure 9 Research Methodology...... 45

Figure 10 Flow chart of Questionnaire survey for SP experiment...... 49

Figure 11 Neighborhoods...... 51

Figure 12 House types...... 52

Figure 13 Examples of Sub-game A and B by Google form...... 53

Figure 14 Computer-assisted face-to-face interview survey...... 54

Figure 15 Data preparation in NLOGIT program...... 62

Figure 16 Household income and personal income of the respondents...... 66

Figure 17 Respondents residence and work/school location...... 67

Figure 18 Vehicle wonership and daily usage of the respondents...... 68

Figure 19 Household income and personal income of the respondents...... 71

Figure 20 Respondents residence and work/school location...... 72

Figure 21 Vehicle ownership and daily usage...... 73

Figure 22 Attitudes of the respondents in face-to-face interview survey...... 75

Figure 23 Attitudes of the respondents in online interview survey...... 76

Figure 24 Respondents’ attitudes of ethin group in face-to-face interview survey. .... 80

Figure 25 Respondents’ attitudes of income group in face-to-face interview survey. 83

Figure 26 Respondents’ attitudes of family size group in face-to-face interview survey...... 85 1

INTRODUCTION

We live in a rapidly urbanizing world. Today, more than half of the world’s population are urban dwellers and this number is expected to reach 70 percent by 2050. Conventional urban planning practice has yielded haphazard and unplanned growth, and its ability to face future urban problems has been questioned by many academics, urban leaders and organizations, and planning professionals. Cities around the world vary in their approaches to accommodating their ever-growing populations while also limiting sprawl. These approaches all share one distinct feature: densification (Hack, Fishman, Yaro, Brenner, & Wachsmuth, 2012). Increasing density appears in all the discussions of sustainable urban form, such as infill development and suburban retrofitting (Williamson, 2013), urban growth boundaries. However, increased density is not a solution in and of itself.

The generalization of preference research results across multiple segments of the population whether based on ethnicity, income, or other socio-demographic variables is a common practice. It helps researchers aggregate research results in a meaningful way to understand a phenomenon and communicate these results with policymakers. A common aggregation approach used in housing research is based on individual life cycle. This approach, however, has been criticized for being linearly determined and for its inaccurate generalization of housing needs for particular segments of the population. Globally, more people move to live in urban areas than in rural areas, with 54 percent of the world’s population residing in urban areas in 2014. In 1950, 30 percent of the world’s population was urban and by 2050, 66 per cent of the world’s population is projected to be urban.

The level of urbanization of ASEAN countries ranks between 0% and 40% and this was the lowest level of urbanization. Thailand also has this low level of urbanization. The total population in the 10 ASEAN countries is over 590 million in 2010, 6.9 billion of world population (Kim, 2018). In Southeast Asian countries such as Thailand, Indonesia and Philippines developed after the 1950s by that time the 2

urbanization of Myanmar also started after that the amount of FDI and GDP largely increased compared to the former period. Along with economic growth, the average annual growth rate of the urban population increased. In Myanmar the remarkable changes to urban development in the two largest cities Yangon and Mandalay started after the market economic liberalization policies of 1988. However, like other countries, Myanmar still has a predominately rural Population of around 70% in Table 1. The urbanization level of Myanmar is mostly the same to Thailand. Singapore ranks the first within the world urbanization level.

Myanmar, often referred to as Burma, is a nation in South East Asia. The country has an estimated population of 5.15 million people. Myanmar is a multi- ethnic country with at least 135 distinct ethnic groups recognized by the government. Bamar, Burmese people are the predominant ethnic group at 68% of the Country population. The other ethnic minorities make up the remaining 32% of the population. Some of the ethnic minorities include Shan 9%, Rakhine 4%, Karen 7%, Mon 2%, Chinese 3%, Indian 2% and others 5%. The Burmese are spread in predominantly urban areas, the bigger towns and cities, and the alluvial plains of the country. The ethnic minorities have settled in the hilly areas along the country’s borders. With urbanization and development, this is likely to change and intermixing of ethnic groups has already increased in larger cities like Yangon and Mandalay where people come in search of work.

Table 1 Population, percentage of urban population in the ASEAN countries in 2011.

Country Total Urban Percentage Largest cities Second Population Popul of urban population largest cities (mill.) ation population (mill.)

Thailand 69.12 32.77 47 Bangkok Nonthaburi Myanmar 47.96 12.33 26 Nay Pyi Mandalay 3

Taw/Yangon Phillippines 93.26 40.99 44 Queson Manila Indonesia 239.871 119.7 50 Jakarta Bandung

Brunei 0.4 0.3 75 Begawan

Singapore 5.09 5.09 100 Singapore Laos 6.2 1.91 31 Vientiane Savannakhet Cambodia 14.14 24.48 17 Phnom Penh Batdambang Vietnam 87.85 27.26 31 Ho Chi Min Hanoi City Malaysia 28.4 20.27 71 Kuala Lumpur Ipoh

Total 592.291 285.1 48.14

Source: World Urbanization Prospect, 2011.

Urban landscaping is an integral part of modern urban construction and also presents the development of economic conditions. One of the most important factors of urbanization is population size. Urbanization is developed rapidly, based on rural- urban migration and natural growth of cities and towns. As urban area develops changes occur in the landscape such as buildings, roads, recreational sites. etc. Although the country’s population remains largely rural because of Myanmar economy is based on agriculture, urban population growth was faster than spatial growth. Myanmar has no large urban areas by East Asian standards but Yangon and Mandalay were medium-sized with populations of one to five million. A considerable number of urbanizing cities around the world, including Mandalay which is the second largest city of the Union of Myanmar and is located in the middle part of Myanmar, besides being a religious and cultural centre, it is a major commercial and industrial hub, are facing great challenges because of their growth has been marked by a substantial increase in auto-dependency, traffic congestion, air pollution, and urban sprawl. One strategy often suggested to reduce these negative effects is the integration of land-use and public transportation through intense residential development. The 4

work of urban planners, urban designers, architects, and policy makers centres on improving the built environment and increasing the quality of people’s lives.

However, their work entails making decisions that are not always in tandem with people’s preferences (e.g., increasing housing density, proposing a mix of land uses in residential neighbourhoods, introducing public transportation close to where people live and work) (Alsaiari, 2018). Moreover, the city has been unabatedly urbanizing over the past few years. This urban growth, however, has been characterized by a low-density sprawling pattern and also has tried many approaches to face this problem. On the other hand, high-density housing should not be developed without insight into how people will respond to such developments and, more importantly is which are the factors influencing on that development.

Mandalay, the second biggest city in Myanmar and, besides being a religious and cultural centre, it is a major commercial and industrial hub. The population of Mandalay was approximately 636,000, growing to its estimated 1.2 million (86%) in twenty-one years. Based on the data obtained from each township, the population in 2014 is estimated to be 1.22 million. Based on 2013 population figures provided by the Department of Immigration, there are an estimated 240,000 households in Mandalay City with an average household size of 5.25 persons per household. Urban expansion started the Mandalay-Yangon highway, Mandalay-Amarapura road. Urban facilities were launched with hospitals, banks, schools and other enterprises. During this period, the Mandalay urban area developed with houses in vacant places. Therefore, many people also migrate from rural areas for their better life and income. Mandalay have now changed into higher priced lands with residential buildings, commercial center, industries, service business and transportation routes with international nations due to the establishments of the new extension townships and industrial zones. In addition, the rapid urban growth can be seen by the high-density housing development in many parts of the city. However, the current development of high-density housing that is going on in the city center may not best match with people’s preference, on the contrary it will even make the problem more severe resulting in traffic congestion and accelerating the urban sprawl. 5

OBJECTIVES

1. To determine factors that influence on residential choice in Mandalay.

2. To analyse the residential preference of different socio-economic groups.

3. To examine the relationship between personal attitudes and residential preferences.

STUDY AREA

The scope of this study is to examine the factors influencing on residential location preference in Mandalay. This study was focused on the people who live in six townships of Mandalay as shown in Figure 1. In each township, original Burmese people, seven majority ethnic groups of people and Chinese people who migrated from their home country were being lived together.

Figure 1 Six townships of Mandalay City.

LITERATURE REVIEW

1. Mandalay

1.1 Socio-economic Overview of Mandalay City

Mandalay is located in the middle of Myanmar, which is about 390 km north of Nay Pyi Taw and about 700 km north of Yangon. Mandalay District, corresponding to Mandalay City Development Committee (MCDC) management area, is about 900 km2 which stretches 46 km east-west and 40 km north-south. It was founded in 1857 as the royal capital. Having the royal palace backed by 240 m high Mandalay Hill on the north, this old capital stretches its boundary south along the Ayeyarwady bank, the second biggest city in Myanmar and, besides being a religious and cultural centre, it is a major commercial and industrial hub. Most of the industries and manufacturing establishments are located in the Pyi Gyi Ta Gon Township. By contrast, Amarapura and Pa Thein Gyi are mostly rural in nature. Mandalay City, administered by MCDC, is currently composed of seven townships, which also form the Mandalay District. Five out of the seven townships are fully urbanized areas which have long constituted the city. Amarapura Township and Patheingyi Township were incorporated into the city in 2011 and 2015, respectively.

Mandalay District with a population of 1.7 million occupies 28% of Mandalay Region in terms of population size (JICA Study Team: Nippon Koei Co., 2016). About 76.4% of the district population are urban residents as shown in Table 2. Under the control of Mandalay City Development Committee, the six townships are separated into three areas: Central Business District (CBD), Old-town and New-town. Table 2 Population of Mandalay District by Township in 2014.

Population (thousand) Urban Region/ District/ Township Populatio Total Urban Rural n Mandalay Region 6,166 2,143 4,022 34.80%

Mandalay District 1,727 1,319 407 76.40% 8

Aungmyaytharzan 266 266 100% Chanayetharzan 197 197 100% Maharaungmye 241 241 100% Chanmyatharzi 284 284 100% Pyigyidagun 238 238 100% Amarapura 238 81 157 34% Patheingyi 264 13 251 4.90%

Source: (JICA Study Team: Nippon Koei Co., 2016).

1.2 Population, Employment and Ethnic groups

Table 3 Population in each township of Mandalay City.

Pop; Zone Township Total Pop; Male Female Density

1 Aungmyaytharzan 265,779 129,959 135,820 9293.0

2 Chanayetharzan 197,175 93,245 103,930 15051.5

3 Mahaaungmyay 241,113 116,903 124,210 16291.4

4 Chanmyatharzi 283,781 137,528 146,253 10999.3

5 Pyigyitagon 237,698 120,794 116,904 9285.1

6 Amarapura 237,618 114,481 123,137 1147.9

7 Patheingyi 263,725 129,004 134,721 444.0

Total 1,726,889

Source: Department (Census Report 2015).

The population and employment of Mandalay city in each township are shown as in Table 3 and Table 4.

Table 4 Employment in each township of Mandalay City. 9

Total Area Emp; Zone Township Male Female Emp; (km2) Density

1 Aungmyaytharzan 120,460 72,991 47,469 28.60 4211.9

2 Chanayetharzan 91,902 52,253 39,649 13.10 7015.4

3 Mahaaungmyay 105,161 63,842 41,319 14.80 7105.5

4 Chanmyatharzi 124,017 77,541 46,476 25.80 4806.9

5 Pyigyitagon 107,247 71,289 35,958 25.60 4189.3

6 Amarapura 108,323 63,164 45,159 207.00 523.3

7 Patheingyi 120,052 72,031 48,021 594.00 202.1

Total 777,162 908.90

Source: Department (Census Report 2015).

Myanmar’s history showed that immigration to be outcome of colonialism after 1824. Prior to 1824, no settlements of foreigners were found in the erstwhile Burma, where over a hundred ethnic groups lived with unity within the country’s borders. Population estimated reveal 3% to be Chinese and 2% Indians residing in Myanmar. The migrants gained access to the country after an administrative link came up between India and Myanmar during British Rule. A large part of the capital used for agricultural expansion in the delta region came from Indian.

Chinese immigration can also be attributed to the countries being neighbors with a 2171 km border between the two nations. Access by road and sea brought the first Chinese from Yunan and the traders from the coastal areas of Guangdong, followed later via the sea route. More Chinese traders to settle in Yangon. Mandalay, the second largest city after Yangon, has 30-40% Chinese in its population.

1.3 History of the Mandalay City

Mandalay has long history of urbanization process. It was established by King Mindon in 1857. In 1862, King Mindon designated at this time the area of Mandalay was 16. 27 sq-mi. After King Mindon, King Thibaw was ascended the throne. Thus, Mandalay was ruled by the royal capital from 1857 to 1885. Then, in 1885 the British annexation was Took over the Upper Myanmar. At the end of 1885 the British captured the entire Myanmar. From this time, Yangon became the capital of Myanmar and Mandalay became the second capital city. From this time the Mandalay was divided into four divisions as East part, West part, South part and North part and then constituted as Mandalay Municipal Committee in 1887. In 1939 the area of Mandalay increased to 25.55 square miles. After the independence from British or after 1948, Mandalay continued to be the cultural capital of Upper Myanmar. During this period the population increased rapidly but the urban area of Mandalay was not remarkable changed. The population also increased from 185,867 in 1953 to 256,541 in 1962.

From 1962-1988, urban expansion started the Mandalay-Yangon highway, Mandalay-Amarapura road. Urban facilities were launched with hospitals, banks, schools and other enterprises. During this period, the Mandalay urban area developed with houses in vacant places and the population became 417,938 according to 1973 census. In 1979 Mandalay was reformed in four townships as South-east (14 wards), South-west (11 wards), North-east (15 wards), and North-west (16 wards). The population of Mandalay reached to 532,948 in 1983. After 1988, the government started to launch the market-oriented economy as put into practice trade opportunity which attention to encourage international trade investment. It was followed the new economic policies which pulled many people reside to urban areas.

According to the economic policy, Mandalay City Development Committee was constituted to set up systematic urban management and development and urban areas were expanded and absorbed the surrounding suburban areas. After 1990, the systematic new industrial zones were established and the reallocation of inner industries replaced to these new industrial zones. In 1992, Mandalay was reformed into five townships and 86 wards by an announcement of the Ministry of Home Affairs and the urban area became 41.35 square miles (Khaing, 2015). At present the 11

total population of Mandalay is 6,145,588 or 12.0 % of the total Myanmar or the second most populated region followed Yangon in 2014.

It was also become the major trading and communications center in northern and central and is also linked with other large cities by rail and highway. Moreover, the industrial zones of Mandalay city were established due to the government policy for a market-oriented economy. Mandalay Industrial Development Committee and various governmental departments have been cooperated in the development of the industrial zones. As the local populations were employed in the industries, the job opportunities they have provided changed their living standard and economic condition. Therefore, many people also migrate from rural areas for their better life and income. According to the establishment of industrial zones, the local population in the industrial zone surrounding areas are obtain infrastructure such as electricity, better education, health care system, more efficient transport and communication system.

Mandalay have now changed into higher priced lands with residential buildings, commercial center, industries, service business and transportation routes with international nations due to the establishments of the new extension townships and industrial zones. Job opportunities due to industrial zones and other trade with international and local economic conditions, it is assumed that Mandalay urban development is pronounced in Myanmar. In Mandalay, there are many social problems concerned with urbanization. The most important negative impact is cultural impact. There are many people migrates from northern Myanmar and neighboring country due to market liberalization policy. Their cultural influenced to the local people especially young people. Some of their cultural is very different to Myanmar traditional cultural.

1.4 Urban and Housing Situation

The current urban area of Mandalay City is composed of the following different characteristic zones. Their features and issues of urban space and buildings are described as follows in Figure 2.

1.4.1 Central Urban Area

The central urban area, which is located around the palace between Nadi Canal in the east, Shwe Ta Chaung Canal in the east, and the 12th Street in the north. The area with grid road pattern extends up to around the Mandalay University and the old airport site. The old downtown area consists of commercial buildings, residential buildings, and mixed buildings. The residential district has relatively good environment and it is equipped with roads with enough width.

The commercial and business buildings, which were constructed as low buildings, are reconstructed and replaced with five to six-storey buildings. Some buildings are reconstructed to seven to ten-storey commercial buildings along the wide arterial roads such as the AH1 and 35th Street. Most of the residential buildings and mixed-use buildings in the downtown area are flat of two to three-storey buildings. Some of these buildings, which face major street with wide grid pattern roads, are being reconstructed and replaced with apartment buildings of four to six floors. The current environment of urban space is relatively good, but it is notable that the lack of public facilities such as parking space, green open space, and other facilities might arise according to progress of their densification.

1.4.2 Surrounding Area of the Old Downtown

In the surrounding area of the old downtown area, the urban area is composed of flat and two to three-storey buildings. These areas spread in the periphery of the central downtown as extension areas. The west side of the Shwe Ta Chaung Canal in the west of the city is a typical urban area of this character. Most of the buildings are wooden and are constructed in narrow housing site. The following are the issues to be improved in their living environment: 13

Housing district which is composed of narrow winding roads, High density residential area with small-scale wooden buildings, and Lack of land use balance between residential and public facilities.

1.4.3 New Housing and Business Development Areas

Residential development is ongoing in the east and south of the old airport. In the east of the old airport, residential area is underdeveloped, which is integrated with business and commercial facilities. Midrise apartment buildings with 3-10 stories and detached housings are constructed in this area. These buildings and facilities are constructed in the planned development area and their living environment is excellent.

In the southeast and south of the city, a new housing development area is continuously extended from the northern central business district (CBD) area and reaches to the surrounding areas of the industrial areas in the south. The residential area is formed by grid pattern streets and reinforced concrete (RC) structures with brick, or flat or two to three-storey wooden buildings are constructed. In some residential blocks in the southeast, five-storey apartment housings are under construction. These areas are orderly developed, but there remain issues to be improved regarding the condition of their infrastructure.

(A) Chanayetharzan Township (CBD)

(B) Maharaungmye Township (CBD)

(D) Amarapura Township (old-town)

(E) Chanmyatharzi Township (new-town)

(F) Pyigyidagun Township (new-town)

Figure 2 Central Business District (CBD) area, Old-town and New-town areas. 17

1.5 Land Use

A land use pattern was developed based on the analysis of the 2015 satellite imagery covering only the urban area, and using the 2012 imagery for other areas. Looking at the land use, built-up area shares 16.7% of Mandalay City; in contrast, it seems to dominate in the five townships. Table 5 and Figure 3 show the land use of Mandalay City and the land use percentage in Mandalay city is as shown in Figure 4.

Figure 3 Current Land Use (General) of Whole Mandalay City.

Built-up Vegetation Water

17%

78%

Figure 4 Land use percentage of Mandalay.

Source: JICA Study Team: Nippon Koei Co. (2016).

Table 5 Current Land Use (General) of Whole Mandalay City.

Type Built-up Vegetation Water Total

Area (ha) 15,165.60 71,042.80 4,746.00 90,954.30

Share 16.70% 78.10% 5.20% 100%

Source: JICA Study Team: Nippon Koei Co. (2016).

Based on the analysis of the 2015 satellite imagery, land use was sub- categorized into 15 types targeting only the urban area of Mandalay (five townships) as shown in the Figure 5 and Table 6 as detailed below.

Figure 5 Current Land Use (Detailed) of Part of Mandalay City (Mainly Urban Area).

Source: JICA Study Team: Nippon Koei Co. (2016).

Table 6 Sub-categorized land use type of area and percentage.

Type Area (ha) Share

Residential 5,716 51%

Commercial 168 2% Built-up Industrial 543 5%

Public 2,055 18%

Transit 950 8% 20

Forest 0 0%

Agricultural 470 4%

Plantation 49 0% Vegetation Green space 72 1%

U.Construct 55 0%

Others 205 2%

River 281 3%

Channel 148 1% Water Lake 338 3%

Pond 122 1%

Total 11,172 100%

Figure 6 Current Land Use (Detailed) of Urban Area of Mandalay (Enlargement).

Source: JICA Study Team: Nippon Koei Co. (2016). 22

1.6 Transport in Mandalay

In Myanmar, most vehicles are registered in Yangon and Mandalay. Moreover, the registered plate of these vehicles has not been properly updated but the vehicles are used in the city outside their registered city. Therefore, it is often difficult to make a correlation between the number of registered cars and actual traffic volume. In Mandalay, the number of registered vehicles correlates with the number of vehicles since the majority of vehicles (89%) are properly registered in Mandalay. The vehicle registration system in Myanmar shall be improved at a national level. Accordingly, the estimated modal rate in Table 7 is considered to be appropriately representing the trend of the mechanized trips. The notable difference of mechanized transportation mode between the two cities is that the citizens of Mandalay and Yangon prefer using motorcycle (trip rate - 88%) and buses (72%), respectively.

Motorcycle ownership has exploded rapidly in Asia. In Viet Nam’s Ho Chi Minh City, vehicle ownership by household is 2.01 motorcycles, 0.76 bicycle, and 0.09 car (total 2.86), with an average household size of 3.6 persons. Ownership levels are roughly similar in Ha Noi, and the average ownership levels of the two cities shows a motorcycle ownership of 560 per 1,000 population. In 2007, the number of registered motorcycles in Denpasar, Bali was 390,000 of the total 457,000 registered vehicles. Motorcycle ownership in Taipei, China is slightly higher at 600 motorcycles per 1,000. Motorcycle ownership in Taipei, China has remained constant for the last 2 years and, as such, could be considered to have reached saturation point. By comparison, Ho Chi Minh City and Ha Noi are also close to saturation.

Table 7 Estimated Trip Rate of Mandalay and Actual Trip Rate of Yangon.

Indicators Mandalay Yangon Total number of trips 1,698,152 5,063,723 Population 1,754,000 5,100,000 Trip Rate excluding NMT 0.97 0.99 Source: Bank (2016). 23

1.6.1 Road and traffic problems

Traffic condition in Yangon and Mandalay is problematic especially in the morning and evening which are the time that many people who live outside municipality area are travelling to work and study in urban area. It causes many vehicles on the road, and then causes the traffic jam. Traffic jam problem is not only happening in the downtown area, but it also happens in the suburbs area especially industrial zones where is full of a suburban residential area. However, the relevant authorities have tried to solve the problem such as constructing more roads or abutments to resolve traffic problems in the bypass road.

In Mandalay City, the current bus system is targeting to the industrial zones and new towns. The actual total bus travel time takes longer and longer by gathering passengers and so the demand is decreasing. Similarly, the advantage of the current bus system is to carry the various things bought by the passengers. In the land-use based accessibility measurement, there is 21% good access, 22% fair access and 57% repair and no service area in Mandalay City. The highest good accessibility level is found in the kind of commercial land use and the least good accessibility level is found in the kind of the land use associated by transportation type. From land use planning approach, the type of commercial land use has the excellent public transport accessibility level in Mandalay City. By increasing the amount of land required for a given amount of development, higher road and parking requirements favour urban fringe development, where land prices are lower. As a result, automobile-oriented planning is self-fulfilling: practices to make driving more convenient make alternatives less convenient and increase automobile-oriented sprawl.

1.7 Bus Service in Mandalay

Transport in Mandalay is dominated by motorcycles. It is estimated that two- wheelers account for 92% of trips, excluding walking in Table 8. Urban bus services play a minor role. The Mandalay region has 30% of the country’s motorcycle registrations with 688,652 registered motorcycles in 2014 in the city alone (1,182,691 24

across the region). This equates to 2.12 motorcycles per household (0.16 for cars and 0.92 for bicycles), and a motorcycle ownership rate of about 400 per 1,000 population.9 The motorcycle ownership rate might appear high, but international comparisons indicate that it could still grow by 50% before saturation point is reached. Considering that the city population is likely to double by 2030, the number of motorcycles may triple.

Table 8 Vehicle Fleet and Modal Share.

Vehicle Number Transport Modal Share (%)

Motor-cycle 688,000 70.2 Bicycle 300,000 21.6 Car 54,000 5.6 Bus 450a 2.6 a refers to daily operations.

Source: Bank (2016).

1.7.1 Use of Transport by Purpose

The motorcycle dominates all trip purposes as illustrated below as shown in Figure 7 and Table 9.

Table 9 Main Modes of Travel for Different Trip Purposes.

Work Shop School Recreation

1. Motor-cycle 1. Motorcycle 1. Motorcycle 1. Motorcycle 2. Bicycle 2. Walk 2. Walk 2. Car 3. Walk 3. Bicycle 3. Bicycle 3. Bus

Source: Bank (2016). 25

Figure 7 Current transport mode share.

The dominance is greater when considering that four times as many people travel to work by motorcycle than by bicycle and twice as many go to school by motorcycle than by walking. The motorcycle is therefore extremely dominant as a transport mode for all purposes within the city with public transport playing a negligent role. Over the recent years there has been particular effort made in the world in creating an image of conventional bus transport emphasizing cleanness and comfort of buses, improving personal safety and access for senior and disabled people at bus terminals and stops. Strategies used to favour bus public transport over the private car transport include: land use planning which locates large traffic generators at sites which are capable of being well served by buses, improving of bus services and introduction of traffic restraints that make car travel more difficult.

Many cities of similar gross domestic product throughout the world have high public transport modal share, 70 to 85 percent in the region. However, in some Asian 26

cities such as those in Viet Nam and Taipei, China, motorcycles in particular compete strongly with public transport and consequently have similar mode shares to that of Mandalay. This has been attributed to different factors including (i) the historic dominance of bicycle as a transport mode; (ii) the absence of a public transport sector because of lack of market identification by the private sector or identification of need by the public sector; (iii) cheap imports of motorcycles with low-cost finance deals and low import duty; and (iv) the desire for personal mobility and the means of achieving it with limited income.

Transport influences the amount of land available for development and the spatial distribution of economic activity. In turn, this has an impact on land prices, housing affordability, business costs, productivity, and, ultimately, economic performance. Land use mix describes the heterogeneity of land uses in geographically defined areas. Measures of land use mix typically include residential, commercial, institutional, industrial, recreational, and agricultural uses. Authors highlight concerns regarding how the concept of land use mix is operationalized and determined, as descriptions often lack specificity, and geographical scale often varies across studies (see Duncan et al, 2010). With regard to the impact on the urban transport, a more heterogeneous mix of land uses is thought to enable easier access to services and facilities, and employment, and to facilitate greater opportunities for active travel, thereby increasing physical activity.

1.7.2 Public Bus Transportation Supply

While the city has a train line and a central train station, the train line plays no urban transport function and, due to its slow and uncompetitive run times, has a limited regional or national function. Consequently, the only means of public transport is via the 57 bus routes that are plied by a fleet of 817 buses. These are made up of 18 city buses (90,000 passengers per month). 383 Dyna (759,600 passengers per month). 15 Hilux (24,300 passengers per month), and 27

401 light trucks (686,700 passengers per month).

Bus routes cover the vast majority of the city. Buses carry approximately 55,020 passengers per day. They operate from 6 a.m. to 6 p.m. and generally make three trips per day. The average route length is 22 kilometers (km) with a range of 16– 35 km. Out of the 817 buses, only 450 are operating daily approximately 3 trips and 120 passengers per day (40 passengers per trip). This is extremely low utilization. Buses are generally overcrowded (typically 25–30 people in Hilux, Dyna, or light trucks), with passengers standing on the steps at the back of the bus and often on the roof. The low utilization is because the number of bus trips is reduced, and the majority of passenger trips (up to 75%) are reported to be from terminus to terminus.

Buses are operated by four branches, which are nongovernment organizations (NGOs) that coordinate individual bus owners or operators. The branches are not geographic and can operate throughout the city. Most public transport vehicles use diesel with some petrol, 163 of which use compressed natural gas (161 Dyna and 2 Hilux). The current use of bus types are as shown in Figure 8.

Figure 8 Current Public bus supply.

1.7.3 Public Transport Institutional and Regulatory Structure

A multilayered bus operation system is in place in Mandalay. It consists of the following: Level 1: Bus owner or operators. Some of them may own several buses. Level 2: Bus associations. They represent groups of individual owners. Level 3: Bus Supervisory Committee. The Bus Supervisory Committee (BSC) is a single body that controls and manages bus operations throughout the city. The BSC is elected from association members and headed by the Regional Minister of Transport. Level 4: Branches. The BSC organizes operations (and associations) into branches. There are four bus branches plus one for taxis and one for veteran operators (who own or operate three vehicles only).

Each branch represents a group of townships (terminals for buses). Buses in Mandalay are owned and operated by private entities. Bus owners must register their vehicles to ensure that they are fit for operation. The Motor Vehicle Law of 1964 requires the registration of vehicles and a transport business license. Registration and licensing is granted by the local offices of the Ministry of Transport and Communications, which act under the supervision of the Regional Ministry of Transport. Bus owners must then apply to operate a bus route to an association, which in turn submits it to the appropriate branch, which forwards it to the BSC for approval.

An application to operate might include the following, details of the vehicle(s) including the registration documents, route proposed, including trip length, terminals, and trip time, days of operation, an organization chart of the bus operation and map showing bus stops. The following are origin points for bus services, with the vast majority terminating in the city center: A - Myayenandar (New Satellite Town) B - Patheingyi C - Industrial Zone D - Yadanarpon (New Satellite Town) 30

E - Amarapura F - Myitnge G - Mandalay Hill H - Tharyargone I - Kywenaphar J - Aungpinlae (New Satellite Town) K – Tampawaddy

Bus fares are distance-based with a minimum fare of MK150 ($0.15) and maximum of MK400 ($0.40). Fares are set by the regional government and the road and local navigation transport organization. By international standards, these fares are within the normal range for a low-income country.

In most cities, fares are distance-based, adopting the principle of correlating cost of provision with payment. The level of payment levied is dependent on many factors including affordability and transport and social policy. A subsidy is required when the mandated fare produces revenue less than the operating cost. If public transport is treated as a public good, then fares should be equitable, i.e., affordable to those without travel choices. If public transport is considered an environmentally and sustainable alternative to private transport, then fares should be set at a level that attracts those that might otherwise be enticed by private transport. Concessions for the most vulnerable travelers in the community is a key consideration. It is not known how bus fares are set in Mandalay. A report commissioned by GTZ demonstrated that there is a general correlation between gross domestic product per capita and urban transport prices. Considering an income-based classification of countries, the average price for normal nonsubsidized fare for a 5-kilometer bus ride was $0.20 in low- income countries, $0.40 in middle-income countries, and $1.60 for European countries.

Fares are collected on the bus usually by the conductor and sometimes by the driver, and handed over to the bus operator who then must allocate fares according to the following percentages: 31

BSC salaries - 30% Investment by BSC - 30% Office management - 15% Bonus for branch associations - 10% Bonus for BSC members and employees - 5% Gratuities - 5% Donation for social purposes - 5%

In addition, the branch and the BSC share MK500 ($0.50) per bus per day. The BSC monitors bus arrivals and departures and ensures that the conditions of the licenses are met (Bank, 2016).

2. Residential Location Choice Model

2.1 Factors Influencing in Residential Location Choice

The generalization of preference research results across multiple segments of the population whether based on ethnicity, income, or other socio-demographic variables is a common practice. It helps researchers aggregate research results in a meaningful way to understand a phenomenon and communicate these results with policymakers. A common aggregation approach used in housing research is based on individual life cycle. This approach, however, has been criticized for being linearly determined and for its inaccurate generalization of housing needs for particular segments of the population. People have different preferences, and these preferences are shaped by many factors, including one’s stage in life, aspirations, and lifestyle. From this standpoint, a stream of residential preference research has focused on several clustering methodologies, often referred to as psychographics (Beamish, Carucci Goss, & Emmel, 2001), to describe lifestyle as a determinant of housing-type and locational choices. However, the lifestyle classification method is also not immune to criticism.

Jansen, Coolen, and Goetgeluk (2011) provided an overview of housing research using lifestyle as the classification method and provides 20 lifestyle typologies used in housing research. After reviewing recent literature, Jansen concludes by questioning the suitability of this criterion because it changes more frequently than the built environment and because lifestyle explained only 5 percent of the variance over and above socio-economic variables in the studies she reviewed.

In Canada, the city of Edmonton has conducted a stated preference survey to consider trade-offs involving a wide range of elements of urban form and transportation, including mobility, air quality, traffic noise, treatment of neighbourhood streets, development densities and funding sources such as taxes (Hunt, 2010). Besides socio-demographics, recent studies have shown that ‘subjective’ (or soft) factors, such as attitudes and environmental awareness, greatly influence residential location decisions (Olaru, Smith, & Taplin, 2011). Although price or affordability is one of the most important factors in determining residential locations, this factor is mostly absent in the previous studies using stated-preference questionnaires (Cao, 2008).

However, Understanding the preference of resident in choosing the house location and what really make people to locate in a certain location is needed. Therefore, to indicate the influences of different attributes for specific groups households were established by estimating standard logit modes for those households using the observations obtained in the survey. The resulting parameters estimates for the logit model indicate the influences of the attributes.

Vichiensan (2010) found that out of the attributes of household and transportation factors considered, price and brand have greatest impacts on residential attractiveness for the typical household. The objective of that study is to examine factors influencing choice of residential location, which trades-off between housing attributes and accessibility. The study area is MRT Purple line in Bangkok. As an analysis tool, a discrete choice model, namely, Logit model, was developed. Stated preference survey on residential location and travel mode choice is conducted. The 33

1,200 respondents living in the study area was randomly chosen for face-to-face interview. At the beginning, each respondent was asked to describe their personal and household information such as gender, age, level of education, number of children, household size, monthly income and etc. By following the concept of stated preference approach, the hypothetical alternatives were made up and shown on an advertisement “House Buying Game” plate. It was found that perceptions on price, size, design, convenience to work, and proximity to the MRT station and such information is important suggesting appropriate policies that promote transit use such as transit-oriented development.

Olaru et al. (2011) proposed that lower income households were constrained to locate in areas farther out of the city, with lower travel/car and housing costs but somewhat longer travel times. Preferences for bigger houses and proximity to ‘‘everything’’ occur among higher income and/or bigger families with school-aged children in in Perth, Western Australia. This study is to estimate the relative explanatory power of key socio-demographic attributes and attitudinal factors in determining location choice by applying discrete choice model with Latent variables. The study design included three categories of attributes, identified through the literature review, which may contribute to a location decision: dwelling, facilities and quality of neighborhood, and travel. The survey was conducted before the opening of the railway corridor. The results indicate that the choice of residence reflects neighborhood and housing attributes, with significant heterogeneity in the populations of the three precincts in terms of their valuation of various housing characteristics, proximity to urban facilities, and transport.

In the paper of Liao, Farber, and Ewing (2015), families with fewer school- age children, low income and renter-occupied households are strong preference for compact development (can accept social heterogeneity and have less desire for privacy). By developing discrete choice model with Latent variables, estimate preferences for compact, walkable and transit-friendly neighborhoods through the application of a discrete choice experiment in the Wasatch Front region in Utah. Analysis of SP data is based on respondents’ choices between hypothetical residential 34

profiles. By comparing respondents’ preferences to their actual residential and travel choices in two contrasting subregions, the author further address the complex relationships between environment, preferences, residential locations and travel.

Tillema, van Wee, and Ettema (2010) stated that location-related factors such as the type of location and the number of bedrooms, turned out to be important factors and also respondents are more sensitive to travel costs (i.e. toll and fuel costs) than to equally high (monthly) housing costs as well. The author investigated that the impact of travel costs, in particular toll costs, on the residential location choice of households by using stated choice experiment among 564 respondents in the Netherlands. Choice data were analyzed by applying discrete choice models both multinomial logit (MNL) and mixed logit (ML) techniques to analyze the trade-offs respondents make between transport costs (toll and fuel costs), travel time and housing and/or location-related factors. In conclusion, toll costs are valued more negatively than fuel costs, although the differences are small.

Ibraimovic and Hess (2018) reported and compared results for two residential location choice models, the base multinomial logit model (MNL) and the Latent Class Choice Model (LCCM). The paper tested the potential of latent class choice models to examine both observed and unobserved heterogeneity in residential choices across ethnic groups. The main data-set used for the analysis is based on the Stated Preferences experiment of neighbourhood choice conducted in the Swiss city of Lugano in 2010. For analyzing the heterogeneity in preferences, a set of socioeconomic and demographic variables were collected in a previously conducted household survey on the same set of respondents. Discrete choice models are used to analyze individual choices made by decision-makers, in this case the choice between different neighbourhoods. The results indicate different ethnic attributes as key choice drivers for households belonging to three latent classes, where the origin of households is the best predictor of class membership. Swiss citizens are mainly concerned about high shares of foreigners, advantaged foreigners favour their co- nationals, while disadvantaged foreigners hold both of such preferences.

Alsaiari (2018) indicated the presence of preference heterogeneity and the emergence of four lifestyle classes that could explain and predict residential preference patterns. Additionally, the results showed a strong preference for low- density housing, even among those who favour living in a TOD; however, increasing density could be mitigated through the presence of other TOD attributes. Using a seemingly homogeneous sample of participants, the analysis incorporated three analytical methods to elicit residential preference: a multinomial logit model, a mixed logit model, and a latent class model in the case study area of the city of Riyadh.

Anwar et al. (2014) analyzed and compared the results of applying these models to a real urban case study using two datasets, 2008/09 and 2010/11 household travel survey (HTS) of Sydney Statistical Division (SSD), and also evaluates the predicted changes of mode choice probabilities based on hypothetical scenarios. Random Parameter Logit model has been chosen to analyze the data due to its some advantages. The RPL model is capable to measure random taste variation and to allow unrestricted substitution pattern and correlation among unobserved factors that help to address the limitations of initially innovated logit models, e.g. multinomial (MNL) and nested logit (NL) models. Three hypothetical scenarios are simulated to forecast the changes that would be relevant to transport policy responses.

Sener, Pendyala, and Bhat (2011) presented a modeling methodology capable of accounting for spatial correlation across choice alternatives in discrete choice modeling applications. A Generalized Spatially Correlated Logit (GSCL) was applied to the analysis of residential location choice behavior using a sample of households drawn from 2000 San Francisco Bay Area Travel Survey (BATS) data set. Model estimation results obtained from the GSCL are compared against those obtained using the standard multinomial logit (MNL) model and the spatially correlated logit (SCL) model where only correlations across neighboring (or adjacent) alternatives are accommodated. In conclusion, the GSCL model offered a rigorous approach for incorporating a continuous spatial correlation structure in discrete choice models of location choice. The comparison of some past studies is as shown in Table 10.

36 36

ior

for spatial USA Ipek N. Sener(2010) San Francisco Bay Area, San Francisco Bay choice behav modeling To presenta capable methodology of accounting correlation choice across alternatives Accommodating spatial Accommodating spatial choice correlation across in discrete alternatives models:choicean modeling to application residential location

(2017) Lugano, Tatjana Ibraimovi Switzerland preferences To examine observed and unobserved in heterogeneity residential choicesacross ethnic groups A latent class A latentclass model of residential choice behaviour and ethnic segregation

- oriented oriented - (2010) Australia Doina Olaro Doina Olaro Perth, Western Perth, Western Residential location location Residential and transit in a development new railcorridor To estimate power explanatory of key socio demographic attributes and attitudinal factors

(2010) related) - Netherlands Taede Tillema The influence of The influence (toll travel costs in residential location ofdecisions households: A stated choice approach To investigate the impact of in travel costs, tollcosts, on residential location choice

friendly friendly - (2014) Wasatch Front Wasatch Front Felix Haifeng Liao Liao Felix Haifeng region in Utah, USA region To estimate for preferences walkablecompact, and transit neighborhoods Compact and development preference in heterogeneity residential location A choice behavior: analysis Latent class

(2010) Thailand Bangkok, Comparison of past studies. of Comparison Varameth

Vichiensan 10 influencing factors in making residential location choice Preference of Preference of residential choice along MRT purple line in corridor Bangkok To examinethe

Table Table

Title Year Author, Objective Study Area

37 37

the 2000 San Travel Survey Travel Survey 702 observations 15,00 households, Francisco Bay Area Area Francisco Bay GSCL, SCL (spatial)GSCL, Household data from Ipek N. Sener(2010)

1346.5 design 0.2423 - Tatjana MNL, LCCM SP, orthogonal 133 households, Ibraimovi(2017) 1626 observations

0.63

travel zone – LCM - 1.250 1.215 (2010)

RP, SP RP, 2355.12 HLCM - 흆 households for Doina Olaro Doina Olaro 50 each 3

2591.5 (2010) 0.2603 - MNL, ML experiment Stated choice Taede Tillema 564 respondents

SP 1.046 0.251 1.071 5468.783 variables - choice sets Liao (2014) Felix Haifeng Felix Haifeng DCM with LatentDCM 1053 samples,200 1053 samples,200

face face

to - size 1200 MNL (2010) 0.0341 Varameth Vichiensan Face respondents, respondents, 2400 sample 2400 sample interview, SP

Table 10 (Continued) 10 Table

ퟐ 흆 Year

Model AIC BIC - Author,

Model fit likelihood Sample size - Survey method Log

MNL

(2010)

Ipek N. Sener 3054.06 3051.17 - - modelThe GSCL offers a rigorous for approach a incorporating continuous spatial correlation structure in discrete choice models location of choice

928.25 0.4663 - Tatjana Ibraimovi(2017) Different ethnic Different ethnic attributes are as key choice drivers for households, Swiss citizens are concerned mainly about high shares of foreigners

money

0.595

1.254 – (2010) 1.2717

2358.18 - 흆 Doina Olaro Doina Olaro housing and characteristics in budget more issues(both and time) Groups with lower Groups with fewer incomes and less vehicles are interested in -

2406.6 (2010) 0.3096 - Taede Tillema related (type of factors location, numbers bedrooms) important of are Travel time plays a less important role, location

age age -

preference occupied occupied 0.261 1.062 1.033 - Liao (2014) Felix Haifeng Felix Haifeng 5394.873 income renter households and strong are - Families fewer school with children, low

(2010) Varameth Vichiensan Price and brand have the greatest impacts on residential attractiveness

ퟐ ퟐ 흆 흆 BIC AIC

likelihood likelihood likelihood likelihood - - - -

Year Log Table 10 (Continued) 10 Table

Log Log Log Author,

Conclusion GSCL SCL LCM ML 39

Abbreviations

SP : Stated Preference RP : Revealed Preference MNL : Multinomial Logit ML : Mixed Logit DCM : Discrete Choice Model

3. Discrete Choice Model

Discrete choice models can be used to analyse and predict a decision maker’s choice of one alternative form a finite set of mutually exclusive and collectively exhaustive alternatives. Such models have numerous applications since many behavioural responses are discrete or qualitative in nature, that is, they correspond to choices of one or another of a set of alternatives. The discrete choice modelling paradigm, and in particular the logit model, have been topics of intense and active research for many years, mainly for applications in the field of transportation choice analysis. Mathematical model, the logit model represents that the behaviour of individuals trading off among the attributes of alternatives when selecting one alternative out of a set of available discrete alternatives (McFadden, 1974).

3.1 Multinomial Logit Model

The specific assumptions that lead to the multinomial logit (MNL) model are (1) the error components are extreme-value (or Gumbel) distributed (2) the error components are identically and independently distributed across alternatives (3) the error components are identically and independently distributed across observations/individuals. The most common assumption for error distributions in the statistical and modelling literature is that errors are distributed normally. The utility function that describes utility values to the new location alternatives, linear form is as below, 40

푈푖푛 = 푉푖푛+ 휀푖푛 (1) where,

푈푖푛 the utility of the alternative 푖 to the decision maker 푛,

푉푖푛 the deterministic or observable portion of the utility estimated,

휀푖푛 the error or the portion of the utility unknown to the analyst.

The mathematical structure known as the Multinomial Logit Model (MNL), which gives the choice probabilities of each alternative as a function of the systematic portion of the utility of all alternatives. The general expression for the probability of choosing an alternative ‘푖’ (푖 = 1,2, … . . , 퐽) from a set of 퐽 alternatives is,

exp (푉푖) 푃푖 = 푗 (2) ∑푗=1 exp (푉푗)

where, 푃푖 probability of the decision-maker choosing alternatives 푖 푉푗 the systematic component of the utility of alternative 푗

The mathematical form of the logit model is relatively simple and convenient

to work with when using empirical data to estimate the values for the parameters in the utility function. Consequently, this formulation is a very attractive one for modelling choice behavior and it continues to enjoy widespread use (McFadden, 1974). When values for the utility function parameters have been estimated, the relative influences of factors can be determined using ratios among the resulting coefficient values.

The significant of differences among estimates can be considered using standard t-statistics and t-ratios, with the t-ratio being the t-statistic for the estimate’s difference from 0. When t-statistic or t-ratio has a value greater than 1.96 in absolute magnitude, this indicates that there is a less than 5% chance that the associated difference is due to random effects only (ANGHS & TANG, 1975), and the difference is said to be significant. 41

3.2 Goodness of Fit

A statistic called the likelihood ration index is often used with discrete choice models to measure how well the models fit the data. Stated more precisely, the statistic measures how well the model, with its estimated parameters, performs compared with a model in which all the parameters are zero (which is usually equivalent to having no model at all). This comparison is made on the basis of the log-likelihood function, evaluated at both the estimated parameters and at zero for all parameters. The likelihood ratio index is defined as (Ben-Akiva, Lerman, & Lerman, 1985),

̂ 2 퐿퐿(훽) 휌 = 1- (3) 0 퐿퐿(0)

where,

퐿퐿(0) log-likelihood for model with zeros for all coefficients ̂ 퐿퐿(훽) log-likelihood for model with estimated coefficients

2 This 휌0 index also takes into account the number of parameters used in the model, favoring more parsimonious model specifications (Ben-Akiva et al., 1985). If the estimated parameters do no better, in terms of the likelihood function, than zero parameters (that is, if the estimated model is no better than no model), then 퐿퐿(훽̂) = 퐿퐿(0) and so 휌2= 0. This is the lowest value that 휌2 can take (since if 퐿퐿(훽̂) were less than 퐿퐿(0), then 훽̂ would not be the maximum likelihood estimate). At the other extreme, suppose the estimated model was so good that each sampled decision maker’s choice could be predicated perfectly. In this case, the likelihood function at the estimated parameters would be one, since the probability of observing the choices that were actually made is one. And, since the log of one is zero, the log-likelihood function would be zero at the estimated parameters. With 퐿퐿(훽̂) = 0, 휌2= 1. This is the highest value of 휌2 can take. In summary, the likelihood ratio index ranges from zero, when the estimated parameters are no better than zero parameters, to one, when the estimated parameters perfectly predict the choices of the sampled decision makers. 42

2 The 휌0 measures the improvement due to all elements of the model, including the fit to market shares, which is not very interesting for disaggregate analysis so it should not be used to assess models in which the sample shares are very 2 unequal. The rho-squared measure with respect to the constant only model, 휌푐 , controls for the choice proportions in the estimation sample and is therefore a better measure to use for evaluating models.

A problem with both rho-squared measures is that there are no guidelines for a good rho-squared value. Consequently, the measures are of limited value in assessing the quality of an estimated model and should be used with caution even in assessing the relative fit among alternative specifications. It is preferable to use the log- likelihood statistic (which has a formal and convenient mechanism to test among alternative model specifications) to support the selection of a preferred specification among alternative specifications.

Another problem with rho-squared measures is that they improve no matter what variable is added to the model independent of its importance. This directly results from the fact that the objective function of model is being modeled with one or more additional degrees of freedom and that the same data that is used for estimation is used to assess the goodness of fit of the model. One approach to this problem is to replace the rho-squared measure with an adjusted rho-squared measure which is designed to take account of these factors. The adjusted rho-squared for the zero model is given by,

̂ 2 [퐿퐿훽]−퐿퐿(0) 휌̅ = (4) 0 퐿퐿(∗)−퐿퐿(0)

퐿퐿훽̂ − 퐾 = 1 − 퐿퐿(0)

where 퐾 is the number of degree of freedom (parameters) used in the model.

It is important that the likelihood ratio index is not at all similar in its interpretation to the 푅2 used in regression, despite both statistics having the same range. 푅2 indicates the percentage of the variation in the dependent variable that is “explained” by the estimated model. The likelihood ratio has no intuitively interpretable meaning for values between the extremes of zero and one. It is the percentage increase in the log-likelihood function above the value taken at zero parameters (since 휌2 =1- 퐿퐿(훽̂)/ 퐿퐿(0) = 퐿퐿(0) - 퐿퐿(훽̂)) /퐿퐿(0)). However, the meaning of such a percentage increase is not clear. In comparing two models estimated on the same data and with the same set of alternatives (such that 퐿퐿(0) is the same for both models), it is usually valid to say that the model with the higher 휌2 fits the data better. But this is saying no more than that increasing the value of log- likelihood function is preferable.

The maximum likelihood estimation involves two important steps; (1) developing a joint probability density function of the observed sample, called likelihood function, and (2) estimating parameter values which maximize the likelihood function. The likelihood function for a sample ‘푇 individuals, each with ‘퐽 alternatives is defined as follows,

훿푗푡 퐿(훽) = ∏∀푡∈푇 ∏∀푗∈퐽(푃푗푡(훽)) (5)

where,

훿푗푡 = 1 is chosen indicator (=1 if 푗 is chosen by individual 푡 and 0, otherwise) and

푃푗푡 is the probability that individual 푡 chooses alternatives 푗.

The value of the parameters which maximize the likelihood function are obtained by finding the first derivative of the likelihood function and equating it to zero. Since the log of the function yields the same maximum as the function and is more convenient to differentiate, we maximize the log-likelihood function instead of the likelihood function itself. The expression for the log-likelihood function and its first derivative is shown in equations (7), 44

퐿퐿(훽) = 퐿표푔(퐿(훽)) = ∑∀푡∈푇 ∑∀푗∈퐽 훿푗푡 × ln(푃푗푡(훽)) (6)

By using NLOGIT software, the model goodness of fit can also consider with the Akaike information criterion (AIC) and the Bayesian information criterion (BIC). AIC and BIC are both penalized-likelihood criteria. AIC is an estimator of out-of- sample prediction error and thereby relative quality of statistical models for a given set of data. Given a collection of models for the data, AIC estimates the quality of each model, relative to each of the other models. Thus, AIC provides a means for model selection and the preferred model is the one with the minimum AIC value (Akaike, 1974). Then the AIC value of the model is used as following, 퐴퐼퐶 = 2푘 − 2 ln(퐿̂). where, 푘 the number of estimated parameters in the model and 퐿̂ is maximum value of the likelihood function for the model. Moreover, BIC is also a criterion for model selection among a finite set of models; the model with the lowest BIC is preferred. It is based, in part, on the likelihood function and it is closely related to the Akaike information criterion (AIC). The BIC is formally defined as the form follow, 퐵퐼퐶 = ln(푛)푘 − 2 ln(퐿̂). where, 퐿̂ the maximized value of the likelihood function of the model , 푛 is number of observations and 푘 is number of parameters estimated by the model.

RESEARCH METHODOLOGY

This research is to study and determine factors that influence on making house location decision in Mandalay. To examine the relationship between personal attitudes and house location preferences, research procedures are as shown in Figure 9.

1. Literature Review

Residential Location Choice Model Mandalay

Urban Development

2. Survey Plan

CBD areas, Old town areas, New town areas

Revise

3. Stated Preference Experiment

Choice experiment by Orthogonal Design

4. Data Collection

5. Analysis

Analyzing Socio-economic characteristics

Finding Factors Influencing on Residential location choice

6. Conclusions and Recommendations

Figure 9 Research Methodology. 46

1. Questionnaire Survey Design

The questionnaire was divided into three parts: (I) Socio-economic characteristics, (II) attitudes and behaviours of considering house location and (III) SP survey for a choice set of housing alternatives. The flow and process of questionnaire survey and SP experiment is as shown in Figure 10. In Part I, the respondents were asked to provide the following socioeconomic information such as

1. Age 2. Gender 3. Marital status 4. Employment status 5. Education 6. Household location 7. Workplace/school location 8. Travel time to do their daily activities 9. Mode usually used for travel to those places 10. Household income per month 11. Personal income per month 12. Number of people in household and numbers of children younger than 11 years old 13. Nature of household tenure (own or rent) 14. Present dwelling type and size 15. Current house price or rental fees per month 16. Number of years live in present house 17. Total number of private vehicles owned by household 18. Plan to move or relocate within 1 year 19. Reason for move or relocate to the new place

In Part II, the attitudes of house location choice considered by respondents were described in terms of the following, 1. The township location of the house 47

2. The price of the house 3. The size of the house 4. Having good prospect future value of the location 5. Living in green environment 6. Having local shops within walking distance 7. Living near to the relatives and friends 8. Low crime rate within neighborhood 9. Less traffic congestion on nearby street 10. Living closer to the main road 11. Closer to work place/school 12. Having parking space

1.1 Stated Preference Experiment

The main data-set used for the analysis is Part III and based on the Stated Preferences experiment. SP surveys are a widely used method for identifying preferences in cases where revealed preferences (RP) data are unavailable or inadequate to identify preferences (Louviere, Hensher, Swait, & Adamowicz, 2000). SP data are based on hypothetical scenarios and can control correlations between alternatives and the definition of choice sets in systematic manner consistent with the research objectives. In order to reveal preferences for the characteristics of the residential location choice, the SP experiment presented respondents with choice scenarios, a hypothetical situation where they were asked to choose one among three alternative residential situations, the first option was the respondent’s actual area of residence and the attribute values corresponded to real observed values. The second and the third house situation were represented by two hypothetical alternatives with attribute levels pivoted around the values of the current residence situation. This setting permitted respondents to recognize a familiar choice situation, thus making the choice experiment more realistic and reliable. The three residential choice situations were described by a number of different characteristics.

The experiment included variables indicating the township location, size, price, travel time to work/school and neighbourhood environment. Table 11 presents the summary details of the experiment. In each option, attribute contained the current residential location of respondents and two additional levels expressed as positive and negative percentage deviations. Besides, the example of considering better neighbourhood and bigger or smaller house size are presented in Figure 11 and Figure 12. Though orthogonal design has been used as the major experimental design method in practice, orthogonality is not that important in the nonlinear discrete choice models (Suel & Polak, 2018).

Table 11 Summary of the experiment.

Attributes Levels Location New location, Current residence location, New location Size -50%, Current house size, +50%

Price -20%, Current house price, +20%

Neighborhood environment -10%, Current environment, +10%

Travel time to work/school -50%, Current travel time, +50%

Experimental design Fractional factorial orthogonal design approach Alternatives Current residence (A) and two hypothetical alternatives (B, C)

Part I Socio-economic characteristics

Part II Attitudes and behaviors of considering house location

Part III, SP experiment

Sub-game A Choice Tasks,

i = 1 to 6 Choice B Choice C Choice A New hypothetical New hypothetical Present house house house

If not choose present house, repeat to another game Choice the present house

If choose present house, force to answer sub-game B

Sub-game B

Choice C Choice B Repeat to another sub-game New hypothetical New hypothetical house house

Figure 10 Flow chart of Questionnaire survey for SP experiment. 50

1.2 Orthogonal Design

Historically, researchers have relied on orthogonal experimental designs, in which the attributes of the experiment are statistically independent by forcing them to be orthogonal (3). In this way, orthogonal designs theoretically allow for an independent determination of each attribute’s influence upon the observed choices. There are several approaches to generate full or fractional factorial orthogonal design (4,5,6,7). While orthogonal design has long been used in practice, researchers in recent years argue the importance of orthogonality in SC data. They doubted the effectiveness when orthogonal design is used to estimate discrete choice model, not to mention whether orthogonality can be kept in reality (8,9). Orthogonality is important in linear models since it avoids multicollinearity problem and also minimize variance- covariance matrix of the estimated model. Unfortunately, discrete choice model is nonlinear. The derivation of its variance-covariance matrix is very different from the way in linear models. Therefore, keeping orthogonality of the parameters has little to do with minimizing their standard errors. Based on the fractional factorial orthogonal design, the experiment contained 6 choice tasks by randomly made as shown in Appendix. The values of attributes describing the alternatives neighborhoods varied across every choice task. The SP choice experiment was conducted by using Google form and Computer-assisted personal interviewing (CAPI), is an interviewing technique in which the respondent or interviewer uses an electronic device to answer the questions.

Moreover, there are several benefits of using google form and CAPI such as reduced time and costs for data input, elimination of errors during data transcription, Google Forms stores the feedback received so we can analyse it in detail, the forms are integrated with Google spreadsheets therefore we can access to a spreadsheet view of the collected data and so on. Each respondent was assigned to choose within 3 choices. If the respondents were selected the present house, they were forced to make a choice between the rest of two choices as presented in the following Figure 13. 51

(A) Worse Neighborhood

(B) Better Neighborhood

Figure 11 Neighborhoods. 52

(A) Smaller house

(B) Bigger house Figure 12 House types. 53

Figure 13 Examples of Sub-game A and B by Google form. 54

2. Data Collection

The questionnaire survey was carried out by computer-assisted face-to-face interview and online survey. In face-to-face interview part, the respondents were randomly chosen at Government offices, Schools, Universities and Private companies/offices by self-administered questionnaire survey. These places were located in three areas (CBD area, old town, new town) of Mandalay City as mentioned above. Survey was carried out during 24th August 2019 to 28th August 2019 and shown in Figure 14. 122 respondents were interviewed since some respondents did not want to interview survey. 283 people were answered in online survey.

Figure 14 Computer-assisted face-to-face interview survey.

2.1 Data codes for analysis

Data collection is divided into types of information, socioeconomic characteristics of the households. The nature of the detailed code of collected data for analysis is as shown in Table 12.

Table 12 Data in questionnaire.

Data Information Characteristics Code Gender Male 0 Female 1 Others 2 Age 18-25 years 1 26-45 years 2 46-60 years 3 Over 60 years 4 Marital Status Single 1 Married 2 Others (Divorced/Widowed) 3 Occupation Student 1 Government officer 2 Company staff 3 NGO/INGO officer 4 Own business 5 Others 6 Education Lower than high school 1 High school 2 Diploma/College 3 Bachelor degree 4 Higher than bachelor 5 Family members Numbers Ethnicity Burmese 1 Kachin 2 Kayar 3 Kayin 4 Chin 5 Mon 6 Rakhine 7 Shan 8 Chinese 9 Household income per month Less than 500,000 Ks 1 56

500,001 - 1,000,000 Ks 2 1,000,001 - 1,500,000 Ks 3 More than 1,500,000 Ks 4 Personal income per month Less than 100,000 Ks 1 100,001 - 200,000 Ks 2 200,001 - 300,000 Ks 3 More than 300,000 Ks 4 Have children (younger than 11 years old) Yes 0 No 1 Number of Children Numbers Number of vehicles Numbers House location Aungmyaytharzan 1 Chanayetharzan 2 Mahaaungmye 3 Chanmyatharzi 4 Amarapura 5 Pyigyidagun 6 Number of years in present house Below 1 year 1 1-5 years 2 6-10 years 3 11-20 years 4 21-30 years 5 30 years above 6 Type of present house Single house 1 Condominium/Apartment 2 Shop house 3 House status Own 1 Rent 2 Government resident 3 Plan to move or relocate within 1 year Yes 0 No 1 Work/School Location Aungmyaytharzan 1 Chanayetharzan 2 Mahaaungmye 3 Chanmyatharzi 4 Amarapura 5 Pyigyidagun 6 Type of vehicle use Walk 1 Motorcycle 2 Passenger car 3 Motorcycle Taxi 4 Bicycle 5 Bus 6 Other 7 Travel time Less than 5 minutes 1 57

5-10 minutes 2 11-20 minutes 3 21-30 minutes 4 More than 30 minutes 5

2.2 Variables used in the model

From the survey data collection, there are various variables that will be used to create the model to analyse the factors influencing on location choice behaviour. That

variables are described as Table 13 below and Y is the dependent variable and X1 –

X11 are independent variables that used in the research.

Table 13 Variables used in study

Variable Description Name

3 choices for house location present house and two new Y CHOICE hypothetical houses

X1 D_AMTZ Location of the old town area (Aungmyaytharzan Township)

X2 D_CATZ Location of the CBD area (Chanayetharzan Township)

X3 D_MHA Location of the CBD area (MahaaungmyeTownship)

X4 D_CMTZ Location of the new area (Chanmyatharzi Township)

X5 D_PGDG Location of the new town area (Pyigyidagun Township)

X6 D_AMARA Location of the old town area (Amarapura Township)

X7 SIZE Size of the house

X8 PRICE Price of the house

Travel time from the house to work/school (two new X9 T_Time hypothetical houses)

X10 D_better Better neighborhood (two new hypothetical houses)

X11 D_worse Worse neighborhood (two new hypothetical houses)

D_worse

1.00 11 X 58

pendent pendent

D_better 1.00 0.45 10 - X

T_time 9 1.00 0.21 0.29 - X

Price 8 1.00 0.12 0.04 0.12 - - X

Size 7 1.00 0.48 0.11 0.00 0.21 - X

D_AMARA 6 0.21 0.11 0.03 0.14 0.19 1.00 - - X

D_PGDG 5 0.23 0.11 0.26 0.14 0.13 1.00 0.13 - - - - X

D_CMTZ 4 0.13 0.14 0.10 0.09 0.07 1.00 0.17 0.27 ------X

Face survey. Face

D_MHA To 3 - 0.10 0.27 0.12 0.11 0.19 1.00 0.25 0.12 0.19 ------X

D_CATZ 2 0.16 0.06 0.10 0.02 0.11 1.00 0.23 0.32 0.16 0.25 - - - - - X

D_AMTZ 1 1.00 0.20 0.15 0.21 0.10 0.16 0.02 0.03 0.04 0.01 0.16 ------X

Size Price relationship relationship between each variable was analyzed by considering the correlation coefficient. If some variables are highly T_time Correlations in three alternatives ofalternatives Face in three Correlations D_better D_MHA D_worse D_worse D_CATZ D_PGDG D_CMTZ D_AMTZ

D_AMARA Relationship between each independent variables between each Relationship The 14

2.3

riables that were expected to affect on residential location on residential to preference. affect that were expected riables

4 5 6 7 8 9 2 3 10 11

related, that variables were eliminated. Table 14, Table 15, Table 16, Table 17 va are shown the correlation coefficient of inde Table

1 X X X X X X X X X X X

D_worse D_worse

59 59 1.00 11 X

D_better D_better

1.00 1.00 10 - X

T_time

9 1.00 0.21 0.21 - X

Price

8 1.00 0.22 0.14 0.14 - - X

Size

7 1.00 0.42 0.02 0.10 0.10 - - X

D_AMARA

6 0.22 1.00 0.10 0.22 0.11 0.22 - - X

D_CMTZ

0.14 1.00 0.21 0.04 0.15 0.20 0.14 - - - X

D_MHAM

3 0.14 1.00 0.14 0.22 0.46 0.13 0.21 0.14 - - - - - X Face survey. Face

To D_CATZ -

0.49 1.00 0.29 0.28 0.45 0.12 0.10 0.32 0.49

- - - - - X

D_AMTZ

0.15 1.00 0.30 0.15 0.14 0.23 0.10 0.16 0.21 0.15 1 ------X

Correlations in two alternatives of Face in two alternatives Correlations

D_AMARA Size Price T_time D_better D_worse D_AMTZ D_CATZ D_MHAM D_CMTZ 15

9 10 11 3 4 6 7 8 1 2

X X X X X X X X X X Table Table

D_worse D_worse 60

60 11 1.00 X

D_better

10 1.00 1.00 - X

T_time

9 1.00 0.22 0.22 - X

Price

8 1.00 0.19 0.12 0.12 - - X

Size

7 1.00 0.38 0.04 0.08 0.08 - - X

D_AMARA

6 0.23 0.23 1.00 0.10 0.17 0.06 - - X

D_CMTZ

4 0.14 0.14 1.00 0.20 0.07 0.11 0.18 - - - X

D_MHAM

3 0.16 0.16 1.00 0.14 0.23 0.47 0.11 0.21 - - - - - X

D_CATZ

0.50 0.50 1.00 0.32 0.28 0.46 0.15 0.06 0.35

- - - - - X

D_AMTZ

0.14 0.14 1.00 0.28 0.14 0.13 0.21 0.07 0.11 0.21 1 ------X

Correlations in three alternatives ofalternatives Online in three survey. Correlations

D_CMTZ D_AMARA Size Price T_time D_better D_worse D_AMTZ D_CATZ D_MHAM 16

8 9 10 11 3 4 6 7 1 2

X X X X X X X X X X Table

D_worse D_worse

1.00

D_better D_better

1.00 0.45 10 - X

T_time

9 1.00 0.21 0.32 - X

Price

8 1.00 0.18 0.03 0.11 - - X

Size

7 1.00 0.42 0.18 0.00 0.22 - X

D_AMARA

6 0.03 0.10 0.17 1.00 0.20 0.13 - - X

D_PGDG

5 0.29 0.08 0.14 1.00 0.15 0.22 0.10 - - - - X

D_CMTZ

4 0.04 0.00 0.12 1.00 0.17 0.25 0.03 0.05 - - - - - X

D_MHAM

3 0.17 0.09 0.18 1.00 0.22 0.13 0.19 0.20 0.17 ------X

D_CATZ

0.11 0.03 0.11 1.00 0.23 0.30 0.17 0.26 0.13 0.04

- - - - - X

D_AMTZ

0.11 0.05 0.17 1.00 0.21 0.16 0.20 0.12 0.18 0.03 0.07

------X

Correlations in two alternatives of Online in two alternatives survey. Correlations

D_PGDG D_AMARA Size Price T_time D_better D_worse D_AMTZ D_CATZ D_MHAM D_CMTZ 17

7 8 9 10 11 3 4 5 6 1 2 X X X X X X X X X X X Table Table

3. Data Preparation in the NLOGIT software

The NLOGIT software package (NLOGIT, 2007) was used to estimate the parameters in this research. NLOGIT is an extension of another very large, integrated econometrics package, LIMDEP, that is used world-wide by analysts of models for regression, discrete choice, sample selection, count data, models for panel data, etc. NLOGIT includes all of the capabilities of LIMDEP plus package of estimators for models of multinomial logit (MNL), multinomial probit (MNP), nested logit, mixed logit and several others. To analyze the factors influencing the residential location choice behavior in Mandalay, it is started by preparing the data according to various variables and independent variable groups as shown in Figure 15. After importing the data, the program will analyze the relationship between various variables and then show the summary.

Figure 15 Data preparation in NLOGIT program.

RESULTS AND DISCUSSION

1. Characteristics of the Respondents

1.1 Face-to-face survey respondents’ characteristics

The general information of the respondents is described using descriptive statistics in Table 18. The sample was mixed in terms of gender 48 male (39.3 percent), 70 female (57.4 percent) and 4 others (3.3 percent). 56 persons (45.9 percent) are young age groups and middle age groups represents 39 persons (32 percent). More than 80 percent of respondents hold bachelor degree, while 6.6 percent is master degree or more. In addition, 17.2 percent have diploma/college level education and 5.7 percent report high school and most are the students (42.6 percent). 28 persons (23 percent) are government officers. As most of the respondents are student, they lived together with their family and 44 persons (36 percent) have four family members. In each household, only 11 persons (9 percent) have the children younger than 11 years old and most have 1 child (72.7 percent). In terms of ethnicity, 72.1 percent represent the original Burmese and 14.8 percent of Chinese people. For house type and house status, 83.6 percent live in Single house type and 69.7 percent own house which is indicated that they have highly preferred in their own house and do not have any plan to move a new location.

Table 18 Descriptive statistics of the Respondents.

Variables Frequency Percentage

Gender Male 48 39.3% Female 70 57.4% Others 4 3.3% Age 18-25 years old 56 45.9% 26-45 years old 39 32% 46-60 years old 25 20.5% Over 60 years old 2 1.6% Education 64

High school 7 5.7% Diploma/College 21 17.2% Bachelor 86 70.5% Higher than Bachelor 8 6.6% Marital Status Single 81 66.4% Married 35 28.8% Other (divorced/widowed) 6 4.8% Occupation Student 52 42.6% Government officer 28 23% Company staff 11 9% NGO/INGO officer 16 13.2% Own business 12 9.8 Others 3 2.4% Family members 1 person 3 2.5% 2 people 3 2.5% 3 people 16 13.1% 4 people 44 36% 5 people 32 26.2% 6 people 11 9% More than 6 people 13 10.7% Having children (younger than 11 years old) Yes 11 9% No 111 91% Number of children 1 child 8 72.7% 2 children 1 9.1% 3 children 2 18.2% Ethnicity Burmese 88 72.1% Kachin 3 2.5% Kayin 3 2.5% Mon 1 0.8% Rakhine 1 0.8% Shan 8 6.6% Chinese 18 14.8% House-type Single-house 102 83.6% 65

Condominium/Apartment 16 13.1% Shop-house 4 3.3% House-status Rent 16 13.1% Own 85 69.7% Government residence 21 17.2%

1.1.1 Household income and Personal income

As shown in Figure 16, 38.5 percent (47 persons) and 13.9 percent (17 persons) are the middle household income groups and 26.2 percent (32 persons) represent the high household income group. In terms of personal income, 41 persons (33.6 percent) are high-income. As most of the respondents are students, 32 persons (26.2 percent) are low personal income and also in the middle-income group, there are 17 persons (13.9 percent) and 32 persons (26.2 percent) respectively.

% of Household Income

Less than 500,000 Ks 26.20% 21.30% 500,001 - 1,000,000 Ks 13.90% 38.50% 1,000,001 - 1,500,000 Ks

More than 1,500,000 Ks

% of Personal Income

Less than 100,000 Ks

26.20% 33.60% 100,001 - 200,000 Ks

13.90% 200,001 - 300,000 Ks 26.20% More than 300,000 Ks

Figure 16 Household income and personal income of the respondents.

1.1.2 Respondents residence and work/school location

The percentage of their current house location and work/school location is as shown in Figure 17. The 34 persons (27.9 percent) live in new-town area and 31 persons (25.4 percent), 27 persons (22.1 percent) live in CBD. The 49 persons (40.2 percent) do their daily activities in CBD area.

% of Respondents current house location

Aungmyaytharzan (old-town) 8.20% 15.60% 0.80% Chanayetharzan (CBD)

27.90% Mahaaungmye (CBD) 22.10% Chanmyatharzi (new-town)

25.40% Pyigyidagun (new-town)

Amarapura (old-town)

% of Respondents work/school location

4.10% Aungmyaytharzan (old-town)

0.80% Chanayetharzan (CBD) 32% 40.20% Mahaaungmye (CBD)

Chanmyatharzi (new-town) 23% Pyigyidagun (new-town)

Amarapura (old-town)

Figure 17 Respondents residence and work/school location.

1.1.3 Respondents vehicle ownership and daily vehicle usage

According to Figure 18, 32.8 percent of the respondents own more than 3 private vehicles and mostly use Motorcycle (61.5 percent). That means people prefer to use private vehicle and they did not care about the public transportation and travel time because of less traffic congestion.

% of Vehicle ownership

4.90%

12.30% No-vehicle 32.80% 1 vehicle 18.90% 2 vehicles

31.10%

% of Daily vehicle usage

4.90% 0.80% 13.90% Walk 0.80% Motorcycle 18% Passenger car Motorcycle-taxi Bicycle 61.50% Car-taxi

Figure 18 Vehicle wonership and daily usage of the respondents.

1.2 Online survey respondents’ characteristics

In the case of online survey, the sample was mixed in terms of gender 116 male (41.0 percent), 166 female (58.7 percent) and 1 other (0.4 percent). 181 persons (64 percnt) are middle age group and 67 persons (23.6 percent) represent the young age group. According to marital status, most of the respondents are single (62.8 percent) and 101 persons (35.7 percent) are married. Most of the respondents work in government, company and NGO/INGO (17.7 percent, 32.4 percent, 7.4 percent) respectively and 75 persons (26.5 percent) work their own business. The family size is also as big as face-to-face suvery, 4 family members, 5 family memebers and 6 family members (28.3 percent, 24.7 percent, 10.6 percent). 234 persons (82.7 percent) did not have the children younger than 11 years old and only 31 persons have 1 child. More than 70 percent of respondents hold bachelor degree, while 6.6 percent is master degree or more. In addition, 17.2 percent have diploma/college level education and 5.7 percent report high school. In terms of ethnicity, 67.5 percent represent the original Burmese and 14.8 percent of Chinese people. For house type and house status, 91.5 percent live in Single house type and 87.3 percent own house which is 69

indicated that they have highly preferred in their own house and do not have any plan to move a new location as shown in Table 19.

Table 19 Descriptive statistics of the Respondents.

Variables Frequency Percentage Gender Male 116 41.0% Female 166 58.7% Others 1 0.4% Age 18-25 years old 67 23.6% 26-45 years old 181 64% 46-60 years old 34 12% Over 60 years old 1 0.4% Education High school 7 5.7% Diploma/College 21 17.2% Bachelor 86 70.5% Higher than Bachelor 8 6.6% Marital Status Single 178 62.8% Married 101 35.7% Other (divorced/widowed) 4 1.5% Occupation Student 35 12.4% Government officer 50 17.7% Company staff 92 32.4% NGO/INGO officer 21 7.4% Own business 75 26.5% Others 10 3.6% Family members 1 person 8 2.8% 2 people 23 8.1% 3 people 52 18.4% 4 people 80 28.3% 5 people 70 24.7% 6 people 30 10.6% More than 6 people 20 7.1% 70

Having children (younger than 11 years old) Yes 49 17.3% No 234 82.7% Number of children 1 child 31 63.3% 2 children 16 32.7% 3 children 1 2% 4 children 1 2% Ethnicity Burmese 191 67.5% Kachin 11 3.9% Kayar 2 0.7% Kayin 11 3.9% Chin 4 1.4% Mon 6 2.1% Rakhine 5 1.8% Shan 11 3.9% Chinese 42 14.8% House-type Single-house 259 91.5% Condominium/Apartment 24 8.5% Shop-house 0 0% House-status Rent 21 7.4% Own 247 87.3% Government residence 15 5.3%

1.2.1 Household income and Personal income

In terms of household income, 92 persons (32.5 percent) and 84 persons (29.7 percent) have the middle income and only 25 persons (8.8 percent) represent the low income. 144 persons (50.9 percent) are the high-income group and most are the middle-income group as in terms of occupation (government officers and company staff). The percentage of the household and personal income of the respondents are stated in Figure 19.

% of Household Income

8.80% Less than 500,000 Ks 29% 29.70% 500,001 - 1,000,000 Ks

1,000,001 - 1,500,000 Ks 32.50%

More than 1,500,000 Ks

% of Personal Income

6% 8.50%

50.90% Less than 100,000 Ks 34.60% 100,001 - 200,000 Ks

200,001 - 300,000 Ks

More than 300,000 Ks

Figure 19 Household income and personal income of the respondents.

1.2.2 Respondents residence and work/school location

As shown in Figure 20, 28.3 percent of the respondents live in CBD area and 21.9 percent, 19.4 percent live in new town area. 48.4 percent of the people do their daily activities in CBD area.

% of Respondents current house location

Aungmyaytharzan (old-town) 8.80% 4.20%

21.90% Chanayetharzan (CBD)

17.30% Mahaaungmye (CBD)

19.40% Chanmyatharzi (new-town)

28.30% Pyigyidagun (new-town)

Amarapura (old-town)

% of Respondents work/school location

Aungmyaytharzan (old- 4.20% 11.00% town) 1.80% 8.8% Chanayetharzan (CBD)

Mahaaungmye (CBD)

25.8% Chanmyatharzi (new-town) 48.40% Pyigyidagun (new-town)

Amarapura (old-town)

Figure 20 Respondents residence and work/school location.

1.2.3 Respondents vehicle ownership and daily vehicle usage

According to Figure 21, 41.98 percent of the respondents own 2 private vehicles and mostly use Motorcycle (67.8 percent). That means people prefer to 73

use private vehicle and they did not care about the public transportation and travel time because of less traffic congestion.

% of Vehicle Ownership

4.62%

18.02% No-vehicle 35.38% 1 vehicle 2 vehicles 3 vehicles More than 3 vehicles 41.98%

% of Daily Vehicle Usage

1.40%

1.10%

30% Walk Motorcycle Passenger car 67.80% Car-taxi

Figure 21 Vehicle ownership and daily usage.

2. Attitudes of the Respondents in considering house location

The respondents were requested to make decisions on three house-choosing situation. During these decision, they have to answer their own attitude and agreements in considering house location. The percentage of each attitudes in both face-to-face and online interview survey are as shown in the following figures.

According to Figure 22, 31 percent of the respondents agree that having parking space is important, 42 percent also agree closer to the work pace/ school is important. 23 percent of the people did not care about closer to the main road and people prefer less traffic congestion in nearest street (33 percent), Although 60 percent strongly agree to live in low crime rate area, they did not care about to live near to their relatives or friends. 31 percent agree that having local shops within neighborhood, the important of living in green environment is strongly agree (66 percent), 30 percent of the respondents strongly agree that having good prospect future value, 32 percent agree that the size of the house is important and the important of the price of the house is 39 percent of the respondents are strongly agree.

33 percent of the respondents disagree that having parking space is important, 30 percent strongly agree that closer to the work pace/ school is important. 23 percent of the people did not care about closer to the main road and people did not care the less traffic congestion in nearest street (23 percent), Although 34 percent strongly agree to live in low crime rate area, they did not care about to live near to their relatives or friends. 24 percent agree that having local shops within neighborhood, the important of living in green environment is agree (35 percent), 21 percent of the respondents strongly agree that having good prospect future value, 37 percent agree that the size of the house is important and the important of the price of the house is 31 percent of the respondents agree, as shown in Figure 23.

6.6% Having parking space 3.3% 7.4% 9.0% 18.9% 31.1% 22.1% 1.6%

Closer to work place/ school 4.1% 6.6% 8.2% 17.2% 41.8% 18.9% 0.0% 3.3%

Closer to the main road 8.2% 9.0% 23.0% 19.7% 9.0% 22.1% 4.9%4.1%

Less traffic congestion 3.3% 8.2% 18.0% 10.7% 32.8% 23.8% 0.8% 2.5%

2.5% 4.9% 0.8% 27.0% 59.8% Low crime rate 1.6% 1.6% 1.6%

Near to the relatives/friends 9.0% 13.1% 18.9% 25.4% 10.7% 16.4% 3.3%3.3%

Having local shops 4.9%4.1% 13.9% 15.6% 17.2% 31.1% 12.3% 0.8%

3.3% 0.8% Green environment 3.3% 15.6% 65.6% 4.1% 7.4%

Good prospect future value 2.5%9.0% 13.1% 17.2% 22.1% 30.3% 4.9% 0.8%

2.5% Size of the house 4.9% 5.7% 11.5% 17.2% 32.0% 25.4% 0.8%

4.1% Price of the house 6.6% 14.8% 14.8% 39.3% 16.4% 4.1%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Strongly disagree Disagree Somewhat disagree Neutral Somewhat agree Agree Strongly agree No idea

Figure 22 Attitudes of the respondents in face-to-face interview survey. 76

Having parking space 4.6% 3.9% 32.5% 17.3% 21.9% 17.0% 1.4% 1.4%

3.5% Closer to work place/ school 2.8% 3.9% 31.1% 20.8% 30.0% 7.1% 0.7%

Closer to the main road 5.3% 18.4% 23.3% 22.3% 7.8% 15.9% 5.7% 1.4%

Less traffic congestion 5.3% 13.8% 27.9% 14.8% 7.4% 19.1% 10.2% 1.4%

Low crime rate 4.2%4.2%13.8% 19.8% 21.6% 33.9% 1.4% 1.1%

Near to the relatives/friends 6.7% 14.1% 23.7% 26.1% 8.8% 14.1% 6.0% 0.4%

Having local shops 4.2%11.0% 25.8% 23.7% 8.5% 18.4% 7.8%0.7%

2.8% Green environment 4.6% 7.1% 15.2% 35.0% 33.6% 1.4% 0.4%

Good prospect future value 3.9% 16.6% 32.2% 13.8% 21.2% 8.1% 2.8% 1.4%

3.5% Size of the house 3.9% 11.0% 27.6% 37.5% 13.1% 2.8% 0.7%

3.2% Price of the house 4.2% 6.4% 16.6% 21.9% 31.4% 15.2% 1.1%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Strongly disagree Disagree Somewhat disagree Neutral

Somewhat agree Agree Strongly agree No idea

Figure 23 Attitudes of the respondents in online interview survey.

3. Results of all sample Face-to-face and online interview

Table 20 presented the result comparison of all sample in sub-game A and sub-game B of Face-to-Face and online survey. There are 510, 238 observations of Face-to-Face and 1482 and 206 of online respectively in Sub-game A and B. It is found that only location variable is not significant in sub-game A, Face-to-Face survey. While considering in two new hypothetical houses, Sub-game B, travel time, travel time, location of CBD area are significant. In Sub-game A, respondents are not care the neighbourhood environment and house location in CBD area. They are likely to live in CBD area with better neighbourhood in sub-game B. This is because of Figure 22, respondents’ attitude in thinking the house situation, they are strongly agreed with living in low crime rate area with green environment and closer to work place or school if they were moved or relocated to new place.

Respondents are not likely to live in CBD area and also do not care the travel time in Sub-game A of online survey result. In addition, they prefer higher price house. In Sub-game B, people like bigger size located in CBD area with better neighbourhood environment. Although being agree in the important of less traffic congestion, they prefer longer travel time. As a consequence, the goodness of fit (AIC, BIC, 휌2 ) values in Face-to-Face results are better than online survey results, this is because of questionnaire survey interview method. In Face-to-Face, the respondents can more understand and answer their present house size, price and neighbourhood environment by asking the interviewer than online survey. Therefore, in this research we discuss about some socioeconomic groups such as ethnic group, high-income and low-income group and big and small family members group of Face- to-Face survey results in Sub-game A.

* * * *

78 78 3.02 5.53 3.65 2.09 - statistic

- t

506 Online 1.2918 1.3336 0.0824 0.0681 y significant.

0.0355 1.4035 1.8969 0.0008 - Estimate

game B

- 2.81 4.82 2.37

Sub -

statistic

- * refers to 95% statisticall t

Face Face -

to * * *

- 238 1.2500 1.3230 0.0983 0..1286 Face 0.0494 1.7546 1.9188 - Estimate Face and Online survey. and Online Face -

* * * - 7.44 6.13 14.48 -

statistic

- t

1,482 Online 0.0954 1.9875 2.0054 0.0985 game B of Face game -

0.0009 0.0352 0.9326 - Estimate

* * * * game A

- 3.73 4.94 4.97

Sub - 13.13 - statistic

- t game A and Sub and game A

Face - -

- 510 0.2387 1.6727 1.7142 0.2432 Face 0.0005 0.0010 0.0608 1.6154 - - Estimate sample in Sub sample in

neighbourhood

2 Results comparison of all comparisonResults of 휌

2 휌

BIC Adjusted Dummy of CBD area CBD Dummy of Dummy of better Number of observations AIC Size Price Travel Time

Table Table

3.1 Burmese ethnic and Chinese or other ethnic groups result of Face-to-face interview

Table 21 Burmese ethnic and Chinese group result of Face-to-face interview in Sub- game A

Burmese Chinese

Estimate t-statistic Estimate t-statistic Size -0.0007 -4.15* Price 0.0010 3.83* 0.0011 3.17* Travel Time 0.0443 3.03* 0.0980 4.21* Dummy of CBD area Dummy of better-neighborhood -1.4926 -10.22* -1.9395 -8.12* Number of observations 330 158 AIC 1.7026 1.6333 BIC 1.7602 1.7220

휌2 0.2389 0.2819 2 Adjusted 휌 0.2251 0.2567 * refers to 95% statistically significant.

Table 22 Burmese ethnic and Chinese group result of Face-to-face interview in Sub- game B

Burmese Chinese

Estimate t-statistic Estimate t-statistic Size 0.4691 2.06* Price -0.0007 -2.11* Travel Time Dummy of CBD area -1.3788 -3.71* 2.0740 3.37* Dummy of better- 3.0609 1.95* neighborhood Number of observations 158 82 AIC 1.3085 1.1945 BIC 1.4054 1.3412 80

휌2 0.1017 0.2263 Adjusted 휌 2 0.0561 0.1383 * refers to 95% statistically significant.

3.3% Chinese 3.3% 13.3% 3.3% 20.0% 43.3% 13.3%

1.8%

Burmese 7.3% 10.9% 16.4% 49.1% 12.7% place/ school place/

Closer to work to Closer 1.8% 3.3%

Chinese 3.3% 10.0% 36.7% 46.7% within

Burmese 7.4%1.9% 24.1% 64.8% neighborhood Low crime rate crime Low 1.9% 3.3% Chinese 3.3% 10.0% 10.0% 6.7% 66.7%

5.5%1.8%

Burmese 1.8%7.3% 20.0% 61.8% environment Living in green in Living 1.8%

Chinese 6.7%10.0% 16.7% 3.3% 30.0% 33.3%

3.7%

Burmese 5.6% 7.4% 9.3% 20.4% 35.2% 18.5% Size of the house the of Size

Chinese 13.3% 10.0% 23.3% 16.7% 30.0% 6.7%

5.5% Burmese 5.5% 18.2% 18.2% 32.7% 18.2%

1.8% Price of the house the of Price 0% 20% 40% 60% 80% 100% 120% Strongly disagree Disagree Somewhat disagree Neutral Somewhat agree Agree Strongly agree

Figure 24 Respondents’ attitudes of ethin group in face-to-face interview survey.

In terms of ethnic group, Burmese and Chinese ethnic group have 330 and 158 observations. People did not consider the location of the district and they prefer their present house. They did not concern about the travel time and neighbourhood quality 81

because of very familiar with their current house lcation. While considering only two new hypothetical house, the Burmese ethnic group does not consider the size of the house and travel time when making choice decision. And they do not likely to live in CBD areas that is they concern about the price of the house. Also, the better- neighborhood environment is not significant for this group of determining the house location decision.

The estimation results for sub-sample of Chinese and other ethnic groups are as shown in Table 22. The decision making with present house and new house, people did not consider the township location and size of the house. Price and travel time variables are significant and they did not mind the longer travel time. The better the environment in the area, the more likely these ethnic groups to live. In addition, the price of the house is not a decision-making factor for these Chinese or other ethnic groups.

3.2 High-income and Low-income groups result of Face-to-face interview

Table 23 High-income and Low-income groups result of Face-to-face interview in Sub-game A

Low-income High-income (≤500,000 – 1,000,000 (>1,000,000 Kyats) Kyats) Estimate t-statistic Estimate t-statistic Size -0.0011 -4.76* Price 0.0014 5.15* 0.0007 2.12* Travel Time 0.0787 4.15* 0.0369 2.37* Dummy of CBD area -1.0532 -3.14* Dummy of better- -1.9005 -9.36* -1.4842 -9.12* neighborhood Number of observations 234 276 AIC 1.5934 1.6851 BIC 1.6672 1.8032 82

휌2 0.2959 0.2627 Adjusted 2 휌 0.2852 0.2505

* refers to 95% statistically significant.

Table 24 High-income and Low-income groups result of Face-to-face interview in Sub-game B

Low-income High-income (≤500,000 – 1,000,000 (>1,000,000 Kyats) Kyats) Estimate t-statistic Estimate t-statistic Size Price 0.0008 2.29* -0.0003 -2.74* Travel Time -0.0609 -2.09* -0.0225 -1.92* Dummy of CBD area 2.1764 3.78* 1.5245 3.16* Dummy of better- 5.4021 2.43* neighborhood Number of observations 107 131 AIC 1.1533 1.3507 BIC 1.2783 1.4604 휌2 0.2361 0.4691 2 Adjusted 휌 0.1826 0.2429

* refers to 95% statistically significant.

10.3% High-income 2.6% 7.7% 25.6% 43.6% 10.3%

2.2% Low-income 8.7% 6.5% 13.0% 50.0% 15.2%

place/ school place/ 4.3% Closer to work to Closer

High-income 2.6% 5.1% 38.5% 53.8%

2.2% within 8.9%

Low-income 2.2% 2.2% 20.0% 62.2% neighborhood Low crime rate crime Low 2.2% 5.1% High-income 2.6% 12.8% 12.8% 64.1% 2.6%

4.3% Low-income 6.5% 19.6% 60.9%

environment 4.3% 4.3% Living in green in Living

High-income 2.6%10.3% 12.8% 17.9% 20.5% 33.3%

2.6% house

Low-income 8.9% 6.7%11.1% 11.1% 44.4% 15.6% Size of the of Size 2.2%

High-income 7.7% 7.7% 28.2% 17.9% 25.6% 10.3% 2.6%

house 2.2% Low-income 8.7% 13.0% 17.4% 39.1% 15.2% Price of the of Price 4.3% 0% 20% 40% 60% 80% 100% 120% Strongly disagree Disagree Somewhat disagree Neutral

Somewhat agree Agree Strongly agree

Figure 25 Respondents’ attitudes of income group in face-to-face interview survey.

In the high-income people group, people are likely to live with higher price with longer travel time. They would not like to live with better neighborhood in CBD areas and they did not care about the travel time and neighborhood environment. They are likely to live in their present house although their house is not in CBD area. The people with low-income are significantly care about the price of the house although they are likely to live in CBD area and less travel time. Same as the high-income group, they did not think about the better neighborhood environment and travel time. 84

3.3 Big family and Small family groups result of Face-to-face interview

Table 25 Big family and Small family groups result of Face-to-face interview in Sub- game A

Big family Small family (≥ 5 members) (< 5 members) Estimate t-statistic Estimate t-statistic Size -0.0003 -2.36* -0.0005 -2.59* Price 0.0008 3.04* 0.0013 4.20* Travel Time 0.0551 3.59* 0.0751 4.03* Dummy of better- -1.4277 -8.18 -1.8338 -10.39* neighborhood Number of observations 234 276 AIC 1.5933 1.5703 BIC 1.6672 1.6359 휌 2 0.2351 0.3078 Adjusted 2 휌 0.2167 0.2964

* refers to 95% statistically significant.

Table 26 Big family and Small family groups result of Face-to-face interview in Sub- game B

Big family Small family (≥ 5 members) (< 5 members) Estimate t-statistic Estimate t-statistic Size Price 0.0005 2.00* 0.0011 3.68* Travel Time Dummy of new town area -1.8966 -3.00* -2.1700 -3.83* Number of observations 102 136 AIC 1.3498 1.1993 BIC 1.4785 1.3064 휌 2 0.1034 0.2109 Adjusted 2 휌 0.0307 0.1682

* refers to 95% statistically significant.

5.1% Big family 10.3% 5.1% 20.5% 48.7% 7.7% 2.6%

school 2.2% Small family2.2% 8.7% 17.4% 45.7% 17.4%

6.5% Closer to work place/ work to Closer

2.6% Big family 13.2% 2.6% 34.2% 47.4%

4.3% Small family2.2% 23.9% 67.4%

Low crime rate Low crime 2.2% within neighborhood within

2.6% Big family 10.3% 2.6% 15.4% 59.0% 2.6% 7.7%

2.2%

environment Small family 8.7% 17.4% 65.2%

Living in green Living in 6.5%

Big family 7.7% 10.3% 17.9% 10.3% 23.1% 25.6% 5.1%

4.4% Small family 6.7% 17.8% 42.2% 22.2%

6.7% Size of the house the of Size

2.6% Big family 12.8% 28.2% 17.9% 25.6% 7.7% 5.1%

2.2% Small family 4.3% 13.0% 17.4% 39.1% 17.4%

Price of the house the of Price 6.5%

0% 20% 40% 60% 80% 100% 120% Strongly disagree Disagree Somewhat disagree Neutral

Somewhat agree Agree Strongly agree

Figure 26 Respondents’ attitudes of family size group in face-to-face interview survey.

According to family size, people did not want to live in small size of the house. New town area was negatively significant in both big and small family members groups. In the big family member group, people did not consider the neighborhood of the house is good or bad and also the travel time. Size, price and travel time variables 86

are statistically significant in the small family members group and preferred the smaller size of the house with high price. The location and neighborhood variable were not significant. In addition, while thinking together with the present house and two new houses, both location and neighborhood were not the important factor and also longer travel time.

CONCLUSION

This study has examined the factors that have influence on residential preference, i.e., attractiveness of attributes with respect to various groups of people in Mandalay. Firstly, to determine factors that influence on making house location decision, we made the questionnaire interview survey the people who live in Mandalay city. The survey included two parts Face-to-Face and online with computer assisted survey. The questionnaire had three parts, socioeconomic characteristics, SP experiment and attitudes of the people in thinking house situation. Face-to-Face results are better than online survey results, this is because of questionnaire survey interview method.

In Face-to-Face, the respondents can more understand and answer their present house size, price and neighbourhood environment by asking the interviewer than online survey. In this research, the latter part is to find the behaviour of the group of people in choosing the residence, the researcher separated different groups based on socioeconomic characteristics such as ethnic group, income group and family members group for data analysis. Both of all groups have two sub groups, Burmese and Chinese ethnic group, High-income and Low-income group and Big family members and Small family members group.

The results support the existence of factors which differ in their housing choice preference. For those people who prefer their present house than the two new hypothetical houses, they apparently do not desire to live in the CBD and spend more time for travelling to work or school. This finding clearly reveals that the district location and travel time are not the main factors for people in Mandalay when choosing house. This phenomenon is due to the less congested traffic situation in Mandalay even in the peak hour.

In the ethnic groups, the neighbourhood environment is not very important in considering their present house with new hypothetical house and also same as in location of CBD area. The case in new hypothetical situation, Chinese people 88

answered that low crime rate neighbourhood and size of the house are important and the price of the house is not the priority. Both of the groups are not care the travel time factor. Low-income group considered that the factors of price and size of the house are very important. For High-income group people, less travel time and locate in CBD area with better neighbourhood environment is very important.

Small family members cared about the price factor that other factors. The bigger house is important in Sub-game B of big family members group although the price is not the important factor. Therefore, travelling from outskirt to the CBD to work or school is not a big deal but can compromise with better neighborhood in the outer area of the city. This was examined by asking people to choose only the two new hypothetical houses. In addition, different socioeconomic groups have different important factors with different personal attitudes on making house location decision. It is found that the Burmese-race respondents mainly consider location although CBD is not preferred, neither pay attention to travel time.

The results reveal that people are considering factors not only house size, house price, but also locational convenience in terms of commuting time, and neighborhood quality. It is also found that different socio-economic groups, i.e., ethnicity, exhibit different location preferences. In contrast, Chinese-race respondents exhibit strong preference to locate in the CBD where the commercial opportunity are there while preferring a larger house locating in better neighborhood but still trade-off with travel time. The effect of other socio-economic characteristic is left for the future study.

LITERATURE CITED

Akaike, H. (1974). A new look at the statistical model identification. In Selected Papers of Hirotugu Akaike (pp. 215-222): Springer. Alsaiari, H. N. (2018). Residential Preference at Transit-oriented Development: A Visual Choice Experiment. Virginia Tech, ANGHS, A., & TANG, W. (1975). Probability concepts in engineering planning and design. Vol. 1, Basic principles. Anwar, A. M., Tieu, A. K., Gibson, P., Berryman, M. J., Win, K. T., McCusker, A., & Perez, P. (2014). Temporal and parametric study of traveller preference heterogeneity using random parameter logit model. Bank, A. D. (2016). Myanmar Transport Sector Policy Note, Urban Transport. Retrieved from Beamish, J. O., Carucci Goss, R., & Emmel, J. (2001). Lifestyle Influences on Housing Preferences. Housing and Society, 28(1-2), 1-28. doi:10.1080/08882746.2001.11430459 Ben-Akiva, M. E., Lerman, S. R., & Lerman. (1985). Discrete choice analysis: theory and application to travel demand (Vol. 9): MIT press. Cao, X. (2008). Is Alternative Development Undersupplied?:Examination of Residential Preferences and Choices of Northern California Movers. 2077(1), 97-105. doi:10.3141/2077-13 Department, P. (Census Report 2015). The 2014 Myanmar Population and Housing Census, Mandalay Region. Retrieved from

Hack, G., Fishman, M., Yaro, R., Brenner, N., & Wachsmuth, M. (2012). Planning Ideas That Matter: Livability, Territoriality, Governance, and Reflective Practice: MIT Press. Hunt, J. D. (2010). Stated preference examination of factors influencing residential attraction. In Residential Location Choice (pp. 21-59): Springer. Ibraimovic, T., & Hess, S. (2018). A latent class model of residential choice behaviour and ethnic segregation preferences. Housing Studies, 33(4), 544-564. doi:10.1080/02673037.2017.1373749 Jansen, S. J., Coolen, H. C., & Goetgeluk, R. W. (2011). The measurement and analysis

of housing preference and choice: Springer Science & Business Media. JICA Study Team: Nippon Koei Co., L., Nine Steps Corporation and International Development Centre of Japan Inc. (2016). Data Collection Survey on Urban Development Planning for Regional Cities (Mandalay, Pathein and MAwlamyine). Retrieved from Khaing, T. T. (2015). Urbanization: the structures of sustainable urban landscape of Myanmar. Paper presented at the Proceedings of the International Conference on Burma/Myanmar Studies, Chiang Mai, Thailand. Kim, Y. (2018). Mandalay, Myanmar: The remaking of a South-east Asian hub in a country at the crossroads. Cities, 72, 274-286. doi:https://doi.org/10.1016/j.cities.2017.09.003 Liao, F. H., Farber, S., & Ewing, R. (2015). Compact development and preference heterogeneity in residential location choice behaviour: A latent class analysis. 52(2), 314-337. doi:10.1177/0042098014527138 Louviere, J. J., Hensher, D. A., Swait, J. D., & Adamowicz, W. (2000). Stated Choice Methods: Analysis and Applications: Cambridge University Press. McFadden, D. (1974). Conditional logit analysis of qualitative choice behavior, Frontiers in econometrics 105-142.

NLOGIT. (2007). LIMDEP and NLOGIT software. Olaru, D., Smith, B., & Taplin. (2011). Residential location and transit-oriented development in a new rail corridor. Transportation Research Part A: Policy and Practice, 45(3), 219-237. doi:https://doi.org/10.1016/j.tra.2010.12.007 Sener, I. N., Pendyala, R. M., & Bhat, C. R. (2011). Accommodating spatial correlation across choice alternatives in discrete choice models: an application to modeling residential location choice behavior. Journal of Transport Geography, 19(2), 294-303. doi:https://doi.org/10.1016/j.jtrangeo.2010.03.013 Suel, E., & Polak, J. W. (2018). Incorporating online shopping into travel demand modelling: challenges, progress, and opportunities. Transport Reviews, 38(5), 576-601. doi:10.1080/01441647.2017.1381864 Tillema, T., van Wee, B., & Ettema, D. (2010). The influence of (toll-related) travel costs in residential location decisions of households: A stated choice approach.

Transportation Research Part A: Policy and Practice, 44(10), 785-796. doi:https://doi.org/10.1016/j.tra.2010.07.009 Vichiensan, V. (2010). Preference of residential choice along the MRT purple line corridor in Bangkok. ATRANS RESEARCH, 1. Williamson, J. (2013). Designing suburban futures: New models from build a better burb: Island Press.

APPENDIX

Appendix Questionnaire Survey Design with Google Form

100

101

102

103

104

105

106

107

108

CURRICULUM VITAE

NAME Theint Htet Htet Aung

DATE OF BIRTH 3 March 1992

BIRTH PLACE Mandalay

ADDRESS No.3/74 Between 37x38 street and 63x64 street, Mahar Aung Myay Township, Mandalay, Myanmar. EDUCATION Bachelor of Engineering (Civil)

SCHOLARSHIP Kasetsart University Scholarships for ASEAN for Commemoration of the 60th Birthday Anniversary of Professor Dr. Her Royal Highness Princess Chulabhorn Mahidol.