A

A Housing demand model: A case study of the Bangkok Metropolitan Region, Thailand

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy

Pon Vajiranivesa

School of Property, Construction and Project Management Design and Social Context Portfolio RMIT University September 2008

DECLARATION

I certify that except where due acknowledgement has been made, the work is that of the author alone; the work has not been submitted previously, in whole or in part, to qualify for any other academic award; the content of the thesis is the result of work which has been carried out since the official commencement date of the approved research program; and, any editorial work, paid or unpaid, carried out by a third party is acknowledged.

Pon Vajiranivesa 29th September 2008

ACKNOWLEDGEMENTS

I am most grateful to my supervisors, Associate Professor James Baxter and Professor Ron Wakefield for encouraging me in this research. Their valuable criticism and opinions helped me to complete the research.

I also thank Naresuan University for supporting me in this PhD study by providing me with a scholarship. I gained valuable experiences from studying abroad in Australia, in a different environment, knowledge-system and language.

For the useful discussions we had, I need to thank my PhD candidate friends. I will never forget the atmosphere of our beautiful Friday night academic discussions. I include Dr. Titus Mbiti from Kenya, Gabriel Nani from Ghana, Dr. Atif Ali from Pakistan, Ehsan Gharaie and Babak Abbasi from Iran, Dr. Ting Kwei Wang from Taiwan, Tracy Ng from Hong Kong and Dr. Supaporn Chalapati from Thailand.

Finally, I would like to thank my family for their emphasis on education, patience, encouragement, support and loe.

TABLE OF CONTENTS

Declaration.............................................................................................................B Acknowledgements................................................................................................C TABLE OF CONTENTS...................................................................................... D LIST OF TABLES..................................................................................................J LIST OF FIGURES ............................................................................................... L LIST OF EQUATIONS ......................................................................................... P ABSTRACT...........................................................................................................R

1 INTRODUCTION ........................................................................................... 1-1 Introduction......................................................................................................... 1-1 The problem being approached........................................................................... 1-2 Traditional housing-demand modelling.......................................................... 1-4 Important and relevance of this research ........................................................ 1-5 Definitions of housing demand and housing needs ........................................ 1-8 History of housing-demand studies and their factors ......................................... 1-9 The effect of demographic factors on housing demand................................ 1-11 The effect of economic factors on housing demand ..................................... 1-20 The effect of social factors on housing demand ........................................... 1-22 Consequence of all factors to the model of this research.............................. 1-23 Model development process ............................................................................. 1-25 Conceptualising the research methodology and the housing-demand model1-26 Undertaking the data collection .................................................................... 1-27 Analyse the results of model simulations ..................................................... 1-28 Research methodology...................................................................................... 1-28 Research objectives and aims ........................................................................... 1-29 Chapter summary .............................................................................................. 1-33 Terminology definition of housing types in this research (for Bangkok Metropolitan Region) ................................................................. 1-35

2 LITERATURE REVIEW .............................................................................. 2-40 Introduction....................................................................................................... 2-40

The context for a review of housing demand literature ................................ 2-40 Housing-demand studies and housing-need studies ..................................... 2-41 History of housing-demand model studies ................................................... 2-43 The distinguishing features of housing (as goods in an economy)............... 2-45 Summary of housing features and previous models ..................................... 2-48 Several approaches to the study of housing-demand study .......................... 2-51 Merging demographic, social and economic factors in housing-demand studies ........................................................................................................................... 2-52 Modelling methodology comparison (between Econometrics and System Dynamics approaches)............................ 2-54 Econometric modelling................................................................................. 2-54 System Dynamics model............................................................................... 2-56 An Example to illustrate the System Dynamics approach............................ 2-57 An econometrics approach to the same scenario .......................................... 2-60 Example for Econometrics approach of the orange model........................... 2-60 Controversial on Econometrics and System Dynamics .................................... 2-63 Housing-demand study dilemma ...................................................................... 2-65 Previous housing research limitations: ............................................................. 2-68 System dynamic approach advantage: .............................................................. 2-70 Gaps identified within the literature ................................................................. 2-73 Approach to System Dynamics......................................................................... 2-75 System Dynamics introduction ..................................................................... 2-76 System Dynamics history ............................................................................. 2-77 System Thinking ........................................................................................... 2-78 Systems Thinking and System Dynamics method........................................ 2-81 iThink software program for System Dynamics modelling.......................... 2-81 System Dynamics for Housing-demand study.................................................. 2-82 Housing demand environment from System-Thinking approach................. 2-83

3 CASE STUDY BACKGROUND AND INFORMATION ........................... 3-87 Introduction....................................................................................................... 3-87 Physical information of case study area (Bangkok Metropolitan Region and Thailand) ................................................. 3-87 Thailands Emerging Economic Policies and Strategies .............................. 3-95

Overview of housing context in Bangkok Metropolitan Region .................. 3-97 History of the housing market in Bangkok Metropolitan Region................. 3-99 Background of housing market in Bangkok Metropolitan Region............. 3-106 The role of housing in the Bangkok Metropolitan Region (BMR)............. 3-107 Housing in the Bangkok Metropolitan Region in the last 10 years ................ 3-108 Proportion of housing type in the market in Bangkok Metropolitan Region from 1996 to 2005................................. 3-112 Price index of housing in Thailand ............................................................. 3-113 Used housing market in Bangkok Metropolitan Region............................. 3-114 Population and Housing growth in the Bangkok Metropolitan Region.......... 3-115 Population related to housing market in the Bangkok Metropolitan Region.......................................................... 3-117 Distribution of housing market in Bangkok Metropolitan Region compared to Thailand as a whole.......... 3-119 Population and household size in Bangkok Metropolitan Region and Thailand.......................................... 3-120 Summary ......................................................................................................... 3-122

4 CONCEPT AND METHODOLOGY OF THE HOUSING DEMAND MODEL ......................................................................................................................... 4-123 Introduction..................................................................................................... 4-123 Establishing a conceptual housing demand model ......................................... 4-123 Consequential concerns with the housing demand model .......................... 4-129 Assumptions and limitations....................................................................... 4-132 An assumption regarding the used housing market .................................... 4-132 A limitation of the model when dealing with housing consumption by wealthy people ....................................................................................... 4-133 Limitations of the data ................................................................................ 4-133 Principle behind the model and the process for building the model............... 4-134 Demographic sector methodology part of the model.................................. 4-135 Flow rate equation of demographic factors ............................................ 4-139 The social sector methodology part of the model....................................... 4-144 Flow rate equation for social factors....................................................... 4-144 Co-efficiency estimation for splitting of single head household data..... 4-145

Data of marriage and divorce co-efficiency estimation.......................... 4-158 Economic sector methodology part of the model ....................................... 4-161 Flow rate equation of economic factors.................................................. 4-161 Equation for economic factors ................................................................ 4-169 Demographic stocks.................................................................................... 4-179 Social stock ................................................................................................. 4-179 Economic stock........................................................................................... 4-179 Check point stock........................................................................................ 4-179 System environment of the model .................................................................. 4-180 Symbols used in the Model............................................................................. 4-182 Chapter summary............................................................................................ 4-184

5 MODEL SIMULATION ............................................................................. 5-186 Introduction..................................................................................................... 5-186 Considerations in running the model .............................................................. 5-186 Description of Underlying Housing Needs Units (UHNUs) .......................... 5-187 Validation of the Model .................................................................................. 5-191 Description of graphical function ................................................................... 5-193 Graphical function of the model between GDP growth index and coefficient of GDP growth index on consumption decision ................ 5-197 Graphical function of the model between housing loan per housing price index and coefficient of housing loan per housing price index on consumption decision...................................... 5-199 Graphical function of the model between housing sales data and coefficient of housing sales data on housing supply............................ 5-200 Benefit of model ............................................................................................. 5-202 Comparing historical data to the simulation run............................................. 5-204 Comparison between simulations of housing needs line and historical housing sales line ................................................................. 5-205 Principles deduced from simulation result.................................................. 5-206 Comparison between simulations of housing demand line and historical housing sales line ................................................................. 5-210 Comparison between the simulated and historic housing sales lines ......... 5-212 Comparison between simulations of housing supply line

and historical housing supply line............................................................... 5-213 Comparison between simulations of housing vacancy line and historical housing vacancy line ............................................................ 5-214 Chapter summary ............................................................................................ 5-215

6 FORECASTING WITH THE MODEL....................................................... 6-216 Introduction..................................................................................................... 6-216 Demographic factors projection...................................................................... 6-216 Population in the Bangkok Metropolitan Region ....................................... 6-217 Adolescents in the Bangkok Metropolitan Region ..................................... 6-218 Mature People in the Bangkok Metropolitan Region projection ................ 6-219 Elderly Popele in the Bangkok Metropolitan Region projection................ 6-220 Social factor projections ................................................................................. 6-221 Economic factor projections ........................................................................... 6-229 GDP growth index in Thailand projection.................................................. 6-229 Projected Housing Loans in Thailand......................................................... 6-231 Projection Housing Price Index in Thailand............................................... 6-232 Simulation results............................................................................................ 6-233 Projected Housing Sales in the Bangkok Metropolitan Region.................. 6-238 Housing Supply Projections for the Bangkok Metropolitan Region .......... 6-240 Projected House Vacancies in the Bangkok Metropolitan Region............. 6-242 Projected house vacancies are shown in the following figure: ................... 6-242 Summary ......................................................................................................... 6-243

7 FINDINGS, CONCLUSIONS AND RECOMMENDATIONS.................. 7-246 Introduction..................................................................................................... 7-246 Summary of main findings.............................................................................. 7-247 Differentiation between housing demand and housing sales data .............. 7-247 Achievement of objectives of this research ................................................ 7-248 Conclusions from conceptual settings in the model ................................... 7-249 Conclusions Linking Housing Demand and Housing Sales ....................... 7-252 Conclusions from the model ....................................................................... 7-255 Suggested principles ....................................................................................... 7-258 Conclusion ...................................................................................................... 7-261

Housing demand forecasting........................................................................... 7-265 The cycle of housing consumption ................................................................. 7-266 Research contribution ..................................................................................... 7-268 Future research................................................................................................ 7-269

References ......................................................................................................... 271

Bibliography ..................................................................................................... 280

Appendix A ....................................................................................................... 287 Terminology....................................................................................................... 287

Appendix B ....................................................................................................... 292 Table of data used in the model ......................................................................... 292

Appendix C ....................................................................................................... 306 The model formulae documentation listing and instruction .............................. 306 Note * Demographic sector ............................................................................... 306 Note * Social sector ........................................................................................... 309 Note * Economic sector ..................................................................................... 312 Note * Validation sector .................................................................................... 316

Appendix D ....................................................................................................... 318 Mathematical prove for the evidence supported the principle in Chapter 7 ...... 318

LIST OF TABLES

Table 1-1 : Number of single-head household of population ........................... 1-19 Table 2-1 : Comparison of the characteristics of Econometrics and System Dynamics, concluded from Myrtveit ... 2-64 Table 3-1 : The list of provinces of Thailand ................................................... 3-92 Table 3-2 : Comparison of housing loans between........................................... 3-98 Table 3-3 : Housing price and units prefer for people .................................... 3-110 Table 3-4 : Comparison of housing loans between Bangkok Metropolitan Region and other regions......................... 3-119 Table 3-5 : Birth rate death rate and increased rate by area and region.......... 3-121 Table 3-6 : Size of household in Thailand and Bangkok Metropolitan Region............................................... 3-121 Table 3-7 : Population in provinces in Bangkok Metropolitan Region .......... 3-122 Table 4-1 : Comparison between historical data of population and simulated data of population.................................................. 4-138 Table 4-2 : Population by age from 2001 to 2003 in Thailand....................... 4-140 Table 4-3 : Number of head household from data of population in municipal area in Thailand in 2005.................... 4-147 Table 4-4 : Number of single head household of population in municipal area in Thailand in 2005.................... 4-148 Table 4-5 : Number of single head household of population in municipal area in Thailand in 2005.................... 4-150 Table 4-6 : Number of single head household of population in municipal area in Thailand in 2005.................... 4-152 Table 4-7 : Splitting (formation) single head household number from data of municipal area in Thailand in 2005 ......................... 4-153 Table 4-8 : Splitting of single head household number by age group from data of municipal area in Thailand in 2005 ... 4-156 Table 4-9 : Data of marriage and divorce ratio of population ........................ 4-158 Table 4-10 : Marriage and divorce number of population in Thailand........... 4-160 Table 4-11 : Housing price index in Bangkok Metropolitan Region and missing data ........................................................................... 4-169

Table 4-12 : minimum and maximum value of GDP growth index and coefficient of GDP on housing affordability & investment... 4-171 Table 4-13 : minimum and maximum value of GDP growth index and coefficient on housing consumption decision........................ 4-172 Table 4-14 : Setting relation of housing loan and housing price index on housing consumption capacity................................................. 4-174 Table 4-15 : minimum and maximum value of fraction of housing loan over housing price index and coefficient on housing consumption decision......................................................................................... 4-175 Table 4-16 : housing sales and housing supply .............................................. 4-177 Table 4-17 : minimum and maximum value of housing sales and coefficient on housing supply ................................................ 4-178 Table 4-18 : Detail of initial values of all stock in housing demand model ... 4-180 Table 5-1 : The comparison surplus and deficit between housing needs and housing sales data ............................ 5-208 Table 6-1 : Splitting of single head household number from data of municipal area in Thailand in 2005 ......................... 6-222 Table 6-2 : Percentage of splitting of single head household for each age group is show below ................................................ 6-225 Table 6-3 : Marriage and divorce number of population in Thailand............. 6-228 Table 7-1 : A comparison between the (equilibrium) linear line and the underlying housing needs number ................................... 7-264 Table in appendix D-1: Comparison between linear line value and housing sales data..323

LIST OF FIGURES

Figure 1-1 : Demographic and social factors impact on housing market ......... 1-15 Figure 1-2 : Development of household types.................................................. 1-16 Figure 1-3 : Ratio and trend of household types in Australia from 1991 ......... 1-17 Figure 2-1 : Relationship between rain and orange flower............................... 2-58 Figure 2-2 : Relationship between bee and orange........................................... 2-59 Figure 2-3 : The Orange model (example) ....................................................... 2-60 Figure 2-4 : Structure of System Thinking approach........................................ 2-80 Figure 3-1 : Map of Thailand from South-East Asia region Map..................... 3-88 Figure 3-2 : Map of regions in Thailand........................................................... 3-89 Figure 3-3 : Map of 72 provinces in Thailand .................................................. 3-91 Figure 3-4 : Map of Bangkok Metropolitan Region, Thailand......................... 3-93 Figure 3-5 : Map of Bangkok............................................................................ 3-94 Figure 3-6 : GDP growth index in Thailand between 1981 and 2005 ............ 3-103 Figure 3-7 : Vacant house units in Bangkok Metropolitan Region ................ 3-104 Figure 3-9 : Percentage of housing supply in Bangkok Metropolitan Region from 1974 to 1989 .................. 3-107 Figure 3-10 : Housing stock in Bangkok Metropolitan Region...................... 3-108 Figure 3-11 : Housing opening in Bangkok Metropolitan Region ................. 3-109 Figure 3-12 : Housing method to build from 1996 to 2005............................ 3-111 Figure 3-13 : Type of housing in Bangkok Metropolitan Region .................. 3-112 Figure 3-14 : Housing type in the market from 1996 to 2005 ........................ 3-113 Figure 3-15 : Housing price index in Thailand from 1977 to 1992 and 2000 to 2005 .......................................................................... 3-114 Figure 3-16 : The ratio of people per house hold (by average) ...................... 3-117 Figure 3-17 : Housing registered in Bangkok Metropolitan Region .............. 3-118 Figure 3-18 : New housing loan from1995 to 2006........................................ 3-119 Figure 4-1 : Kinetic box diagram of housing demand model ......................... 4-125 Figure 4-2 : Housing demand production process .......................................... 4-129 Figure 4-3 : Population data in Bangkok Metropolitan Region...................... 4-136

Figure 4-4 : Proportion of adolescent population in Bangkok Metropolitan Region ................................................. 4-141 Figure 4-5 : Proportion of mature population in Bangkok Metropolitan Region ................................................. 4-142 Figure 4-6 : Proportion of elder population in Bangkok Metropolitan Region ................................................. 4-143 Figure 4-7 : show trend of marriage proportion in Bangkok Metropolitan Region ................................................. 4-158 Figure 4-8 : Trend of divorce proportion in Bangkok Metropolitan Region.. 4-159 Figure 4-9 : Equation trend of GDP growth index in Thailand ...................... 4-165 Figure 4-10 : Equation trend of housing loan in Thailand.............................. 4-166 Figure 4-11 : Equation trend of housing price index in Bangkok Metropolitan Region ................................................. 4-167 Figure 4-12 : coordinate of GDP growth index on affordability and investment.................................................... 4-171 Figure 4-13 : coordinate of GDP growth index on housing consumer confidence. ................................................ 4-172 Figure 4-14 : coordinate of housing loan and housing price on housing consumer capacity...................................................... 4-175 Figure 4-15 : coefficient of housing sales data on housing supply................. 4-178 Figure 4-16 : System of housing demand environment .................................. 4-181 Figure 4-17 : Map of housing demand mode.................................................. 4-185 Figure 5-1 and Figure 5-2: Comparison of the proportion of the underlying housing-needs composition with housing sales............................ 5-188 Figure 5-3 : The function between GDP growth index and coefficient of GDP growth index on Affordability & Investment ...................... 5-195 Figure 5-4 : The function between GDP growth index and coefficient of GDP growth index on consumption decision.. 5-197 Figure 5-5 : The function between fraction of housing loan over housing price index and coefficient of the fraction of housing loan per housing price index on consumption decision.............................. 5-199 Figure 5-6 : The function between housing sales data and fraction of housing sales data over housing supply ................................... 5-200

Figure 5-7 : Comparison between simulations of housing needs line and historical housing sales line ................ 5-205 Figure 5-8 : Housing sales data and linear trend compare with underlying housing needs number ......................... 5-206 Figure 5-9 : Comparison between simulations of housing needs line and historical housing sales line with overlap area and underlap area 5-207 Figure 5-10 : Comparison between simulation of housing demand line and historical housing sales line ................................................... 5-210 Figure 5-11 : Comparison between number of simulated housing demand line and simulated housing sales line .................................................. 5-211 Figure 5-12 : Comparison between simulation of housing sales line and historical housing sales line ................................................... 5-212 Figure 5-13 : the comparison between simulation of housing supply line and historical housing supply line ................................................ 5-213 Figure 5-14 : comparison between simulation of housing vacancy line and historical housing vacancy line.............................................. 5-214 Figure 6-1 : Population data in Bangkok Metropolitan Region projection .... 6-217 Figure 6-2 : Adolescent data in Bangkok Metropolitan Region projection.... 6-218 Figure 6-3 : Mature in Bangkok Metropolitan Region projection.................. 6-219 Figure 6-4 : Elder in Bangkok Metropolitan Region projection..................... 6-220 Figure 6-5 : Marriage rate in Bangkok Metropolitan Region projection........ 6-226 Figure 6-6 : Divorce rate in Thailand projection ............................................ 6-227 Figure 6-7 : GDP growth index in Thailand projection.................................. 6-229 Figure 6-8 : Housing loan in Thailand projection........................................... 6-231 Figure 6-9 : Housing price index in Thailand projection................................ 6-232 Figure 6-10 : Projection of housing needs in Bangkok Metropolitan Region 6-234 Figure 6-11 : The first scenario projection of housing demand in Bangkok Metropolitan Region ................................................. 6-236 Figure 6-12 : The second scenario projection of housing demand in Bangkok Metropolitan Region ................................................. 6-237 Figure 6-13 : The first scenario projection of housing sales in Bangkok Metropolitan Region ................................................. 6-238 Figure 6-14 : The second scenario projection of housing sales in Bangkok Metropolitan Region ................................................. 6-239

Figure 6-15 : The first scenario projection of housing supply in Bangkok Metropolitan Region ................................................. 6-240 Figure 6-16 : The second scenario projection of housing supply in Bangkok Metropolitan Region ................................................. 6-241 Figure 6-17 : The first scenario projection of housing vacancy in Bangkok Metropolitan Region ................................................. 6-242 Figure 6-18 : The second scenario projection of housing vacancy in Bangkok Metropolitan Region ................................................. 6-243 Figure 7-1 : Flow pattern of housing demand model...................................... 7-250 Figure 7-2 : The comparison between simulated housing demand and simulated housing sales ......................................................... 7-254 Figure 7-3 : The comparison between housing supply data and simulated housing supply....................................................... 7-254 Figure 7-4 : Example graph for spring force principle explanation................ 7-261 Figure 7-5 : Housing sales data and the linear trend line compare with the underlying housing-need line........................... 7-262 Figure in appendix D-1: Mathematical prove of linear function for distance issue......... 319 Figure in appendix D-2: Mathematical prove of linear function for area issue... 320 Figure in appendix D-3: Mathematical prove of linear function for triangle area issue ......320 Figure in appendix D-4: Housing sales data and the linear trend line compare with the underlying housing-need line.. 322

LIST OF EQUATIONS

Equation 1-1..1-14 Equation 2-1..2-44 Equation 2-2..2-44 Equation 2-3..2-44 Equation 2-4..2-46 Equation 2-5..2-46 Equation 2-6..2-46 Equation 2-7..2-55 Equation 2-8..2-60 Equation 2-9..2-61 Equation 4-1....4-135 Equation 4-2....4-135 Equation 4-3....4-141 Equation 4-4....4-142 Equation 4-5....4-143 Equation 4-6....4-145 Equation 4-7....4-148 Equation 4-8....4-149 Equation 4-9....4-154 Equation 4-10..4-154 Equation 4-11......4-154 Equation 4-12......4-157 Equation 4-13......4-159 Equation 4-14......4-160 Equation 4-15......4-162 Equation 4-16......4-165 Equation 4-17......4-166 Equation 4-18......4-167 Equation 5-1....5-190 Equation 5-2....5-190

Equation 5-3....5-209 Equation 6-1....6-223 Equation 6-2....6-224 Equation in appendix D-1...........321

ABSTRACT

Housing as a special product- distinguishes the behaviour of its demand and supply. An imbalance of housing supply against demand can be a crucial part of economic crises, as happened in Thailand between 1996 and 1997. Can the housing market be controlled in a robust and rigid system? A secure market depends on balancing demand and supply dynamics; therefore, any demand has to be quantified.

This research demonstrates how housing demand can be modelled by using a System Dynamics approach. This modelling concept has been set up, using the root causes which generate housing demand. Causal factors which influence housing demand are collated, and mapped. A model simulating housing demand was developed. Keys to this are demographic, social and economic factors. This model is presented with a view to pursuing new approaches for housing demand modelling. Conceptual ideas are developed on how to quantify housing demand, and the result of the simulation can then be used as a basis for policy and decision making in housing markets.

The main housing market in Thailand is in the Bangkok Metropolitan Region (BMR), the primate city. It has a population of more than 15 million which is about 20% of the countrys population. Bangkok plays a critical role in the Thai economy with its net real estate worth, being an important national consideration. The BMR -which incorporates central Bangkok and five surrounding provincesforms the case study in this research. The importance of the real estate sector was

underscored in the economic crisis of 1996 -1997, spreading from Thailand and affecting the economies of most Southeast Asian countries.

The housing demand model developed from this research depends on many interrelated factors. These factors can be categorized into three broad groups, following precedent set by a review of available literature. Initial factors included demographics which deal with population number, age structure, including migration, birth and death rate. Next, social factors, in terms of marriage, divorce and splitting-household rate (i.e. household formation rate) play a major role in creating housing needs. The last factor is that of economics: conversion of housing needs units into housing demand units. Income (substituted by GDP growth index in the model), housing loans, and housing price indices, used in the modelling, are the main economic factors which constrain housing demand.

Simulations from the model clearly show that the underlying housing needs line was the crucial factor that limits the capacity of housing consumption in the housing market. Based on 30 years of historical data (from 1977 to 2006), the research also showed that the surplus reached its maximum when the housing sales unit was twice that of the underlying housing needs (as happened between 1994 and 1996). Furthermore, the deficit reached its maximum when the housing sales unit was half that of the underlying housing needs (as happened between 1999 and 2002). With respect to accumulated numbers, the surplus of housing sales units exceeded that of the underlying housing needs and generally reconciled with that housing vacancies in the market.

The evidence from the model simulations also supports a view that the oscillated movement of housing sales is constrained by the balance of reaction between underlying housing needs and housing sales. The cycle of housing sales appears to be running over and under the underlying housing needs line because of the limit of housing consumption capacity, constrained by underlying housing needs.

This research provides new criteria in respect of housing demand research for both private and government sectors. Decisions surrounding investment in the housing market by entrepreneurs -and the development of housing policy by government- will be more efficient and cost effective when housing demand can be quantified. The consequence of this can be anticipated as a significant social benefit through the provision of better housing where and when needed.

This research provides a new and different approach to the traditional econometric modelling. System Dynamics, as used here, allows a more complex approach to modelling to be undertaken, more closely mimicking reality. It provides a basis on which further work can be undertaken within both government and private sectors with a view to achieving better long-run outcomes for the community.

1-1

Chapter 1

INTRODUCTION

Introduction

Housing is a unique market product because of its special features, including durability, fixity and heterogeneity. Its market mechanisms are different from those of other goods. Consequently, the special features of housing distinguish the behaviour of housing demand and supply from the demand-and-supply curves for other goods or commodities. An imbalance of housing supply against demand can be a crucial part of an economic crisis as happened in Thailand during 19961997. In order to balance demand and supply dynamics and to secure the housing market, it is necessary to quantify housing demand.

In this research, System Dynamics is the method used to model housing demand and to seek an accurate method for determining the quantum of such demand. The root of causal factors for the demand is therefore examined, leading to tentative conclusions in order to show how a demand model can be mapped and tested for accuracy. This research also aims to present a study of demographic, social and economic factors which affect housing demand and to formulate a housingdemand model. The data used in this study are sourced from the Bangkok Metropolitan Region in Thailand.

1-2

The problem being approached

In 1997 Thailand faced a serious economic crisis. The causes of the economic crisis were complex, but they had far-reaching consequences. The crisis has particular importance to this research. Within two years, fifty-four financial companies collapsed. Consequently, nearly all housing projects in Thailand (proposed, underway or newly completed) experienced severe financial problems. Housing consumption in the country therefore dropped very quickly. An oversupply of housing then constituted a serious problem for Thailands economic system, exacerbating an already dire situation.

As a general rule when the economic situation is good, housing affordability increases as income rises. Thus, housing markets respond by increasing investment in housing projects. However, when an economy collapses, the capacity to purchase housing in the market is diminished, as clearly demonstrated in Thailand post-1997.

In these circumstances, an oversupply problem within the housing market develops, causing serious problems in the housing markets and for the housing industries. This research, then, starts with the question: Is it possible to capture housing demand numerically at a particular time and in a particular area?

A review of substantive works dealing with this housing demand question Anikeeff (1999); Anily, Hornik & Israeli (1999); Arimah (1992); Banerjee et al. (2004); Carlinger (1973); Charles (2003); Di & Liu (2006); Fu (1991); Gat (2000); Gibb (2000); Hui (2001); Ioannides & Zabel (2003); Liu et al. (1996); Macpherson & Sirmans (1999); Malpezzi & Mayo (1987); Mayo (1994); Miron

1-3

(2004); Pogodzinski & Sass (1990); Serlen (2004); Skaburskis (1999); Wheaton (1990); White (2003); Zhou (1997); Ziegert (1988) informed the development of a methodology for this research, focusing on experiments with different approaches to modelling rather than those traditionally taken and seeking a way to achieve more accurate results.

To model housing demand in the Bangkok Metropolitan Region or anywhere else, all factors influencing that demand must be identified and defined with regards to when, how and to what extent they impact on the level of housing needs within the population and consequent housing demand.

A rational approach might suggest that every household should have one house. Because of the complexity of demand and supply factors, however, it has become accepted that there are no reasonable or accurate methods for determining the precise level of housing demand at any particular point within economic cycles.

The lack of dwellings to meet demand may come about in two main ways. The first cause is either decay or depreciation of housing stock over time, or through natural disaster. It is relatively easy to count the number of houses damaged in these ways, and we can assume that this count is equal to the number of new houses required. The second cause is that a lack of dwellings within any particular region is from an increase (possibly rapid) in housing demand, generated through a population increase and combined with other social and economic factors. These two causes are often complicated in reality, and it is not easy to specify what the housing demand will be at any particular time. It is this second cause on which this research focuses with the intention of developing a means of more accurately tracking housing demand over time.

1-4

From a government viewpoint, it is beneficial to derive information about housing demand. Data on predicted and forecasted housing demand are also necessary for city planning and fiscal policy. The benefit of housing demand data can further suggest to government for land-use planning and building-control policy.

Traditional housing-demand modelling

Current knowledge is unsatisfactory. Housing demand is replaced by the number of housing sales as a proxy. Econometric equation modelling has used housing sales as a proxy for the dependent, variable housing demand. This assumes that, housing sales are equal to housing demands in the market, but it will be shown that this is not the case.

The purpose of this current research is to establish a housing-demand model which gives a similarity to the real world housing markets. It seeks to substantially alter the basic concepts of housing-demand modelling knowledge. Whilst this is an initiate model, the principles and results can be further developed and studied to achieve greater accuracy within the model. For example, there are at least three main processes which produce housing demands as listed:
x

Processional flow from demographic data to curve or forecast housing needs

Processional flow from housing needs data to the generation of an accurate forecast of housing demand

Processional flow from housing demand data to a conversion figure for housing sales

These processes can be further developed to sharpen their accuracy and basic knowledge in order to undertake exploratory research in the future. The benefit of

1-5

this System Dynamics model is the provision of a grounded concept for housingdemand modelling.

Important and relevance of this research

Data on housing demand is necessary for many aspects of public and private sectors in terms of planning and management. In the private sector (considered as a property market), housing comprises the major part of property markets. Housing demand data is a key element in decision-making about property investment. The success or failure of property businesses is related to the ability of entrepreneurs to understand and predict future demand for housing.

In the public sector, housing demand is a key indicator for government in the development and implementation of housing policies and for urban-planning facilities. Inadequate housing is a major problem in the life of ordinary Thais who traditionally concern themselves with four basic needs: housing, clothing, food and medicine. For them, adequate housing is fundamental to individual and family security. From a broader social perspective, a sufficient housing supply promotes economic stability; and good quality housing will improve health and wellbeing.

For these reasons, it is important to develop a clear response to the question: How many houses will be needed at any particular time, and where will they be needed? However, the methods of available statistical analysis do not appear to adequately satisfy the researchers needs or adapt current knowledge to develop reliable housing demand data.

1-6

In housing research, econometric analysis is commonly used to develop demand projections. The number of housing sales is taken to represent housing demand. In reality, the number of housing sales and the demand for housing are two distinct matters. Nonetheless, the distinctions between these two concepts, under differing conditions, are explained.

In simple terms, where the housing supply is less than the demand for housing, it is argued that housing sales are constrained by the housing supply, and demand will exceed the number of house sales. The reality is, however, more complex, and many other factors affect the actual demand curves. These are dealt with in detail later in Chapter 2, and in Chapters 4 and 5 which deal with the model.

Where housing supply is greater than housing demand, it is argued that housing sales are not constrained by the housing supply, resulting in demand that, in all probability, will be equal to; more than; or less than, the number of house sales. In Thailand, the government does not provide housing for rent to low-income earners. It does, however, develop and sell housing to those people. The housing market therefore becomes fully interactive between housing demand and supply without any intervention from the government sector. The Thai government does provide housing for low-income people by developing and selling houses, so in this circumstance the housing supply by government is considered as the same mechanism of housing supply from private sectors in the whole housing market. In either case, the demand for housing can be unequal to the number of house sales when those who are ready to buy delay their purchase for whatever reason.

At any given time, the assumption that housing sales can be used to determine housing demand is imprecise because the relationship between the two is rarely

1-7

equal. Housing demand is a theoretical concept that which is difficult to capture numerically. A new approach to research on housing demand is therefore needed, to more accurately predict demand and supply curves and more accurately model reality.

To capture housing demand numerically, this research project builds a demand picture by examining the root causes of housing demand. The System Dynamics method is used to examine the factors that produce housing demand which allows demand to be counted from the initial sequence through to housing sales as the end point.

The System Dynamics method used in this research is a widely accepted tool for analysing systems. It has the capability to manage the following features: x x x Change through time Non-linear out-put data Feedback loops (Yucun, 2002)

Simulations generated through a System Dynamics methodology can produce predictions about how many housing units are required in a particular area. The benefit from the knowledge contributed by this research is that it offers a means for government agencies to track demand and formulate responsive housing policies. It also provides the private sector with a tool for decision-making about housing investment.

1-8

Definitions of housing demand and housing needs

The terms housing demand and housing needs are often confused. There are some key differences between housing demand and housing needs that must be clarified. The meaning assigned to both terms is erroneously similar in many documents. The following statement is extracted from the study of housingdemand models published by the Housing Branch in Hong Kong which asserts that:

Housing needs is defined as the number of existing or new households requiring adequate housing. An

adequately housed household is one that lives in selfcontained living quarters made of permanent material.

(Liu, Wu, et al. 1996)

They also proffer a workable definition of housing demand, viz:

Housing demand is defined as the number of households actually seeking accommodation. In the public sector, demand is assumed to be equal to housing needs. In the private sector, demand is constrained by affordability.

The Australian Bureau of Statistics states that Housing satisfies a fundamental human need for shelter, security and privacythe extent to which it suits the needs of the people it houses affects the quality of life of the occupants (ABS, 2007). Most housing-need studies are produced by governmental or public organizations (such as Non Government Organizations, or NGOs). The majority

1-9

of these studies attempt to understand or quantify and solve accommodation problems, treating them as basic social problems.

Housing-demand studies are usually undertaken within the private sector. In such studies, demand is often assumed on the basis of a defined number of people wanting to buy or rent accommodation. Certainly, this responds directly to the housing market, and also affects basic economics if a broad view point is taken.

In summary, housing need is a reflection of the basic requirement of obtaining a house to provide adequate living standards. Housing need can be created without any constraints or limits. It has traditionally been seen as a subject within disciplines of social studies. Conversely, housing demand has usually been studied as part of economic research. As affordability is a constraint to housing demand, it can be considered as a general demand for goods (houses) in the economic context. Factors contributing both to housing needs and to housing demand are explained in the next section.

This research then aims to focus on the topic of housing demand only, and specifically on how it is influenced by demographic, social and economic factors. However, housing need produces (or limits) housing demand. As a consequence, any model must be able to account in a robust manner for the dynamism of realworld shifts in demand.

History of housing-demand studies and their factors

Housing demand has been studied for at least thirty years. Despite that long period of research activity, housing analysts still have difficulty accurately assessing housing demand Megbolugbe, Marks et al. (1991). The early model

1-10

used to estimate housing demand was that of Neoclassic Consumer theory, which was later found to be imperfect due to the unique features of housing such as durability, heterogeneity and spatial fixity. Concerns have been expressed about the imperfections of current knowledge about housing demand and available housing models. Megbolugbe et al (p. 381) put the following views:

The fully developed economic theory of the housing market for analysing housing decisions is the

neoclassical consumer theory of housing demand. The reviews of the various modifications that have been made to better operationalise the imperfect and non-

competitive features of the housing market show that these modifications have been introduced in several partial models. These models include tenure choice models, search models, mobility models, and housing trait models. There is currently no single model that incorporates all of the modifications attempted in these partial models. In fact, it may be impossible to operationalise and incorporate all of the modifications of the neoclassical model into a single model. Therefore, the most feasible and conceptually correct research strategy to advance our understanding of housing consumption decisions is to analyse the impact of demographic and social processes on housing consumption decisions (Megbolugbe, Marks et al. 1991)

1-11

Megbolugbe et al also suggest that the most useful way to advance knowledge about housing-consumption behaviour is to study demographic and social factors affecting housing demand. These factors include attitudes, preferences and perceptions of housing consumers. Their ideas are a central element in the approach taken in this research to develop a housing-demand model. Also, their work brings a new approach to capturing housing demand. It provided the initial inspiration to undertake this present research which seeks applicability in the Bangkok Metropolitan Region. In addition to studying demographic and social factors, this research also studies economic factors in developing a housingdemand model.

The effect of demographic factors on housing demand

Demographic factors (e.g. population and immigration) are the related factors that create housing demand Brown (1989); Martin (1989); Wright (1993); Tuccillo (1995); Filardo (1996); Vidaurri (1996); Berson (1997); Prybolsky (1997); Chapman (1999); Emrath (1999); Porter (1999); Blackwell and Partners (2001); Yates (2001); Brown and John (2002); Scholl (2002); Brett, Chen et al. (2003); Sanchata (2003). Wunsch and Termote (1978, p.274) proposed the following definition of demography:

Demography is the study of population, its increase through births and immigration, and its increase through deaths and emigration. In a broader context, demography is also the study of the various determinants of population change and of the impact of population on the world around us. Population factors may therefore be studied

1-12

from a variety of viewpoints, and population sociology or the economics of population.

Recently it has been proposed that understanding housing demand is a key to knowing how much accommodation is needed (Brett, 2003; Brown, 2002; Tuccillo 1994.). Authors such as Berson, Martin and Myers suggest that the two main factors which directly influence housing demand are demographic and economic. In this research, social factors are also included, following other works by Megbolugbe, Filardo, Grimes and Pitakteeratum (1994).

Housing-demand models developed to date are imperfect primarily because they have not embraced demographic factors (Megbolugbe et al.,1991). Others such as Emrath (1999), Berson (1997b), Prybolsky (1996) indicate that demographic factors are crucial, but have been neglected. It is an obvious statement to say that every person needs to have adequate shelter. The Universal Declaration of Human Rights and the International Covenant on Economic, Social and Cultural Rights (Shankardass, 2001) state that;

The right to adequate housing as a component of the right to an adequate standard of living is enshrined in many international instruments. Most notable among these are the Universal Declaration of Human Rights and the International Covenant on Economic, social and Cultural Rights. During the 1990s, the right to adequate housing gained increasing recognition. Since 1996, many Governments have adopted or revised housing policies to include various dimensions of human right.

1-13

As earlier indicated, housing remains a cultural priority in Thailand. Also, as a developing nation, economic conditions are often problematic, making it difficult for governments to plan properly. Some straightforward examples provided below demonstrate how demographic factors need to be considered first if housing demand is to be understood.

In any particular geographic area:

x

If there is one person, then that person represents housing demand of one unit.

If there are five people, then they represent housing demand of between one and five units.

In this case, the estimation of housing demand is starting from one unit because it is possible that all five people would like to live together in one unit.
x

If there are 10 people then they represent housing demand of between two and ten units.

In this case, the estimation of housing demand is starting from two units because it is unlikely possible that all 10 people would like to live together in one unit.
x

If there are N people then they represent housing demand between X and Y units.

Consequently, the basic way to calculate number of a housing unit can be expressed in the following equation from Yucun (2002) and with reference to Tse, (1995):

1-14

N =

P n*h

Equation 1-1

Where, N = number of living quarters P = population n = number of household per unit h = average members per household

There is also a point of concern that age structure (or age profile) will impact on households. The Adelaide City Council has undertaken some major work on social factors. Its website shows how current-age structure is one of the social factors driving population growth and housing markets (City of {Anonymous, 2008 #571}.

At the small area level, the primary drivers of population change are the age structure of the existing population, the housing markets attracted to and away from an area and their associated demographic characteristics (fertility patterns, household types etc.) and the supply of residential land (or potential for rezoning of land), which impacts on the dwelling stock of the area.

1-15

Source: City of Adelaide website (2008-01-28)

Figure 0-1 : Demographic and social factors impact on housing market

It is also asserted that age structure is an important social factor influencing household formation:

The age structure of the local population impacts on the household types and size, the likelihood of the local population having children and to die, as well as the propensity for people to move. Age specific propensities for a population to have children or die are applied to each small area's base population. An older population will have fewer births, more deaths, while a younger population will have vice versa.

The figure from Adelaide City Council website also describes the relation of household types and population changes: The sorts of households that people live in and changing preferences over time affects the way in which a population changes. As people grow from children to adults and into old age, they change the sorts of

1-16

households that they live in. The traditional path has been to start as a child in a family household, move into a group or lone person household as a youth, becoming a part of a couple relationship within 5-10 years. Rearing of children is followed by an empty-nester period and ultimately being a lone person, as partners die. Capturing the changes that people make by age through their life is a key driver and the way in which this is changing, with a greater preference to live alone or as a couple without children.

Source: City of Adelaide website (2008-01-28)

Figure 0-2 : Development of household types

1-17

Salt (2005) confirms that the proportion of household types in Australia is changing. The major proportion of households was nuclear families in 1991, but by 2031, families and couples are expected to be replaced by singles.

Source: http://www.bernardsalt.com.au/ (2008-2-13)

Figure 0-3 : Ratio and trend of household types in Australia from 1991 to 2031

From the figure above, it can be said that demographic factors, such as population, age profile (structure) and household structure are logically main factors in determining the extension of a household, the type of household and, consequentially, housing demand. In other words, changes in population size (either an increase or decrease in the number of people) directly affects the extension of family and also housing demand (Borrie, 1995).

1-18

The chief causes of population increase and decrease are stated below.
x Population increase is caused by:

- Fertility rates which represent the actual level of childbearing within a population (Marshall,1994) - Immigration, the process of entering another area in order to live there (Oxford Dictionary 2002)
x Population decrease is caused by:

- Mortality rates which represent deaths standardized, usually by age and sex to facilitate comparison between areas and social groups (Oxford Dictionary 2002) - Emigration, the process of leaving a country or area in order to live in another area or country (Oxford Dictionary 2002)

Some studies dealing with demographic change (Chapman,1999; Yates, 2002) state that other factors should be considered, including qualitative factors (such as moving to a new job), lifestyle, age structure and policy change. Such studies report that the factors which most influence housing demand are the number of teenagers in a household, the number of working people and the number of elderly people. Thomas and Malmberg (2005) found evidence in Sweden that:

We find as expected that large groups of young adults are associated with higher rates of residential construction. More interestingly, we find a significant negative effect from the age groups above 75.

1-19

Theoretically this may be explained by a demand pattern for age groups derived from a life cycle with finite life and uncertain time of death combined with the strongly increasing, mortality rates at that age. The data collected from Thailand shows the similar trend of negative effects in the age group of 55-59 and above 70. This can be seen clearly in Table 1-1, where the numbers of change per age group (the last column) are minus number.

Table 0-1 : Number of single-head households in BMR, Thailand, 2005

Single Age Group household head (thousand) Total 0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90 1,949.8 3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4

Change per age group (thousand)

3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 -58.0

Source: composite figures from governmental sources.

1-20

Myers and Vidaurri (1990) support the influence of demographics: the effect of changing demographics is felt in markets for all types of real estate. In housing markets demographic factors shape the number of households formed and the type of housing selected.

Demographic factors are one of the three main factors initially affecting housing demand. The demographic factors considered in this research are fertility, mortality and migration. Demographic factors alone, however, will not produce a viable model of housing demand.

The effect of economic factors on housing demand

While demographic factors influence housing needs, economic factors such as income, housing price and interest rate also affect housing affordability (Carlinger, (1973); Kau and Keenan (1980); Fik, Ling et al. (2003); Francke (2004); Gelfand (2004); Miron (2004); Wang (2004)).

An explanation of the relationship between demographic and economic factors on housing-consumption decisions is found in the Canadian Housing Observer (Howell , 2003):

Housing market activity is strongly influenced by demographic and socio-economic trends. Both the growth of the population and its characteristics influence the rate of household formation which, in turn, is a key driver of housing demand. While rapidly growing populations tend to generate more housing demand than slow-growing populations, housing needs and preferences are also

1-21

shaped by the characteristics of the individuals in the population, particularly by their age and family status. Demographic changes are not the whole story, however, for in order to act on their preferences, people must have sufficient financial resources either income or wealth or a combination of the two.

The relationship between economic factors and housing demand has become more important. Many countries, including Thailand, have a capitalist economic system whereby supply and demand of any good are determined by market forces. In this circumstance, the influence of economic factors, such as household income, housing prices, required repayments on housing loans, and interest rates, play a significant role in determining housing demand (Ellis, 2003). These variables are taken into account in the construction of the model used in this research.

Liu, Wu et al. (1996) found that those variables were critical in their housingdemand study. Although their work was based on Hong Kong, their basic approach is considered important to the context of this research as their social and economic factor approach is similar. Their research involved four steps to establish the dwelling requirement for the territory. The first stage in their research design was to formulate the accommodation generation rate as a means of estimating housing needs. The second stage was to apply a splitting ratio by separating total housing needs into public and private housing needs. The third stage was to convert housing needs into housing demand. Finally, the fourth stage was to convert units of housing demand into the required number of dwellings.

1-22

However, even though the research of Liu et al, which includes social and economic factor variables such as economic growth, property price indices are not parameters in their model. The model experimented in this research then includes those variables to concern the influent against housing demand.

The effect of social factors on housing demand

The second main group of factors effecting housing demand are social (Filardo, 1996). These factors concern household formation which is influenced by marriage, divorce and other circumstances, such as young people leaving home. These factors may lead to an increase or decrease in housing demand depending on social behaviour.

Moreover, Megbolugbe, Marks et al. (1991) support the argument that there is a need to undertake research on how to include demographic and social constructs in meaningful housing-demand studies that capture the attitudes, preferences, and perceptions of the consumer.

In Thailand, Pitakteeratum (1994) undertook some pioneering research in Thailand prior to 1994. According to her research, the factors affecting housing demand are the housing-price index, population, housing repayments (and therefore interest rates on loans) and government policy. From her research, which used secondary data only, she recommended that further research should collect primary data and try to account for other factors to improve the accuracy of housing-demand models.

1-23

From these basic studies of housing-demand factors, and an analysis of methodologies employed to date, it can be seen that it is critical to analyse the influence of demographic, economic and social factors, including their interrelated impact on housing-demand patterns. This should then enable valid theoretical bases to be formulated, specifying the impact of all these factors on housing demand and on the sale of existing housing. The most recent paper on housing demand in Thailand, presented at the Urban Research Symposium 2007 Conference, states that not only demographic and socio-economic characters determining the potential demand for housing, but also current housing characteristics and their property rights determine the potential demand for housing (Choiejit and Houbcharaun, 2007). Their research used Binary Logistics Regression to estimate housing demand in Bangkok and its vicinity.

However, in this model, the variety of characteristics of housing in the market is assumed (i.e. from holistic viewpoint) to match the preferences of the housing buyer. Thus, this research aims to investigate the relationship between these three groups of factors (i.e. demographic, social and economic) against housing demand and to develop and modify a housing-demand model based on the principles of System Dynamics.

Consequence of all factors to the model of this research

An initial preview shows that there are three main factor groupings affecting housing demand. They are demographic, social and economic factors. Each group affects housing need and demand in different ways. System Dynamics is the methodology used for processing the model in this research. The related

1-24

discussion on the justification of using System Dynamics as the methodology is found in Chapter 2.

Beginning with the influence of population numbers, demographic factors are an originating cause of housing needs in terms of the number of people and their characteristics. If housing has been specifically built for the number of people seeking housing, then it would be the case that the number of people seeking housing has exactly influenced the number of houses built.

Next, social factors set the timing for peoples housing needs. People prefer to live privately and separately when they grow up, marry or divorce. Therefore, adults form the most influential group in determining housing needs (Liu, Wu et al. (1996); Thomas and Malmberg (2000)). Economic factors determine affordability: a function that acts as a constraint on housing demand (Liu, Wu et al. 1996).

In this research, the three factors are sequentially brought together to generate a number that quantifies housing demand (i.e. expressed in units) by using the System Dynamics method. Results are then compared with real data from 1977 to 2007. The simulation run is then projected into the future for 20 years (i.e. from 2008 to 2027) in order to predict housing trends over that period in the Bangkok Metropolitan Region.

1-25

Model development process

The research model development process was divided into six stages. Each stage is described below.
Stage 1

aimed to develop a comprehensive picture of housing demand based on an exhaustive survey of the available literature on housing demand. During this stage, it was planned to define the scope of research and to establish a structure for the research.

Stage 2

was to establish a structure for the housing-demand model which is incorporated each factor. This stage was intended to identify factors or groups of factors for modelling. However, in their final state, the factors used in the model may vary from their initial state, due to calibrations performed to secure the aim of the modelling.

Stage 3

selected the method of modelling. In this stage, it was planned that the researcher would review the knowledge base of modelling to choose a suitable method for modelling housing demand in Bangkok Metropolitan Region. It was intended that the advantages and disadvantages of various methods would be identified. At this stage, data about selected factors would be collected from the field. It was also intended that a decision would be made at the same time about the method of research modelling pursued. In this research, the data and modelling method were synchronised in order to match each others conditions or limitations.

1-26

Stages 4

aimed to set, test and calibrate the model. It was recognised that this stage may take place over a long period and require many revisions until a robust model is established producing valid results.

Stage 5

verified the validity of the model. The criteria for verification of the model are dependent on the models type and concept of used modelling. These are described in Chapter 6.

Stages 6

the findings were written up,

The processes outlined above provide a rigourous way of determining an answer to the primary research question. However, the main idea was to establish a model that could be readily conceptualised. Detail in respect of that conceptualisation is found in Chapter 4.

Conceptualising the research methodology and the housing-demand model

Once a picture of the housing-demand context in the Bangkok Metropolitan Region was established, it became necessary to identify factors for inclusion in the housing-demand model. This required the mapping of relationships between the factors, which were then represented diagrammatically. Mapping and data collection took place over the same period of time. Where missing data were encountered, they were substituted by using results from linear regression equations or trends. In the case of linear-regression equations, data giving the highest coefficient of determination (R2) were utilised. Where trends were employed, the data providing outcomes that could be reconciled with historical data were substituted.

1-27

Undertaking the data collection

An exhaustive process of exploratory research was then undertaken, using the secondary data in various combinations to arrive at a model that best reflected reality. This model was then tested and retested. Data were run through the model to test it in various ways. This was done to determine the ideal combination of factors and relationships to achieve the best fit for the model. The results from experimentation with the model explained the relationship between different combinations of final factors in the model. The final set of factors used in the model were selected and arranged to reconcile with historical data. All data used in this research are secondary data. Population data (i.e. from the National Statistical Office of Thailand from 1990-2006) was used because it is more precise, serving as a better proxy than any primary sample group data. The benefit of using population data rather than sample-group data is that the population data can provide full and real information from the field, while sample-group data need the deductive paraphrase for information used. As a general principle, the larger the sample size, the more accurate the result. (Benedict and Szatrowski, 1989).

This research was conducted as a case study of the Bangkok Metropolitan Region (BMR) for which the most complete data sets are available and (secondary) population data offered from BMR is the most complete and robust. Even though some data were unavailable or missing, the proxies used reflected similarity and were therefore adopted. Another reason for choosing BMR as a case study is because it occupies the main market share of housing markets in Thailand. As Bangkok is a primate city, that is a major city that works as the financial, political and education hub of the country, it is not rivaled in similarity of these aspects by

1-28

any other city in Thailand. Normally, a primate city means the city which must be at least twice as populous as the second largest city in the country. The ratio of market share of housing loan (i.e. from Government Housing Bank) between BMR and all other regions is, by average, about 60 % from 1995 to 2004.

In terms of data collection, population and social data were gathered from sources in Thailand including National Statistical Office (NSO). Economic data were gathered from Office of the National Economic and Social Development Board, National Housing Authority and Department of Business Economics.

Analyse the results of model simulations

Modelling requires equation setting, data analysis, management of missing data and calibration of the model. Scenarios are designed to run through the model as simulations, testing the results and correcting errors in the simulations.

The System Dynamics model was selected as the modelling method for this research. iThink software was the System Dynamics program used in modelsimulation testing. Further discussions on comparative benefits of this approach are undertaken later in this chapter and in Chapter 4.

Research methodology

The purpose of this research is to establish a new approach by experimenting with new concepts to model housing demand. The research combined several research strategies as follows:
x

Theoretical research which intends to study in the frame of a science theory to expand or develop a new knowledge or theory

1-29

Empirical research which applies a principle or theory to explain how is the behaviour of any focal factor changing and its relation

x x

Applied research which has its principal use in city planing and policy Experimental research which uses data for trial by computer-simulation program (iThink software program) to test whether which factors are the best logical and influential to housing demand

This research is also identical to exploratory and predictive research in terms of its modelling processes and forecasting ability.

Research objectives and aims

In this research, a model based on the System Dynamics method is utilised to show stock numbers and their flow rate for elements of each factor. The System Dynamics approach is adopted in the research because the behavioural pattern associated with housing demand is similar to the behavioural pattern associated with the stock-flow models.

In the first step, the number set for the stock of population is the number of total population, which flows through the valve (or equations) that converts them to a number or group of population (i.e. households) and those who need houses (i.e. housing-need number). The equation calculates this number from the stock of population to the number of household (and also any other steps). This is clearly demonstrated in the discussion found in Chapter 4. The unit is changed from

people to households. This number then flows through the valve again to

1-30

become the number of population (or household) - those who need houseing and can afford to buy a house (i.e. they are then recognised as a housing demand number).

Housing demand units in this model are assumed to be homogeneous features in terms of number only. The feature of heterogeneity of housing still exists in the concept of this model. However, the variety of housing is assigned to meet the various ability of each person to purchase. For example, people, who have housing needs and the ability to pay but want to buy a different type of housing than the one they can afford, are assigned the same unit number as those who can afford to buy other particular housing types. Therefore, one housing demand unit in this model means one person or one group of people who are willing and able to buy one housing unit as they require at a particular time.

Thus, this research aims to imitate the housing demand environment in BMR in order to show how that demand is created and which people or groups play key roles which impact on movements in housing demand. The model seeks to capture and explain all the main factors in realistic simulations.

The currently unsatisfactory situation is that housing demand is substituted by a housing sales number. Housing sales have often been used as a proxy for housing demand, as dependent variables in econometric equation modelling. In this case, housing sales are assumed to be equal to housing demand in the market. The imperfection of using housing sales to represent housing demand as a proxy is mentioned further in the Literature Review (Chapter 2).

1-31

The purpose of this research is to establish a principle for housing-demand modelling which is the most similar to reality in housing markets. The benefit will be to redefine basic concepts in housing-demand modelling knowledge. As a pilot trial, the principles and results of this research can be further developed in the future to suit many purposes. For example, there are at least three main processes which produced housing demand as listed below.
x x x

the use of demographic data to generate housing needs the use of housing needs data to generate housing demand the use of housing demand data to achieve a forecast of housing sales

These processes can be developed further to sharpen accuracy and achieve an initial basic knowledge for exploratory research in the future. The benefit of this research (i.e. model) then provides the grounded concept of housing-demand modelling. The aims and objectives of this research study are set out below.

The aims of this research are therefore:

To study the housing demand life cycle from its initial state to its end state.

To clearly understand housing demand behaviour and the implications of this behaviour for government policy, so that better control is exerted over future supply imbalances in the housing industry.

1-32

The objectives of this research are therefore:

To investigate the relativity of three factors (i.e. demographic, social and economic factors) that influence housing demand.

x x

To calculate an accurate figure for housing demand. To classify main factors which, when modelled, can best accord with reality.

To develop a housing-demand model that reflects the impacts of those demographic, social and economic factors for the Bangkok Metropolitan Region.

Research questions and sub-questions:

How do demographic, social and economic factors affect housing demand in Bangkok Metropolitan Region, Thailand?

How do the three groups of factors interrelate to influence housing demand in Bangkok Metropolitan Region, Thailand?i

From these research questions, the following subsidiary questions arise:

x x x

What demographic factors impact on housing needs? What social factors impact on housing demand? What economic factors impact on housing demand and housing sale?

Relevant literature was investigated to determine how each group of factors might be best arranged in a model of housing demand. The literature also provided a broad overview of housing circumstances in conditions, similar to those in the Bangkok Metropolitan Region. Demographic factors selected for modelling,

1-33

following review of the literature, include birth rate, death rate and migration rate. Social factors selected include demographic age structure, social characteristics which determine household formation (i.e. marriage, divorce and splitting singlehead household formationcalculated from proxy data described in Chapter 5). Economic factors selected for modelling include the GDP growth index, further housing policy (i.e. housing loan from bank) and the housing price index. Previously, conducted studies dealing with housing-demand models and the history of the housing market in Bangkok Metropolitan Region were consulted as part of the review in Chapter 3.

Chapter summary

This chapter has provided an overview of the background to problems, research questions, research methods and benefits of this research. The background to housing demand is described in this chapter as information of the case study area (i.e. Bangkok Metropolitan Region). The research is detailed in the following six chapters:

Chapter 2Literature review

This chapter reviews the extant knowledge on the subject and depicts the background to bridge the gap apparent in that knowledge. It also suggests the method and programming of the model developed in this research.
Chapter 3Background and information of the case study area

This chapter depicts all information involving Bangkok Metropolitan Region as a major and important housing-market share in Thailand. Some basic background

1-34

about Thailand is also described in this chapter. Recent developments within the housing market environment in the Bangkok Metropolitan Region are examined.
Chapter 4Overview and development of a housing-demand model

This chapter explains what was found in this research, such as housing-demand model and the research methodology used. It also deals with concepts, assumptions, processes and limitations, including the map of the model.
Chapter 5Simulation of the housing-demand model

In Chapter Five, the model simulations are illustrated and the verifications of the model are presented. Comparisons between simulation run and historical data are incorporated in the discussion.
Chapter 6Model forecasting and scenario analysis

The sixth Chapter explains model forecasting. The forecast is presented in two scenarios. The first scenario is set on the basis that all factors grow in line with historical data trends. The second scenario assumes that economic factors will repeat the cycle from the last ten years (i.e. between 1998 and 2007). The data used in the model are projected for the next 10 years to show the trends. Then, a description of the models forecasting ability is incorporated showing where the housing market may be expected to be over the next 20 years (i.e. from 2008 2027).
Chapter 7Conclusions and recommendations

1-35

The final chapter deals with conclusions and recommendations. Two theories found in this research (i.e. from the model) are also displayed. Research contribution is suggested as well as future directions concerned by this research.

Terminology definition of housing types in this research (for Bangkok Metropolitan Region)

To ensure a uniformity of understanding, this section provides definitions of the terms used in this thesis. Terminologies of housing types in Bangkok Metropolitan Region are defined. Housing types in Bangkok Metropolitan Region can be classified into five types described below:
Detached house (Single house)
x x

a house on its own block of land containing between two and seven bedrooms (there are only a small number of properties with more than five bedrooms)

usually having a fenced backyard and fenced front yard

Semi -detached houses (Twin house)

x x x x x

also known as half-a-house one building usually contains two houses located on its own block of land and units divided by a common wall each unit contains one to three bedrooms, and, each unit has a fenced front or back yard
Attached house (Town house)

x x x

three or more units built next to each other and divided by common walls generally two floors each unit contains between one and four bedrooms, and,

1-36

each unit usually has its own small fenced yard or courtyard
Commercial building and shop house building

x x x x x

often multi-storey, commercial use on ground floor less than ten units built next to each other and divided by common walls generally two to four floors each unit contains between one and four bedrooms, and, each unit has its front face to a road or street
Apartment (i.e. flat and condominium)

x x x x x

contains between one and three bedrooms usually in a complex of two or more floors each apartment is on one level some complexes have a communal laundry, and, generally has a communal yard/garden Many documents and information sources consulted for this research classify housing into three main types. The first type comprises houses, which include detached houses, semi-detached house, and attached houses. The second type consists of commercial buildings and shop-house buildings. The third type comprises apartments, which include flats and condominiums.

This research however uses particular terms in the following manner (A-Z):

Absolute housing demand is the term used to refer to the housing

demand number, counted by using the principles and calculations applied in this research.

1-37

Adolescent is the term used to refer to people from 0 to 19 years of

age.
x x

Attached house is the term used to refer to town house.

Condominium is the term used to refer to apartments, flats and condominiums.

Couple head household family is the term used to refer to a family

which consists of two parents (in most case it appear as father and mother) who both take care of their family.
x x x

Detached house is the term used to refer to a single house. Elder is the term used to refer to people 60 years of age and over. Existing housing market is the term used to refer to the re-sale

housing market.
x

Extended family is the term used to refer to a family group that

consists of parents, children, and other close relatives, often living in close proximity or a group of relatives, such as those of three generations, who live in close geographic proximity rather than under the same roof.
x

House is the term used to refer to detached houses, semi-detached and attached houses.

Housing affordability has no uniform definition. A widely accepted implicit definition states that monthly housing costs for adequate housing should be no more than 30% of household income. As such, housing affordability can be measured as a ratio of housing payments to household income, adjusted for household size. Housing affordability in this research then is the term used to refer to purchasing power of people with willingness to buy housing unit.

1-38

Housing demand is the term used to refer to willingness to buy a

housing unit, matched with the peoples financial ability to buy a house or houses.
x

Housing demand unit (HDU) is the term used to refer to number of

person or persons who need to buy a house for living or investment and having the financial ability to buy.
x

Housing needs unit (HNU) is the term used to refer to number of a

person or persons who need to buy a house for living or investing even have or have no financial ability to buy.
x

Housing needs without affordability unit (HNwoA) is the term used

to refer to the number of a person or persons who need to buy a house for living or investing but have no ability to buy due to lack of founds.
x

Housing needs is the term used to refer to needs of people to have a

house to live in.

x

Housing sales unit (HSU) is the term used to refer to number of

houses that have been sold to persons in housing demand unit.

x

Housing unit is the term used to refer to all types of housing, counted as the same unit.

x x

Housing is the term used to refer to all types of housing.

Investment housing demand or Speculation housing demand is

the term used to refer to housing demand unit of people who want to buy a house for investment or speculation purpose.
x x

Mature is the term used to refer to people from 20 to 59 years of age. Nuclear family is the term used to refer to a family unit consisting of a

mother and father and their children.

1-39

Semi detached house is the term used to refer to two houses sharing a

party wall.
x

Shop-house or block is the term used to refer to commercial buildings and shop-house buildings.

Single head household family is the term used to refer to family

which consist of only one person or one parent who take a leadership role in that family.
x

Bangkok Metropolitan Region or BMR is the term used to refer to

Bangkok and five provinces around Bangkok: Samut Prakarn, Nontaburi and Patum Thanee, Samut Sakorn and Nakorn Patom. Altogether, it covers an area of 7,761.50 km, and with a registered population of 9,988,400 (as of April 4, 2007) it has a population density of 1,286.92 per km.
x

Underlying housing needs units(UHNUs) is the term used to refer to the number of housing needs that has been counted by using the principle and calculation in this research to simulate number of persons who need to buy a house for living in or investment even though they have or have no financial ability to afford a house. The principle (i.e. from this research) to produce the underlying housing needs units comes from two main factors: demographic and social factors.

2-40

Chapter 2

LITERATURE REVIEW

Introduction

Housing is one of the basic needs in human life. It occupies approximately 75% of urban area (Pongsatat, 1998). In other words, housing is both the most significant element in urbanization and the main cause of the spread of suburban areas, which impacts in many ways on people. Housing demand research supports government housing policy planning and development, and it is a key input to private sector decision making in respect of housing investment (Pitakteeratum, 1994). This literature reviews examines the history of housing knowledge as well as highlighting gaps in current knowledge. A new proposal for a housing-demand model is depicted and developed in this chapter before trial and test in Chapter Five.

The context for a review of housing demand literature

Housing demand has been the subject of research for at least three decades, and researchers have established many theories and models, and produced much evidence about it. There have been many discoveries and much argument about the most accurate way of obtaining housing demand estimations and predictions for any specified area. This review surveys previous research and introduces the concept of housing-demand investigations.

There is some confusion or lack of clarity, among themes; and differences

2-41

between the definitions of housing demand and need. Prior to discussing the housing demand, the first necessary step is to determine what defines housing demand and need, and what differentiates the two notions.

Housing-demand studies and housing-need studies

While the terms housing needs and housing demand appear regularly in the literature, they are studied for different purposes and in different ways. Most housing-demand studies have been undertaken in developed countries. This is because in these countries, there seems to be a sufficiency of dwellings for the vast majority of people. In developing countries, housing studies focus on housing needs because of the relative poverty of the people. Studies of housing needs often involve issues about inadequate shelter, homelessness and poverty. They are directly concerned with standards of living, Moreover, some believe that the provision of housing for people in developed countries is the responsibility of governments (Hudson, 1997). In contrast, housing-demand studies emphasize economic factors and housing markets. Housing-demand studies have found that demand is varied and impacted by economic factors, such as income and housing prices (Apps, 1971).

In social science terms, housing satisfies fundamental human needs for shelter, security and privacy. The quality of life of occupants is influenced to the extent to which it suits their needs. Therefore, most housing-need studies are produced by governments or by public organization, such as Non Government Organizations (NGOs) in an effort to understand the social problem that housing represents and to solve it.

2-42

On the other hand, most housing-demand studies are performed by economists from the private sectors viewpoints. In this case, it is assumed that housing demand can be quantified as the number of people wanting to buy or rent accommodation. Certainly, housing demand responds directly to housing markets and affects the broad economic situation.

Although studies of housing needs and housing demand are different in many respects, both have similarities in that they struggle to identify workable definitions and measures. There has been such a debate in Australia and overseas about the definition of housing needs. However, some direct quotes from literature to clarify and conclude that argument are shown next.

Often, the purpose for examining housing needs determines what elements are included in a measure (Karmel, 1998). Karmel neatly and clearly limits the scope of the study of housing needs in the following manner:

People have housing needs if they cannot afford their current housing or their current housing is not appropriate and adequate and they cannot afford to rent appropriate and adequate housing: or, despite being able to afford it, they cannot obtain appropriate and adequate housing due to environment conditions such as discrimination or lack of suitable accommodation.

People are not considered to have housing needs if they can afford appropriate and adequate housing, but choose to live in an inadequate dwelling.

2-43

Therefore (in this research), housing affordability is the main factor in the model to concerning the change between housing needs and housing demands.

Furthermore, Megbolugbe, Marks et al. (1991) state that, despite thirty years of modelling housing markets, housing analysts still have difficulty in assessing housing demand accurately. In order to bridge this gap in housing demand research, it is necessary to identify the imperfections in our knowledge, which is the task of the following section.

History of housing-demand model studies

A review of literature revealed housing-demand studies that analyse factors which, directly or indirectly, influence housing demand and consumption. Methodologies from these studies can be adopted, and the results can be used to support housing market analysis (Brecht, 1998).

The neoclassical consumer theory of housing demand was the first economic theory of housing markets which supported analysis of housing consumption (Megbolugbe, Marks et al. 1991). The neoclassic economic model uses three main factors to calculate the equation: income factors, price factors and taste factors.

The neoclassical economic theory of consumption was applied subsequently to housing demand within the framework created by Muth (1960) and Olsen. (1969). First, the household decision making is assumed to be similar to the consumer decision making. Second, the object of consumer decision-making is not considered as the observable heterogeneous commodity of housing, but an unobservable homogenous commodity or housing services. Third, the competitive housing market is assumed to be a perfect market. In addition, it is expected to

2-44

operate in a tax-free world where capital and asset markets are perfect in equilibrium (Megbolugbe, Marks et al. 1991).

The general form of the housing-demand equation put forward by Megbolugbe, Marks & Schwartz (1991) is:

Q = q (Y, Ph, Po, T), Where, Q Y Ph Po T = housing consumption = household income = the relative price of housing = a vector of price of other goods and services = a vector of taste factors

Equation 2-1

Several studies include a vector of household characteristics (e.g. age, marital status, household consumption) as factors which can capture differences in consumer preferences (Megbolugbe, Marks et al. 1991).

If Where, T H = a vector of taste factors

T = t (H),

Equation 2-2

= a vector of household characteristics

Then

Q = q (Y, Ph, Po, H),

Equation 2-3

2-45

Where, Q Y Ph Po H = housing consumption = household income = the relative price of housing = a vector of price of other goods and services = a vector of household characteristics

The housing economics literature also interestingly addresses some special features of housing demand. Several authors including Quigley, Follain & Jimenez, Smith & Rosen, Fallis and Megbolugbe, Marks et al. (1991) presented a modification of simpler neoclassical equilibrium models of housing demand drawing on the special features which distinguish housing .

The distinguishing features of housing (as goods in an economy)

Housing is a special good. There are some characteristics indicating that market mechanisms for housing are different from those of other goods or services. These factors determine demand and supply in the market. To model housing demand, a clear knowledge of conditions affecting housing is required. The distinguishing features of housing are listed and described below.

Durability of housing

Durability is the dominant characteristic of the standing stock of housing units. This specific feature is an important reason for considering the housing market in two separate sectors. Firstly, demand for housing services is driven by consumers whose demand is modelled in consumption theories which can be similar to any

2-46

other consumable goods demand. Secondly, demand for housing stock is driven by investors whose demand is modelled in investment theories (Richard and Paul (1993); Boehm and McKenzie 1982; Linneman 1986)).

The second sector is the distinguishing feature due to durability of housing. Different from other consumable goods, houses can be rent out after the first client buys it. With this consequent, demand for housing can be concern into two aspects which are buying and renting.

Lee and Trost (1978) combined two characteristics in the tenure choice model which is a substantial advance on the explanatory power of the traditional neoclassic model. They use the following equation to explain the relationships between housing ownership and renting:

I =

I (X) + u

Equation 2-4

Qo = qo (Zo) + uo

Equation 2-5

Qr = qr (Zr) + ur

Equation 2-6

The tenure choice equations above, describes:

x

The unobservable index, I, the households likelihood of owning versus renting, as a function of the independent variables, X.

The owners demand for housing services, Qo is function of variable Zo.

2-47

The renters demand for housing services, Qr is function of the variable Zr.

The Us are disturbance terms.

To sum up, the feature of housing as a durable good allows itself to be utilized or consumed over a long period of time. Houses can be rented to suit the housingservice needs of tenants. Then, they are the goods that make profit to the owner (i.e. landlord) from the rental fee. This process creates the second tier of housingdemand units which, as a segment of the market, is for housing-investment purposes. These two conditions lead to an important concern regarding economic factors in this researchs model. However, renters demand is not the main point of this model. It has been considered as a temporary dwelling for those people or household who do not meet the financial ability to buy housing.

Heterogeneity of housing

Heterogeneity is an inherent characteristic of housing. It has been raised as a weakness of previous housing-demand models which have been characteristically based on standard econometric analysis and applied to a composite commodity whose units are homogeneous (Muth, 1969). For example, housing units which have the same price can differ in size, age and capacity of service. Models which emphasize the heterogeneity of housing are not the focus of this research. However, it is worth noting that the housing heterogeneity model is based on the postulate that households value goods for their characteristics (Becker (1987), Lancaster (1966) and Muth (1966)). The model set out in this paper assumes that

2-48

housing is a good which responds to housing service and housing investment in homogeneous way.

Spatial fixity of housing

In housing markets, location is the main point of interests for people wanting to buy or rent accommodation. Previous housing-demand models group location factors, relying mostly on the characteristics approach (Linneman, 1986). Although fixity of housing has been seen as a part of housing heterogeneity, there are many location specifications affecting housing demand and must be considered. These include distance from important locations like the central business district, shopping areas and transportation routes; the nature of land use in neighbourhood of the house; the socio-economic character of neighbourhood; the physical nature of the neighbourhood environment (Megbolugbe, Marks et al., 1991).

Summary of housing features and previous models

The specific features of housing and the limitations of the previous models (i.e. neoclassical models) mean that those models are based on inadequate information when trying to reach conclusions about housing demand. Even though many models have developed from neoclassic economic theories of housing demand, most seek to solve the equation by specific limitations on variation. Some models attempt to prove the influence of relationships between housing demand and some specific factors such as location and neighbourhood choice Ioannides and Zabel (2003). Others Carlinger (1973); Dooley and Ann (1977); Malpezzi and Mayo (1987); Wheaton (1990); Fu (1991); Arimah (1992); Zhou (1997); Skaburskis (1999); Fik, Ling et al. (2003); Ioannides and Zabel (2003); Banerjee, Gelfand et

2-49

al. (2004); Case, Clapp et al. (2004); Francke (2004) all try to develop the accuracy of many sets of variables as an example economic variable set such as income and price.

All of these models, however, consider factors which can be categorised into two groups. The first group includes factors that specify housing characteristics. The second group includes economic factors (such as income, price, and housing loans) because most of the models were developed from the original neoclassicaleconomic theories of housing demand, and emphasise the capture of data on consumer behaviour in housing consumption and the condition of the housing market.

Such housing-demand models attempt to calculate housing demand by selecting significant variables which are developed from the initial economic model. The assumptions behind these models are questionable because most of them overlook the main demographic factors. In fact, most of the housing consumption decisions that households face over their life cycle are predictable (Megbolugbe, Marks et al. 1991)

The system dynamics model used in this study attempts to overcome the shortcomings of the previous modelling, thereby quantifying demand more accurately. The system dynamics model connects together many equations as one long equation. When the input data or the equation is wrong, the result is obviously wrong. The system dynamics model has the special feature of showing all steps leading to the results. Therefore, it is easy to detect any error. The equations for every stock-unit and flow-unit in the system dynamics model can be

2-50

calibrated: the models calibration requires the user to input the data and set the equation in a reasoned way.

Existing research into housing-demand models fails to account for the root causes that produce housing demand. A system dynamics method to develop a housingdemand model from the bottom up approach with a view to achieving a more realistic and accurate outcome is considered to be an improvement on the traditional thinking. The steps of each consequential setting in this research model are based on evidence from previous research. In this study, all factors affecting housing demand are grouped under one of three main factors: demographic, social or economic. As all factors are dealt with consequentially, this research model begins with demographic factors, and then deals with social factors before finally dealing to economic factors. It therefore mimics reality far more closely than other methods do.

This research approaches housing demand from different points of view. Beginning with demographic aspect, children do not create housing demand, but they create demand when they leave their parents households (Di and Liu, 2006), marry, have children, divorce, retire or change jobs (Leary, 2004). These activities can stimulate housing demand in housing markets. Research in many countries is concerned with the effect of demographic factors on housing demand (Aluvaala, 1993). The next section introduces several approaches from recent research which includes demographic factors.

2-51

Several approaches to the study of housing-demand study

This section reviews some seminal housing-demand studies based on analysis of demographic and social factors.

The Hong Kong study of housing demand used a model published in 1996 by the Working Group Housing Demand (WGHD) which was traces of conceptual and methodological origins to the Working Group on Long Term Housing Strategy Review (LTHSR). This model (used counting approach) calculates the specific level of housing demand by reference only to demographic data, such as first marriages, divorces, immigration and splitting rate of any type of family household. This model then, is different from several models using economics. The model, described by Liu, Wu et al. (1996), comprises four steps as set out below.

x x x x

Projection of total housing needs Categorization of housing needs into public and private housing Derivation of housing demand from the sectored housing needs Estimate of sectored flat production requirements

Liu, Wu et al. (1996) explain that, according to the Housing Panel of the Legislative Council of Hong Kongs document referred to their research, economic variables such as economic growth and property price index are not included as the parameters in their model and are assumed to keep similar with past trends. The authors nonetheless concede that economic variables, such as GDP growth, property-price index and income growth may affect housing demand.

2-52

In Thailand, Maneekeatpaiboon (1997) analysed several housing-demand models in her research. One of these models A Household Model for Economic and Social Studies (HOMES) was built to capture the structure of households in order to find relationships between households and housing demand. This model emphasized only demographic variables affecting housing demand and did not include economic factors.

The study by Burmham Campbell and Nipon Poapongsakorn (1998), which used 1970 census data from Thailands Department of Interior, showed that the primary factor influencing housing demand is the number of people in the population. They concluded that economic factors such as average income and the housing price index do not influence housing demand.

While some studies stress only the influence of demographic and social factors on housing demand, some authors consider both demographic and economic factors.

Merging demographic, social and economic factors in housing-demand studies

Yates (2001) asserted that demographic, lifestyle, economic factors and policy changes have all shaped and reshaped housing market.

According to Myers and Vidaurri (1990) over the decade prior to their study, observers had come to an almost universal appreciation of the importance of demographic factors in shaping demand for housing. Population growth has always been recognized as a primary force driving demand, but the new

2-53

recognition has centred on the ages of potential home buyers not simply their number.

Lindh (2000), who did demographic and housing-demand research, using time series regressions on Swedish data, found that a large group of young adults are associated with higher rates of residential construction. A significant negative effect from the age groups above 75 was found. Theoretically, this may be explained by a demand pattern for age groups derived from a life cycle optimization model with finite life and uncertain time of death, combined with the strongly increasing mortality rates at that age.

While some only studied the effect of social factors on housing demand and other studies considered both demographic and economic factors (White, 2003), all pose the question: At what age do people create housing needs? (Kamara, 1994).

Housing-demand studies based on different assumptions and concepts, using variations in the groupings of data, have generated great variations in results depending on the source of input data. For example, when housing demand is calculated with different models that use the data from the same area, the models can produce divergent results. In reality, the precise quantification of housing demand for a particular time and for a particular area could have only one accurate result.

This is the gap in housing-demand studies which this research seeks to fill by answering the following question: Is it possible to merge demographic, economic and social factors in one model in order to calculate an accurate figure for housing demand? What is the suitable methodology to capture housing demand quantity?

2-54

To conclude, population growth generates housing needs which will become housing demand when individuals and families can afford housing. Housing needs are first created through social activities like marriage, divorce and changing jobs. Three sets of factors demographic, social and economic are sufficient on their own to explain housing demand and to produce logical housing-demand estimations and predictions.

Modelling methodology comparison (between Econometrics and System Dynamics approaches)

There are limited methods to model housing demand. The main traditional approach for housing-demand modelling has been econometric. System Dynamics is the selected method to be used for this research for a new and improved approach to problem-solving. The different approach and concept between Econometrics and System Dynamics method are described as follow:

Econometric modelling

Econometrics is derived from many subjects and disciplines. Those include mathematics, economics, statistics economic statistics, and economic theory. There are two aims of Econometrics. First is to provides econometric theory empirical data and to empirically validate it. The result can be measured and then proven to compare. However, this method still has several criticisms. For example, a relationship between two variables is linearly measured when it should be curved. The popular tools using in econometrics is regression analysis, crosssectional analysis and Time-series analysis.

2-55

Econometrics is concerned with the tasks of developing and applying quantitative or statistical methods to the study and elucidation of economic principles (Frisch, 1993). Econometrics combines economic theory with statistics to analyze and test economic relationships. Theoretical econometrics considers questions about the statistical properties of estimators and tests, while applied econometrics is concerned with the application of econometric methods to assess economic theories (Pesaran, 1987).

Econometric analysis may also be classified on the basis of the number of relationships modeled. Single equation methods model a single variable (the dependent variable) as a function of one or more explanatory (or independent) variables (Wright, 2001). A simple econometrics model can be like the following regression analysis equation.

HD

E 0  E1 pop  E 2 social  E 3 econ  H .

Equation 2-7

Where, HD = Housing demand = Constant value = Coefficient of population factors = Coefficient of social factors = Coefficient of economic factors = Error value

E0
E1

E2

E3
H

2-56

Regression analysis can be used to estimate the unknown parameters 0, 1, 2 and 3 in the relationship of an example using data on population, social and economic factors. The model could be tested for statistical significance as to whether an increase of population, social and economic factors is associated with an increase in the quantity of housing demand, as hypothesized: 1 < 0.

System Dynamics model

System Dynamics is the method of modelling systems that has special features to count on feed back loop. It was invented by Jay W. Forrester, at the Massachusetts Institute of Technology, in 1961.

System Dynamics deals with how things change through time, which includes most of what most people find important. It uses computer simulation to take the knowledge we already have about details in the world around us and to show why our social and physical systems behave the way they do. System Dynamics demonstrates how most of our own decision-making policies are the cause of the problems that we usually blame on others, and how to identify policies we can follow to improve our situation. (Forrester, 1991)

The models methodology, using System Dynamics processes, is as follows:

x x

identifying a problem developing a dynamic hypothesis explaining the cause of the problem

2-57

building a computer simulation model of the system at the root of the problem,

testing the model to be certain that it reproduces the behaviour seen in the real world

devising and testing in the model alternative policies that alleviate the problem

implementing this solution

This different approach of Econometrics and System Dynamics cannot be easily understood without giving examples. The example is assumed that someone would like to model the orange that produced in a particular time and area. Both method approaches can be simply modelled like the following:

An Example to illustrate the System Dynamics approach

For ease understandable, an orange case would be raised to be and example because the real case about housing demand is too complicate for demonstration. As, we know that oranges come from orange trees, so the initial factor which produced the orange is the orange tree. The first main factor of this orange model then starts with a stock of (quantity of) orange trees.

Step one Assuming that there are 1,000 orange trees in a particular area, the stock of orange trees is set in system dynamics model that 1,000 is the initial number. Assuming that for an orange tree to produce oranges, it has to be between 3 and 8 years of

2-58

age (assumed for this example that the life expectancy of an orange tree is 10 years, and for the first 2 years it is too young to produce oranges and for the last 2 year, it is too old to produce oranges) and assume that the farmer plants 100 trees per year beginning 10 years ago.

Step two At the 11th year, the model shows that there are 600 orange trees mature enough to produce oranges this year. At this year, there are 200 trees too young to produce oranges and other 200 trees 2-year too old to produce oranges. Therefore, from all 1,000 orange trees in the farm, only 600 orange trees can produce oranges each year.

The next factor is (for example) water for growing them well. Assuming that the rain can stimulates the growth of orange flowers. More rain means more orange flowers on each tree. Assuming that the historical data can be collected and the rate of rain that generates orange flowers can be known, as presented in the Figure 2-1.
Rain

Orange flower

Figure 2-1 : Relationship between rain and orange flower (Assumption for an example)

2-59

Then the rain data in this year can be inputted to calculate how many orange flowers per tree will result from that rain. After inputting rain data to the model, it then calculates the number of orange flowers as a result.

Step three By assuming that bees are the main factor in pollination the flowers and therefore setting fruit, the number of working bees directly affects the sequence of converting orange flowers to fruit. The number of working bees and the success of fertilizing the oranges are assumed the relationship as the Figure 2-2 below.

Bee

Orange Figure 2-2 : Relationship between bee and orange (Assumption for an example)

After the number of bees in this year from that particular area is inputted, the resultant quantity of oranges is generated from the model. The System Dynamics model map of orange production can therefore be presented as follows (see Figure 2-3):

2-60

Figure 2-3 : The Orange model (example)

An econometrics approach to the same scenario

If the same scenario was to be modelled by using econometrics, the variable set might be expressed as the following equation for a regression analysis approach:

Example for Econometrics approach of the orange model

The comparison of Econometrics in the same scenario can be described. The set of variable can be given as equation below from regression analysis approach:

Y= a+bX1+ cX2+dX3+ e

Equation 2-8

Where,

2-61

Y X1-3 A b, c and d e

= dependent variable = independent variable = constant value = coefficient = error value

Then, this case can be set all factors influent to the quantity of orange as the equation below:

O = a(Ta)+b(Ty )+c(R)+d(B)+e

Equation 2-9

(In this case) where, O Ta Ty R B a, b and c e = number of orange = number of all orange trees = number of young (mature) orange trees = quantity of rain = number of bees = coefficient = error value

From the equation above, it can be found that there are some problems in this equation regarding the following points:
x

A simple regression equation, as used in econometrics, cannot deal with sequential calculations from the first step to the last step.

Multi-collinearity presents a problem in dealing with the chunk of independent variables that can act in concern to increase nominal orange

2-62

production. Those data will create the highly statistically insignificant value, in terms of values such as the t-statistic value and f-statistic value.

From this simple example, it can be seen that there are advantages in using realistic modelling, such as system dynamics methods, to imitate the real environment. Although such models are better at dealing with an imitation of reality, there are limits attaching to the confirmation of precise prediction.

This example model may lead to the picture of product supply side. However, in this research, housing demand is looked as a production which is produced from population (as compared to orange trees which are producing oranges). Without population, housing demand will not exist. The orange trees can be deduced as households units which produce new households by splitting new household units to the market.

The feedback loop feature of a System Dynamics model can be considered in the context of this research (i.e. housing-demand model). However, the concept of this researchs model could not capture the strong evidence showing in the feedback loop in the housing demand system itself (i.e. refer to figure 4-2).

Forecasting resulting from system dynamics models is still dependent on many factors, such as the precision of the initial data and the influence of those factors inputted into the model. System dynamics models need several points of relevant research in order to achieve a precise model to achieve reasonable forecast aims.

2-63

Controversial on Econometrics and System Dynamics

There are comparisons on Econometrics and System Dynamics Model by Myrtveit (2005) who asserts that Econometricians base their work on statistical analysis data.Data by itself cannot be used to determine causality. Therefore the focus of econometrician is on correlation (as opposed to causality).

Econometrics is used to prove economic theories. It may involve implicit or explicit a priori assumption about cause-effect relationships which can describe correlations in the observed data. Also, the statistical analyses applied as part of the econometrics (by itself) are not suited for determining causality. Therefore, an econometrician does not explain why this happens, but only that it happens and under what circumstances (Myrtveit, 2005).

The purpose of System Dynamics is to improve the general understanding of a problem, and to identify working policies for improving the performance of systems. It is also a problem oriented in the way that researcher can study the reasons and possible solutions to problem that prove themselves in system. Which open function feature, the researcher is free to select any mathematical principle for model simulation Myrtveit (2005). However, -since System Dynamics is chosen for the research- the housing demand model in this research is not used in the sense of system performance testing but modelling housing demand. The policies identify then would not be focused in this work. The table of comparisons and all characteristics of both methods as Myrtveit described can be concluded as the Table 1 below.

2-64

Table 2-1 : Comparison of the characteristics of Econometrics and System Dynamics concluded from Myrtveit
Characteristics

Econometrics

System Dynamics

Purpose

Give empirical content to economic theory Empirically verify economic theory Reductionist perspective; focus on a small number of variables and keep the remaining variables constant

Generate understanding of causes and effects creating a problem Identify solutions, in the form of policies, to problems. Holistic perspective; include all factors that have significant impact on how the problem develops over time Long-term (relatively speaking); study both short- and long-term consequences

Short-term perspective in forecasting

Scope

Closed system view (few or no exogenous variables); both source of problem and solution are found Often disconnected from time (in the case of equilibrium models) within the modelled system itself Aggregated system view; details are removed through averaging, and sequences of events are grouped into continuous flows

Open system view (many exogenous variables) High level of aggregation Build on economic theory about the nature (e.g., linear) of given economic relationships Results are based on correlations identified within data.

Behaviour can be reproduced from initial state and system structure, without the need for exogenous input. The qualitative behaviour of a system is generated by a relatively low number of feedback loops constructed from interconnected stocks and flows. Conservation of mass

Assumptions

2-65

Quantitative prediction Determination of the equilibrium state Main focus is on the product (i.e. estimates and hypotheses tests).

Results

Qualitative (as opposed to quantitative); patterns of behaviour rather than predictions, forecasts or prognosis Focus is on the process (learning) rather than on the product (model/results).

Source: conclude from Myrtveit( 2005)

Housing-demand study dilemma

In all markets -including housing- demand is an economic term quantifying peoples purchasing power at any particular time. Normally, demand increases when the price of goods is cheap and decreases when the price is high. In theory, there is a quantifiable demand which can be expected to move over time. Housing demand is the same, but until now there has been no reliable method to capture the exact number.

Although taking housing sales numbers and delineating them equal to the housing demand is the assumption in previous studies, it is not comprehensively satisfactory. This is because the housing sales number is only equal to the number of those people ready and able to buy a house (i.e. the number of people or house holds who have actually bought a house). But it does not include the number representing those who are ready, but who may not have found an available house in the market. Nevertheless the use of substituting a figure purportedly representing demand for housing by reference to the number of housing sales has persisted until now. Therefore, the model in this research considers housing demand and housing sales separately in order to find out the actual housing demand. The actual housing

2-66

demand means both the demand by those ready and able to buy a house and the demand by those ready but not yet meeting criteria leading to satisfaction with an available house in the market at a particular time. The unmet demand is called in this research as the latent, unsatisfied demand.

Specifying an actual-demand number is important in many respects. Entrepreneurs use demand projections for making decisions on investment. Governments use demand projections to determine housing policy. Demand projections also benefit house buyers seeking to invest in the housing market.

There are ways to determine the real demand number for any good, at any particular time. If we conceptualise demand as meaning the willingness of people to buy, coupled with their ability to buy, a method of quantifying demand in a particular time and a particular area can be demonstrated.

Whilst other researchers have focussed on housing-demand models, none appear to have studied the sequencing of factors that generate housing demand. This research aims to investigate that sequence of factors that creates housing demand from the root of apriori cause, and therefore provide a new tool for decision makers within that jurisdiction. Bangkok Metropolitan Region is selected as a case study for this research.

Most research on housing-demand models around the world have been undertaken by economists using econometric methods. These include Carlinger (1973); Lee and Trost (1978); Mayo (1980); Malpezzi and Mayo (1987); Dicks (1988); Myers and Vidaurri (1990); Megbolugbe, Marks et al. (1991); Kamara (1994); Filardo (1996); Liu, Wu et al. (1996); Vidaurri (1996); Maneekeatpaiboon (1997);

2-67

Prybolsky (1997); Chapman (1999); Macpherson and Sirmans (1999); Zhang (1999); Thomas and Malmberg (2000); Tiwari and Hasegawa (2000); Hui (2001); Brown and John (2002); C.Goodman (2002); Charles (2003); Howell (2003); Ioannides and Zabel (2003); Gross (2004); Miron (2004); Smith (2004)

The purpose of such research tends to focus on the relationship between various independent factors against housing demand as a dependent factor. Models are then used to predict or forecast future housing demand. However, there remain difficulties with such methodology in reconciling results and adaptation (Megbolugbe, Marks et al., 1991).

Housing demand research undertaken by economists is intended to identify the relationships between housing demand and other several factors, and also to find out the elasticity of those relationships. The aim of those research studies is not intended to capture a precise quantification number of housing demand. A range of articles were reviewed which outlined research purposes of this aim Straszheim (1975); Lee and Trost (1978); Mayo (1980); Arimah (1992); Berson (1997); Maneekeatpaiboon (1997); Kan (2000); Tiwari and Hasegawa (2000); Ioannides and Zabel (2003); Ramrattan and Szenberg (2004); Guest (2005); Di and Liu (2006). The data used for housing demand variables in the research of this group is adopted from different sources. This variation in data sources produces different result and different forecasts (Malpezzi and Mayo, 1987).

The housing demand estimations or predictions issuing from econometric models are changed if variables in equations are changed or different data sources are used. For example, each varied palette of factors affects housing demand in

2-68

different ways. The relationship between variables depends on the set of data used in the model or location of the case study. Therefore, it can be concluded that, the effect of housing demand from those independent factors is changed if the studys parameter is changed.

Some of the limitations of the previous housing research methodology and the advantage of the new housing demand research approached by System Dynamics are set out in the following points:

Previous housing research limitations:

Variation may result depending on the equations used. There are many types of equations in mathematics that can be adopted. Researchers use the same data to calculate their models, but the output can be significantly different depending on what type of equations is used. Furthermore, different types of equation will provide different coefficients of factors in the same model. Then, the reliability of results is still at question and impinges on practical utility. Many researchers on housing demand update results using the same methodology. This means a robust set of principles in housing demand theory is not yet formulated. For example, Pitakteeratum, (1994) studied factors affecting demand for housing in Bangkok Metropolitan and Vicinities showing that the coefficient for each factor used in her model changed and depending on other component factors used in the model. The method for her models calibration is to select the set of variables in her research which give the best result of statistical values. At the end of her research, she had chosen

2-69

the model (from multiple linear regression method) which include housing price index, population, housing loan (or housing credit) and the government housing policy. The statistic values that she concerned include the coefficient of determination (R2), the adjusted coefficient of determination(R-2), F-statistic value, t-statistic value, Durbin-Watson statistics value.

Much of the research is based on a definition of the relationship between influential factors and housing demand, using a delta value measurement between independent variables and dependent variables. The coefficient of variation in those econometric models is calculated mathematically. This methodology tends to focus on the relationship between variables and minimises the error value (such as the ordinary least square method) rather than looking to the real and consequential causes of housing demand.

There are several palettes of items that can be chosen to calculate housing demand. With econometric methods, researchers are able to establish a varied palette of factors to model housing demand. Those factors tested (by econometric methods) are not being forced that they must come from root causes of housing demand. In some cases, the causal variables do not have direct relationship with housing demand such as interest rates or government housing policy.

2-70

System dynamic approach advantage:

This research introduces a System Dynamics model which follows mathematic and counts sequential factors to calculate results. For example, if there is a population of 100 in a specified area with 20 houses, it can be assumed that an average of five people lives in each house. Therefore, if there are ten couples married in a given year, it can be assumed that housing needs will be generated for these ten newlyweds. It can also be understood that all ten housing needs would be created in that same year. Actually, some of newlyweds may simply live with one set of parents. The rate of splitting of new household is then applied into this model to deal with this function (details of splitting of household are described in chapter 4). This scenario shows how the sequencing of factors, using System Dynamics, is used in this research to model housing demand. The concepts of System Dynamics modelling as used in this research are described in Chapter 4 and the model is described in Chapter 5.

System Dynamics methods can be used to imitate real situations in every scenario. The function of calculation in a System Dynamics model can be used, starting from simple methods, such as simple counting, through to involved complex mathematical equations. In this research the assumption is that a number representing housing demand could be derived by counting the demand influenced from the root of causes of

2-71

demand, finally yielding a precise calculation of housing demand. The conceptual detail of the model is depicted in Chapter 4.

Economists usually or perhaps traditionally estimate housing demand by reference to housing sales. Housing sales do not necessarily represent housing demand accurately because some people who have a demand for housing (ability and willingness to buy) may not buy since the housing market is not a perfect market. Information on the availability of housing may not be perfect. People may also require a type of housing not currently on the market. Additionally, the process of buying housing takes a considerable period of time.

Key comparisons between econometric approaches used in housing research in Bangkok Metropolitan Region, and the System Dynamics approach, are set out below.

Another way of looking at the dilemma is to suggest that all econometric approaches use the number of house sales to represent housing demand. This is inaccurate in some scenarios, for example:

Housing demand will exceed (be greater than) the number of housing sales when the housing supply is insufficient to meet market demand.

Housing demand will exceed (be greater than) the number of housing sales when available housing supply does not satisfy the needs of all consumers in the market.

As argued above, the logic of assuming housing sales to be a direct proxy for housing demand is unreliable because it fails to include the latent, unsatisfied

2-72

demand. It is reasonable to criticise earlier research based on econometric analysis that uses this rationale (the setting of housing sales as housing demand proxy).

Econometrics is a statistical and mathematical method for finding the relationship between an independent variable and a dependent variable in one sequence. The System Dynamics method, however, can link factors together along in sequential steps, expressed as a set of equations. It is argued that such an approach more closely mirrors reality, and is more responsive to changes over time.

It is also noted that the econometric method which produce a linear equation is less realistic than nonlinear equations. The System Dynamics method has a feedback loop feature which is a feature of real-world situations. It also provides the feature for inputting unlimited types of equations into the model.

The System Dynamics model used in this study attempts to overcome the shortcomings of previous modelling, thereby quantifying demand more accurately. The System Dynamics model connects together many equations as one long equation. When the input data or the equation is wrong, the result is obviously wrong. The System Dynamics model has the special feature of showing all steps leading to the results. Therefore, it is easy to detect the error. The equations for every step in the System Dynamics model can be calibrated and the models calibration constrains the user to input the data and set the equation in a reasoned way.

In conclusion, existing research into housing-demand models fails to account for the root causes that produce housing demand. This research will introduce and demonstrate the use of the System Dynamics method to develop a housing-

2-73

demand model from the bottom up approach with a view to achieve a more realistic and accurate outcome. The steps of each consequence setting in this research model are based on evidence from previous research. In this study, all factors affecting housing demand are grouped under one of three main factors: demographic, social or economic. As all factors are dealt with sequentially, this research model begins with demographic factors and then deals with social factors before dealing finally to economic factors. The sequencing and detail of the model is described in chapters 4 and the models simulation is described in chapter 5.

Gaps identified within the literature

As mentioned, housing-demand studies have been undertaken for at least three decades. Most research into housing-demand models has been undertaken by economists using econometric approaches. The purpose of such research tends to focus on the relationships between the various factors leading to housing demand and which influence the reliability of housing demand forecasts. The limitations of this research are as follows:

Variations in housing demand forecasts can arise depending on the equation used. There are many types of equations which can be used in regression analysis including linear, logarithmic, polynomial and exponential equations. Even if a researcher uses the same data source to input into a model, the outcome may differ if different types of equations are used. This variation in outcomes undermines the utility of such models. In addition, different equations yield different coefficients of factors affecting housing-demand models.

2-74

The relationship between factors influencing housing demand and housing demand itself emerges from measuring the average summation value from each data point. Coefficients of factors in the equation (used in the model using an econometric method) are calculated mathematically. Rather than looking at real causes and effects, this approach tends to look at the relationships between numbers and to determine ways of minimising errors (such as through using the ordinary least square method).

This approach may be an appropriate way to model housing demand, but its adequacy relies on the researchers careful and well-reasoned selection of factors and methodology.

Most housing studies use regression analysis (Banerjee, (2004); Anikeeff, (1999); Schulz, (2004); Kan, (2000); Case, (2004); Richard, (1993); Di, (2006); Ramrattan, (2004); Guest, (2005); Wang, (2004); Arimah, (1992); Parsons, (1986); Ziegert, (1988); Lee, (2001); Jantarakolica, (1996); Malpezzi, (1987); Tiwari, (2000); C.Goodman, (2002); Carlinger, (1973); Chapman, (1999); Charles, (2003); Dicks, (1988); Filardo, (1996); Gross, (2004); Howell, (2003); Ioannides, (2003); Kamara, (1994); Lande, (2002); Lee, (1978); Macpherson, (1999); Maneekeatpaiboon, (1997); Mayo, (1980); Miron, (2004); Murray Brown, (2002); Myers, (1990); Schulz, (2004); Smith, (2004); Vidaurri, (1996); Zhang, (1999); Zhou, (1997)). There are a few research studies which used other methods (Liu, Wu et al. (1996); Macpherson and Sirmans (1999); Skaburskis (1999); Gat (2000); Blackwell and Partners (2001); Waddell (2002); Yucun (2002); Banerjee, Gelfand et al. (2004); Leary (2004); Leary (2004); Chiu and Ho (2005)) but all of them rely on calculations of housing demand which use relative computations.

2-75

In addition, the main purpose of those research studies is not to find out what the absolute housing demand number is, but rather to observe relationships between specific factors and housing demand. So the housing demand data used represents proxy housing demands numbers taken from different sources. This is just one of the reasons why research into housing-demand models continues to be done, and has been done for many years. However, differing results will ensue from a regression analysis, depending on which proxy factor is selected to act as the group of independent variables.

This research will introduce a methodology that produces absolute numerical values for housing demand at any point in time. The study will isolate the factors that are the root cause of housing demand and identify the sequential manner in which they produce demand. The study will also describe the constraints and impacts of these root causes on housing demand. This research was done in the Bangkok Metropolitan Region as the case study area. The conclusion and principles found from this researchs model experiment are also described at the end of this research.

Approach to System Dynamics

As mentioned, the type of of data to be used for a housing-demand model can cause a variety of relationship and result. With econometric methods, researchers are allowed to use a wide choice of factors for modelling housing demand. On the other hand, System Dynamics model can be used to count a number in each factor before transferring to the next factors. The equation for every step in the System Dynamics model can be calibrated, and the model causes the user to input data and set the equation in a realistic way.

2-76

This literature review found that existing research into housing-demand models fails to account holistically for the root causes that produce housing demand. This research will demonstrate the use of the System Dynamics method to develop a housing-demand model from the bottom up. The steps for each consequence setting in this research model are based on evidence from previous research. In this study, all factors affecting housing demand are grouped under one of three main factors: demographic, social or economic. As all factors are dealt with in steps; this research model begins with demographic factors, and then deals with social factors before finally turns to economic factors. The stepwise progression imitates the real world situation. The consequence and detail of the model is discussed in the following chapter.

System Dynamics introduction

The term System Dynamics refers to a method that, by using numbers for measurement, supports the management of complex systems in which the elements of the systems affect each other in forward and backward directions. For the researcher, System Dynamics can provide a way of dealing with number as part of quantitative research. System Dynamics is a methodology that supports the modelling and simulation of any system in which number changes over time (Forrester, 1991).

System Dynamics is the method used to model interactions among multiple factors within a system. The dynamic and complex nature of the model requires a specific and complex computer program. System Dynamics has been applied in many fields of study including in business and government policy, economic,

2-77

military, environmental and medical studies (Yucun, 2002). The study of scenarios that involve interactions between variables can benefit from this method. It can be applied whenever problems can be expressed as variable behaviours through time.

Along with this numerical method, the feedback loop is a further distinguishing feature of System Dynamics. Feedback systems can explain when factor X affects factor Y and when factor Y in turn affects factor X, allowing the tracking of causes and effects. One cannot study the link between factor X and factor Y and then forecast how the system will behave. Only the study of the total system as a feedback system will bring one close to simulate the real world situation (Forrester, 2006).

System Dynamics history

Jay W. Forrester, and his colleagues at the Sloan School of Management, Massachusetts Institute of Technology (MIT), invented System Dynamics in 1960. It was developed initially from the idea of applying feedback control theory to the study of industrial systems. Forresters famous 1960 model was Urban Dynamics. The model explained population growth and subsequent decline in many cities in America such as Manhattan, Detroit, Chicago and Boston. The simulation showed the interaction between industries, housing and people (Forrester, 2006).

Forresters book Industrial Dynamics (1961) is still a significant statement of philosophy and methodology in the field. Since its publication, the span of

2-78

applications has grown extensively and now encompasses work in many fields, including:
x x x x x x x

corporate planning and policy design public management and policy biological and medical modelling energy and the environment theory development in the natural and social sciences dynamic decision making complex nonlinear dynamics

The great influence of System Dynamics started in 1972 when a group of MIT scientists wrote The Limits of Growth which later became an international bestseller. The book conveyed a scenario in which the worlds population consumed resources at a rate that would produce seriously adverse consequences. It influenced many people to develop ecological concerns. With this book, the word sustainability entered the worlds agenda and remains at the forefront today (Meadows, Randers et al. 2004).

System Thinking

There are many complex systems in our civilization. As human populations grow, consume and extend the built environment, these complex systems impact in even more complex ways. Systems Thinking offers a way of understanding complex systems through the use of System Dynamics methodology (Hannon, 1999). On the assumption that elements in a system are dynamic and interactive with other elements (whether in positive or negative ways and whether in forward or backward loops), a Systems Thinking approach then can show the movement of

2-79

any element in any model which is established. A numerical method from this System Dynamics approach is used to present how, how many and when the dynamic elements interact with other elements. With this methodology, any movement in a real situation can be captured and observed, providing us with knowledge about the effects of elements in a defined system and allowing the trial of different impacts through running simulations.

2-80

Source: Coyle( 1996)

Figure 2-4 : The System Dynamics Model Process - After Coyle (1996)

2-81

Systems Thinking and System Dynamics method

In any real world system, dynamic situations can be explained by using a numerical approach. A simple example is when people have a meal. The energy gained can be calculated by analysing the food consumed and, similarly, the nutrition and weight outcomes of the meal can be calculated and analysed through counting how many calories they have consumed. If the aim is to know how many calories are required, a backward process can be undertaken by identifying activity levels, calculating the number of calories needed and choosing the food to suit that allocation of calories. However, the number of calories required alters with age, so needs change over time. System Dynamics is a method that can calculate actual numbers and accounts for the changes over time. This tool can also be used for forecasting purposes.

In the same way, to precisely quantify housing demand at a particular time the factors relevant to housing demand need to be identified and their relationships to it defined. In this situation, housing demand will be set as one component in a system circumstance. All factors relevant to housing demand will then be set in the model to explain the system and make it robust.

iThink software program for System Dynamics modelling

iThink is a computer simulation program chosen for modelling in this research. Usually this program is used to simulate business scenarios and other similar things. The reason for choosing this program was to provide an easy-tounderstand graphical interface for observing the interaction of factors within a model. Apart from iThink, Powersim, DYNAMO and Vensim are other three software programs commonly used by modellers (Yucun, 2002). However,

2-82

researcher choice seems to largely depend on personal reasons because the basic features of all these programs are similar when modelling, and do not lead to any significant difference.

The advantage offered by the iThink program include the capability of:
x x x x

Building models that simulate all scenarios Creating what-if scenarios for decision support Reducing the risk of policy or process change Developing shared understanding across functional teams

The graphical function is one of the special features offered in System Dynamics computer software. This feature deals with non-linear correlation between factors in the model. The graphical function used in this research are shown and described in Chapter 5.

System Dynamics for Housing-demand study

The review of available literature also showed that there is some research done using System Dynamics to model housing demand. A Study of System Dynamics for Urban Housing Development in Hong Kong was undertaken by Hu Yucun in 2002. His research established a descriptive simulation model for Hong Kong urban housing development based on System Thinking. The projection of the model simulates the development trend and demand of the housing industry in Hong Kong.

His research adds to the body of knowledge on how housing demand is generated by different groups of factors. This model uses similar methodology (System Dynamics) to simulate the reality of the root causes emphasised just on housing

2-83

demand only. All factors in this researchs model are suggested by conclusions from the literature review. Factors in the model are the results of trial and error. This research introduces conceptual and methodological approaches to quantifying housing demand in a particular area and time. The processes involved are described in Chapters 4 and 5.

Housing demand environment from System-Thinking approach

From an holistic view, looking to the root causes of the system, the picture which can be seen provides reasoned elements picture-composed in that system. Starting with the understanding of the whole complex system, this leads to the comprehension of dynamic causes until reaching an ability to model. Systems Thinking is the method of thinking and System Dynamics is the methodology that would offer ways to reach this aim.

From the bottom up, looking at the root causes of housing demand, the originating factor driving housing demand is people (i.e. population) (Berson, 1997). The number of the total population is not a good proxy for modelling housing demand. Research suggests that the young adult group has the most significant effect on creating housing needs and housing demand (Goldscheider and Vanzo (1985); Heer, Hodge et al. (1985); Borsch-Supan (1986); Dicks (1988); Haurin, Hendershott et al. (1993); Skaburskies (1994); Ermisch and Salvo (1997); Goldscheider (1997); Ermisch (1999); Masnick and Di (2000)). The period of social activity of people (population) who create housing needs is the key to track and create the functions that contribute to the state of housing needs (in the model). Furthermore, the financial status of people (population) who create

2-84

housing demand is the key for calculating from the relationship between housing needs and housing demand.

This research proceeds by extracting the effective number of housing needs in a given population. The number of the total population would be classified between the number of people who do not create housing needs and the number of people who create housing needs. The groups which do not create housing needs include children and the elderly. The method simply requires setting the range of age groups within a population using results from previous research as evidence for the selected range. The first step of this housing-demand model is completed through identifying the number of people who are generating housing needs at particular time.

The first step isolates the number of people who need housing from those who do not. The next process relates to household formation in a particular area and at a particular time. Data from the Bangkok Metropolitan Region (BMR) were used in this research. Causes of household formation include marriage, divorce and growth to adulthood, all of which produce new households or families. These causes were used in the Hong Kong housing-demand model study by Liu, Wu et al. (1996) in order to predict precise housing demand using a bottom- up approach.

Next, the number of people who need housing in a particular area, and at a particular time, is identified. From this, the number of people who can afford housing plays an important role in the modelling, because this will limit the

2-85

number of housing demand units produced from overall housing needs. The number of people with a housing need is selected by using economic factors which govern the capacity of people to buy a house. There are several factors influencing housing affordability. In this research, the main economic factors chosen for testing are taken from previous research on housing demand in Bangkok (Pitakteeratum (1994); Maneekeatpaiboon (1997); Pongsatat (1998); Sheng and Kirinpanu (2000); Calhoun (2003)). These include factors such as:
x x x

Income data - peoples average income at the year () Expense data peoples average expenses at the year () Savings data - peoples average savings at the year or income expense ()

x x x

Distribution of saving Housing price index () Home loan incentives, data on the percentages borrowed on home loans (%)

Government incentives, data on the effect of government policy on housing demand (%)

The factors listed above are the prospective experimental factors in this researchs model; however, the models final factors may differ from prospective factors as a result of the limitations of available data. These data and this method will produce the number for housing demand units in a particular area and at a particular time. To imitate the real market, the total number of housing demand units per year is derived by subtracting from the number of housing sales units in that year. In reality, housing supply continually responds to housing demand every day, but the

2-86

data available for this research is provided on a yearly basis. Therefore, years are set as the time basis for this models simulation, although the time period for simulation runs in this model can be set as daily, weekly, monthly and yearly basis. Housing sales data for the Bangkok Metropolitan Region were first collected in 1977. The initial value used is therefore 1977. All others data are projected backward to the same year. The simulations run through to 2007 (present time) and then project for the next twenty years to 2027 for a forecasting trail.

The results generated from model simulation are then compared with historical data. Errors were observed and detected, then logically calibrated to minimise the difference between historical data and simulation data. Suggestions for further research are made at the final chapter.

While model precision is desirable, the most important outcome of this research is to demonstrate the quality of fit between the model and the real world. System Dynamics is considered most likely to be the best method for imitating and depicting the real world, even though limitations of this application were found by Richmond (1993). The model in this research was calibrated as much as it could be to fit reality before accepting the results of simulation runs and forecasts. Nevertheless, the unpredictable nature of economic factors and the emergence of unanticipated factors may influence acceptance of the model because these factors will affect its accuracy.

3-87

Chapter 3

CASE STUDY BACKGROUND AND INFORMATION

Introduction

Bangkok has been the capital city of Thailand for more than 200 years. The Bangkok Metropolitan Region (BMR) is the case study area for this research. The major market share of housing market in the country is found in this area. This chapter examines all the relevant information involving the Bangkok Metropolitan Region which has the major housing market share in Thailand. Some basic background about Thailand is also described in this chapter. The recent situation within housing market in the case-study area is found in the concluding sections of this chapter.

Physical information of case study area (Bangkok Metropolitan Region and Thailand)

Thailand is located at the centre of south-east Asia. It shares borders with Burma, Laos, Cambodia and Malaysia. In the southern part of Thailand, there are long coastlines on the Andaman Sea and also the Gulf of Thailand (formally the Gulf of Siam, the old name of Thailand). Bangkok, as the capital city of Thailand, is located in the central region, as shown in the following figure (3.1).

3-88

Source: http://www.ticketfox.de/Flug/Thailand-map.gif
Figure 3-1 : Map of Thailand from South-East Asia region Map

Thailand is officially divided into five rural regions and one urban region. Those five rural regions (in figure 3.2, Following) include the northern region (shown in dark green on the map below), the north-east region or E-san region (shown in pink), the eastern region (shown in blue), the western region (shown in yellow) and the southern region (shown in light green). Thailands main urban region Bangkok - is located roughly in the centre of the country and is called the central region (shown in orange on the map below)

3-89

Source: http://eng.konbaan.com/index.php>
Figure 3-2 : Map of regions in Thailand

Although apparently one large city, the BMR actually covers five provinces surrounding the capital province. It covers an area of 7,761.50 km and the registered population in 2007 was 10,061,726. It therefore has, a population density of 1,296.36 per km.

As the capital city of Thailand, Bangkok (like many large cities) has seen a great flux of immigrants who are not listed by the city's administration. The population

3-90

is therefore more realistically in the 12-15 million band . The Bangkok special administrative area covers 1,568.7 km (i.e. 606 square miles). Because much of this area is considered to be the city of Bangkok it is therefore one of the larger cities in the world.

The map and table below show the list of total 72 provinces around Thailand (more recently expanded to 76 provinces). In this research, Bangkok and the five provinces around it - including Nakhon Pathom, Nonthaburi, Pathum Thani, Samut Prakan and Samut Sakhon is called the Bangkok Metropolitan Region (shown in the dot box in the map in Figure 4 below).

3-91

The Bangkok Metropolitan Region

16. Bangkok 24. Nakhonpathom 32. Nonthaburi 33. Pathumthani 52. Samutprakan 53. Samutsakhon

Source: http://www.visit-thailand.info/information/map-of-thailand-regions.htm (2008-01-29)

Figure 3-3 : Map of 72 provinces in Thailand

3-92

Table 3-1 : The list of provinces of Thailand

1. Ang Thong 2. Buriram 3. Chachoengsao 4. Chai Nat 5. Chaiyaphum 6. Chanthaburi 7. Chang Mai 8. Chang Rai 9. Chon Buri 10. Chumphon 11. Kalasin 12. Kamphaeng Phet 13. Kanchanaburi 14. Khon Kaen 15. Krabi 16. Krung Thep Mahanakhon (Bangkok) 17. Lampang 18. Lampang 19. Loei 20. Lop Buri 21. Mae Hong Son 22. Maha Sarakham 23. Nakhon Nayok 24. Nakhon Pathom 25. Nakhon Phanom 26. Nakhon Ratchasima 27. Nakhon Sawan 28. Nakhin Si Thammarat 29. Nan 30. Narathiwat 31. Nong Khai 32. Nonthaburi 33. Pathum Thani 34. Pattani 35. Phangnga 36. Phatthalung

37. Phayao 38. Phetchabun 39. Phetchaburi 40. Phichit 41. Phitsanulok 42. Phra Nakhon Si Ayuthaya 43. Phrae 44. Phuket 45. Prachin Buri 46. Prachuap Khiri Khan 47. Ranong 48. Ratchaburi 49. Rayong 50. Roi Et 51. Sakon Nakhon 52. Samut Prakan 53. Samut Sakhon 54. Samut Songkhram 55. Sara Buri 56. Satun 57. Sing Buri 58. Sisaket 59. Songkhia 60. Sukhothai 61. Suphan Buri 62. Surat Thani 63. Surin 64. Tak 65. Trang 66. Trat 67. Ubon Ratchatani 68. Udon Thani 69. Uthai Thani 70. Uttaradit 71. Yala 72. Yasonthon

Source: http://www.visit-thailand.info/information/map-of-thailand-regions.htm (2008-01-29)

3-93

The Bangkok Metropolitan Region

16. Bangkok 24. Nakhon Pathom 32. Nonthaburi 33. Pathum Thani 52. Samut Prakan 53. Samut Sakhon

Source: http://de.wikipedia.org/wiki/Bild:Thailand_Bangkok_and_vicinity.png (2008-01-29)

Figure 3-4 : Map of Bangkok Metropolitan Region, Thailand

As a modern, large city, Bangkok is seen to be rising vertically and growing horizontally. Administratively it is split up into 50 districts (i.e. khet), but these are rarely used in practice so the conceptual physical division below from Tan (2007) is well understood and more practical .

3-94

Source: http://bangkokview.blogspot.com/2007/05/districts.html
Figure 3-5 : Map of Bangkok

1. Sukhumvit (light blue) The long Sukhumvit Road, changing its name to

Ploenchit Road and Rama I Road going west, is Bangkok's modern commercial core, full of glitzy malls and hotels. The Skytrain intersection at Siam Square is the closest thing Bangkok has to a centre.
2. Silom (pink) To the south of Sukhumvit, the area around Silom Road and

Sathorn Road is Thailand's sober financial centre by day, but Bangkok's primary party district by night when quarters like the infamous Patpong comes alive. This area is the central business district of Bangkok and full of high rise buildings.
3. Rattanakosin (dark blue) Between the river and Sukhumvit lies the densely

packed "Old Bangkok", home to Bangkok's best-known wats (Thai Buddha temple). Yaowarat (Chinatown) and sights around the Chao Phraya River are also included here. Bangkok's backpacker area (Khao San Road) and the surrounding district of Banglamphu are located on the northern part of Rattanakosin.

3-95

4. Thonburi (green) This is the quieter west bank of the Chao Phraya River,

with many small canals and some offbeat attractions.

5. Phahonyothin (orange) The area around Phahonyothin Road and Viphavadi

Rangsit Road is best known for the Chatuchak Weekend Market and Don Muang Airport.
6. Ratchadaphisek (yellow) This district, north of Sukhumvit, is centered on

Ratchadaphisek Road (part of which is called Asoke), and stretches from Phetchaburi Road to Lat Phrao. This area has really opened up recently as the new metro line following Ratchadaphisek Road.

Thailands Emerging Economic Policies and Strategies

The economy and politics of Thailand from the 20th to 21st centuries is well described by Fry (2004):

Prior to 1939, Thailand was known as Siam. Through the visionary leadership of the early kings of the Chakri Dynasty, which was established in 1782, Siam opened its borders to both Chinese immigrants and Western missionaries. Such visionary decisions helped Siam avoid colonialism and significantly influenced the nature of the contemporary Thai political economy. King Chulalongkorn, who reigned from 1868 to 1910, was a major reformer who initiated the policy of sending academically-promising Siamese abroad for study. This educational endeavour by Chulalongkorn contributed to unanticipated consequences, namely the overthrow of the absolutist monarchy in June 1932. Since that time, Thailand has been a constitutional monarchy. During World War II, the country

3-96

demonstrated its skill in bamboo diplomacy by simultaneously collaborating with both Japan and the Allies. Consequently, Thailand suffered less than virtually any country in the Asian region during the war. During the Vietnam War, Thailand became a staunch ally of the United States and became a land-based aircraft carrier for the intensive US bombing of Vietnam, Cambodia, and Laos. Such foreign policy contributed to the beginning of the Thai economic boom, which fully blossomed in the 1980s and early 1990s.

Fry also describes clearly the causes of the economic crisis in Thailand in 19961997 as follows:

Prior to 1997, Thailand had been one of the worlds hottest economies. From 1984 to 1995, its economic growth averaged an impressively high 8.5 percent; with a peak growth rate of 13.2 percent in 1988 The liberalization of Thai capital markets, encouraged by the West and Japan in the late 1980s, planted the seeds for the subsequent crisis. Japanese and Western banks offered relatively low interest rates and relaxed capital and foreign exchange controls so that the Thai private sector could borrow abroad and to invest in the booming Thai economy While it may have been rational for individuals to engage in such economic behaviour, collectively such actions resulted in huge private borrowing from overseas Despite an excellent reputation for professional and technocratic independence in the 1970s and 1980s, the Bank of Thailand failed to regulate excessive private sector borrowing in the 1990s and fell under

3-97

the undue influence of politicians with vested economic interests, who tried to save economic face by protecting the overvalued Thai Baht. The root causes of the Thai economic crisis were excessive private overseas borrowing and inadequate regulation of the financial sector by the government and Bank of Thailand.

Thailands economy in recent times could be categorized as that of a developing nation, heavily export-dependent, with exports accounting for 60% of GDP. Thailand ranks midway in the wealth-spread in South-east Asia as its 4th richest nation per capita, after Singapore, Brunei, and Malaysia. Thailand's recovery from the 1997 Asian financial crisis relied on exports, based on external demand. Agriculture has always been the traditional income generator, but has declined in relative terms in recent years as overall exports increased. Tourism has been on the rise as well, but not without negative consequences. With the instability surrounding the recent military coup in September 2007, however, the GDP growth of Thailand has settled at around 4% from previous highs of 5-7% under the previous administration. Locals, as well as foreign companies, are holding back on investment due to the political uncertainty. The lower and middle income groups within the country are most likely to demand housing, and are actively seeking better quality living conditions.

Overview of housing context in Bangkok Metropolitan Region

This section describes the importance of the housing demand problem in the Bangkok Metropolitan Region (BMR) and also provides background information on housing in Bangkok. The content in this section leads on to discussion of the

3-98

research topic and sites the importance of this study in relation to housing demand studies.

The housing market in the BMR relative to Thailand as a whole

The Bangkok Metropolitan Region is the primary centre in Thailand for business, communication, education and employment. Houses in Bangkok are spread around the periphery in lower density development whilst high rise developments are becoming more common in the central business district (CBD).

Housing loans in the Bangkok Metropolitan Region account for more than half of all loan amounts in Thailand (see Table 3-2). As this table indicates, the ratio of housing loans in the Bangkok Metropolitan Region to other regions has changed over time. This is due to the 7th social and economy development plan (i.e. from 1992 to 1996) which dealt with income distribution policy and development in regional and rural areas.

Table 3-2 : Comparison of housing loans between Bangkok Metropolitan Region and others regions Housing loans All others regions (with out Bangkok Metropolitan Region) Housing loans (M Baht)

Year

Bangkok Metropolitan Region

Housing loans (M Baht)

Rate of growth (%) 29.2 35.7 5.6 -6.0 -1.7 0.2 3.3

1995 1996 1997 1998 1999 2000 2001 2002

94,719 122,337 165,963 175,296 164,706 161,871 162,115 167,465

44,825 73,439 109,841 118,715 115,450 112,892 111,365 126,656

Ratio between housing loans in Bangkok Metropolitan Rate of Region and all growth other region (%) (%) 67.9 63.8 62.5 49.6 60.2 8.1 59.6 -2.8 58.8 -2.2 58.9 -1.4 59.3 13.7 56.9

3-99

2003 2004

184,737 214,132

10.3 15.9

147,428 178,907
(2005)

16.4 21.4 average

55.6 54.5 59.4

Source: Government housing Bank

History of the housing market in Bangkok Metropolitan Region

A housing market, in the sense that the western world may understand it, was developed in the Bangkok Metropolitan Region in the 1960s and 1970s. Its success has been the catalyst for a developing housing industry. The housing industrys history in the Bangkok Metropolitan Region started probably about 40 years ago. Bangkok was initially dependent on agriculture. The spread of urban development can be traced to the beginning of the housing development projects in the 1960s. From that point there has been a recognisable housing market. Before then there were formal arrangements in place for the sale of land. These formal arrangements for the sale of land came about when an entrepreneur bought a large parcel of land, subdivided it into many plots of land and then sold them to people who were seeking to build a house. This is distinct from a typical modern housing development project, where an entrepreneur buys a big parcel of land, subdivides it into many lots and then sells them to people who already have houses or alternatively may build houses on the subdivided land and sell them as completed housing units.

Pornchokchai (2006), dates the prehistory of the market to around 1890 with a land subdivision project in the Rangsit area in Patumthanee province in the northern part of Bangkok. This project was only for agricultural allocation purposes, unlike present housing projects. The modern housing market genesis was about 10-15 years before 1932, the year that Thailand changed from a system

3-100

of absolute monarchy to a democratic system. At that time there were some small land projects undertaken in the central area of Bangkok (Sathorn road, Wireless road, Rajdomri district, Plernjit district and Phyathai district). Most of them were for royal executive officers. There was, however, no evidence of formal housing projects (like those in recent times) for the general populace.

Data on the housing market in the Bangkok Metropolitan Region was first formally collected in 1968. During 1968-1969, there were about 40 housing development projects in the region. The Thai government, through the Revolutionary Council, made an announcement (Revolutionary Council No 286 (i.e. Por Vor 286)) about building control on 24 November 2515 (i.e. 1972 in the Western calendar). The purpose of this law was to control building construction and housing-and-land allocation. This law was cancelled, by the force of an act of legislation of land allocated, in 2543 (1990).

In 1973 economies across the world faced the first oil shock. As the emergent housing market in the Bangkok Metropolitan Region market was small, there was no strong evidence of a housing market collapse at that time. The subsequent recession did, nevertheless, affect what may be viewed as the first period of the Bangkok Metropolitan Region housing market. Until 1976, the market remained static. Townhouses were first introduced in 1976 as the recession ended, and they were a popular housing type until the second oil shock in 1980.

The first condominium law named An Act of Legislation of Condominium Control 2522 (i.e. 1979 in the Western calendar) for controlling condominium development was passed by parliament on 21 April 1979. Its purpose was to recognise condominium property deeds and management. In 1981, condominiums

3-101

were introduced to the housing market. In the period 1984-1985, many condominium projects were completed, and the units were sold. The total number of housing sales increased to 50,000 units per year in 1985, and reached a plateau at that level (i.e. see Figure 8).

The second condominium law, An Act of Legislation of Condominium Control 2534 (i.e. 1991 in the Western calendar) was passed on 30 September 1991. The purpose was to allow foreigners to own condominium property deeds. The aim of this second act was to promote foreign investment in the real estate sector with an attempt to stimulate the Thai economy. The Bangkok Metropolitan Regions real estate market boomed from 1986 to 1990. After 1990, it stagnated due to the economic effects of the war in the Persian Gulf. The recession which followed the military coup in Thailand also affected commercial property markets (i.e. office buildings and shopping centres).

Between 1993 and 1995 condominium sales peaked along with the booming Thai economy. Condominium sales were about 50,000 units in 1993; 75,000 units in 1997; and rose to about 50,000 units in 1998. Condominium presale rates were high, but the intent of most of buyers was speculation in the real estate sector rather than for a personal dwelling.

Housing sales in the Bangkok Metropolitan Region were therefore at their peak between 1994 and 1996. This period could perhaps be regarded as the golden period within the housing history of the Bangkok Metropolitan Region. However, the period is more commonly recalled as an economic bubble. The housing market was still growing until it faced the major economic crisis of 1997. The

3-102

crisis had a disastrous effect on the broad economic situation. The GDP growth index dropped from 9.2% in 1995 to -10.5% in 1998 (see Figure 3-6).

In 1997, the Thai Baht was devalued from 25 Baht per US dollar to approximately 50 Baht per US dollar. This currency devaluation effectively doubled the value of private sector debt on short term loans owed to foreign banks. Fry (2004) described the economic crisis as follows:

By 1996, Thailand had a staggering US$120 billion foreign debt. Thai borrowers also perceived no foreign exchange risk, since the baht visa-vis the dollar was probably the most stable in the world between 1960 and 1997. During that period, the Baht-Dollar exchange rate was stable within the narrow band between 20:1 and 25:1. With the rapid devaluation of the Baht after the crisis, the value of overseas loans to be repaid doubled Another factor to the economic crisis was the nature of the investments undertaken by Thai borrowers. Many loans were invested in activities that did not generate foreign exchange, and investors participated in property speculation that resulted in many non-performing loans.

Consequently, many private companies faced insolvency. Consumption capacity shrank. Many real estate companies confronted severe liquidity problems and could not meet their debt obligations. It should be noted that real estate companies in Thailand develop housing projects and are also involved in their sale). Many banks faced financial problems due to a sharp increase in non-performing loans. Fifty-six finance companies were closed by government order due to bankruptcy.

3-103

GDP index in Thailand

15.00

10.00

5.00 growth rate

0.00

20 01

19 97

19 99

19 85

19 87

19 89

19 93

19 81

19 83

-5.00 year -10.00

-15.00 GDP index in Thailand

Source: Government housing Bank (2006)

Figure 3-6 : GDP growth index in Thailand between 1981 and 2005

Many people had bought housing as an investment, and many had speculated on the property market during the economic bubble which preceded the 1997 Asian financial crisis. This consumption of housing units over real housing needs left a large number of vacant housing units in Bangkok Metropolitan Region. There were about 300,000 vacant house units in 1995. This figure rose to 350,000 in 1998. Statistics show that the high vacancy rate prevailed until 2003 (refer to Figure 3-7) as shown below.

19 91

19 95

20 05

20 03

3-104

vacant house unit

400000 350000 vacant house unit 300000 250000 200000 150000 100000 50000 0 1996 1997 1998 1999 year 2000 2001 2002 2003

Source: Government housing Bank (2007)

Figure 3-7 : Vacant house units in Bangkok Metropolitan Region between 1996 and 2003

The market was in deep recession from 1996 until it started to improve slowly in 2000. In 2004, a number of housing developments were launched into the market signifying the end of the recession. This development benefited from a low interest rate policy an economic measure was put in place by the Thai government in 2002 to assist in the housing markets recovery. However, from 2004 to 2007, the number of housing units sold has shown only slight growth. This suggests that interest rate policy alone is not the only factor required to boost housing demand.

A timeline reconstructing this remarkable period 1977 to 2006 within the housing market is presented in Figure 3-8.

3-105

housing sale history data

0 50,000 100,000 150,000 200,000

In 1979, the government passed a law controlling condominium. After that the number of condominiums being built rose dramatically. The period from 1984 to 1985 was when the first condominiums were completed and sold.

1977

1978

1979

1980

1981

In 1987, the ratio of houses built by the private sector compared with housing built by owners on their own land increased from 14.9% in 1983 to 57% in 1987 and 72% in 1989. From 1988 to 1989, the condominium market expanded again.18,776 units were completed across Thailand. Half were built in the BMR. In the meantime, an increasing number of low price condominiums were constructed in the BMR. In 1989, 80,031 housing units were completed - an all-time record for the BMR. There was transformation in financing, with expanded capital loans from Bangkok International Banking Facilities. By 1991, land prices had increased to 21.5 times the price in 1985.

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

In 1993, 23 % of people who bought houses or units and 31 % of people who bought condominiums at the Housing and Condominium Exhibition did so for investment and speculation. In 1994, this number increased to 50 % of people booking housing units and 65 % of people booking condominium units in this exhibition doing so for investment and speculation.

1992

1993

1994

1995

In 1995 there were about 300,000 units vacant in the Bangkok Metropolitan Region (rose to 350,000 units in 1998). In 1996 it was expected that 17% of current year housing projects and 40.7% of housing projects in1997 would collapse as a result of over supply problems. In 1997 there was an economic crisis stemming from devaluation of the Thai Baht. Banks stopped giving loans for housing projects and faced non performing loan problem. There was shrinkage of housing demand consequent of the economic problems. The rest of South East Asia was also dramatically affected.

1996

1997

1998

1999

2000

2001

2002

2003

From 2001 to 2003, the government subsidised people buying houses by reducing interest rates and fees and taxes on housing purchase. From 2004 to 2006, the political situation was unstable. The government was ousted by a military coup in September; 2006.A new constitution was drawn up in 2007 and a new government was installed in 2008, following the 2007 election.
Figure 3-8 : time line of housing market in the Bangkok Metropolitan Region

2004

2005

2006 year

3-106

Background of housing market in Bangkok Metropolitan Region

The property market in Thailand boomed from 1985. Land prices in Bangkok jumped five fold within four years and up to 20 fold in 1991. At that time, there was considerable investment in housing projects in many BMR areas. Housing prices were going up day by day. The land price increased to 30 times the 1985 price due to developments in the economic system moving from agriculture as Thailand emerged as a Newly Industrialising Country (NIC). Housing projects were built at that time until this circumstance caused the housing market to face the oversupply problem.

The Tom Yum Kung economic crisis in 1997 (later simply called the Tom Yum Kung crisis) resulted in the reduction in the value of the Thai Baht against the US dollar in July of that year. This situation led to decreasing land prices. They dropped to about one-fifth of the price earlier in the same year. The new housing project ratio was reduced to one out of five.

The crisis continued for two to three years. The economy has recovered steadily since 2000, and along with it, the housing market. The value of new housing project development in 1998-1999 was about 5,000 million baht per year. This increased to 13,000 million baht in 2000.

From 2000 to 2003, new housing unit numbers doubled in each year. The value of investment in housing projects went from 150,404 million baht in 2003 to 200,875 million baht in 2004 - or about 30 % each year.

3-107

The role of housing in the Bangkok Metropolitan Region (BMR), Thailand

Central Bangkok by itself has a population of about 6 million with another four million living in five adjacent provinces. The Bangkok area functions as the central business area and also as a centre of education. People who live close to Bangkok in the metropolitan region commute from the five surrounding provinces to work and study in Bangkok. Most commute to Bangkok city every working day. Those areas include Samut Prakan, Nonthaburi, Pathum Thani, Nakhon Pathom and Samut Sakhon. Housing projects have spread to those provinces around Bangkok. The housing market of Bangkok and the greater metropolitan region (i.e. the BMR) now works as one market. Housing loans in the BMR constitute more than half of all housing loans in Thailand.

The proportion of the housing supply provided by private sector projects increased from 4.3% in 1974 to 72% in 1989 (Pornchokchai (2006)). Therefore, the study of housing in the Bangkok Metropolitan Region reflects the main housing market of Thailand (see Figure 3-9).
Ratio of housing project in BMR
% 80 70 60 50 40 30 20 10 0 1974 1983 1987 Housing project 1988 1989 year

Source: Housing information centre of Thailand

Figure 3-9 : Percentage of housing supply in Bangkok Metropolitan Region from 1974 to 1989

3-108

Housing in the Bangkok Metropolitan Region in the last 10 years

The growth rate of housing in the BMR has risen dramatically in the last decade. The number of applications for building permits in BMR during 1996-2005 was 722,673 units cumulatively. These data indicate growth in housing unit numbers of approximately 18.5% within the last ten years.

Housing register in BMR 200000 176,616 148,631

150000

unit

100000 67,054 56,085 50000 33,880 37,182 29,029 33,074 69,050 72,072

0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Year

Housing register in BMR

Source: Housing information centre of Thailand

Figure 3-10 : Housing stock in Bangkok Metropolitan Region

In 1995 about 176,616 housing units were built. After the economic crisis in Thailand in 1997 the number of housing units built decreased dramatically from 148,631 in 1997 to 29,029 in 1999. Since 2000, the number of housing units built rose slowly from 33,074 units to 72,072 units in 2005 (refer to Figure 3-10 above).

3-109

Growth of housing market in the Bangkok Metropolitan Region from 1993 to 2005

Over the period 1993 to 1995, the growth of the housing market in the BMR can be divided into two parts. The first period, from 1993 to 1995, was a boom cycle in the housing market. Some 508,332 units were completed in this period, or about 169,444 units per year. The peak was 253,159 units in 1994. This level of housing supply was the highest point in the history of the BMR housing market.
New housing launch to market
unit 300000 253,159 250000 200000 150000 113,150 100000 64,909 50000 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 year 41,300 1,071 1,357 3,040 6,157 16,000 142,023

New housing launch to market

Source: Government housing Bank

Figure 3-11 : Housing opening in Bangkok Metropolitan Region from 1993 to 2002

Refer to Figure 3-11, the second period is between 1996 and 2002. In 1997, there was an economic crisis in Thailand and South East Asia. In 1997, the number of new housing units completed decreased to 41,300 units. Supply fell dramatically to 1,071 units in 1998. After that, the market remained steady until the graph rose slightly to reach 21,493 units in 2004 and then dropped away again to 18,963 units in 2005.

3-110

Housing price and units prefer for people in Bangkok Metropolitan Region

Survey data collected by the Centres of Housing Information and Home Buyer Guide Company from people who attended the Home and Condominium Exhibition 2003 showed that most people wanted a house priced in a range between $AUD17,000 and $AUD35,000 (500,000-1,000,000 Thai baht at an exchange rate of 30 baht per $AUD). Furthermore, the majority of people (71%) wanted to buy a house priced between $AUD 17,000 and $AUD 103,448 (that is, between 500,000 Thai baht and 3,000,000 Thai baht) (see Table 3-3).

Table 3-3 : Housing price and units prefer for people

Price (Thai Baht) <500,000 500,001 - 1,000,000 1,000,001 -1,500,000 1,500,001 - 2,000,000 2,000,001 -3,000,000 3,000,001 -4,000,000 4,000,001 - 5,000,000 5,000,001 - 7,000,000 >7,000,000 Non price expected Total Units 1,432 4,797 3,855 3,163 3,294 1,618 1,042 1,014 418 556 21,189 People (%) 6.8 22.6 18.2 14.9 15.5 7.6 4.9 4.8 2 2.6 100

About 71 % is the result from the price between 500,000 to 3,000,000

Source: Government housing Bank and Home buyer guide Company (2003)

3-111

Housing provider and type of housing in Bangkok Metropolitan Region

Based on data collected between 1996 and 2005, housing built by owners on their own land represents about 30% of the market share. Housing projects built by entrepreneurs make up 70% of market share (see Figure 3-12).

Graph of housing by type of house build

housing builed by owner 30% housing builed by developer 41% condominium builed by developer 29%

condominium builed by developer

housing builed by developer

housing builed by owner

Source: Government housing Bank

Figure 3-12 : Housing method to build from 1996 to 2005

3-112

Type of housing in the market in the Bangkok Metropolitan Region

The types of housing can be separated into four main groups: detached houses, attached houses, condominiums and block housing (townhouses and residential shops).
Graph of housing types in BMR (data from 1996 to 2005)

attached house 1% block 24% condominium 30% detached house 45%

detached house

condominium

block

attached house

Source: Government housing Bank

Figure 3-13 : Type of housing in Bangkok Metropolitan Region

The proportion of each housing type in the BMR is shown in the pie chart above. Detached houses form 45% of all types of housing, condominiums form 30% with the remainder being block housing or shop-houses (about 25%) and attached housing (at 1%) - (see Figure 3-13).

Proportion of housing type in the market in Bangkok Metropolitan Region from 1996 to 2005

The proportions within the housing market described above are changing. In 1996, the majority of dwelling units were condominiums (40%), 35% was block housing and 25% was detached housing. By 2005, detached houses constituted

3-113

65% of all dwelling types in the BMR with block housing standing at about 20 % and condominiums down to 15 % (see Figure 3-14).

Graph of ratio between house block and condominium

% 80
69.9 70.8

70
60.2

67.2 61.7 64.1 64.6

60 50
40 42.3 46.2

40 30 20 10 0 1996 1997 1998 1999 house 2000 2001 block 2002 2003 2004 2005 year
34.2 25.4 29.3 27.8 17.7 8 8.2 13.7 13.8 35.8 31.2 23.1 20.2 14.9 18.9 13.2 22.3 12.2 19.2 15.1

condominium

Source: Government housing Bank

Figure 3-14 : Housing type in the market from 1996 to 2005

The growth in the number of detached houses reflects the situation in the housing market. In recent years (eg. 2007), many housing projects have been undertaken in suburban areas where land prices are lower than in the city. The spread of housing projects around Bangkok has pushed up the number of detached houses in the BMR housing market.

Price index of housing in Thailand

The trend line of the housing price index in Thailand was slightly up from 2000 to 2005. The index was at 131.2 in 2000 and grew to 156.0 in 2005. The most recent data between 2004 and 2005 shows that the housing price index rose about 9% from 143.0 to 156.0 (The base price index of 100% being set in 1986).

3-114

percent of index (1986 as 100%)

180 160 140 120 100 80 60 40 20 0

Housing Price Index in Thailand

19 77

19 79

19 81

19 85

19 87

19 89

19 91

19 93

19 83

19 95

19 97

19 99

20 01

20 03

year (set 1985 as 1st year) Housing price index

Source: Government housing Bank

Figure 3-15 : Housing price index in Thailand from 1977 to 1992 and 2000 to 2005

Used housing market in Bangkok Metropolitan Region

Used houses are categorised as those houses sold by the previous owner/occupier. These can include all types of housing. Data for the used housing market show that about 70% of used houses in BMR are sold by the owner. Data about used housing supply are not collected systematically. Some say that the used housing supply exceeded new housing supply number in 2002. However, the real information was estimated from Government Housing Bank of Thailand sources to be about 20,000-30,000 units. It is thought that less than 15% of the number of used houses was traded on the market through an agency (Government Housing Bank of Thailand Journal).

Recently, in 2007, the housing market in the Bangkok Metropolitan Region is mainly comprised of new housing projects. The majority of housing sales relate to new houses which were built by housing or real estate companies. The used

20 05

3-115

housing market is now just a minor part of the market share. These phenomena show that the number of housing needs and housing demand units have increased as a result of increasing population, itself a result of the immigration factor. This is because housing suppliers in the market supply housing units to the market by building new projects (i.e. new housing projects built in the suburban area surrounding the Bangkok Metropolitan Region), while the proportion of the used housing market takes a decreasing market share. There is no recorded number for people who immigrate to the Bangkok Metropolitan Region from other regions of Thailand. The strong evidence of this immigration phenomenon is however obvious in Bangkok Metropolitan Region society.

Population data projected in this thesis shows that the mature BMR population is still growing at the same rate post-2007. It can be inferred, then, that upsizing will be an important feature of housing needs and housing demand and that the new housing market might expand. It can also be expected that the used housing market would expand if the number of housing vacancies remains high. The information from ERA Franchise Systems (Thailand) Co., Ltd, - the leading agency for used houses in the Thai market - shows that there were only about 460 units sold in 2001 at an average price per unit of approximately 2.5 million baht. The proportion of used houses on the market is therefore small, but seems to be growing.

Population and Housing growth in the Bangkok Metropolitan Region

Tangmatidham, (2006), suggests that new housing unit supply is increasing because the population is increasing, leading to new household formation and the purchase of housing units as investments In recent years, the population in

3-116

Bangkok has remained steady while Patumthanee province in the metropolitan area has increased significantly. An increasing population in any area can be expected to stimulate the extension of housing projects.

Land prices in Bangkok are very high; thus developers prefer to build housing projects in surrounding areas, to provide housing at an affordable price. Furthermore, there is limited space in Bangkok city. The recent average number of people per household in Thailand is 3.28, and the trend is towards smaller numbers per household than in the past.

3-117

Population related to housing market in the Bangkok Metropolitan Region

In Thailand the rate of housing growth is higher than that of population growth along the same timeline. The reason is that the rate of people per household has decreased over time (see Figure 3-16). Tangmatidham (2006) describes the relationship between the population and the increasing number of new houses in the market. The most recent ratio of people per household in the Bangkok Metropolitan Region is 2.56 people per household (data collected in 2005). It has decreased from 3.6 people per household in 1993.

Source: Department of Provincial administration, Thailand (2006) in Tangmatidham 2006 p.16

Figure 3-16 : The ratio of people per house hold (by average) from 1993 to 2005

The number of households in the Bangkok Metropolitan Region increased by 1.63% in 2005 (the most recent available data). The population of the BMR is

3-118

about 9.79 million (as registered). Therefore the population growth rate from 2004 to 2005 was about 160,000 people per year. This number can be divided by the average number of people per household in 2005 (2.56). The result is that 62,500 new housing units would be required, but the real data in the BMR housing register in 2005 shows 71,301 units (see Figure 3-17). Therefore, the number of houses registered in that year was higher than the number of new housing requirement as derived from average household size.

Source: Housing information centre of Thailand in Tangmatidham 2006 p.16

Figure 3-17 : Housing registered in Bangkok Metropolitan Region

This surplus can be explained by the fact that many buyers were not buying houses to live in. Rather, they purchased them for investment or speculation. This finding is similar to the evidence presented by Vanichvatana (2006) which showed that from 1994 to 1996 many bought houses for investment purpose.

3-119

Distribution of housing market in Bangkok Metropolitan Region compared to Thailand as a whole

Table 3-4 : Comparison of housing loans between Bangkok Metropolitan Region and other regions Housing loan All others regions (with out Bangkok Metropolitan Region) Rate Housing loan of growth (M Baht) (%) 44,825 73,439 63.8 109,841 49.6 118,715 8.1 115,450 -2.8 112,892 -2.2 111,365 -1.4 126,656 13.7 147,428 16.4 178,907 21.4
average
Source: Government housing Bank (2005)

Bangkok Metropolitan Region Year

Housing loan (M Baht)

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

94,719 122,337 165,963 175,296 164,706 161,871 162,115 167,465 184,737 214,132

Rate of growth (%) 29.2 35.7 5.6 -6.0 -1.7 0.2 3.3 10.3 15.9

Ratio between housing loan in Bangkok Metropolitan Region and all other region (%) 67.9 62.5 60.2 59.6 58.8 58.9 59.3 56.9 55.6 54.5 59.4

Graph of new housing loan from 1997 to 2006 M Baht 350,000 300,000 241,172 250,000 200,000 164,851 150,000 103,733 100,000 50,000 0 1995 1996 1997 1998 1999 2000 year 2001 2002 2003 2004 2005 2006 64,301 112,611 108,886 223,408 202,720 296,661 294,403 279,392 262,638

Source: Housing information centre of Thailand

Figure 3-18 : New housing loan from1995 to 2006

3-120

From 1995 the total value of new housing loans rose from 223,408 million baht to reach its first peak at 241,172 million baht in 1996. After 1997 (i.e. the year of Thailands economic crisis), the number of housing loans dropped sharply to touch a bottom of 64,301 in 1999. The total value increased slightly to reach a second peak at 294,403 million in 2004. This 2004 figure does not relate only to the housing market in the Bangkok Metropolitan Region, but also reflects a picture of holistic Thai market (see Tables 3-4 and Figure 3-18 above). There has not yet been a repeat of the 1996 boom.

Population and household size in Bangkok Metropolitan Region and Thailand

Census data were first collected in Thailand in 1909. At that time, the population was about 8 million. The rate of population increase rose from 1.22 per 1,000 people in 1919 reaching a peak of 3.15 per 1,000 people in 1960. After the Thai government launched its birth control policy, the rate of increase dropped to 0.88 per 1,000 people in 2000. The present population of Thailand is about 63 million, and about 10-15 million people were estimated to be living in the BMR. Birth rates, death rates and the increase in population rate, are shown as Figure 3-5.

3-121

Table 3-5 : Birth rate death rate and increased rate by area and region between 1985-1986, 1991 and1995-1996 1985-1986 Increase Birth Death rate rate rate by (per 1,000) (per 1,000) nature
(per 100)

Area

Birth rate
(per 1,000)

1995-1996 Increase Increase Death Death Birth rate rate rate rate rate by by (per 1,000) (per 1,000) (per 1,000) nature nature
(per 100) (per 100)

1991

Thailand 23.87 Bangkok Metropolitan 18.92 Region

6.44 3.84

1.74 1.51

20.19 14.65

5.93 3.31

1.43 1.13

17.90 14.24

6.02 3.92

1.19 1.03

Source: The National Statistical Office of Thailand

The most recent official data on household size in the Bangkok Metropolitan Region shows that there are 3.2 people per household (see Table 3-6).

Table 3-6 : Size of household in Thailand and Bangkok Metropolitan Region

Size of household in Thailand and Bangkok Metropolitan Area 1996 1997 1998 region 1999 2000 2001 2002

Thailand Bangkok Metropolitan Region

3.7

3.7

3.7

3.6

3.6

3.5

3.4

3.2

3.4

3.3

3.2

3.2

3.3

3.2

Source: The National Statistical Office of Thailand

3-122

Summary

The Bangkok Metropolitan Region covers an area of 7,761.50 km and has a registered population of 9,988,400 (as at April 4, 2007). Its population density is 1,286.92 per km, though more than 13 million people are in the metropolitan area during the day. The Bangkok metropolitan area covers the provinces listed in the table 3-7 below.

Table 3-7 : Population in provinces in Bangkok Metropolitan Region Area Provinces km in 2007 /km Population Pop. Density

Bangkok Nonthaburi Samut Prakan Pathum Thani Samut Sakhon Nakhon Pathom
Total

1,568.20 622.30 1,004.50 1,525.90 872.30 2,168.30

7,761.50

5,704,723 1,006,660 1,115,026 871,627 465,450 824,447

9,988,400

3,637.75 1,617.64 1,110.03 571.22 533.59 380.44

1,286.92

Source: Thai Department of Provincial Administration

4-123

Chapter 4

CONCEPT AND METHODOLOGY OF THE HOUSING DEMAND MODEL

Introduction

The review of the literature described in Chapter 2 confirmed that there are three main factors impacting on housing demand. The literature review also explored theories, methodologies and applications for modelling (i.e. System Dynamics). This chapter describes how this research brings together the three main factors and how System Dynamics is adopted to model housing demand in the Bangkok Metropolitan Region. The phenomena of housing demand is conceptualized; and the methodology utilised is described step by step in this chapter.

Establishing a conceptual housing demand model

The purpose of this research is to quantify housing demand precisely by building a model comprising three main factors. Figure 4-1 describes the conceptual basis for the model. Each box represents the quantity for each variable shown in the figure below.
x x

The 1st box represents the number of people in the total population The 2nd box represents the number of people in the mature population

The 3rd box represents the number of people in the mature population who have housing needs

4-124

The 4th box represents the number of people in the mature population who have housing needs and can afford to purchase a house (housing affordability)

The last box (Dashed line box) represents the number which quantifies the housing supply and which responds to the housing demand number.

4-125

Figure 4-1 : Kinetic box diagram of housing demand model

4-126

The fourth box is a subset of the first, second and third boxes. Similarly, the third box is a subset of the second and the first boxes, and the second box is a subset of the first box.

These boxes represent numbers (as explained earlier) and these numbers change over time. For example, the first box represents the population in a specific area. As people mature, they are represented in the population box as well as in the mature population box. Similarly, when people in the mature population box have housing needs, they are reflected in the mature population with housing needs. In the same way, when people in mature population with needs box (housing needs) can afford a house, they are represented in mature population with needs and affordability box. The sequence of moving people from the first box to the fourth box represents the phenomena which form this model. The unit in each box is changed from the first and the second box to the third and the fourth box, from the unit of people to the unit of household. To sum up, the absolute number of people contained within each box therefore moves from the first to the fourth box, depending upon certain conditions.

The arrows represent factors that can expand (i.e. pull factor) or shrink (i.e. push factor) the boundary of each box. This means that these factors can affect an increase or decrease in the numbers in each box by changing the size of each box because this size represents the amount of population in each box. Therefore, the bigger boxes represent more population than the smaller ones, respectively. Following the extent literature, there are three groups of factors which can impact each box in different ways (i.e. see Figure 4-1).

4-127

As the figure shows, all the factors are categorized into three groups. The first group represents demographic factors which play an important role in determining the boundary from the first to the fourth boxes. The second group of social factors controls the third and the fourth box. The last group comprises the economic factors which also impact on the third and the fourth boxes. Figure 4-1 therefore pictorially describes the conceptual basis for setting a housing demand model that deals sequentially with demographic, social and economic factors.

The dashed line box represents the housing supply or the number of houses in the market. The area of the dashed line box covers the third box and the fourth box. The area of the dashed line box covering the fourth box represents the number of houses supplied; in effect, the number of housing sales. The area of dashed line box that covers the third box also represents the number of houses absorbed by those people who cannot afford to purchase a house; in other words, the number of rental houses. The rental housing market is not considered in this research, as it unnecessarily complicates the model for the purposes of the discussion being undertaken. It does, however, need to be further considered and should be the subject of complimentary research at a later stage.

For the purposes of this research, the numerical value of housing demand is a number that represents all those people who need a house to live in and who can finance the purchase of a house. This mirrors the situation that occurs in the Bangkok Metropolitan Region where this demand is largely met by governmental provided housing.

People who have housing needs but who do not have the ability to buy a house are not therefore counted as a housing demand unit. Likewise, people, who have the

4-128

financial capacity to buy a house but have no housing needs or have no interest in buying a house, are not counted as a housing demand unit. However, people who do not need a house to live in but who can afford to purchase a house for investment purposes, and wish to make such an investment, can be counted as a housing demand unit.

Theoretically, demand is a technical economic term. Changes in demand are affected by price mechanisms and supply in the market. Both demand and supply react against each other to result an equilibrium price in the competitive market.

From this aspect of economics, demand cannot be easily captured due to movements in supply and the constraints of the inherent price mechanism. However, in reality, there would be a certain number of households which would finally occupy the houses. Thus, this process can be categorized and counted as a housing demand number.

The manner in which these elements of housing demand are counted by reference to the life cycle of housing demand is explained in the Figure 4-2 below.

4-129

Figure 4-2 : Housing demand production process

Consequential concerns with the housing demand model

A special feature of the System Dynamics model is the feedback loop. However, the housing demand model constructed for this research does not show any strong evidence of feedback loop action. The housing demand model has been constructed carefully to reflect the sequence of factors within the three key groups: beginning with demographic factors, moving to social factors and ending with economic factors.

By starting with the number of people (population) in a particular area, this number is, by necessity, impacted upon by social constraints. These social

4-130

constraint factors include marriage rates, divorce rates and new household formation rates. Through this process, the resultant model will show the number of households in need of housing. This number is filtered again by economic constraints. Those economic constraint factors include GDP growth index, income, housing loan and housing price index. Thus the number revealed by the model will reduce from the first result to the last result (i.e. see Figure 4-2). This is similar to the real world, where the number representing housing demand should be less than the number for the total population. This is because not all of the population buys or contributes to the demand for housing at the same time. The flow (or the speed) of those numbers are then constrained by the three groups of factors referred to previously.

In respect of demographic factors, the simulation for the Bangkok Metropolitan Region uses the 1977 population of four million as an initial value. This population is divided into three age groups: adolescent, mature and elderly. The numbers within each group changes without feedback from other groups; each generation grows freely. Thus, the changing of any groups quantity does not necessarily affect movements in other groups.

The social factors used in the model include the formation rate of new households, as well as marriage and divorce rates. Each is influenced by natural social behaviour and does not feed back to other factors. Both marriage and the formation of new households flow directly from population factors. While the annual divorce rate is influenced by the number of marriages, it does not feed back in a way that would change the number of marriages in the following year.

4-131

Finally in respect of economic factors, the GDP growth index, housing loans and the housing price index are included in the model. All economic factors are set as independent factors which directly affect housing demand and housing sales.

The GDP growth index is an economic factor that is measured by economic aspects. GDP represents the quantity of gross domestic product within one year. The GDP growth index is therefore set as an independent factor in this model to be a proxy representing the wealth of the population. The assumption then is set as:
x x

An increase in GDP represents an average increase in peoples income. A decrease in GDP represents an average decrease in peoples income.

Housing loans are influenced by many other factors, including income distribution, government policy and economic conditions. These factors are not included in this housing demand model, so in this models construction, housing loans are set as an independent factor whose movement depends on historical data. The housing loan factor in this model assumes that the data on housing loans (by itself) is the result of the interplay all these factors.

The housing price index is the last economic factor used in this model. It is a factor influenced by the housing market mechanism.

In conclusion, all factors flow in one direction from demographic factors to social factors and then ending with economic factors as shown in Figure 4-17. All forecasts in this model result from the values of the factors set for each scenario.

4-132

Regression equations are used to project (forward and / or backward) all of these demographic, social and economic factors. These projections are described furthering more detail later in this chapter.

Assumptions and limitations

There are two key assumptions underlying the model, and there are two key limitations of the model. These are described below.

An assumption of homogeneous feature of housing demand

In this research, each housing demand unit in the BMR is assumed to be homogeneous; that is, each housing demand unit has the same features. In reality, housing demand can be divided into many types. Housing types in BMR are divided into five categories: detached house, duplex, attached house (the formal term may be 'row-house' or 'townhouse'), apartment (unit, flat and condominium) and shop house. There are also varying levels of quality and cost, and hence value.

As the objective of this research is to find the quantity of housing demand as a whole, therefore, this model does not differentiate the units type of housing demand. This is because the housing demand (as a whole) can be satisfied by any type of housing. A housing demand unit in this model refers to the demand for every type of housing in the BMR.

An assumption regarding the used housing market

When used houses are sold they infer an exchange of a supply unit for a demand unit. The number of housing demand units is not affected by the sale of used

4-133

houses. For example, when someone sells their own house, housing demand is decreased by one unit, but at the same time a new unit of demand is created as the seller becomes a new demander. The assumption is that when the people in the used housing market sell or buy houses to each other, the number of housing demanders remains the same. This is because the selling and buying are actually just a swap between the house owners. The total housing demand number still remains the same. Thus, in this model, the influence of the used housing market is considered as a non-action factor and not included in the model.

A limitation of the model when dealing with housing consumption by wealthy people

This housing demand model does not capture the behaviour of wealthy people whose capacity to purchase housing is not limited by questions of affordability. The behaviour of this group is unpredictable. This behaviour is included in the error value which appears between the simulated data and historical data.

Limitations of the data

There are limitations of the data in this research. These limitations are related to data collection, analysis and calibration.

This study used only secondary data for the simulation of the model. Data from the case study area (Bangkok Metropolitan Region) was used. Where datasets were incomplete or unavailable, statistical methods were employed to estimate missing data or data trends. Secondary data for the case study area and population was collected. The reason for doing this was to follow the principle espoused by

4-134

Benedict and Szatrowski (1989): in general .the larger the sample size, the more accurate the result.

Data about migration of people to the BMR is unavailable. People do move in and out of the Bangkok Metropolitan Region without registering. People commute freely within the country for work and study purposes. As mentioned earlier, the population data is influenced by birth rates, death rates and migration rates. Therefore, due to the lack of migration data, this research cannot use only birth rates and death rates to determine the change in population without a migration rate. Thus, this study uses the data of the whole population adjusted each year to represent data which include all influential factors (i.e. birth rate, death rate and migration rate). These limitations are further discussed in the section dealing with demographic aspects of the housing demand model in Chapter 5.

The economic data used in the main factors in this model are from the GDP growth index, and the housing loan and housing price indices. It should be noted that these factors are inter-dependent upon many other factors. It is assumed that the economic factors result from influences outside the scope of the model. These influences are not included in this model. Data are collated on a yearly basis.

Principle behind the model and the process for building the model

The main aim in constructing the model is to imitate reality as closely as possible. To achieve this purpose, the sequence of input factors was carefully considered in order to imitate steps in the process that generates housing demand. The model is divided into three parts: demographic, social and economic, following the evidence offered in the literature review. Each part links to the numbers which flow from the demographic part to the social part and lastly the economic part.

4-135

The unit of the model begins with people (as a unit) from the demographic part and then changes to housing unit in the social part and the economic part. The model begins in 1976 and runs to 2007 on actual data, and then projects to 2017. The principle and process of this housing demand model for the BMR is described below.

The period of the model is somewhat limited by available housing sales data. Housing sales data, (which is finally compared with simulated housing demand data in the model) are available from 1977. Thus the period of the model is from 1977 to 2007. The projection for forecasting is from 2008 to 2027.

Demographic sector methodology part of the model

The intention in the demographic sector of the model is to imitate changes in population numbers in the BMR at particular times. The data used and method of calculating population numbers for simulation are explained below.

Population data is available from 1990 to 2006, and this data is used to generate the simulated data. The simulated population data is projected backward to 1976 as the initial year for modelling. The most similar data comparison is provided by the exponential equation which gives the highest coefficient of determination (R2) at 0.9295, as shown in equations 4-1, 4-2 and figure 4-3 below.

y = 4E+06e0.033x Or, y = 4,000,000*e0.033x

Equation 4-1

Equation 4-2

4-136

Where, y x e = number of population = number of year = 2.71828 (Euler's number)

As mentioned earlier, changes in population numbers over time are assumed to include all factors which influence population size. These factors consist of birth rates, death rates and migration rates. Thus, simulated numbers for population data were calculated from the population factors set out above. The method for establishing population trends is to apply the exponential equation in Figure 4-3 below.

population 12,000,000 10,000,000 8,000,000 6,000,000 4,000,000 2,000,000 0

Population data in BMR

y = 4E+06e0.033x R2 = 0.9295

Figure 4-3 : Population data in Bangkok Metropolitan Region

After inputting of population data, the model then calculates population numbers from 1976. Once this calculation is completed, the next step is to isolate the

19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 20 06
year (set 1990 as first year) population data to make equation for theory simulated data of population Expon. (population data to make equation for theory)

Source: From excel program

4-137

number of mature people within the total population. The age structure data for each age group is needed to generate the number of mature people within the population (see Table 4-2, the proportion by age from 2001 to 2003 in Thailand).

The population numbers simulated by the model are compared with BMR population data between 1990 and 2006. All factors influencing population numbers (such as birth rate, death rate and migration factors) are assumed. The changing of population in each year is a result of all these factors. The rate of population change is calculated using data from 1990 to 2006 shown in the Table 4-1 below.

4-138

Table 4-1 : Comparison between historical data of population and simulated data of population Population in Bangkok Metropolitan Region year * Historical data n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a 6,198,000 6,349,000 6,495,000 6,635,000 6,778,000 6,919,000 7,061,000 7,204,000 7,348,000 7,496,000 7,637,000 8,737,000 9,677,000 9,812,000 9,638,000 9,792,000 9,943,000 ^ Simulation data (4,000,000) 4,134,202 4,272,907 4,416,265 4,564,433 4,717,572 4,875,850 5,039,437 5,208,513 5,383,261 5,563,873 5,750,543 5,943,477 6,142,884 6,348,981 6,561,993 6,782,151 7,009,696 7,244,875 7,487,945 7,739,169 7,998,823 8,267,187 8,544,556 8,831,231 9,127,523 9,433,756 9,750,264 10,077,391 10,415,492 10,764,938

1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Source: The National Statistical Office of Thailand

(*data from historical data, ^data from calculation of Excel program, (number) initial data for the model)

Population numbers are then separated into three age groups: adolescent (0-19 years), mature (20-59 years) and elder (60 years or more). The proportion of each

4-139

population group is categorized and calculated by using the linear equation which gives the highest coefficient of determination (R2). These proportions are calculated from Thai population data from 2001 to 2003 (the most recent data available). These equations of adolescent, mature and elder are shown in equation 2, equation 3 and equation 4 respectively.

Flow rate equation of demographic factors

Using historical population data for the BMR, this research adapts the proportion of age structure of population in Thailand from 2001 to 2003 as a prototype for the ratios within the total population of each of the three age groups. Details are shown in the Table 4-2 below.

4-140

Table 4-2 : Population by age from 2001 to 2003 in Thailand

Population by age from 2001 to 2003 in Thailand 2001 % (Total) Age 0 - 4 5 - 9 10 - 14 15 - 19 Average of adolescent 20 - 24 25 - 29 30 - 34 35 - 39 40 - 44 45 - 49 50 - 54 55 - 59 Average of mature 60 - 64 65 - 69 70 - 74 75 - 79 80 - 84 85 - 89 90 - 94 95 - 99 over 100 Average of elder
Unknown
2,946,599 4.73 62,308,887 100

2002 Total
100 62,799,872

2003 total
100 63,079,765

%
100

%
100

Total
100

4,207,497 4,813,614 4,614,429 4,928,170

6.75 7.73 7.41 7.91 ^29.79 (31.19)

4,042,604 4,821,676 4,700,093 4,780,924

6.44 7.68 7.48 7.61 ^29.21 (30.56)

3,902,496 4,754,166 4,759,358 4,661,217

6.19 7.54 7.54 7.39 ^28.66 (29.93)

^ 30.577349 % 5,341,014 5,513,728 5,655,524 5,341,771 4,590,483 3,781,378 2,874,844 2,085,444 8.57 8.85 9.08 8.57 7.37 6.07 4.61 3.35 ^56.47 (59.14) 5,282,247 5,479,986 5,663,969 5,413,362 4,739,965 3,930,215 3,039,054 2,148,128 8.41 8.73 9.02 8.62 7.55 6.26 4.84 3.42 ^56.84 (59.47) 5,208,192 5,406,128 5,563,479 5,505,556 4,889,452 4,060,211 3,202,542 2,286,141 8.26 8.57 8.82 8.73 7.75 6.44 5.08 3.62 ^57.26 (59.87)

^ 59.496846 % 1,815,101 1,491,784 1,027,704 622,331 338,959 170,554 73,441 33,649 40,869 2.91 2.39 1.65 1.00 0.54 0.27 0.12 0.05 0.07 ^9.01 (9.43) 1,872,448 1,523,598 1,106,138 657,745 354,636 179,575 74,995 33,839 42,715 2.98 2.43 1.76 1.05 0.56 0.29 0.12 0.05 0.07 ^9.31 (9.74) 1,855,785 1,580,908 1,145,133 715,638 380,257 190,802 71,214 30,612 34,784 2.94 2.51 1.82 1.13 0.60 0.30 0.11 0.05 0.06 ^9.52 (9.95)

^ 9.7103632 % 2,911,960 4.64 2,875,694 4.56

Source: The National Statistical Office of Thailand, ^ calculated data, (number) = include average unknown number

4-141

The Figure 4-4 below presents the equation to calculate for trend equation.

Adolescent proportion
35 percentage of adolescent in BMR 30 25 20 15 10 5 0 2001 2002 2003 2004 2005

y = -0.5676x + 30.356 R2 = 0.9998

2006

2007

year(set 2001 as a 1ST year) Adolescent proportion Linear (Adolescent proportion)

Source: Excel program

Figure 4-4 : Proportion of adolescent population in Bangkok Metropolitan Region

A linear equation is used to calculate the trend for adolescent group numbers. The details of equation are shown below.

Adolescent population trend equation (0 to 19 year of ages)

Y = -0.5676x + 30.356 R2 = 0.9998 Where, Y x = percentage of adolescent population at time = year (set 2001 as the 1st year)

Equation 4-3

4-142

Mature proportion
60 percentage of mature in BMR 59 58 57 56 55 54 53 52 51 50 2001 2002 2003 2004 2005

y = 0.3981x + 56.062 R2 = 0.9989

2006

2007

year(set 2001 as a 1ST year)

Mature proportion

Linear (Mature proportion)

Source: Excel program

Figure 4-5 : Proportion of mature population in Bangkok Metropolitan Region

A linear equation is used to calculate the trend for mature group numbers. The detail of equation is shown below.

Mature population trend equation (20 to 59 year of ages)

Y = 0.3981x + 56.062 R2 = 0.9989 Where, Y x = percentage of mature population at time = year (set 2001 as the 1st year)

Equation 4-4

4-143

Elder proportion
12 percentage of elder in BMR 10 8 6 4 2 0 2001 2002 2003 2004 2005

y = 0.2547x + 8.7703 R2 = 0.9905

2006

2007

year(set 2001 as a 1ST year)

Elder proportion

Linear (Elder proportion)

Source: Excel program

Figure 4-6 : Proportion of elder population in Bangkok Metropolitan Region

A linear equation is used to calculate the trend for elder group numbers. The details of the equation are shown below.

Elderly population trend equation (60 and over 60 year of ages)

Y = 0.2547x + 8.7703 R2 = 0.9905 Where, Y x = percentage of elderly population at time = year (set 2001 as the 1st year)

Equation 4-5

After dividing population numbers into the three age groups, the next step is to build social factors into the model. Social factors include the marriage rate, the divorce rate and the new household formation rate.

4-144

The social sector methodology part of the model

The formation of new households, marriage and divorce are the social factors that affect housing needs. Thus, data for these three social factors are used to generate a numerical value for housing needs. The literature review shows that the mature age group produces a significant majority of new households. In this model, numerical data of those people who married, divorced and moved out from old households to make new households are calculated from data collated by the Thai statistic bureau.

Flow rate equation for social factors

Population data for municipal areas in Thailand are used to set the flow rate equation for the formation of new households.

The number of new household generated from population number is from three factors:
x x x

The marriage number where one married couple creates one housing need The divorce number where one divorced couple creates one housing need The formation of new households number where one household splits into at least two, thus creating one new housing need

These phenomena can be explained in terms of three activities:

x

People create housing needs when they get married. This is a traditional and common social activity in almost every society around the world. The newlyweds need to have their own private and separate accommodation is common.

4-145

People create housing needs when they get divorced. This happens because of the need to live separately from their ex-husband or ex-wife.

People create housing needs when they move out from their existing household. This can happen for many reasons, such as seeking for job, requiring privacy or setting up a new household.

In this research, the first activity is called the formation of couple head household while the second and third activities are called the formation of single head household. The phrase the formation of new households contains both the formation of couple head household and the formation of single head household.

Co-efficiency estimation for splitting of single head household data

The population data for municipal areas in Thailand is adopted as the base data for calculating a number for the formation of new households. This data is used because firstly sound population data in the BMR is not available and, secondly the behaviour of splitting households to produce new households is assumed to be similar to the behaviour of populations in municipal areas, rather than in rural areas. From this data, the number of single head households is assumed to equal the number of total households minus the number of couple head household (see equation 4-6 below). Resulting single-head household numbers are shown in Table 4-5, and the data used to calculate this is shown in Table 4-4.

NSHH = NTH - NCHH

Equation 4-6

4-146

Where, NSHH NTH NCHH = number of single head household = number of total household = number of couple head household

4-147

Table 4-3 : Number of head household from data of population in municipal area in Thailand in 2005 Total population in municipal area (thousand) 19,376.7 Total household (thousand) 5,288.8 Marriage (couple) head household (thousand) 3,338.9

Group of ages Total

0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90

4,021.5 1,423.6 1,640.4 1,856.2 1,861.7 1,760.6 1,606.9 1,410.0 1,092.8 807.8 1,124.6 591.8 152.4 26.2 *

5.6 55.8 217.5 404.9 558.0 629.6 667.5 657.8 561.9 445.6 662.2 337.2 76.7 8.5

1.9 34.7 174.5 342.2 453.9 486.7 450.0 427.1 330.8 235.7 280.0 109.0 11.3 1.0

Source: The National Statistical Office of Thailand

4-148

Table 4-4 : Number of single head household of population in municipal area in Thailand in 2005 Single head household (thousand) 1,949.8 3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4

Order of cohort

Group of ages Total

1 2 3 4 5 6 7 8 9 10 11 12 13 14
Source:

0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90 *

The National Statistical Office of Thailand

To be able to capture the rate of growth (or change) in numbers for each cohort, this research simply applies the delta method as shown in equation 4-7 and equation 4-8.

NSSHH = NSHH

Equation 4-7

Where,

4-149

NSSHH NSHH

= Number of splitting of single head household = Number of single head household

And, NSHH = NSHH (2) - NSHH (1)

Equation 4-8

Where, NSHH NSHH (2) NSHH (1) = Number of single head household = Number of single head household of older cohort = Number of single head household of younger cohort

From this equation, the result of number of growth per group of ages are provided as a Table 4-5 below

4-150

Table 4-5 : Number of single head household of population in municipal area in Thailand in 2005 Single head household (thousand) 1,949.8 Number of growth (change) of the formation of single head household per group of ages (thousand) 3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 -58.0

Order of cohort

Group of ages Total

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90

3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4

Source: calculation of Excel program

The data set out in Table 4-5 show strong evidence of growth in the formation of single-head households withn the mature group, between the ages of 30 and 44 years, and in the older group between the ages of 60 and 69 years. As the available data groups the elderly into a 10 year range as the one group, the data cannot be divided into sub-groups of, say, 60 to 64 and 65 to 69. Therefore, for the group of 60 to 69, the number of changes per age group would be divided by

4-151

10 to get the number of change per year range. Similarly, the number in the 0-14 age group is also divided by 15 for the same reason.

As the data reported in the table show, there are some negative values in the 55 to 59 and 70 and over age groups. In fact, it is possible for the heads of single households to be dying out or moving in with others. This is to be expected in the high age groups. However, to avoid the problem of using a minus number in calculating an equation to be used in this work, this research considers only those groups with positive values. The negative values are considered as non-positive or zero number in the model. The reason is because that the negative value means that no household is produced. Where the household growth production is less than one it infers a zero number for all circumstance.

Next the number representing the splitting of single-head households will be divided by the range of age in each group to get the number representing the splitting of single head household per age range. The results are shown in the Table 4-6 below.

4-152

Table 4-6 : Number of single head household of population in municipal area in Thailand in 2005 Single household head (thousand) 1,949.8 Number of change per group of ages (thousand) Number of change per each cohort group (thousand)

Order of cohort

Group of ages

Total

1 2 3 4 5 6 7 8 9 10 11 12 13 14

0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90

3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4
Source: calculation of Excel program

3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 -58.0

0.26 3.48 4.39 3.92 8.28 7.76 14.92 2.64 0.08 -4.22 17.23 -30.81 -32.54 -11.60

Finally, the number of splitting single head households in each age group is calculated. This is to find out the result as the ratio of splitting population in percentage. The results are shown below in Table 4-7.

4-153

Table 4-7 : Splitting (formation) single head household number from data of municipal area in Thailand in 2005
*
Group

*
Total number of population in municipal area (thousand)

of ages

* * Number Marriage head of household head (thousand) household

^
Single head household (thousand)

(thousand)

^ ^ Number Number of of change change per per group each (thousand) cohort group

^ Ratio of the splitting

(formation)

of single head (thousand) household per total

population

Total 0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90

19,376.7

5,288.8

3,338.9

1,949.8

(percent) 3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 -58.0 0.26 3.48 4.39 3.92 8.28 7.76 14.92 2.64 0.08 -4.22 17.23 -30.81 -32.54 -11.60
0.01 0.24 0.27 0.21 0.44 0.44 0.93 0.19 0.01 -0.52 1.53 -5.21 -21.35 -44.20

4,021.5 1,423.6 1,640.4 1,856.2 1,861.7 1,760.6 1,606.9 1,410.0 1,092.8 807.8 1,124.6 591.8 152.4 26.2 *

5.6 55.8 217.5 404.9 558.0 629.6 667.5 657.8 561.9 445.6 662.2 337.2 76.7 8.5

1.9 34.7 174.5 342.2 453.9 486.7 450.0 427.1 330.8 235.7 280.0 109.0 11.3 1.0

3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4

Source: The National Statistical Office of Thailand (*data from historical data, ^data from calculation of Excel program)

This research uses the percentages shown in Table 4-7 to calculate in the model. The equations to calculate the ratio of splitting of single head household per total population are shown as the equations below.

4-154

RSHHpTP =

'(NHH - NMHH)  RofGOA * TP

Equation 4-9

Or,

RSHHpTP =

'(NHH - NMHH) RofGOA TP

* 100

Equation 4-10

Where, RSHHpTP NHH NMHH RofGOA TP = ratio of splitting of single head household per total population = number of total head household = number of married head household (couple head household) = range of each group of ages = total population in municipal area

And (NHH NMHH) = NHH NMHH Where, NHH NMHH = NHH2 NHH1 = NMHH2 NMHH1
Equation 4-11

And NHH2 NHH1 NMHH2 = number of total head household at time 2 = number of total head household at time 1 = number of married head household (couple head household) at time 2 NMHH1 = number of married head household (couple head household) at time 1

4-155

The splitting ratio of single head households to the total population will be used (as a rate) to calculate and to generate a number for single head household formation in the model. The population group in the age bracket from 55 to 59 and over 70 which have negative values of splitting single head household will be calculated by using zero value because negative numbers for the splitting of single head households would be interpreted as they merge their household with others. From a demographic perspective, population in this model are divided into three groups (i.e. adolescent, mature and elder). Splitting single head households and their formation are shown in Table 4-8 below.

4-156

Table 4-8 : Splitting of single head household number by age group from data of municipal area in Thailand in 2005 * Group of ages *
Total population in municipal area (thousand)

^
Single head household (thousand)

^ Number of change per group

(thousand)

^ Number of change per each cohort group (thousand)

^ Ratio of the splitting of single head household per total

population

(percent)
Total 19,376.7 1,949.8

^ Ratio of the splitting of single head household by age group (percent) 4.27

0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89 90
Source:

4,021.5 1,423.6 1,640.4 1,856.2 1,861.7 1,760.6 1,606.9 1,410.0 1,092.8 807.8 1,124.6 591.8 152.4 26.2

3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 7.4

3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 -58.0

0.26 3.48 4.39 3.92 8.28 7.76 14.92 2.64 0.08 -4.22 17.23 -30.81 -32.54 -11.60

0.01 0.24 0.27 0.21 0.44 0.44 0.93 0.19 0.01 -0.52 1.53 -5.21 -21.35 -44.20

0.25

2.49

1.53

The National Statistical Office of Thailand (*data from historical data, ^data from calculation of Excel program)

To estimate the co-efficienct for finding the ratio of the formation of single head household number in each age range per year, the population data in 2005 for Thai municipal areas (the most recent data) is used. The ratio of the data is used to

4-157

approximate the rate of splitting of single head household in each group. The data is used in the equation shown below in Equation 4-12.

NSSHHpY

X * (0.29/100) Y * (2.49/100) Z * (1.53/100)   19 40 40

Equation 4-12

Where, NSSHHpY X (0.29/100) 19 Y (2.49/100) = number of splitting single head households per year = number of adolescents from 0 to 19 = proportion of adolescent creating a single household per year = age range for the adolescent group (from 0 to 19) = number of mature people from 20 to 59 = proportion of mature people creating a single household per year 40 Z (1.53/100) 40 = age range for the mature group (from 20 to 59) = number of elders from 60 to 100 = proportion of elders creating a single household per year = age range for the in elder group (from 60 to 100)

In the next step, the BMR marriage and divorce data from 1996 to 1999 is fed into the model. This model uses these data to set a trend of the data calculated from the regression equations. This is because the data is available only for the four year period from 1996-1999. The coefficients to determine a value (R2) from regression equations in these cases do not provide a high level of confidence. However, the trend for both factors follows the data for Thailands population. Details for marriage and divorce are reported in Table 4-9, Table 4-10, Figure 4-7, Figure 4-8, Equation 4-13 and Equation 4-12.

4-158

Data of marriage and divorce co-efficiency estimation

The most recent marriage and divorce data for BMR, on which the following table is based, were collected for the four year period 1996-1999.

Table 4-9 : Data of marriage and divorce ratio of population in Bangkok Metropolitan Region * Population in Bangkok Metropolitan Region 7,348,000.00 7,496,000.00 7,637,000.00 8,737,000.00 * Number of marriage ^ Percent of marriage

Year

* Number of divorce

^ Percent of divorce

1996 1997 1998 1999

Source:

64,179 64,229 52,626 58,499

1.65 1.62 1.30 1.42

14,742 16,483 17,699 17,042

22.97 25.66 33.63 29.13

*The National Statistical Office of Thailand (*data from historical data, ^data from calculation of Excel program)

% of population 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1996 1997

Marriage proportion in BMR

y = 1.68x -0.1486 R2 = 0.6215

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

year set 1996 as first year Mariage proportion of mature population in BMR Power (Mariage proportion of mature population in BMR)

Source: Excel program

Figure 4-7 : Marriage in the Bangkok Metropolitan Region

To estimate the coefficient for the number of marriages for a particular time period, a regression equation method was used (by the power equation) with the

4-159

coefficient of determination (R2) at 0.6215. Then this method provides the equation as shown below (see Figure 4-7).

MN =

TP *1.68 * Y 100

-0.1486

Equation 4-13

Where, MN TP Y = marriage number = total population = year of marriage & divorce equation (set 1996 as the 1st year)

Divorce ratio of marriage People in BMR

% of m arriage 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 1996 1997 1998 1999 2000 2001 2002 2003 2004

y = 23.031x 0.2263 R2 = 0.6855

2005

2006

2007

year set 1996 as first year

Divorce ratio of marriage

Power (Divorce ratio of marriage)

Source: Excel program

Figure 4-8 : Divorce in the Bangkok Metropolitan Region

To estimate the coefficient for the number of divorces for a particular time period, a regression equation method was used (by the power equation) with the

4-160

coefficient of determination (R2) at 0.6855. Then this method provides the equation below (see Figure 4-8).

DN =

TP * 23.031 * Y 100

0.2263

Equation 4-14

Where, DN TP Y = divorce number = total population = year of marriage & divorce equation (set 1996 as the 1st year)

The accuracy of trends established for marriage and divorce was confirmed through application of the power equation. Comparisons were then made with marriage and divorce data for the population of Thailand (see Table 4-10). These comparisons are shown in Table 4-10. Comparison confirms the trend projection for marriage and divorce in the Bangkok Metropolitan Region. This table shows that the number of marriages decreased, while divorce numbers and rate (percentage of divorce) increased from 1993 to 1995.

Table 4-10 : Marriage and divorce number of population in Thailand Table of marriage and divorce in Thailand year Marriage (people) Divorce (people) percent of divorce 9.69 10.77 11.38 12.98 15.72

1993 1994 1995 1996 1997

484,569 435,425 470,751 436,831 396,928

46,953 46,903 53,560 56,718 62,379

4-161

1998 1999 2000 2001 2002 2003 2004 2005 *

324,262 354,198 337,140 324,661 291,734 328,356 365,721 345,234

67,551 61,377 70,882 76,037 77,735 80,886 86,982 90,688

20.83 17.33 21.02 23.42 26.65 24.63 23.78 26.27

Source: The National Statistical Office of Thailand

Economic sector methodology part of the model

Economic factors used this model include GDP growth index, housing loans and housing price index. Before final selection, the full range of economic factors suggested from the literature review (including GDP growth index, housing loan, housing price index, housing loans interest rate, income index and housing supply) were inputted. A number were found to have no influence and were therefore deleted from the models input. The three selected economic factors selected were found to provide best-fit results in the model. These three factors also play an important role in converting housing demand to housing sales. The flow-rate equation uses the graphical function of System Dynamics (iThink

program). This is because the coefficients between economic factors and housing needs in the model have a nonlinear function. Most required economic data are available. Where data were found to be unavailable a regression equation to provide missing data was used.

Flow rate equation of economic factors

After social factors are included in the model, the model generates a numerical value for underlying housing needs. The subsequent incorporation of economic factors generates a numerical value for housing demand by taking the numerical

4-162

value for underlying housing needs and modifying that value with reference to data on the capacity of people to finance a house purchase, and on their interest in purchasing a house. Interest in buying a house is counted for both those who need a house to live in and those who wish to buy a house as an investment. The process is described as follows.

First, the group of people who need houses is assumed to have income: the distribution of income depending on condition of economic factors. TheGDP growth index is taken as the factor representing the movement of peoples income and wealth required to buy a house in this model.

Next, the number of any housing needs unit that cannot be converted to be housing demand units for the first time (i.e. year) is generated and included in the number of housing needs units in the next round (i.e. year). The balance of housing needs equation is as follows.

RNHN = NHN NHD Where, RNHN = the remaining number of housing needs NHN NHD = number of housing needs = number of housing demand

Equation 4-15

Converting housing demand into housing sales relies upon people needing a house, and having the money to buy one. The number of houses bought is converted to the housing demand and then becomes housing sales in the model.

4-163

This part of the model uses economic factors as indicators of capacity to pay for a house purchase. At this point all of the housing demand units in the model are assumed to be based on a willingness to buy a house. These demand units are converted to housing sales when the people in those units have sufficient finances to make the purchase (housing affordability). Economic factors which influence capacity to purchase a house are GDP growth index, housing loans from banks and housing price index.

The number of people who have the capacity to purchase can be determined by reference to income levels and that willingness to buy. GDP, housing loan and housing price indices in tandem represent these conditions. The relationships between these factors and housing demand are non-linear. This research uses a graphical function to calculate the fraction of efficiency for those who have the capacity to pay. The three economic factors and their relation to housing demand are described below.
x

GDP - the Gross Domestic Product growth index -

is an index for

measuring economic growth. When GDP increases, it implies that income is growing within a population. Psychologically, people are more likely to spend when economic conditions are good. For these reasons and also strong evidence from historical data of relationship between GDP and housing sales from 1994 to 1998, the GDP growth index is taken as the first indicator of capacity to pay (housing affordability).
x

Housing loans or home loans are the main financial mechanism for buying

housing, and housing loan availability has become an important tool for controlling housing demand. Most house buyers need banking support to finance their purchases. Home loans approved are assumed to then convert housing

4-164

demand to housing sales. Home loans therefore bring housing demand into direct relationship with housing sales.
x

The housing price index has a key impact on the number of housing sales.

The indirect relationship between the housing price index and the rate of conversion from housing demand to housing sales assumes that when housing prices increase, the number of people able to buy a house decreases. In contrast, when housing prices decrease, housing affordability increases, provided that finance is available at reasonable rates.

The housing loan and housing price indices are combined as one fraction in this model. The reason is that both factors affect each other. The number of housing loans approved by the bank does not translate to the type of household unit, however the available data on housing loans is collected in terms of Thai currency unit (million Baht). Thus the housing price index is adapted from the fraction of housing loan over housing price index. Therefore, when housing loans are increased, the fraction of housing loan over housing price index also increases. (see Figure 4-13). To sum up, this model uses the GDP growth index and the fraction of housing loan over housing price index as a coefficient function to generate the numerical values from housing demand as housing sales. The historical performance, and trends, for the three economic factors are shown in Figure 4-9, Figure 4-10 and Figure 4-11.

4-165

GDP 15.00 10.00 5.00 0.00

19 80 19 82 19 86 19 84

GDP in Thailand

y = -1.0436Ln(x) + 8.4333 R2 = 0.0344

19 92

19 90

19 88

19 98

19 96

19 94

20 04

20 02

20 00

-5.00 -10.00 -15.00

year (set 1980 as 1st year)

GDP data

Log. (GDP data)

Source: Excel program

Figure 4-9 : Equation trend of GDP growth index in Thailand

The projection equation of GDP growth index (to be input to the model for forecasting from 2008 to 2027 for the first scenario) uses the logarithmic equation at the coefficient of determination (R2) at 0.0344 as shown below.

Y = -1.0436 Ln(x) + 8.4333

Where, Y x Ln = GDP growth index = year (set 1980 as the first year) = mathematical value (Formally, Ln (a) may be defined as the area under the graph of 1/x from 1 to a, as the integral)

20 06

Equation 4-16

4-166

Loan (M Bath) 350,000 300,000 250,000 200,000 150,000 100,000 50,000 0

Housing loan in Thailand

y = 39295x 2 R = 0.5772

0.6

20 01

20 05

19 85

19 87

year (set 1985 as 1st year) Housing loan Power (Housing loan)

Source: Excel program

Figure 4-10 : Equation trend of housing loan in Thailand

The projection equation of housing loan (to be input to the model for forecasting from 2008 to 2027 for the first scenario) uses the power equation at the coefficient of determination (R2) at 0.5772 as shown below.

Y = 39295 *x0.6

Equation 4-17

Where, Y x = housing loan = year (set 1985 as the first year)

20 07

19 89

19 93

19 91

19 97

19 95

19 99

20 03

4-167

percent of index (1986 as 100%)

180 160 140 120 100 80 60 40 20 0

Housing Price Index in Thailand

y = 93.081x
2

0.1362

R = 0.9249

19 79

19 95

20 01

19 99

19 89

19 91

year (set 1985 as 1st year) Housing price index Power (Housing price index)

Source: Excel program

Figure 4-11 : Equation trend of housing price index in Bangkok Metropolitan Region

The projection equation of housing price index (to be input to the model for forecasting from 2008 to 2027 for the first scenario) uses the power equation at the coefficient of determination (R2) at 0.9249 as shown below.

Y = 93.081 *x0.1362

Where, Y x = housing price index = year (set 1985 as the first year)

20 05

Equation 4-18

20 07

19 85

19 81

19 87

19 93

19 77

19 83

19 97

20 03

4-168

The assumptions relating to the impact of economic data on housing demand in this model are defined as follows:x

Income is assumed to follow the same direction as the GDP growth index.

Capacity of housing buyers to purchase a house is assumed to follow the same direction as rises or falls in the number of housing loans.

Capacity of housing buyers to purchase a house is assumed to follow the opposite direction as rises or falls in the number of housing price index.

The housing price index in Thailand has been set as a base of 100% for 1986. Housing price data missing between 1991 and 2000 are replaced by results from the power equation shown in Figure 4-11.

4-169

Table 4-11 : Housing price index in Bangkok Metropolitan Region and missing data * Housing price index 53.90 54.20 56.20 63.40 68.20 78.50 83.30 88.90 94.60 100.00 106.90 111.80 117.10 121.10 124.00 n/a n/a n/a n/a n/a n/a n/a n/a 131.20 130.90 131.80 135.70 143.00 155.80 ^ Data from equation for price index 123.56 125.55 127.37 129.03 130.57 132.00 133.34 134.60 -

year 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Source: National Housing Bank (*data from historical data, ^data from calculation of Excel program)

Equation for economic factors

To define the relationship between the GDP growth index and capacity to pay, the graphical function is used in the model. Because of the non-linear relationship

4-170

between GDP growth index and the capacity to pay, the assumption set is that the higher the growth recorded by the GDP growth index, the more capacity to pay is stimulated.

The graphical function is specified by 11 data points. The range of the x axis is divided into 10 spans. The process of graphical setting functions is: 1. capturing the fraction between X factor and Y factor at minimum and maximum points 2. considering the slope or fraction of X factor and Y factor from minimum to maximum points (with in 10 spans) 3. linking all points with a line to make a smooth curve

All graphical function ratios have to be calibrated in order to fit the historical data. These functions and their relationships are described below.

GDP is the first economic factor used in this model to determined housing affordability (for both housing service and housing investment purpose).The minimum and maximum value of GDP from 1980 to 2005 is considered for setting the graphical function. The coefficient of GDP on housing affordability is calibrated from simulation trial to fit the historical data. The graphical function and table of co-ordinates for this factor (i.e. coefficient of GDP on affordability and investment) are shown in Table 4-12 and Figure 4-12 below.

4-171

Table 4-12 : minimum and maximum value of GDP growth index and coefficient of GDP on housing affordability & investment

Minimum co value efficient GDP growth index (%)

(1998)

Maximum value
(1988)

co efficient

Purpose on affordability & investment (from housing needs to housing demand)

-10.5

0.12

13.2

10.3

Source: iThing program (* this table adopt historical data from1977 and 2007, (xxxx) = year)

Source: simulation program

Figure 4-12 : coordinate of GDP growth index on affordability and investment

The GDP growth index also affects housing consumer confidence. The relationship between GDP growth index and housing consumer confidence assumes that when the GDP growth index falls, housing consumer confidence drops. Generating a factors coefficient uses the graphical function setting as

4-172

described previously. The graphical functions and coordinates are shown in Table 4-13 and Figure 4-13 below.

Table 4-13 : minimum and maximum value of GDP growth index and coefficient on housing consumption decision Minimum co Maximum value efficient value GDP growth index (%)
(1998) (1988)

co efficient

Purpose on consumption decision (from housing demand to housing sales)

-10.5

0.075

13.2

10.1

Source: iThing program (* this table adopt historical data from1977 and 2007, (xxxx) = year)

Source: simulation program

Figure 4-13 : coordinate of GDP growth index on housing consumer confidence.

Next, housing loans and the housing price index are set as a fraction to derive a coefficient of the capacity to buy housing (housing affordability). The assumption

4-173

is that when the fraction of housing loan over the housing price index increases, the housing-consumed rate increase. Conversely when it decreases, the housing consumed-rate also decreases. In the market, when the housing price index increases, the value of this fraction decreases automatically as it is a fraction of housing loan over housing price index. This principle follows the mathematical role of normal fraction of the numerators and denominators. The graphical function and coordination are show below (see Table 4-14, Table 4-15 and Figure 4-14).

4-174

Table 4-14 : Setting relation of housing loan and housing price index on housing consumption capacity Housing loan Housing price index 53.90 54.20 56.20 63.40 68.20 78.50 83.30 88.90 94.60 100.00 106.90 111.80 117.10 121.10 124.00 ^123.56 ^125.55 ^127.37 ^129.03 ^130.57 ^132.00 ^133.34 ^134.60 131.20 130.90 131.80 135.70 143.00 155.80 141.81 ^Ratio of housing loan on housing price index

year

1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

^29,000 ^29,000 ^36,000 ^41,000 ^45,000 ^40,000 ^35,000 ^30,000 ^35,000 ^35,000 ^30,000 ^40,110 53,736 59,368 78,306 111,346 147,829 209,811 223,408 241,172 202,720 103,733 64,301 108,886 112,611 164,851 296,661 296,025 279,407 279,392

538.03 535.06 640.57 646.69 659.82 509.55 420.17 337.46 369.98 350.00 (minimum value) 280.64 358.77 458.89 490.24 631.50 901.15 1,177.45 1,647.26 1,731.44 1,853.41 1,535.76 777.96 477.72 829.92 860.22 1,250.77 2,186.15 (maximum value) 2,070.10 1,908.90 1,970.19

Source: ^ Excel program

4-175

Table 4-15 : minimum and maximum value of fraction of housing loan over housing price index and coefficient on housing consumption decision Minimum co Maximum co value efficient value efficient Housing loan over housing price index (ratio)

Purpose

(1986)

(2003)

350

1.164

2186.15

1.771

on consumption decision (from housing demand to housing sales)

Source: iThing program (* this table adopt historical data from1977 and 2007, (xxxx) = year)

Source: Excel program

Figure 4-14 : coordinate of housing loan and housing price on housing consumer capacity

In conclusion, the number of housing sale and housing supply is not the main objective of this research, but is required to give an holistic view of the housing demand model. The relationships between housing sales and housing supply are included. The assumption is that, as housing sales increase, supply is stimulated

4-176

and will also increase. The significant point of these relationships, based on historical data, is that housing supply increases dramatically when housing sales reach 100,000 housing units per year (as shown in Figure 4-15). The historical data for housing sales and housing supply from 1977 to 2006 is shown in Table 416, below.

4-177

Table 4-16 : housing sales and housing supply year *Housing *Housing sales supply 63,163 n/a 48,119 n/a 63,775 n/a 63,652 n/a 45,892 n/a 41,881 n/a 26,297 n/a 50,638 n/a 50,086 n/a 33,722 n/a 53,353 n/a 67,451 n/a 80,031 n/a 102,335 n/a 129,688 81,657 108,001 58,497 134,086 113,150 171,254 253,028 172,419 155,522 166,785 120,671 145,355 36,537 63,864 2,407 33,382 4,068 32,028 6,789 34,023 13,126 37,182 29,217 56,085 55,539 69,050 68,579 70,072 63,579 75,000 66,118

1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

Source: Government housing bank of Thailand

At the point of converting housing-sale stock and housing-supply stock in the model, the number of housing-sales does not feed into the stock of housingsupply as earlier processes might suggest (i.e. from population factors to economic factors). The intention of making this link was for verifying the housing market behaviour in the model and comparing it to the real world. The assumption was that whenever housing sales increase, they stimulates housing developers supply to the market to that particular order. The coefficient of housing sales on

4-178

housing supply increases significantly from 100,000 housing units to 160,000 housing units, but after that the increase in housing supply starts to slow as shown in the Figure 4-15. The graphical function and coordination are shown below (Table 4-17 and Figure 4-15).
Table 4-17 : minimum and maximum value of housing sales and coefficient on housing supply Minimum co Maximum co value efficient value efficient Housing sales (unit)
(1983) (1995)

Purpose on housing supply (from housing sales to housing supply)

26297

0.33

172419

1.91

Source: iThing program (* this table adopts historical data from1977 to 2007, (xxxx) = year)

Source: simulation program

Figure 4-15 : coefficient of housing sales data on housing supply Stock equations

4-179

The stock equations are set as an accumulated number each year to draw a graph that compares the numbers simulated through the model with year-by-year historical data. Four equations represent the demographic data: one equation represents the social and economic data; the other three equations represent the check points for verifying the models accuracy.

These are listed below (see Figure 4-17). Each equation set is then shown in Table 4-18 below.

Demographic stocks

x total population x adolescents from 0 to 19 year of age x mature people from 20 to 59 year of age x elders from 60 and over year of age

Social stock

x housing needs (sHS needs)

Economic stock

x housing demand (sHS demand)

Check point stock

x housing sales (sHS sales) x housing supply (sHS supply) x housing vacancy (sHS vacancy)

4-180

Details of initial values of all stocks are provided in Tables 4-18 below.

Table 4-18 : Detail of initial values of all stock in housing demand model initial values (1976) 4,000,000 794,407 1,467,125 229,515 60,000 60,000 60,000 60,000 0

Stock

Total population Adolescents from 0 to 19 Mature people from 20 to 59 Elders from 60 and over Housing needs Housing demand Housing sales Housing supply Housing vacancy (accumulated value)
Source: iThink program

System environment of the model

The model is best described in Figure 4-16 which shows the sequence of the model. The model is separated into three parts: demographic, social and economic. Factors in the model are grouped into two types: causal factors and resultant factors. Causal factors are represented by rectangles and resultant factors are represented by ovals. Arrows show the direction of flow for numbers and its influence factors. Arrows which point to other arrows indicate the causal factor which influences the rate of flow between the factors linked by the arrow.

4-181

Figure 4-16 : System of housing demand environment

4-182

Symbols used in the Model

This research used the iThink software program, Version 8, for modelling. The modelling process is best understood with reference to the map of the model. The composition of the model is shown below.

Compose of model can be described as follow:

The box figure represents stock of factors.

The valve figure represents inflow rate.

The valve figure represents outflow rate.

The circle figure represents factors which influence to inflow or outflow rate.

4-183

The arrow represents the influence between factor and inflow or outflow rate.

The map of the iThink program model shows the sequences and the relationships between all factors in the model. The models map begins with the stock of total population which is then divided into three cohorts: adolescent, mature and elder. The splitting of single head household factor uses the data from those three stocks for calculation. Total population stock also provides data for the calculation of the marriage factor, and data for marriage is used to calculate the divorce rate.

Social factors are the summation number of those factors, which includes the formation of single head household factor, the marriage factor and the divorce factor. These are combined to represent the housing needs stock. Economic factors include the GDP growth index, housing loans and housing price index. A graphical function is used to perform calculations and conversions in sequence from housing needs to housing demand and then to housing sales.

The stock of housing needs and housing sales are used to calculate housing supply stock and housing vacancy stock. The results of housing demand model simulations are presented in the next chapter (Chapter 5).

4-184

Chapter summary

This chapter explains how the model used in this research was derived. Evidences from the literature review were used to develop the conceptual idea for this model. The data used were described step by step, from the demographic part to the social part and to the economic part. All the equations used in this model were also shown. The concepts underlying the model are illustrated in Figure 4- 17 below.

4-185

Source: iThink program

Figure 4-17 : Map of housing demand mode

5-186

Chapter 5

MODEL SIMULATION

Introduction

This chapter will describe model simulation and its accurate calibration. The chapter also considers some of the models limitations. The objective of the model is to capture elements in the Bangkok Metropolitan Region (BMR) environment that produce housing demand. Producing a model that generates forecasts of housing demand is one outcome sought from this study. This chapter does not contain an exhaustive review of the process of trial and error prior to achieving a final model. This process is, however, recorded in Appendix B.

Considerations in running the model

The demographic and social factors fed into the model produced a figure which represents the Underlying Housing Needs Units (UHNUs). This figure captures the number of housing needs units at a particular time. In theory, if the number of housing sales is the same at all times as the number of UHNUs, and if there are no financial constraints among those with housing needs, then all UHNUs would be converted to housing sales in a short period of time. The market would then be considered to be in an equilibrium at that particular time and in that particular area. The following section provides further detail on UHNUs.

5-187

Description of Underlying Housing Needs Units (UHNUs)

The number of UHNUs is produced by combining three factors: the number of married couples at a particular time (i.e. 1 couple = 1 UHNU); secondly, 50% of divorced people (i.e. on the assumption that the other 50% will keep living in the same dwelling and 50% will move to another dwelling); and, thirdly, the number of people who split from their household to make a new household. This last figure is derived by using age structure data. These data used for each of these factors is explained in previous chapter (i.e. Chapter 4). The total number of UHNUs is, therefore, the actual number of housing needs. The assumption is that the number of UHNUs contains both housing needs number (without reference to the capacity to finance the purchase of a house) and the housing demand number (which is defined by capacity to finance a house purchase).

The ratio between housing needs and housing demand moves with the market. Housing demand increases when the economy booms and decreases when the economy slows down.

Theoretically, this UHNUs - which contain both the number of housing needs units and of housing demand units - would have a higher proportion of housing needs and less housing demand, when the actual housing-sales line is running under the underlying-housing-needs line.

On the other hand, it would have a lower proportion of housing needs and higher housing demand when the actual housing-sales line runs over it. The reason for

5-188

this is that housing sales are converted from housing demand. When housing demand is higher, more housing sales will follow. Both scenarios are drawn in the Figure 5-1 and 5-2 below.

Source: simulation program

Figure 5-1 and Figure 5-2: Comparison of the proportion of the underlying housing-needs composition with housing sales

Thus, important characteristics of the UHNUs are: x Most underlying housing needs will be soon converted into housing demand and later into housing sales.

5-189

The total number of UHNUs will change as social factors change over time

Mature people, aged 20-59 years (and especially those aged 30-44 years), will have the greatest impact on the number of UHNUs.

Married couples are counted as one UHNU, and a divorcee is counted as one UHNU.

Houses or dwellings are built for people to live in. The number of people who need a house is the basic factor affecting the housing market. This research converts housing needs into UHNUs as a measure of absorbtion in the housing supply.

The number of underlying housing need units is a number showing the capacity to buy and to utilise houses as goods. It functions to absorb (i.e. buy) and digest (i.e. live and utilize) housing supply in the market. Thus, it works as a process from (firstly) buying and (later) living in (or moving into) that housing unit. For example, at the time when people buy a housing unit, most have to wait for completion of the building or interior decoration. In many cases, when people buy housing for investment purposes occupancy is similarly delayed. Because they do not intend to buy housing to live in they may leave those houses vacant. Such vacancies are represented as houses for rent or for sale. In this case, it is assumed that the number of vacant houses in the market is equal to the number for investment (or speculation) of housing demands. This leads to the following equation.

5-190

NVH = NHS NHN Where, NVH NHS NHN = number of vacant house = number of housing sales = number of housing needs

Equation 5-1

This equation is used to demonstrate and verify the comparison between the simulated housing vacancy line and historical housing vacancy line (see Figure 514).

Another point of verification is the relationship to housing supply. In the free market housing is not a product which is controlled by government. Housing companies or developers undertake housing projects or supply housing to the market as they choose. They are driven by profit motives. Therefore, in this model, it is that the housing supply number is set on the basis that as housing sales rise, they stimulate housing supply to the same degree. The equation is therefore as follows.

NHSP = NHS * C of NHS on NHSP

Equation 5-2

Where, NHSP NHS C of NHS on NHSP = number of housing supply = number of housing sales = co efficient of number of housing sales on Number of housing supply

5-191

This equation will be used to verify the correctness of this model (see Figure 513: the comparison between simulated housing supply line and historical housing supply line).

Validation of the Model

The concept behind this model is that, by imitating the real system via application of logical principles, it will be possible to support effective policy-making for and regulation of - the housing market. This research seeks to imitate housing demand in the Bangkok Metropolitan Region as a prototype for modelling housing demand. The precision of the model depends on the adequacy of the data fed into it. Given limitations of the available data, many data sources are used in order to produce a reasonable and logical simulation and credible results. The behaviour reproduction test method was chosen for this work. This is because data for housing demand is not counted in real time. The important indices relating to housing demand (such as housing needs, housing sales, housing supply and housing vacancy) are used in comparison with results from model simulation. Comparisons between historical reality and simulation runs are then used to verify the model. In order to verify model accuracy it is necessary to compare historical and simulated data. Modelling, beginning with data input, trend equation selection, factors setting in the model, testing and calibration, follows a logical process, using the evidence and the rationale for data inclusion.. System Dynamics, statistics and mathematical techniques are used in this research to identify and

5-192

improve the process and the experiment to form rigid reasoning to explain the housing market system and its dynamics. The model can be validated as follows. x Verification of the logical assumptions which create the equations used in the model x Verification of the logical assumptions which create the graphical functions used in the model x x Verififying and cross-checking results from model simulation Verifying and comparing the similarity between historical data and model simulation

Similarity of historical data compared with simulation result may not always be the arbiter of model, however, simulation runs of this model have shown satisfactory correlation. This is because even if the model is focused on modifying the simulation line for the purpose of similarity between historical data and simulated data only, the result may not make sense. Rather, the intention of this research is to provide a logical explanation of housing factors as elements in a dynamic flow system. The balance of calibration of the model then emphasizes on both process and result.

This is because all factors influencing housing demand in the real context of a housing market can be divided into two types. Uncontrolled factors emerging from the demographic social factors are not necessarily governed by the housing market. There are, however, controlled factors, such as those economic factors influenced by decision-making in the housing-policy sector. Government sector

5-193

housing policy (largely economic) can shape housing markets. Both factor groupings work in tandem to produce housing demand.

This research draws on the body of knowledge about the housing system which shows that there are many factors that potentially alter results generated by this model when comparing those results to historical data. These minor factors do shape housing demand, but it is not possible to capture them in quantitative data collection. They include individual preference, which affects the delay of purchasing power in the real market (i.e. housing style, type and location).

Description of graphical function

Graphical function is a powerful means of dealing with non-linear relationships between factors. Inputs represent every type of relationship from the first point to the end, with an unlimited range. Determining the relation between factors in the model requires that the selected graph pattern in the graphical function is supported by the literature review, historical data and logical principles concluded by the researcher. In this research, four graphical functions are arrived at from the three economic factors previously identified: GDP growth index, housing loans and housing price index. These factors are the main economic factors that control capacity of people in a housing market to finance a house purchase.

This model uses the GDP growth index as a measure of the population wealth in general. This index is used in two graphical functions. Firstly is to determine the coefficient that relates the number of housing needs to that of housing demand.

5-194

Secondly to determine the coefficient that converts the numerical value of housing demand into housing sales.

Housing loans and the housing price index are represented as a fraction, with housing loan over housing price index. This fraction converts the number of housing demand to a number representing housing sales. This fraction, with the GDP growth index fraction, impacts on the consumption decision. This is because a housing loan increases the capacity of people to buy housing. An increase in the housing price index decreases the capacity of people to buy housing. Both factors therefore work together to grow housing sales.

Another factor used is the number of housing sales which simulate the actual number of housing supply units. The comparison of both numbers is plotted as a test point of the model accuracy. The assumption for this relationship is made as more housing sales lead to an increase in housing supply. Housing sales statistical data are used to achieve a coefficient of housing sales over housing supply. These graphical functions are explained in detail in following sections of this chapter.

5-195

Graphical function of the model between GDP growth index and coefficient of GDP growth index on Affordability & Investment

Source: simulation program

Figure 5-3 : The function between GDP growth index and coefficient of GDP growth index on Affordability & Investment

Figure 5-3 shows that, when the GDP growth index rises, the conversion capacity of housing need units to housing demand units also rises. This does not mean that all of housing needs units will convert to housing demand units when the GDP growth index rises individual circumstances vary markedly. The potential for conversion is, however,,increased. When the GDP growth index rises more than 3 points in a year, the value of the coefficient is more than one. This does not mean that, when the GDP growth index rises 3 points, each housing needs unit will be able to buy more than one house. It means some people who do not need housing still have enough money to become housing demand units. In this research, this group is included in the number of housing demands as investors. As a result,

5-196

GDP growth index is a factor that boosts housing demand from non-housing needs units to become housing demand units for investment or speculation.

5-197

Graphical function of the model between GDP growth index and coefficient of GDP growth index on consumption decision

Source: simulation program

Figure 5-4 : The function between GDP growth index and coefficient of GDP growth index on consumption decision

Figure 5-4 is derived from an historical data base. The graph shows that when the GDP growth index is below six points per year, there is a limited impact on willingness to buy housing. Housing purchases are stimulated when the GDP growth index rises from 6 to 14 points but slow growth rates occur after that. If the GDP growth index rises 8 to 10 points a year it reflects very good economic conditions. From this graph, when the GDP growth index is 13 points per year, the coefficient is equal to one. This means that the number of people deciding to buy houses (in the model) is influenced only by housing loan availability and housing price index. Another way to explain this is that all of the housing demand numbers in that particular time (i.e. when the coefficient of GDP on consumption decision is equal to one) will depend on the fraction of housing loan over housing price

5-198

index. This scenario happens only in the process of converting the housing demand to the housing sales.

5-199

Graphical function of the model between housing loan per housing price index and coefficient of housing loan per housing price index on consumption decision

Source: simulation program

Figure 5-5 : The function between fraction of housing loan over housing price index and coefficient of the fraction of housing loan per housing price index on consumption decision

Figure 5-5 shows the impact of housing loans on housing affordability, and housing price index on housing affordability., Both factors, however, work in different ways. When the number of housing loans rises, then housing consumption capacity increases. When it drops, the number of houses purchased decreases. When the housing price index rises, consumption drops. When the housing price index drops, house purchases increase.

5-200

Therefore, given this reasoning this factor has been set as a fraction of housing loans over the housing price index. If the housing loan and the housing price index increases or decreases at the same rate, the ratio of this fraction would be constant.

Graphical function of the model between housing sales data and coefficient of housing sales data on housing supply

Source: simulation program

Figure 5-6 : The function between housing sales data and fraction of housing sales data over housing supply

This factor determines the impact on housing supply. The assumption is that when the number of housing sales increases, entrepreneurs are stimulated to increase supply. Furthermore, when the number of housing sales drops, the impact is to deaden the housing supply number. Calibration of output against historical data shows a sharp curve at 100,000 units of housing sales (see Figure 5-6). This means that when the number of housing sales in the BMR reaches 100,000 units per year, the coefficient increases dramatically. This was observed

5-201

in the past when housing supply increased dramatically after housing sales reached 100,000 units per year. Data collated in 2006 shows that the normal situation has been that an average of 60,000 new housing units were sold annually in the BMR, below the boom levels triggered by that 100,000 figure.

5-202

Benefit of model

There is strong evidence that the three main groups of factors studied in this research impact on the housing market system in the BMR. Demographic and social factors sequentially produce housing needs units, converted in this model to Underlying Housing Needs Units (UHNU). Economic factors also influence housing demand. Those factors expand or reduce, delay or bring forward the ability of people as a household unit to afford a house. Understanding these mechanisms and the processes by which they affect the housing environment helps to define ways to manage the housing market effectively and to prevent the collapse of housing markets due to oversupply.

Housing is a special good, unlike other goods. The special features include spatial fixity, durability, heterogeneity and expense. These features, particular to housing, increase the difficulty of decision making for an entrepreneur looking to investment. This knowledge also benefits investors by helping to minimise risk associated with housing project investment.

Finally, by monitoring the underlying-housing-needs units, government can use economic factors as a tool for controlling housing markets. Housing loans are the main key by which government can control, through housing (home) loan interest rates, the conversion of housing demand units into housing sales. Housing supply and house prices are controlled by market mechanisms. From a social perspective, people should have and expect access to adequate housing and housing security these are basic needs. From an economic perspective, the advantage of suitable controls linking housing demand and housing supply

5-203

benefits the related industries. These industries include building supplies, labour, and financial service providers. Balanced responses to the management of housing supply and housing demand mean long-run efficiency, stability and sustainability in the housing market.

5-204

Comparing historical data to the simulation run

Trend results which can be compared with historical data were achieved through simulation runs. There are five points of comparison used in this study. The purpose of these comparisons is to check the models accuracy. All simulation lines move in a similar way to the historical data. The numerical differences between the simulation data and historical data are acceptable. Comparisons between simulated data and historical data of all the graphs show similar movement for most of the time. The comparisons are as follows: x Comparison between simulation of the housing needs number and the historical housing sales number (Figure 5-7) x Comparison between simulation of the housing demand number and the historical housing sales number (Figure 5-10) x Comparison between simulation of the housing sales number and the historical housing sales number (Figure 5-12) x Comparison between simulation of the housing supply number and the historical housing supply number (Figure 5-13) x Comparison between simulation of the housing vacancy number and the historical housing vacancy number (Figure 5-14)

5-205

Comparison between simulations of housing needs line and historical housing sales line

Source: simulation program

Figure 5-7 : Comparison between simulations of housing needs line and historical housing sales line

Constant increases in both mature population and total population create new housing needs which grew from 57,528 units in early 1977 to 113,988 units in 2007. Housing sales data exhibit falls and rises either side of the line depicting underlying housing needs (see Figure 5-7 above). The surpluses and deficits of the oscillation grew from 1977 to 2007. This growth in housing sales data correlates with an increase in underlying housing needs.

5-206

Principles deduced from simulation result

The interplay of time and economic factors eventually converts underlying housing needs into housing demand and then to housing sales. Housing needs units do not convert immediately into housing sales. The speed with which housing needs convert to housing sales is controlled by economic factors. The linear trend lines for both housing needs units and housing sales data bear in the same direction. They are quite similar in accumulated quantity (see Figure 5-8 and Figure 5-9).

y = 936.66x + 62202 Housing sale data and linear trend 2 R = 0.0334 compare with underlying housing needs number 200,000 180,000 160,000 140,000 housing unit 120,000 100,000 80,000 60,000 40,000 20,000 0 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year Housing sale Linear (Housing sale) Simulated housing needs number

Source: Excel program

Figure 5-8 : Housing sales data and linear trend compare with underlying housing needs number

Whenever the housing sales line deviates (over or under) significantly from the underlying housing needs line, or whenever the accumulated number of housing sales exceeds housing need through market mechanisms, the accumulated surplus of housing sales over housing needs translates to housing vacancies. Then there is

5-207

an accumulation of vacant housing. A large number of housing needs have been supplied, and thus there are few housing needs remaining in the market to be fulfilled. This means that houses do not sell well, and a dramatic drop in housing sales will be observed; sales fall beneath the housing needs line and remain there until housing needs accumulate, leading to the next cycle.

Figure 5-9 (below) shows the historic surpluses and deficits in this cyclical pattern. This is the key finding which had highlighted for -as appeared similarity on- Figure 7-4.

Source: simulation program

Figure 5-9 : Comparison between simulations of housing needs line and historical housing sales line with overlap area and underlap area

5-208

Table 5-1 : The comparison surplus and deficit between housing needs and housing sales data

year
1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

* housing sales data

63,163 48,119 63,775 63,652 45,892 41,881 26,297 50,638 50,086 33,722 53,353 67,451 80,031 102,335 129,688 108,001 134,086 171,254 172,419 176,616 148,631 67,054 29,029 33,074 33,980 37,182 56,085 69,050 70,072 75,000

^ housing needs
43,258 44,849 46,412 47,998 49,627 51,306 53,042 54,835 56,689 58,605 60,586 62,634 64,751 66,940 69,203 71,542 73,960 76,460 79,045 78,663 77,267 77,381 78,187 79,407 80,912 82,638 84,545 86,609 88,815 91,153

# surplus or deficit
19,905 3,270 17,363 15,654 -3,735 -9,425 -26,745 -4,197 -6,603 -24,883 -7,233 4,817 15,280 35,395 60,485 36,459 60,126 94,794 93,374 97,953 71,364 -10,327 -49,158 -46,333 -46,932 -45,456 -28,460 -17,559 -18,743 -16,153

# total of surplus and deficit

56,192

-82,820

570,047

-279,120

total

2,301,616

2,037,317

264,299

264,299

Source: * historical data, ^simulated data, # calculation

Table 5-1 shows that the difference between the accumulated surplus (or the accumulated deficit) of the underlying-housing-needs line and the housing-sales line is quite similar to the accumulated housing-vacancy number. Housing sales

5-209

exceeded housing needs by 264,299 units, yet over the same period the number of accumulated housing vacancy in the Bangkok Metropolitan Region was 352,645 units (the most recent data in 2003). The equation for this is shown below (Equation 5-3).

NHV = NHS NHN

Equation 5-3

Where, NHV NHS NHN = number of housing vacancy = number of housing sales = number of housing needs

In reality, the number of housing sales is controlled by economic factors which are not constant when the economic situation is good, housing sales, swinged upwards as the capacity to finance a purchase, increase across the population.

Accordingly, whenever housing sales rise above actual housing needs, it is predicted that sales will soon fall below housing needs units to regain balance balance. This balance between the number of accumulated housing needs and the number of accumulated housing sales was discussed earlier.

5-210

Comparison between simulations of housing demand line and historical housing sales line

Source: simulation program

Figure 5-10 : Comparison between simulation of housing demand line and historical housing sales line

The assumption here is that increases or decreases in the GDP growth index influence the creation of housing demand with an average distribution effect over time. When the GDP growth index increases the effect is distributed equally to each housing demand unit.

The reason for this assumption is that the GDP growth index reports the rate of economic growth which reflects the productivity or the wealth of the total population. The GDP growth index also reflects income which represents the capacity of people to finance a house purchase. The graphical function defining the relationship between the GDP growth index and housing demand as its coefficient value is non-linear.

5-211

Although the housing demand line is not the same as the housing sales line, it is highly likely to have similarities in term of number and trend (see figure 5-10). This is because of the close relationship between housing demand and the capacity to finance the purchase of a house. When people buy, their housing demand units are converted to housing sales.

The number of housing sales should not exceed the number of housing demand units at any time because the number of housing sales is derived from the number of housing demand units. In reality, however, a conversion delay can occur which leads to housing sales exceeding housing demand units. This situation occurred in 1989 in the model (see Figure 5-11).

Source: simulation program

Figure 5-11 : Comparison between number of simulated housing demand line and simulated housing sales line

5-212

Comparison between the simulated and historic housing sales lines

Source: simulation program

Figure 5-12 : Comparison between the simulated line and historic housing sales lines

The simulated housing demand line is converted to housing sales by the influence of housing loan availability and housing price mechanisms. From model testing it has been shown that when the housing price index rises the number of decisions to buy falls and housing loans therefore impact differently. When the number of housing loans grows, the capacity to buy will increase. Therefore, housing loans and housing price are the main economic factors converting housing demand units into housing sales. The historic housing sales data is graphed against the simulated housing sales to check accuracy. The simulated housing sales line mostly appears under the housing demand line, and the oscillations of simulated housing sales line are similar to the line of real housing sales data (see figure 512).

5-213

Comparison between simulations of housing supply line and historical housing supply line

Source: simulation program

Figure 5-13 : the comparison between simulation of housing supply line and historical housing supply line

From the simulated housing sales data it appears that there is a relationship between housing sales and housing supply in the housing market. The assumption is that as more housing is sold, more housing will be produced (supply). This is because sales success translates to good profits. In reality, entrepreneurs cannot accurately predict the number of housing demand units in the marketplace. Their only guide to the condition of the housing market is housing sales data. Therefore, the simulated housing supply is the line produced by the influence of the simulated housing sales data.

The simulated-housing-supply line oscillates with the real-housing-supply data but with a minor lag (Figure 5-13). During the economic crisis between 1996 and

5-214

1997- all project owners immediately stopped their projects. A drop in simulated housing supply causes a reaction in the simulated housing sales number, albeit slower than the decrease in investment by entrepreneurs in the real market.

Comparison between simulations of housing vacancy line and historical housing vacancy line

Source: simulation program

Figure 5-14 : comparison between simulation of housing vacancy line and historical housing vacancy line

Another comparative check to validate the model was to compare the number of vacant houses in the market and that suggested in the output. There are data for vacant houses in the BMR from 1996 to 2007 (Figure 5-14). Evidence needed to prove the assumption is that vacant houses represent the surplus of housing sales over the number of housing needs. Some people may buy housing for investment purposes, and some may buy a house before they need to live in it. Vacant house data in the BMR is available from 1996, based on a survey of the number of houses which are not using electricity.

5-215

The number of simulated vacant houses is the result of the number of simulated housing sales minus the number of simulated housing needs (as shown in equation 5-3). The number of simulated vacant houses is very similar to the number of real vacant houses in terms of graph curves and quantity. This may be because, in reality, some vacant houses are absorbed by the rental market. From 1997, when the simulated housing sales dropped below the number of housing needs (see Figure 5-9), it appears that the number of accumulated simulated vacant houses dropped. This number (i.e. vacant house) is, however, set as a one-way flow only. This is because the number of housing needs cannot absorb or replace the number of vacant houses due to the fact that owners of those vacant houses did not sell them to others. They retained ownership. When the result from equation 5-3 is a minus number, it indicates that housing sales are less than housing needs. Therefore all housing sales would be absorbed by housing needs. This explains why thehousing vacancy appears as zero, or it would be considered as zero value at various times.

Chapter summary

All of the simulation results and their explanations entered into in this chapter were used to validate the model. Through experiment some important principles were derived from the simulated results. These underlying principles will be further discussed, leading to the conclusions presented in Chapter 7.

6-216

Chapter 6

FORECASTING WITH THE MODEL

Introduction

After tuning the System Dynamics model until it provided results correlating to historical data, it can be claimed that the model simulates real housing market behaviour. The precision of forecasting is, however, not the main purpose of this research, rather it was the projected results from the factors calculated from the trend equations (see Chapter 4). The researchs forecasting adopted in two scenarios:
x The first scenario where economic conditions mean that the trend lines of

all factors, including GDP growth index, housing loan and housing price, are are following the trends assumed in the model.
x The second scenario is where the economic factors, including GDP

growth index, housing loan and housing price, are assumed to repeat the cycle observed over the last ten year period (1998-2007).

Demographic factors projection

As earlier explained, four main demographic factors are used in this model. They include the total population - expressed in adolescent, mature and elder groupings. The number in ach group is projected by using regression equations. The detail of each factor is provided below.

6-217

Population in the Bangkok Metropolitan Region

Source: Excel program

Figure 6-1 : Population in the Bangkok Metropolitan Region

Historical population data for the Bangkok Metropolitan Region (BMR) are available from 1990 to 2006. Projections to 2017 have been made using an exponential equation which provides the best coefficient of determination value (R2) at = 0.9295 (see Figure 6-1).

The population trend for the Bangkok Metropolitan Region in the next decade is for a slow increase from about 10 million people in 2007 to 14 million people. However, not all age groups will increase in numbers. Details for each cohort are depicted below.

6-218

Adolescents in the Bangkok Metropolitan Region

Adolescent proportion
percentage of adolescent in BMR 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00

y = -0.5676x + 30.356 R2 = 0.9998

Figure 6-2 : Projected Adolescent data in the Bangkok Metropolitan Region

The data for the Bangkok Metropolitan Region adolescent group (0-19 years) is available only from 2001 to 2003. Projections are made up to 2017 by using a linear equation which provides the best coefficient of determination value (R2) at = 0.9998 (see Figure 6-2).

The trend for the adolescent proportion of the total BMR population is for a slight drop from 30 % in 2001 to 25 % in 2010, and then a further decrease of about 20 % in 2017. The adolescent trend follows recent social phenomena in Thailand where the number of marriages and births is decreasing while the number of divorces is increasing.

20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17

year(set 2001 as a 1ST year) Adolescent proportion Linear (Adolescent proportion)

Source: Excel program

6-219

Mature People in the Bangkok Metropolitan Region projection

Mature proportion
64 percentage of mature in BMR 62 60 58 56 54 52

y = 0.3981x + 56.062 R2 = 0.9989

Figure 6-3 : Mature People in the Bangkok Metropolitan Region

Data for mature people (20-59 years) in the Bangkok Metropolitan Region is available from 2001 to 2003. Projections are done up to 2017 by using a linear equation which provides the best coefficient of determination value (R2) at = 0.9989 (see Figure 6-3).

The trend for the mature proportion of the total BMR population is increasing slowly from 56 % in 2001 to reach 60% in 2010. The trend line still grows constantly to reach approximately 63 % in 2017.

20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17

year(set 2001 as a 1ST year) Mature proportion Linear (Mature proportion)

Source: Excel program

6-220

Elderly Popele in the Bangkok Metropolitan Region projection

Elder proportion
14 percentage of elder in BMR 12 10 8 6 4 2 0

y = 0.2547x + 8.7703 R2 = 0.9905

Figure 6-4 : Elderly People in the Bangkok Metropolitan Region projection

Data in respect of older people (60 years or more) in the Bangkok Metropolitan Region is available from 2001 to 2003. Projections are done up to 2017 by using a linear equation which provides the best coefficient of determination value (R2) at = 0.9905 (see Figure 6-4).

The trend for the older proportion of the total BMR population is also increasing slightly from 9% in 2001 to 11% in 2011, and then rising up to 13 % in 2017. In this model, one section of this cohort aged 60-69 years had an evident positive effect on both housing needs and housing demand. This may be because, after turning 60 years of age, people may be likely to buy a new house for their

20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17

year(set 2001 as a 1ST year) Elder proportion Linear (Elder proportion)

Source: Excel program

6-221

retirement (see Table 4-5: Number of single head household of population in municipal area in Thailand in 2005).

Social factor projections

There are three main social factors used in this model. These factors include the formation (splitting) of new household, marriage and divorce. Each factor is projected by using regression equations. The details for each factor are provided below (from Table 6-1 to Table 6-3, Equation 6-1 to Equation 6-2 and from Figure 6-5 to Figure 6-6).

6-222

Table 6-1 : Single-head household formation in Thailand, 2005

*
Group

* Total population in municipal area (thousand)

* Number of total household head (thousand)

* Married household head (thousand)

^ Single household head (thousand)

^ Number of change per group (thousand)

^ Number of change per each cohort group (thousand)

^ ratio of splitting of single head household per total population (percent)

of ages

Total 0-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-69 70-79 80-89

19,376.7 4,021.5 1,423.6 1,640.4 1,856.2 1,861.7 1,760.6 1,606.9 1,410.0 1,092.8 807.8 1,124.6 591.8 152.4

5,288.8 5.6 55.8 217.5 404.9 558.0 629.6 667.5 657.8 561.9 445.6 662.2 337.2 76.7

3,338.9 1.9 34.7 174.5 342.2 453.9 486.7 450.0 427.1 330.8 235.7 280.0 109.0 11.3

1,949.8 3.7 21.1 43.1 62.7 104.1 142.9 217.5 230.7 231.1 210.0 382.2 228.1 65.4 3.7 17.4 22.0 19.6 41.4 38.8 74.6 13.2 0.4 -21.1 172.3 -154.1 -162.7 0.26 3.48 4.39 3.92 8.28 7.76 14.92 2.64 0.08 -4.22 17.23 -30.81 -32.54 0.01 0.24 0.27 0.21 0.44 0.44 0.93 0.19 0.01 -0.52 1.53 -5.21 -21.35

90

26.2

8.5

1.0

7.4

-58.0

-11.60

-44.20

Source: The National Statistical Office (* Data from historical data, ^ Data from calculation)

Data for the single-head household formation rate in the Bangkok Metropolitan Region in 2005 (see Table 6-1) was calculated from the number of each cohort

6-223

group (i.e. adolescent, mature and elderly), multiplied by the ratio of single people becoming a household head. The equation for the number feeding to the stock of housing needs is shown below.

(Adolescent from 0 to 19 from equation trend*0.25/100)/19 + (Mature from 20 to 59 from equation trend*2.49/100)/40 + (Elder from 60 to over 60 from equation trend*1.53/100)/40
Equation 6-1

Where, Adolescent from 0 to 19 from equation trend 0.25/100 (i.e. equation 4-3)

= ratio of single head households formed (split) per total population of adolescents

19

= number of age range from 1-19

Where, Mature from 20 to 59 from equation trend 2.49/100 (i.e. equation 4-4)

= ratio of single head households formed (split) per total mature population

40

= number of age range from 20-59

And where, Elder from 60 to over 60 from equation trend (i.e. equation 4-5) 1.53/100 = ratio of single head households formed (split) per total elderly population

6-224

40

= number of age range from 60-99

The data of the formation of single-head household are not available. This research therefore uses the data for the number of households in each age group (see Table 6-1) to calculate the number of the single-head households formed. The equation is shown below.

SHH =

' (NHH - NMHH) NRAG

Equation 6-2

Where, SHH NHH = percentage of the single-head households formed for each age group = number of total household head

NMHH = number of married household heads NRAG = number of range in each age group

6-225

Table 6-2 : Percentage of single-head households formed for each age group

Group of age Total 0 14 Adolescent 15 19 20 24 25 29 30 34 Mature 35 39 40 44 45 49 50 54 55 59 60 69 70 79 Elder 80 89 90 and over

Single head household (household) 1,949,800 3,700 21,100 43,100 62,700 104,100 142,900 217,500 230,700 231,100 210,000 382,200 228,100 65,400 7,400

Households formed (percent) 2.77 0.01 0.28 0.27 0.21 0.41 0.47 0.93 0.19 0.01 Negative value 1.53 Negative value Negative value Negative value

Total of each group (percent) 4.27

0.25

2.49

1.53

Source: The National Statistical Office (* Data from historical data, ^ Data from calculation)

6-226

% of population 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00

Marriage proportion in BMR

y = 1.68x -0.1486 R2 = 0.6215

19 98

19 96

20 04

20 02

20 06

20 10

20 12

20 00

20 08

year set 1996 as first year mariage proportion of mature population in BMR Power (mariage proportion of mature population in BMR)

Source: Excel program

Figure 6-5 : Projected Marriage Rates in the Bangkok Metropolitan Region

Bangkok Metropolitan Region marriage rate data are available only from 1996 to 1999. These data were used to project to 2017 using a power equation which provides the best coefficient of determination value (R2) at = 0.6215 (see Figure 6-5).

Data on marriage rates are available as the percentage of Bangkok Metropolitan Region population that marries each year. The trend for marriage is decreasing; in 1996, about 1.65% of the Bangkok Metropolitan Region population married. This dropped to 1.40% in 1999. Marriages are predicted to decrease to 1.2% in 2007 and 1.1% in 2017, respectively.

20 14

20 16

6-227

Divorce ratio of marriage People in BMR y = 23.031x0.2263

% of m arriage 50.00 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00
19 98 20 02 19 96 20 06 20 00 20 12 20 14 20 10 20 04 20 08 20 16

R2 = 0.6855

year set 1996 as first year Divorce ratio of marriage Power (Divorce ratio of marriage)

Source: Excel program

Figure 6-6 : Projected Divorce Rates in Thailand

Bangkok Metropolitan Region divorce rate data are also available only from 1996 to 1999. These data are also projected up to 2017 by using a power equation which provides the best coefficient of determination value (R2) at = 0.6855 (see Figure 6-6).

Data on divorce rates in the Bangkok Metropolitan Region is presented in the form of fraction with the number of divorces over the number of marriages , calculated as a percentage per year. The data shows that in 1996 about 22 % of marriages ended in divorce, but by 1999 it was higher at about 29 %. The trend line prediction suggests that this percentage will grow to about 40% in 2007 and about 45% in 2017. It is known However that many societies produce separated couples without them being divorced. People who divorce do not necessarily create the need for two

6-228

houses. This phenomenon is noted but it is difficult to capture a precise number and include it in the model. The data collected for divorce rates in this table are the only evidence to show change in the actual divorce rate.

Table 6-3 : Marriages and divorces in Thailand

Year 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Marriages 484,569 435,425 470,751 436,831 396,928 324,262 354,198 337,140 324,661 291,734 328,356 365,721 345,234

Divorces 46,953 46,903 53,560 56,718 62,379 67,551 61,377 70,882 76,037 77,735 80,886 86,982 90,688

% of marriages ending in divorce 9.69 10.77 11.38 12.98 15.72 20.83 17.33 21.02 23.42 26.65 24.63 23.78 26.27

Source: The National Statistical Office

The historical data on marriage and divorce in the Bangkok Metropolitan Region is available from 1996 to 1999 (see Figure 6-6). Table 6-3 confirms the projections for a growing divorce rate in Thai society. The data from this table show a strong trend of an increasing divorce rate: 9.69% in 1993 to 26.27% in 2005.

6-229

Economic factor projections

There are three main economic factors used in this model. These include GDP growth index, housing loans, and the housing price index. Each factor is projected by using regression equations. The detail for each factor is provided below.

GDP growth index in Thailand projection

GDP 15.00 10.00 5.00 0.00 -5.00 -10.00 -15.00

GDP in Thailand

y = -1.0436Ln(x) + 8.4333 R2 = 0.0344

Figure 6-7 : GDP growth index in Thailand projection

The GDP growth index data for Thailand is available from 1980 to 2005. Projections are made up to 2017 by using a logarithmic equation which provides the best coefficient of determination value (R2) at = 0.0344 (see Figure 6-7).

19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 20 06 20 08 20 10 20 12 20 14 20 16

year (set 1980 as 1st year)

GDP data

Log. (GDP data)

Source: Excel program

6-230

The GDP growth index depends on many factors. Given the unpredictability of economic fluctuations, it is impossible to predict values for the GDP growth index. This research adopts the most suitable equation which provides the best coefficient of determination value (R2) to project the trend of the GDP growth index. Reliability of the projections is not the aim of this research, and forecasting can only be based on projected data. If this factor changes in a different way from the trend line projection, the precision of this models forecast would vary.

6-231

Projected Housing Loans in Thailand

Loan (M Bath)

Housing loan in Thailand

y = 39295x 0.6 R2 = 0.5772

350,000 300,000 250,000 200,000 150,000 100,000 50,000 0

19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 20 07 20 09 20 11 20 13 20 15 20 17

year (set 1985 as 1st year) Housing loan Power (Housing loan)

Source: Excel program

Figure 6-8 : Projected Housing Loans in Thailand

The data for housing loans from all banks in Thailand is available from 1985 to 2004. Projections are made up to 2017 by using a power equation which provides the best coefficient of determination value (R2) at = 0.5772 (see Figure 6-8).

Because housing loan data for the Bangkok Metropolitan Region are unavailable, data for housing loans from all banks across Thailand are used in this model. The ratio of housing loans in the Bangkok Metropolitan Region takes the major proportion of the entire housing loan market share in Thailand. Predictions for annual growth in the value of housing loans from are from approximately 280,000 million baht in 2005 to 330,000 million baht in the next decade.

6-232

Projection Housing Price Index in Thailand

percent of index (1986 as 100%)

Housing Price Index in Thailand

180 160 140 120 100 80 60 40 20 0 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

y = 93.081x 0.1362 R2 = 0.9249

2009

2011

2013

2015

year (set 1985 as 1st year)

Housing price index

Power (Housing price index)

Source: Excel program

Figure 6-9 : Projected Housing Price Index in Thailand

Data on the housing price index for Thailand is available from 1977 to 1991 and from 2001 to 2006. Projections are made up to 2017 by using a power equation which provides the best coefficient of determination value (R2) at = 0.9249 (see Figure 6-9).

This housing price index data is used to represent the housing price index for the Bangkok Metropolitan Region. The unit of this data is a percentage. The base year for the housing price index is 1986. The trend for the housing price index shows a slight annual increase, reaching 160% in 2017.

2017

6-233

Results from all data projections were inputted to the System Dynamics model and simulations were run. The results of these simulations are as follows:

Simulation results

The method of projecting trends by using regression equations was adapted to the model. The forecasting results appear as a smooth curve. It is impossible to accurately predict fluctuations in rapidly changing or unstable economic conditions and therefore it is impossible to accurately capture trends for economic factors in such circumstances. To overcome this difficulty two scenarios were developed.

First, it was assumed that in reality the factors would follow the trend lines simulated by the regression equations. Second, that the economic situation would repeat its cycle observed over the last 10 years (from 1998 to 2007). Economic factors that use the same set of data during the last 10 years include the GDP growth index and housing loans. The housing price index was set as a regression trend because it is unlikely to fluctuate. This statement is supported by the evidence which appears in historical data (see Figure 6-9). The housing price index was still rising even during the period of the worst economic crisis.

Model simulation precision was is not the primary purpose of this research and, rather, forecasting trends show a possible result from the model. Reliability of forecasts would be more precise if all economic factors move in the way the equation trends have projected. Five simulation results for this model are

6-234

provided below. Most graphs show both the projected line and the historical housing sales line. Details of the comparisons between them were discussed in the previous chapter.

Projected Housing Needs in the Bangkok Metropolitan Region

Source: simulation program

Figure 6-10 : Projected Housing Needs in the Bangkok Metropolitan Region

Model simulation shows that in the decade 2008-2017, housing needs will increase more sharply than they did during the previous decade 1997-2007. Simulation suggests that in the next two decades housing needs will continue to rise, reaching 150,000 housing units per year (see Figure 6-10). Housing needs projection resulted from the two scenarios developed are the same because the underlying housing needs unit line is dependent on demographic and social factors. Economic factors do not have any influence to the number of underlying housing needs unit in this model. All these economic factors only play an important role in the process which converts housing needs units to housing sales

6-235

units. Thus, the results of two scenarios developed for the economic factors can be explained and compared.

Comparison between Housing Demand Projections in the Bangkok Metropolitan Region

Housing demand projections from the two scenarios are compared at this point. The first scenario developed was that all factors would follow the trend lines simulated from the regression equations. The second scenario developed

assumed that the economic cycle for the last 10 years (from 1997 -2007) would be repeated. The comparative output is explained by reference to figures 6-11 and 6-12, following.

6-236

Source: simulation program

Figure 6-11 : First scenario projection for housing demand in the Bangkok Metropolitan Region (housing sales data)

Model simulation for the first scenario shows that projected housing demand in the Bangkok Metropolitan Region for the decade 2008-2017 will grow from about 125,000 units per year in 2007, to reach 160,000 units per year by 2017 and 200,000 units per year by 2027 (see Figure 6-11).

6-237

Source: simulation program

Figure 6-12 : Second scenario projection for housing demand in the Bangkok Metropolitan Region (housing sales data)

If the economic cycle repeats itself, the second scenario shows that the fluctuation in the housing demand line would repeat its previous pattern, but there would be a magnification in the amplitude of the graphs swing. This is because the number of housing needs, which play an important role as the initialising factor, is increasing along the timeline. Thus, the increased, but fluctuating, trend of upward housing demand would correspondingly follow the increase in the housing needs line (see Figure 6-12).

6-238

Projected Housing Sales in the Bangkok Metropolitan Region

Source: simulation program

Figure 6-13 : First scenario projection: housing sales in the Bangkok Metropolitan Region

Model simulation for the first scenario shows that forecast housing sales will grow slowly from 60,000 units in 2007 to about 75,000 units in 2017 and 90,000 units in 2027 (see Figure 6-13).

6-239

Source: simulation program

Figure 6-14 : Second scenario projection: Housing Sales in the Bangkok Metropolitan Region

The second scenario shows a recurrence of the pattern seen between 1998 and 2007. Movement fluctuates repeating the previous pattern from the first cycle to the second cycle, respectively. Housing demand leads to a greater swing in sales on a 10-year cycle (see Figure 6-14).

6-240

Housing Supply Projections for the Bangkok Metropolitan Region

Source: simulation program

Figure 6-15 : First scenario projection: Housing Supply in the Bangkok Metropolitan Region

The number of new housing supply units forecast under the first scenario is seen to grow slowly from 40,000 units per year to 50,000 units per year in 2007, reaching 60,000 units per year in 2027 (see Figure 6-15).

6-241

Source: simulation program

Figure 6-16 : Second scenario projection: Housing Supply in the Bangkok Metropolitan Region

The second scenario leads to a spike in the forecast housing supply (see Figure 616) (i.e. between 2008 and 2027). Housing sales lead to this supply line, following a similar fluctuating patter in the future.

6-242

Projected House Vacancies in the Bangkok Metropolitan Region

Projected house vacancies are shown in the following figure:

Source: simulation program

Figure 6-17 : First scenario projection: House Vacancies in the Bangkok Metropolitan Region

The projection for vacant houses in the Bangkok Metropolitan Region in the next two decades for the first scenario remains quite constant (see Figure 6-17). This means that the housing needs number and housing sales number arre still

growing at similar rates. The number of vacant houses is derived from the housing needs number, minus that for housing sales.

6-243

Source: simulation program

Figure 6-18 : Second scenario projection: House Vacancies in the Bangkok Metropolitan Region

The forecasted housing vacancy rate (see Figure 6-18) over the next 20 years in this scenario is comparable to the first scenario, with housing vacancies only slightly increased. This is because the number of housing vacancies is the remainder of the number of housing sales minus the number of housing needs. The remainder is similar for the first and the second scenarios, resulting in the same projection for housing vacancies (see Figure 6-18).

Summary

All of the forecasts from simulation runs show that, if all independent factors change in line with the trend equation projections, the housing market will grow slowly with constant growth occurring in the first scenario. In the case of the second scenario, model simulation shows that housing demand, housing sales and housing supply change are similar to the oscillation rhythm demonstrated by

6-244

economic fluctuations. The observed cycle over a 10-year period would be expected to repeat itself.

Only the number of vacant houses in the Bangkok Metropolitan Region would remain the same in both scenarios. The similar numbers of vacant houses in both scenarios leads to a conclusion that, in the real market, other factors play an important role in the fluctuation of the housing demand line. It is clear that the housing demand is mainly influenced by economic factors, but there are some limits to their impact. Initial demographic and social factors constrain the average trend of housing demand in the long run. These constraints control the rate of housing consumption in the market for both scenarios.

Population in the Bangkok Metropolitan Region is increasing constantly. Where there are no strong fluctuations in social and economic factors, then housing needs, housing demand and housing sales would be expected to grow in a constant pattern. The precision of model forecasts is strongly dependent on that the accuracy of factors input to the model. Fluctuations affecting housing demand and supply are always occurring in real circumstances. The method for capturing those factors that influence the fluctuations of all economic data remains complex, and individual forces are not easy to capture. This is because there are other external factors which are not included in this model. These factors are unpredictable and uncontrollable, such as foreign investment, politics, and the global and national economic circumstances. Consequently, the precision of model forecasting is not completely reliable due to the unpredictable nature of the data estimation in the real market.

6-245

7-246

Chapter 7

FINDINGS, CONCLUSIONS AND RECOMMENDATIONS

Introduction

The System Dynamics model developed in this research is a laboratory replica utilising real historical data from the Bangkok Metropolitan Region (BMR). This modelling approach in effect, a computerised laboratory experiment is used to investigate how and why changes occur in housing demand in Bangkoks Metropolitan Region This research has produced a prototype model which attempts to capture the root causes in the housing environment that governs the housing demand in the BMR. It is done with a view to providing accurate prediction of demand. This chapter describes the conclusions and makes recommendations for future research.

It should be noted that there are a number of data limitations which affect the precision of the model simulations and forecasts. A principal limitation impinging on reliability of results is the lack of available data from the BMR. This is a simple reflection on the manner in which data is collected and made available in Thailand. When accurate a data were available, it was shown that the accuracy of model results significantly improved. With improved data collection future prediction will therefore be significantly more accurate.

7-247

Summary of main findings

This research has shown that the main factors influencing housing demand confirm findings in previous research. This research has, however, generated a new concept of how the main factors affecting housing demand work sequentially. Population size was found to play a fundamentally important role in generating the number of housing needs. There is strong evidence that mature people (aged 30-45 years) create the greatest number of housing needs (and also housing demand for housing units). It was also foud that social factors determine the timing of housing needs. Economic factors were also found to play a vital role, establishing the financial conditions which boost or depress the conversion of housing demand into housing sales. When an economy is growing or declining quickly, it has a huge impetus to rapidly move the demand for housing (quantifiable as housing units) and housing sales up or down. In stable economic periods, the movement of housing demand will likely depend more on demographic and social factors. This model developed in this research showed that housing demand can be quantified (both the concept and methodology). The research undertaken began with the differentiation between the number of housing demands and that of housing sales, and also provided further information about the model.

Differentiation between housing demand and housing sales data

This research has shown that the number of houses demanded and housing sales are not fully reconciled with each other. Furthermore, housing sales are not a perfect indicator to represent housing demand data. The model also shows that the actual number of housing sales is the result of housing consumption - the

7-248

decisions made by people - starting from housing demand units. Theoretically, the number of housing sales will always be less than the quantifiable housing demand number. Nevertheless, in this model, the housing demand number is less than the number of housing sales between 1992 and 1998. This period represented a peak in Thai housing markets. Figure 6-11 showed the spiky shape of the housing-sale line at that period. It was also shown that there was strong evidence that investment and speculation was the housing buyers chief purpose at that time.

The overlap of housing sales over housing demand in this period showed that surplus is not a normal phenomenon in housing markets. The GDP growth index showed its maximum fluctuation between 1992 and 1998. It delineated a high point, and also the lowest one, in Thai history - in 1995 a maximum GDP growth index figure of 9 points; and in 1998 the low at minus ten.

Achievement of objectives of this research

The model produced as a result of the research has four objectives, all of which are held to have been achieved: x The first was to systematically model the housing demand environment using three key factors (demographic, social and economic) in order to capture and study individual factors influencing housing demand in the BMR. This objective was achieved by modelling, which was shown to be valid. x The second objective was to verify the models accuracy in forecasting housing demand in the BMR and to analyse scenarios that will improve the body of knowledge in the housing field. This objective was achieved

7-249

by the comparison between simulation data and historical data. The results of these comparisons reconcile with historical data. It was also found from the model simulation runs are described in two points. A suggested principle for forecasting housing demand from this research is described and named the housing-consumption cycle. x The third objective was to determine the major factors affecting housing demand and housing sales. This objective is achieved by the calibration of the model described in Chapter 5. x The last objective was to create a housing demand model showing housing demand for Bangkok Metropolitan Region. This objective is achieved by developing a System Dynamics model shown in figure 4-17. Knowledge from this research can contribute to control and policy making in respect of the BMR housing market.

Conclusions from conceptual settings in the model

This research uses a bottom-up approach, starting at the root of housing need - the need for shelter. The model first considers the role of demographic factors in creating the number which quantifies housing needs. Social factors then determine the timing of these housing needs. Finally, economic factors are considered because of their impact in determining the ability of people to finance a housing purchase. Sequentially, these factors produce housing demand. Factors controlling housing demand can be grouped into two. The first group comprises factors which are outside the control of the housing market. This group includes demographic and social factors, such as the rate of population increase or decrease, the birth, death, migration, marriage and divorce rates, and household formation rates. The

7-250

second group comprises economic factors. Some economic factors are not fully controllable, such as the GDP Growth index. Other factors, including housing loans and housing price, are more easily controlled by government. Thus, in this case, government can increase or decrease housing affordability. This sequential influence of all factors influences housing demand by the concept can be seen in Figure 7-1 as follows:

Figure 7-1 : The Housing Demand Models Flow Pattern

7-251

Housing sales data was found to swing around the line plotting underlying housing needs (previously established in Figure 6-10). This occurs because economic factors are influenced by many events outside the housing market. In periods of economic boom, the valve of housing affordability is opened widely in the model. The majority of housing needs quickly convert into housing demand and then into housing sales. By contrast, in periods of economic downturn, investment in housing projects reduces as the flow of housing demand shrinks and, consequently, the number of housing sales decreases.

In boom times when the number of housing sales is greater than that of housing needs, there are vacant houses in the market. However, in periods of economic decline, if the number of housing sales is less than that of housing needs, then there is a build up of unmet housing needs. This shows that those housing needs units which are building up are the result of declining house affordability. People would then be waiting for the next period of economic boom. This phenomenon can be explained as the spring force theory of housing demand.

Where people cannot afford to buy, renting may reduce the number of vacant houses. This research, however, does not include the rental market, and further research in this area is recommended. Data from 2002 show that the proportion of used houses in the BMR market was greater than that of new houses on the market, but the rate of sale for used houses was small, about 5-10 % of the housing market sector. This is why the number of vacant houses still appears high in both real and simulated data. In a developed city or a city where all the land has been fully developed, the housing market relies on used houses

7-252

(Buranathanung,,2004). In 2007, however,, the housing market in the BMR continued its expansion to the five surrounding provinces. Therefore, it can be expected that the used housing market will grow in the near future.

Conclusions Linking Housing Demand and Housing Sales

In this model the focal point of concern was housing demand and its sequences. Other factors around housing demand were found to provide sequential impact as causal factors with important results. Findings from previous research, such as the study by Pitakteeratum (1994), adopted the number of housing sales to represent housing demand number as a proxy. In this current research it has been found that it is not possible to accurately capture a number representing housing demand by using sales data.

Demand as stated in economics theory is influenced by price and supply. All three factors (i.e. demand, price and supply) react to each other to determine the equilibrium price in a perfect, competitive market. In a practical sense, when referring to housing markets, the perfect competitive market does not exist. The heterogenous features of housing, as a good, lead the market to an imperfect state. In imperfect competitive markets, the market price or price index can be assumed as an average of the equilibrium price. Thus, an assumption is made - to use the housing price index as equilibrium prices for this model - as that the housing price index is the proxy of the equilibrium price. A housing price index is used as one of the economic factors to determine the housing demand in this model.

7-253

There is no easy or accurate method for collecting data on the housing demands by the total population in any particular area. However, in this model, the number of housing demands is not derived from the number of housing sales. Rather, the number of housing sales is the result of the reaction between the accumulated number of housing demand and that of housing supply in the market at a particular time and area. This reaction is influenced by the three economic factors:

x x

GDP growth index which represents the wealth of people housing loan which represents a financial tool to flow or constrict housing affordability

housing price index which represents the limitation against housing affordability

These three economic factors were found to provide the best fit to the historical data from the case-study area in this research (i.e. Bangkok Metropolitan Region).

The difference in results between the simulated housing demand number and the housing sales number in this model is that the number of the simulated housing demands is more than the number of housing sales for most of the time. The model suggests that the number of housing demands runs (by average) about 1040% above the number of housing sales. Only in 1989 was the housing supply a little higher than the number of housing demands (see Figure 7-2). The number of simulated housing demands shows the surplus over the number of simulated housing sales over the experiment period (from 1977 to 2007). In the period of economic crisis in Thailand between 1996 and 2000, both lines dropped sharply

7-254

and seem to converge, but the number of housing demands remained greater than that of housing sales (see Figure 7-2 and also Figure 7-3 below).

Source: simulation program

Figure 7-2 : The comparison between simulated housing demand and simulated housing sales

Source: simulation program

Figure 7-3 : The comparison between housing supply data and simulated housing supply

7-255

Conclusions from the model

Efficient management is a benefit in every respect. Housing, from a broad view, also benefits from efficient management. To achieve efficient management a knowledge of housing demand and supply behaviour is necessary.

System Dynamics was used in this research to model housing demand in the BMR. The basic rationale for using Systems Dynamics in this research was to imitate reality of housing demand behaviour as closely as possible. All factors used in the model were selected logically, using evidence from the body of knowledge surrounding the topic. The precision and reliability of the model was demonstrated from comparisons between model simulation and historical data (or reality).

The final model was created after extensive System Dynamics experimentation. It validated knowledge derived from the literature in respect of housing demand, and the factors influencing it. These factors were found to be demographic, social and economic.

Demographic factors are basic to the creation of housing needs. This model illustrated that the number of housing needs in a particular time and area is generated by the number of people who create new household units. The majority of this group is the group of mature people which can be subdivided into the type of single head household type and the couple head household type.

7-256

Social factors determine periods when households have housing needs. Marriage, divorce and household formation are the social activities which change the status of people from a member of a household to be a head of household. Sometimes people do not marry; but still move out from the family to get a job, study, or privacy. When people move out to form a new household or a new family, this research considers and captures this social activity as the splitting new household. All of these social activities create the number of housing needs. Furthermore, the number of housing needs accumulates, to create the underlying housing needs unit used in this model.

After establishing the housing needs number, it is necessary to consider the capacity of people to afford that housing. The data about underlying housing needs is, however, unable to capture the real situation. Therefore, with the lack of data about affordability, economic data applying to the total population are adopted for calculations in this model. These economic data include the GDP growth index, housing loans and housing price index. The GDP growth index controls the conversion of housing needs numbers into housing demand numbers and then into housing sales. Housing loans and the housing price index work together in the model (as a fraction of housing loan over housing price index) to control the conversion of housing demand into housing sales. All economic factors adopt a graphical function to generate the rate of flow in each unit.

The results of the model simulation correlate with the historical data; all of the factors used in this model appear to impact on housing demand as the models sequence suggests. From a broad view the fluctuation of housing supply over time

7-257

in the BMR shows inefficiency in housing management policy. Housing oversupply occurred in the economic boom period while housing supply shortages occurred in economic recession. Considering housing needs and the housing demand side, it was found that housing needs grow in relation to population influences while housing demand swings according to economic conditions. The full control of economic factors affecting housing seems impossible; however, oscillations in housing demand can be controlled by housing financial policy.

This may occur even though economic theory suggests that demand is a number which results from the interaction between price and quantity of supply in the market. In an unstable competitive market in which price and quantity of supply keep changing this theory is unable to capture the real market demand number.

This research considered housing as both a special commodity and a necessary commodity for people. It is possible that every household could finally occupy a house (housing unit) to live in. Housing is the most expensive commodity that people buy in their life time. It is something that family or household may buy only once or twice in a life time (by average). Considering this fact, the number of households can be counted. Housing demand is therefore the sequence of an economic activity between housing needs and sales

As the underlying housing-needs rise constantly, housing demand could, in theory, be controlled to grow at a similar rate to housing needs. This would result in a reduction in the fluctuation of housing sales. Thus, the housing-supply side is more likely to produce housing to match demand in the right place at the right

7-258

period., It will also support efficient housing policy thereby contributing to sustainability. Whilst booms may cause people to buy for speculation, housing need remains the basic factor which absorbs housing in the market.

Suggested principles

Two principles were found from the experimentation with the model in this research. Firstly, that the capacity of housing consumption in the market is constrained by the underlying housing needs number. The underlying housing needs determine or limit the capacity of housing consumption in the market over time. The underlying housing-need line shows the constraint of the oscillating housing-sale line. This line bisects surplus and deficit as shown in Figure 7-5).

This phenomenon happens in a sequence of logical steps. The evidence suggests that the underlying housing needs number is the number of household units requiring housing units to live in. By itself housing is a commodity that is produced for households to live in. Therefore, the number of households is strongly correlated to the number of housing units.

The number of housing units in a housing market can be divided into occupied housing units and vacant housing units. The occupied housing units are taken and lived in by household units. The vacant housing units are owned by a person who is not living in that housing unit. The number of vacant housing units in this research was determined by those housing units which have zero electricity-use bills.

7-259

The difference of the number of households and the number of existing housing units at any particular time can happen due to delay or haste between supply and demand in the market. The imbalance of consumption and production can occur in housing markets is similar to the principle derived from the Beer Game (Senge, 1994). However from a holistic view point and through the theory of supply and demand, while the housing stock for sale is greater than the underlying housing needs, the delay in housing sales can be a signal for the housing supplier to decrease housing production. Conversely, when the stock of housing for sale is lower than the underlying housing-needs number, the accumulated housing demand would increase. The signal to entrepreneurs (i.e. housing suppliers) would then be through market mechanisms in this case by the increasing price. This would stimulate entrepreneurs to produce and launch more housing into the market.

Because the main purpose of housing is to live in, the underlying housing needs number is a key index reflecting an occupation number for housing units. The equilibrium line of housing sales, compared with the underlying housing-needs line, adopts a mathematical function (i.e. linear trend-line function) for the calculation. The mathematics underlying Appendix D. this calculation are described in

The second principle found from this model is the spring force behaviour (i.e. the delay or haste) of housing consumption. Sequentially from the first principle, when housing clients consume housing units more or less than the underlying housing needs number, the imbalance mechanism from the first affects to housing

7-260

sales (i.e. the housing sale here is considered as the housing consumption proxy). This force would try to pull or push back the line of housing sales number to equal the underlying housing-need line in order to balance housing consumption. The further that housing sales swing from the underlying housing-need the greater the force to equilibrium will pull or push housing sales to balance the underlying housing needs number.

This principle can be compared with the behaviour of a spring when it is pushed or pulled from of its normal position. When it has been pushed, then it has a force to rebound against the force. In the same way, when it is pulled, it has a reactive force against that pulled force. One thing which has to be considered carefully is that the neutral range of the spring is changed over time (equating to the changed housing-need line). Thus, underlying housing-needs determine the real housing needs at any particular time. Accumulated housing needs control consumption capacity of those people seeking to buy houses at the same time.

Ideally, the housing-sale line should not oscillate across the underlying housingneed line, but rather grow constantly and parallel to the underlying housing-need line. This would be the most efficient situation allowing an interaction between demand and supply in the market. The situation can be explained in Figure 7-4 below.

7-261

Housing units
Deficit area

Over-surplus area

Surplus area

Over-deficit area

Underlying housing-need line Housing-sale line Spring box and its forces

Time

Figure 7-4 : The Spring-force Principle

Conclusion

Following Geffner (2008) it was found that housing markets are not all alike, although economists note that they share some basic similar patterns. These principles are reflected in the evidence used in this research model which is based on data for a particular area and time. The swing movement is quite evident in housing sales. The underlying housing-need line presents as a neutral curve. Even though the results of this research suggest that economic factors are the main factors driving housing sales from housing needs through to housing demand a limit on housing sales still exists because of the constraint applied by housing needs.

7-262

Housing is an expensive commodity. It can be bought for investment purposes without a need to live in it. It is more likely, however, to be about one housing unit per household. The number of housing sales would not normally be expected to be greater than the underlying housing needs number. Only once, in 1996, did the historical data show that the number of housing sales was about 2.2 times greater than underlying housing needs. In that year it occurred as a result of Thailands rapid economic growth from 1993 until 1996. Housing sales dropped dramatically after 1996; the ratio between housing sales and underlying housing needs dropped to 0.35 in 1999.

y = 936.66x + 62202 Housing sale data and linear trend 2 R = 0.0334 compare with underlying housing needs number 200,000 180,000 160,000 140,000 housing unit 120,000 100,000 80,000 60,000 40,000 20,000 0 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year Housing sale Linear (Housing sale) Simulated housing needs number

Source: simulation program

Figure 7-5 : Housing sales and underlying housing-needs

Interestingly, the different accumulated numbers between the underlying housingneed line and the linear housing sales line nearly approximates the accumulated number of housing vacancies in the BMR (see Figure 7-5). This indicates that more housing is consumed than the underlying housing needs number, some

7-263

houses in that particular area are likely to remain vacant, as shown in Table 7-1 below.

7-264

Table 7-1 : A comparison between the (equilibrium) linear line and the underlying housing needs number

year 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 total

*Accumulated vacant housing n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a n/a 275,039 336,084 341,925 348,308 347,810 353,800 350,259 352,645 n/a n/a n/a 352,645

^housing needs 43,258 44,849 46,412 47,998 49,627 51,306 53,042 54,835 56,689 58,605 60,586 62,634 64,751 66,940 69,203 71,542 73,960 76,460 79,045 78,663 77,267 77,381 78,187 79,407 80,912 82,638 84,545 86,609 88,815 91,153 2,037,317

Accumulated of simulated housing vacancy 230,334 234,940 240,367 245,925 251,201 255,907 259,824 262,771 264,591 265,142 264,290 264,290

Over- needs (Equilibrium) housing linear line consumption 19,880 19,226 18,600 17,951 17,259 16,516 15,717 14,860 13,943 12,964 11,919 10,808 9,627 8,375 7,049 5,647 4,165 2,601 954 2,272 4,605 5,428 5,558 5,275 4,706 3,917 2,947 1,820 551 -851 264,290 63,139 64,075 65,012 65,949 66,885 67,822 68,759 69,695 70,632 71,569 72,505 73,442 74,379 75,315 76,252 77,189 78,125 79,062 79,999 80,935 81,872 82,809 83,745 84,682 85,619 86,555 87,492 88,428 89,365 90,302 2,301,607

Source:* historical data from Government housing Bank of Thailand, ^ simulation of the model

From the historical data shown here, it can be seen that housing sales fall below the housing-need line, then rise and then fall again. In this 30-year data sequence

7-265

there are three points at which housing sales run across the underlying housingneed line - in 1980, 1987 and 1997. The deficit periods between 1981 and 1987, and between 1997 and 2007, show the rebound from an over-supplied market (see Figure 7-5).

From these phenomena, it can be concluded that housing market behaviour in the BMR is constrained by the underlying housing needs. Moreover, when housing has been supplied or has consumed more than the capacity of underlying housing needs, the result will be a limit to growth of housing sales constrained by the underlying housing needs. Thus housing consumption maintains equilibrium with the accumulated, and underlying, housing needs.

The underlying housing needs set a constraint on housing consumption. As stated in Chapter 6, the historical data provides evidence to support this principle. The rise and fall of housing-sale line, compared to the underlying housing needs, demonstrates the oscillation. Results from simulation runs discussed in Chapter 5 also showed similar performance.

Housing demand forecasting

The above findings lead to the second suggested principle, called the spring force behaviour of housing consumption. The durability of housing drives consumption in two ways; firstly, housing consumption for housing service purposes; and secondly, housing consumption for investment purposes. This leads to the following: x Housing services consumption is limited by underlying housing needs.

7-266

Consequentially, housing investment consumption is limited by housing

service consumption.

The cycle of housing consumption

The cycle of housing consumption can be explained in four stages:

The first stage happens when housing sales are greater than the underlying housing demand. The accumulated number of vacant housing starts increasing.

The second stage is when the accumulated number of vacant housing is significantly increased. At this stage, housing consumption reaches its maturity, and housing sales pass their peak. Housing sales then start to decrease and thus a delay in housing supply and housing production in the market occurs. This delay happens because of the price mechanism.

The third stage happens when housing sales drop lower than the underlying housing-need line. The accumulated number of vacant housing units is consumed due to the shortage of housing supply in the market. At this stage, housing sales reach their bottom point.

The fourth stage happens when the accumulated number of housing vacancies is absorbed by buyers or renters. This consumption can occur until it reaches the point where vacant housing cannot satisfy housing clients. Even though a number of accumulated housing vacancies will still remain in the market, the shortage of supply will stimulate prices, which will go up. The indication of the high-price housing at this stage will stimulate the entrepreneur to produce and launch new housing product

7-267

onto the market. As a consequence, housing sales start growing until they reach the underlying housing-need line.

However, the fourth stage of the housing-consumption cycle will not stop at the point of the underlying housing-need line. This happens because the momentum of the price mechanism takes the stage beyond the housing-need line, and hence the housing-consumption cycle starts again.

When this consumption behaviour happens as a full cycle, the housing-sale line would be an explosive oscillation when the underlying housing needs increase. On the other hand, when the underlying housing-need line decreases, the oscillation of housing-sale line would reduce in magnitude and gradually approach the underlying housing-need line.

If the housing-consumption cycle happens in the stated sequences forecasting of housing sales, using the historical trend based on the non-cycle principle, may not be reliable. However, to prove the correctness of this housing-consumption cycle, one full cycle of housing-consumption experimentation is necessary. This would require a lengthy longitudinal study because a full cycle of housing consumption will take decades to complete. Multiple studies, using repetitive tests to prove its robustness, would be required. This may serve as a stimulus, challenging research fellows to study the housing field.

7-268

Research contribution

This research introduces principles found from the housing demand modelling experiment. From the assumption that a number representing housing demand exists and that it changes over time - and at any particular time - the housing demand number will vary, depending on specified influencing factors. This principle can also be applied for modelling and simulating the demand in many areas, such as particular foods or goods.

Such modelling relies on identification of the influencing factors. For example, to know the demand for food in a particular area, the model can calculate the numerical value (i.e. quantity) of that demand, using the same principle deployed in this research, starting with the number of people identifying the rate of food consumption each day for each cohort, and so on.

Economics is a discipline concerned with managing and allocating the worlds limited resources to suit the unlimited needs of humans. If there was a perfect match between demand and supply, it would lead to efficient resource allocation, supporting the concept of sustainability. The principle behind this research can be applied to other policies concerned with resource allocation.

Setting sequences to mimic behaviour remained an important concern. The system within the model should cover all factors involved, or at least those critical factors which influence the target variable. It is also important to compare simulation results with historical data to ensure model accuracy.

7-269

The graphical function from System Dynamics simulation software is also crucial. In the case of non-linear relationships between factors, the graphical function can be adopted. To set the graphical function, the minimum and maximum ratio of the factors would be expanded until they cover the probable range of all factors and simulation results.

Future research

Future research recommendations are as follows. 1. When data on housing demolition is available in the Bangkok Metropolitan Region case study area , the housing system in the BMR should be reviewed, using this data to add to the model in order to better reflect factors that influence housing demand. 2. The rental market can be relevant in some respects and could be added to factors in the housing demand model. This addition would reflect choices between buying and renting housing and therefore better reflect consumption behaviour. 3. In the near future, the used housing market is most likely to grow bigger than the new housing market. If that occurs, the overall BMR housing market will be shaped by the used housing market. Housing system scenarios should then be redefined in the simulated System Dynamics modelling method. 4. A further development of the housing demand model would be to review each type of housing demand separately. To identify factors

7-270

that influence choices for each housing type, the researcher would, however, need to carefully define the factors influencing housing consumers in terms of housing types, preferences and constraints. This would require significant qualitative research using primary data. 5. The principle of modelling from the bottom up can be applied in many fields. For research on any topic using System Dynamics model, the map of the model is dependent on the conceptualisation. Thus, every field can apply this method to study how some things change and affect others. 6. Distribution of housing ownership is an important factor that can shape housing demand. If these data become available, the housing demand model can be sharpened and made more accurate. 7. Housing supply is an interactive factor with housing demand. Models depicting housing supply and demand relationships with feed back loops are suggested for such further research. 8. If the economic life cycle can be captured, other economic factors useful for forecasting housing demand can be built in, thus making the model more precise.

271

References

<http://www.visit-thailand.info/information/map-of-thailandregions.htm>14/02/2007. ABS. (2007). "Measuring wellbeing: frameworks for Australian social statistics."fromhttp://www.abs.gov.au/Ausstats/abs@nsf/66f306f503e529a5ca2 5697e0017661f/1598003. Anonymous. (2008). How did we do the forecasts? What are the components of population change?, Adelaide City council. Aluvaala, A. I. (1993). "Problems Involved in estimating demand in housing Market." Construction review July1993: 8-9. Anikeeff, M. A. (1999). "Estimating the demand for senior housing and home health care." Journal of Real Estate Portfolio Management 5(3): 247. Anily, S., J. Hornik and M. Israeli (1999). "Inferring the distribution of households' duration of residence from data on current residence time." Journal of Business & Economic Statistics 17(3): 373. Apps, P. (1971). A residential model 3 demand equations. Cambridge, center for land use and built form studies in the university of Cambridge. Arimah, B. C. (1992). "An empirical analysis of the demand for housing Attributes in a Third World City." Land economics 68(4): 366. Banerjee, S., A. E. Gelfand, et al. (2004). "Spatial modeling of house prices using normalized distance-weighted sums of stationary processes." Journal of Business & Economic Statistics 22(2): 206. Benedict, N. R. and T. H. Szatrowski (1989). Statistical primer for real estate problem solving. U.S.A, American Society of Real Estate Counsellors of the National Association of Realtors. Berson, D. W. (1997). "The importance of demographics in economic analysis: The unusual suspects." Business Economics 32(1): 12. Blackwell, J. and a. A. i. a. w. R. T. Partners (2001). Population, labour force and housing demand projections, The national spatial strategy Borrie, W. D. (1995). "The founding of Australian demography: A Tribute to W. D. Borrie" Population Studies 49(3): 559-560.i Borsch-Supan, A. (1986). "Household formation, housing prices, and public impacts." Public Economics 30: 145-164.

272

Brett, L. Chen and B. Clarke (2003). "Demographics drive demand." Journal of Property Management(2): 66-67. Brown, M. (1989). "Housing trends reflect national economy, Local Demography." Indianapolis Business Journal 10(19): 6B. Brown, M. and John (2002). Many factors make province a hot property. Financial Times: 05. Buranathanung, N. (2004). Housing price Index: the Thai experience. C.Goodman, A. (2002). "Estimating equilibrium housing demand for stayer." Journal of Urban Economic 51(1-24). Calhoun, C. A. (2003). "Property valuation models and house price indexes for the provinces of Thailand: 1992-2000." Housing Finance International 18 (3): 11-31. Carlinger, G. (1973, 1973/11//). "Income of elasticity of housing demand." Review of Economics & Statistics Retrieved 4, 55. Case, B., J. Clapp, R. Dubin and M. Rodriguez (2004). "Modeling spatial and temporal house price patterns: A comparison of four models." Journal of Real Estate Finance & Economics 29(2): 2004,. Chapman, J. (1999). "Recent migration trends." Housing economics 47(8): 11. Charles, E. (2003). Housing demand remains strong in fairfield. New York Times: 11.7. Chiu, R. L. H. and M. H. C. Ho (2005). Estimate of elderly housing demand in an Asia city: Methodological issues and policy implications. Habitat International, Centre of Urban Planning and Environmental Management, The University of Hong Kong, Hong Kong: 1-12. Choiejit, R. and A. Houbcharaun (2007). Housing demand for the poor in Bangkok and its vicinity. Urban Research Symposium 2007, World Bank, Washington, DC. Coyle, R. G. (1996). System Dynamics modelling. Melbourne, Chapman & Hall. Di, Z. X. and X. Liu (2006). "The Effects of housing push factors and rent expectations on household formation of young adults." The Journal of Real Estate Research 28(Apr-Jun): 149. Dicks, M. J. (1988). "The demographics of housing demand: Household formation and the growth of owner-occupation." Bank of England Discussion Paper 32.

273

Dooley and Ann (1977). "DP model forecasts housing trend for next 10 years." Computerworld 11(18): 29. Ellis, M. (2003). Shortage of properties keeps market on boil. Northern Echo: 12. Emrath, P. (1999). "Seniors' housing: supply and demand." Housing Economics 47(4): 8. Ermisch, J. (1999). "Parents and young people's household formation." Urban Economics 47: 47-71. Ermisch, J. and P. D. Salvo (1997). "The economic determinants of young people's housing formation." Economica 64: 627-644. Fik, T. J., D. C. Ling and G. F. Mulligan. (2003). "Modeling spatial variation in housing prices: A Variable Interaction Approach." Real Estate Economics 31(4): 623. Filardo, A. J. (1996). "The outlook for housing: The role of demographic and cyclical factors." Economic Review - Federal Reserve Bank of Kansas City 81(3): 39. Forrester, J. W. (1991). System Dynamics and the lessons of 35 years, Sloan School of Management Forrester, J. W. (2006). "Origin of System Dynamics." from http://systemdynamics.org/DL-IntroSysDyn/orig_f.htm. Francke, M. K. (2004). "The hierarchical trend model for property valuation and local price indices." Journal of Real Estate Finance & Economics 28(2/3): 179. Frisch, R. (1993). Econometrica 1. Fry, G. (2004). "Recovery through reform Culture Matters in the Thai Turnaround " Harvard International Review 26 (3): 1. Fu, Y. (1991). "A model of housing tenure choice: comment." The American Economic Review 81(1): 381. Gat, D. (2000). "Israel's housing market dynamics: A tale of two sectors." Journal of Real Estate Literature 8(2): 131-152. Geffner, M. (2008). How housing cycles usually work. Chicago sun-time. Chicago: 5. Gelfand, A. E. (2004). "The dynamics of location in home price." Journal of Real Estate Finance & Economics 29(2): 2004,.

274

Gibb, K. (2000). Modelling housing choice and demand in a social housing system: the case of Glasgow Program on Housing and Urban Policy. Goldscheider, F. (1997). "Recent changes in US young adult living arrangements in comparative perspective." Family issue 18(6): 708-724. Goldscheider, F. and J. D. Vanzo (1985). "Living arrangements and the transition to adulthood." Demography 22(4): 545-563. Gross, D. (2004) "The demographics of housing demand." Volume, DOI: Guest, R. S. (2005). "A life cycle analysis of housing affordability options for first home owner-occupiers in Australia." Economic Record 81(Sep): 237. Hannon, B. (1999). "The revolution of dynamic http://web.aces.uiuc.edu/aim/model/day1text1.htm. modeling." from

Haurin, D., P. H. Hendershott, et al. (1993). "The impact of Real Rents and Wages on Household Formation." Economics and Statistics 75: 284-293. Heer, D., P. H. Hendershott and D. Kim. (1985). "The clustered nest: evidence that young adults are more likely to live at home now than in the Recent Past." Sociology and Social Research 69(3): 436-441. Howell, L. (2003). Demographic and socio-economic influences on housing Demand: 1. Hudson, R. (1997). National housing policy: reform and social justice. Melbourne, Ecumenical Housing Inc. Hui, E. C.-m. (2001). "Planning and development control through lease conditions." Habitat International 25(4): 599-615. Ioannides, Y. M. and J. E. Zabel (2003). "Neighbourhood effects and housing demand." Journal of Applied Econometrics 18(5): 563. Jantarakolica, T. (1996). Econometric estimation of regression models of the housing market in Bangkok, Thailand. Economy. United States, Indiana, University of Notre Dame. PhD: 177. Kamara, D. (1994). "The Effect of the Probability of Marriage on Housing Demand for Single Women." Housing Economics 3: 296-311. Kan, K. (2000). "Dynamic modeling of housing tenure choice." Journal of Urban Economics 48(1): 46. Karmel, R. (1998). "Some issue in estimating housing needs." Just Policy 12.

275

Kau, J. B. and D. Keenan (1980). "The theory of housing and interest rates." Journal of Financial and Quantitative Analysis 15(4): 833. Lande, R., S. Engen, B. E. Saether, F. Filli, E. Matthysen and H. Weimerskirch . (2002). "Estimating density dependence from population time series using demographic theory and life-history data. (Statistical Data Included)." The American Naturalist 159(4): 321. Leary, J. O. (2004). Modelling household formation via a microsimulation model. 12th Biennial conference of the Australian Population Association. Canberra, Australia. Leary, J. O. (2004). Recent trend in household formation in Melbourne. 12th Biennial conference of the Australian Population Association. Canberra, Australia. Lee and R. D. Trost (1978). "Estimation of some limited department variable models eighth application to housing demand." Journal of Econometrics 8: 35782. Lee, G. S., P. Schmidt-Dengler, B. Felder and C. Helmenstein. (2001). "Austrian demography and housing demand: Is there a connection." Empirica 28(3): 259. Liu, E., J. Wu, V. Lee and J. Lee. (1996). Study of housing demand model. Hong Kong, Research and Library Services Division Legislative Council Secretarial. RP01. Macpherson, D. A. and G. S. Sirmans (1999). "Forecasting seniors housing demand in Florida." Journal of Real Estate Portfolio Management 5(3): 259. Malpezzi, S. and S. K. Mayo (1987). "The demand for housing in developing countries: empirical estimates from household data." Economic Development and Cultural Change 35(4): 687. Maneekeatpaiboon, A. (1997). An econometric analysis of financial factors on housing demand: A case of government housing bank. Economics thesis. Bangkok, Chulalongkorn university. Master degree. Martin, W. B. (1989). "Demographic trends and the demand for residential rentals." The Journal of Real Estate Development: 51. Masnick, G. S. and Z. X. Di (2000). Updating and extending the joint center house hold Projections Using New cencus Bureau Population Projections, Joint center for Housing Studies of Harvard University. Mayo, S. K. (1980). "Theory and estimation in the economics of housing Demand." Journal of Urban Economics 10: 95-116.

276

Meadows, D., J. Randers, et al. (2004). Limits to growth. Vermont, Chelsea Green Publishing Company. Megbolugbe, I. F., A. P. Marks, et al. (1991). "The economic theory of housing demand: A critical review." The Journal of Real Estate Research 6( 3): 381393. Miron, J. R. (2004). "Housing demand, coping strategy, and selection bias." Growth & Change 35(2): ring2004,. Muth, R. (1960). The demand for durable goods. Chicago, University of Chicago Press. Myers, D. and L. Vidaurri (1990). "Real demographics of housing demand in the United States." The Lusk Review for Real Estate Development and Urban Transformation 2(1): 55-61. Myrtveit, M. (2005). The world model controversy, The System Dynamics Group, Department of Geography, University of Bergen P.O. Box 7800, N5020 Bergen, Norway: 1-27. Olsen., F. (1969). "A competitive theory of housing market." American Economic Review 59:4(6): 12-22. Parsons, G. R. (1986). "An almost ideal demand system for housing attributes." Southern Economic Journal 53(2): 347. Pesaran, M. H. (1987). "The new pal grave: A dictionary of economics." Econometrics 2: pp. 8-22. Pitakteeratum, K. (1994). Factors affecting demand for housing in Bangkok metropolitan and vicinities. Economic thesis. Bangkok, Chulalongkorn university. master degree. Pogodzinski, J. M. and T. R. Sass (1990). "The Economic Theory of Zoning: A Critical Review." Land Economics 66(3): 294. Pongsatat, M. (1998). Housing Policy Bangkok. Pornchokchai, S. (2006). "Thai housing market life cycle." Government Housing Bank 47: 38-43. Porter, L. (1999). "Demographic change " Australian Property Journal 35(5): 389-396. Prybolsky, R. P. (1997). "Housing the immigration wave." Mortgage Banking 58(1): 14. Ramrattan, L. and M. Szenberg (2004). "The sensitivity analysis of the FHA technique of housing market analysis: The effect of ratios and variables, and

277

their perturbations on family and elderly demand estimates." American Economist 48(Spring): 61. Richard, D. and W. W. Paul (1993). "The demand for housing: theoretical considerations." Journal of Economic Theory 61(1993/10): 120. Richmond, B. (1993). "Systems thinking: critical thinking skills for the 1990s and beyond." System Dynamics review 9(2): 113-133. Salt, B. (2005). The big shift. Sanchata, M. (2003). Tokyo flat prices are anything but: commercial land prices may have plummeted but changing demographics and more independent women are driving up the cost of housing in Japan, says Mariko Sanchanta:. Financial Times: 14. Scholl, T. J. (2002). Demand is stronger than supply in Barstow, Calif., Housing Market. i. Schulz, R. (2004). "A state space model for Berlin house prices: estimation and economic interpretation." Journal of Real Estate Finance & Economics 28(1): 37. Sector of housing information government housing bank (2002). Second hand House market situation in 2002 and trend of year 2003. 4/2002. 31. Bangkok: 27-39. Senge, P. M. (1994). The fifth discipline. Sydney, currency and doubleday. Serlen, B. (2004). "Corp. housing providers switch to Franchise model." Business Travel News 21(5): 8. Shankardass, S. (2001). The right to adequate shelter in international instruments the Universal Declaration of Human Rights and the International Covenant on Economic, social and Cultural Rights. Sheng, Y. K. and S. Kirinpanu (2000). "Once only the sky was the limit: Bangkok's housing boom and the financial crisis in Thailand." Housing study 15 (1): 11-27. Skaburskies, A. (1994). "Determinants of Canadian headship rates,." Urban Studies 31(8): 1377-1389. Skaburskis, A. (1999). "Modelling the choice of tenure and building type." Urban Studies 36(13): 2199. Smith, H. (2004). "Robust housing demand continues in Las Vegas." Knight Ridder Tribune Business News: 1.

278

Straszheim, M. R. (1975). An econometric analysis of the urban housing market. New York, Columbia University Press. Tan (2007). Districts <http://bangkokview.blogspot.com/2007/05/districts.html> Tangmatidham, P. (2006). "housing market in 2006 " Government Housing Bank 44: 11-17. Thomas, L. and B. Malmberg (2000). Demography and housing demand--what can we learn from residential construction data? paper presented to Research Seminar of the Institute for Housing Research, Gavle. Thomas, L. and B. Malmberg (2005). Demography and housing demand--what can we learn from residential construction data? paper presented to Research Seminar of the Institute for Housing Research, Gavle. Tiwari, P. and H. Hasegawa (2000). "Effective rental housing demand in the tokyo metropolitan region." RURDS 12(1): 54-73. Tuccillo, J. A. (1995). "Housing in the long term: Trends and prospects." Business Economics 30(2): 51. Vanichvatana, S. (2006). "Cycle of real estate of Thailand why? how?" Government Housing Bank 47: 44-48. Vidaurri, D. (1996). "Real demographics of housing demand in the united states." The Lusk Review for Real Estate Development and Urban Transformation 2(1): 55-61. Waddell, P. (2002). "UrbanSim: modeling urban development for land use, transportation, and environmental planning." Journal of the American Planning Association 68(3): 297. Wang, Z. (2004). "Dynamics of urban residential property prices -- A case study of the manhattan market." Journal of Real Estate Finance & Economics 29(1): 99. Wheaton, W. C. (1990). "Vacancy, search, and prices in a housing market Matching Model." The Journal of Political Economy 98(6): 1270. White, M. (2003). "Land-use planning and the housing market: a comparative review of the UK and the USA." Urban Studies 40(5/6): 953. Wright, L. E. (1993). "Demand and supply: a look at multifamily housing's future." Pension World 29(10): 30. Wunsch, G. J. and M. G. Termote (1978). Introduction to demographic analysis principles and method. New York, Plenum Press.

279

Yates, J. (2001). "Australia's housing choices: retrospect and prospect." Urban Policy and Research 19(4): 415-24. Yates, J. (2002). "The limits to choice in the private rental market Initial version of this paper were presented at the Which Way Housing Policy?" Just Policy 25: 32-48. Yucun, H. (2002). Study of System Dynamics for urban housing Development in Hong Kong. The Research office. Hong Kong, The Hong Kong Polytechnic University. PhD: 214. Zhang, X. (1999). A macro study of the housing demand of the United State. Waco Texas, Department of Economics Hankamer School of Business Baylor University. Zhou, Z.-g. (1997). "Forecasting sales and price for existing single-family homes: A VAR model with error correction." The Journal of Real Estate Research 14(1/2): 155. Ziegert, A. L. (1988). "The demand for housing additions: an empirical analysis." AREUEA Journal 16(4): 479.

280

Bibliography

Andrew, DF, Fickel, PJ & Hampe, lFR 1972, Robust Estimates of Location:Survay and Advances., Princeton University Press, Princeton. Anonymous 1986, 'A New Look at the Homeownership Decision', Housing Finance Review, vol. 5, no. 1, pp. 59-87. ---- 1993, 'Survey predicts a bright future for residential', National Real Estate Investor, vol. 35, no. 6, p. SS20. ---- 1996, 'Japan's changing housing market: Growing popularity of imports', East Asian Executive Reports, vol. 18, no. 11, p. 15. ---- 2000, SAAP National Data Collection, Annual Report, 1999-2000, vol. 50, AIHW Cat, Australia Institute of Health and Welfare, Hou. ---- 2001, 'Demand for Homes Continues to Grow', National Mortgage News, vol. 25, no. 46, p. 2. ---- 2002, 'Housing demand spurred by growing immigrant population', Community Banker, vol. 11, no. 9, p. 68. ---- 2003a, 'Housing activity', Housing Economics, vol. 51, no. 2, p. 18. ---- 2003b, ''Why is it Important to Boost the Supply of Affordable Housing in Australia--and How Can We Do it?' Urban Policy and Research, vol. 21, no. 4, pp. 413-35. ---- 2005, Types of housing, <http://www.housing.qld.gov.au/renting/indigenous/applying/accomm.htm>. ---- 2008, <http://eng.konbaan.com/index.php>. Armstrong, JS 1978, Long-range forecasting from Crystal Ball to Computer, John Wiley & Sons, Inc., Canada. Bajari, P & Kahn, ME 2002, Estimating Housing Demand with an Application to Explaining Racial Segregation in Cities, Stanford University Tufts University, 30 August 2002. Berry, M 2002, ''Investing in rental Housing'', Just Policy. Bock, RD 1975, Multivariate statistical Methods in Behavioural Research., Mcgraw_Hill, New York.

281

Bolding, G, Davis, M, Sherr, L, Hart, G & Elford, J 2004, Use of gay Internet sites and views about online health promotion among men who have sex with men. Branch, ER 1993, 'Short run income elasticity of demand for residential electricity using consumer expenditure survey data', The Energy Journal, vol. 14, no. 4, p. 111. Buckley, Robert, Ermisch & John 1982, 'Government Policy and House Prices in the United Kingdom: An Econometric Analysis', Oxford Bulletin of Economics and Statistics, vol. 44, no. 4, p. 273. Bull, SS, Lloyd, L, Rietmeijer, C & McFarlane, M 2004, Recruitment and retention of an online sample for an HIV prevention intervention targeting men who have sex with men: the Smart Sex Quest Project. Bull, SS, McFarlane, M, Lloyd, L & Rietmeijer, C 2004, The process of seeking sex partners online and implications for STD/HIV prevention. Burrows, BC 1980, 'Urban and regional planningA systems view', Long Range Planning, vol. 13, no. 1980/4, p. 67. Caldwell, JC 2004, 'Demographic theory: a long view. (Notes and Commentary)', Population and Development Review, vol. 30, no. 2, p. 297. Cleary, JP & Levenbach, H 1982, The Professional Forecaster: The Forecasting Process through Data Analysis, Wadsworth,Inc., U.S.A. Cobb-Clark, DA & Crossley, T 2003, 'Econometrics for Evaluations: An Introduction to Recent Developments*', Economic Record, vol. 79, no. 247, p. 491. Cuticchia, AJ, Fasman, KH, Kingsbury, DT, Robbins, RJ & Pearson, PL 1993, The GDB super(TM) human genome data base anno 1993. Davis, M, Bolding, G, Hart, G, Sherr, L & Elford, J 2004, Reflecting on the experience of interviewing online: perspectives from the Internet and HIV study in London. Delano, D 1999, 'Housing market grows despite fading consumer confidence', Industrial Distribution, vol. 88, no. 1, p. S1. Devineni, T & Blanchard, EB 2005, A randomized controlled trial of an internet-based treatment for chronic headache. Drummer, R 1999, 'Inland Empire Focus Demand outstripping housing stock', The Business Press, Jul 19, 1999, p. 1.

282

Eberhardt, JI 1981, Housing Demand in Victoria, University of Melbourne Institue of Applied Economic and Social Research, Melbourne, Victoria. Ehrhardt-Martinez, K 1998, Social determinants of deforestation in developing countries: a cross-national study., University of North Carolina Press. Evans, ML 1993, '1993 national real estate forecast survey', National Real Estate Investor, vol. 35, no. 2, p. SS1. Fernandez, MI, Varga, LM, Perrino, T, Collazo, JB, Subiaul, F, Rehbein, A, Torres, H, Castro, M & Bowen, GS 2004, The Internet as recruitment tool for HIV studies: Viable strategy for reaching at-risk Hispanic MSM in Miami? Filardo, AJ 1996, 'The outlook for housing: The role of demographic and cyclical factors', Economic Review - Federal Reserve Bank of Kansas City, vol. 81, no. 3, p. 39. Fitzgerald, V 2004, 'Australia's ageing population: implications for saving and retirement income. [Paper in: Australia's Ageing Population: Fiscal, Labour Market and Social Implications.]', Growth (Melbourne), vol. 51, pp. 89-102. Ford, CM 2005, 'What Factors Are Fueling Strong Housing Demand?' The Wall Street Journal. Forrester, JW 2006, System Dynamics <http://www.albany.edu/cpr/sds/index.html>.

Society,

September

2006,

Friedland, J 1992, 'Building Confidence: Japan's Housing Sector Shows Signs of Recovery', Far Eastern Economic Review, vol. 155, no. 39, p. 82. G.Willis & AGTaK 1991, Housing the poor in the developing world, vol. 1, Routledge, New York. Glaeser, EL & Gyourko, J 2005, 'Urban Decline and Durable Housing', The Journal of Political Economy, vol. 113, no. Apr, p. 345. Goodman, A 2002, 'Estimating Equilibrium Housing Demand for Stayer', Journal of Urban Economic, vol. 51, no. 1-24. Grootaert, C & Dubois, JL 1950, The Demand for Urban Housing in the Ivory Coast, Washington D.C. U.S.A. Groth, M 1996, 'Economic models: perspectives for the professional', Journal of property Finance, vol. 7, no. 2, pp. 45-57. Hamecs, MC & Benedict, M 1999, 'Housing finance trends', Housing Economics, vol. 47, no. 9, p. 6.

283

Hirshfield, S, Remien, RH, Humberstone, M, Walavalkar, I & Chiasson, MA 2004, Substance use and high-risk sex among men who have sex with men: a national online study in the USA. Holmes, TP, Vlosky, RP & Carlson, J 2004, An exploratory comparison of Internet use by small wood products manufacturers in the North Adirondack Region of New York and the State of Louisiana. 'Housing sector & Housing finance - An Overview', 2008, Accommodation times, 1 Feb 2008. Hughes, BK 1984, 'Futures Markets: Metals Decline on Signs Of Softening Demand; March Orders Fell 2.2%', Wall Street Journal, Apr 25, 1984, p. 1. Ihlanfeldt, KR 2004, 'The Use of an Econometric Model for Estimating Aggregate Levels of Property Tax Assessment Within Local Jurisdictions', National Tax Journal, vol. 57, no. 1, p. 7. Jager, T, Jager, T, Crommentuijn, T, van Gestel, CAM & Kooijman, SALM 2004, 'Simultaneous Modeling of Multiple End Points in Life-Cycle Toxicity Tests.' Environmental Science & Technology, vol. 38, no. 10, p. 2894. Jensen, HE 1999, 'The development of T.R. Malthus's institutionalist approach to the cure of poverty: From.' Review of Social Economy, vol. 57, no. 4, p. 450. Kachur, RE 2004, The Internet Alert Project: spreading the word about highrisk sexual activities advertised on the Internet. Kelejian, HH & Prucha, IR 2004, 'Estimation of simultaneous systems of spatially interrelated cross sectional equations', Journal of Econometrics, vol. 118, no. 1-2, pp. 27-50. Keller, SN, LaBelle, H, Karimi, N & Gupta, S 2004, Talking about STD/HIV prevention: a look at communication online. Klausner, JD, Levine, DK & Kent, CK 2004, Internet-based site-specific interventions for syphilis prevention among gay and bisexual men. Krittayanawach, P 2006, 'Introduction to Real Estate Cycle', Government Housing Bank, vol. 47, pp. 38-43. Lee, SY 1995, 'An integrated model of land use/transportation system performance: System dynamics modeling approach', no. 1995. Leeuw, Fd & Stuyk, RJ 1975, The Web of urban Housing Analyzing Policy with a Market Simulations Model, The Urban Institute, Washington, D.C. Li, T & Hsiao, C 2004, 'Robust estimation of generalized linear models with measurement errors', Journal of Econometrics, vol. 118, no. 1-2, pp. 51-65.

284

Malu, MK, Challenor, R, Theobald, N & Barton, SE 2004, Seeking and engaging in internet sex: a survey of patients attending genitourinary medicine clinics in Plymouth and in London. Mayo, SMSK 1994, A Model Design for a Developing Country Housing Market Study, University of Wisconsin Center for Urban Land Economic Research. McClintock, J 2003, 'Fewer Able to Afford Homes as California's Housing 'Crisis' Threatens Economy', Knight Ridder Tribune Business News, p. 1. McFarlane, M, Ross, MW & Elford, J 2004, The Internet and HIV/STD prevention. McKenzie, T 1981, 'Inflation,Taxes and the Demand for Housing', AREUEA Journal, vol. 10, pp. 25-38. McNemar, Q Psychological Statistics, 4 edn, Psychological Statistics. Meertens, P 1986, Out East Housing "Needs Study" Outer East Regional Housing Council, vol. 1, Melbourne, Victoria. Meloni, I, Guala, L & Loddo, A 2004, 'Time allocation to discretionary in-home, out-of-home activities and to trips.' Transportation, vol. 31, no. 1, pp. 69-96. Meloni, I, Spissu, E & Bez, M 2007, 'A Model of the Dynamic Process of Time Allocation to Discretionary Activities', Transportation Science, vol. 41, no. 1, pp. 15-28. Moore & Jeppson, V 1997, 'Application of a system dynamics model to regional planning'. Nichols, JD, HINES, JE, LEBRETON, J-D & PRADEL, R 2000, 'ESTIMATION OF CONTRIBUTIONS TO POPULATION GROWTH: A REVERSE-TIME CAPTURE-RECAPTURE APPROACH.' Ecology, vol. 81, no. 12, p. 3362. No, SC & Zapata, HO 2004, 'A Dynamic Econometric Modeling of the U.S. Rice Market', Journal of American Academy of Business, Cambridge, vol. 4, no. 1/2, p. 384. Norman, J 2005, The artefacts of the internet era may not look like antiques, but they are still collectable treasures. O'Connor, K 1998, 'Understanding metropolitan Melbourne: without being confused by coffee and doughnuts', Urban Policy and Research, vol. 16, no. 2, pp. 139-45. ---- 1999, 'Misunderstanding modern suburban development', People and Place, vol. 7, no. 4, pp. 21-5.

285

Office, TNS 2005, Population and housing data, <www.nso.go.th >. Olfson, E & Laurence, J 2005, Accessibility and Longevity of Internet Citations in a Clinical AIDS Journal. Parnsoonthorn, K 2004, 'Thai Property Firm Land & Houses Seeks to Maintain 20 Percent Growth Goal', Knight Ridder Tribune Business News, p. 1. Parsons, JT, Koken, JA & Bimbi, DS 2004, The use of the Internet by gay and bisexual male escorts: sex workers as sex educators. Pindyck, RS & Rubinfeld, DL 1981, Econometric Models and Economic Forecasts, 2 edn, 2 vols., Mc Graw-Hill.Inc, U.S.A. Porter, L 1999, 'Demographic change [in Melbourne. Originally presented as a paper to the 7th Annual Australian Property Institute Conference in August (1998)]', Australian Property Journal, vol. 35, no. 5, pp. 389-96. Reynolds, JP, Libby 1998, 'Melbourne's inner city revival', Urban Policy and Research, vol. 16, no. 1, pp. 63-8. Richard, D & Paul, WW 1993, 'The Demand for Housing: Theoretical Considerations', Journal of Economic Theory, vol. 61, no. 1993/10, p. 120. Rodda, DT, Rodda, DT, Lam, K & Youn, A 2004, 'Stochastic Modeling of Federal Housing Administration Home Equity Conversion Mortgages with LowCost Refinancing.' Real Estate Economics, vol. 32, no. 4, p. 589. Ross, MW, Rosser, BRS & Stanton, J 2004, Beliefs about cybersex and Internetmediated sex of Latino men who have Internet sex with men: relationships with sexual practices in cybersex and in real life. Schuppen, V & H., J 2004, 'System theory for system identification', Journal of Econometrics, vol. 118, no. 1-2, pp. 313-39. Shaw, K 2003, 'Discretion vs. regulation and the sorry case of Melbourne City Plan 2010', Urban Policy and Research, vol. 21, no. 4, pp. 441-7. Sibayan, K 2004, 'It's been said before, there is no housing bubble.' Asset Securitization Report, pp. plement. Sourell*, AGSKAAR 2006, 'The housing fulcrum: balancing economic and social factors in housing research and policy ', New Zealand Journal of Social Sciences Online, vol. 1, pp. 65-79. Sperber, B & Sperber, B 2003, '3-D Software Models Land Uses.' Professional Builder, vol. 68, no. 12, p. 34. Suggu, K 2001, 'The real state of real estate', India Abroad, Jul 6, 2001, p. 24.

286

Supan, B, Heiss & Seko 2001, Housing Demand in Germany and Japan, November 2003. Trinidad, M 2003, 'Postcode poverty, service integration and inclusion: [Paper presented at: AIFS Conference (8th: 2003: Melbourne, Vic.). Steps Forward for Families: Research, Practice and Policy]', paper presented to Australian Institute of Family Studies Conference, D 2003. Valente, J, Valente, J, Wu00, S, Gelfand, A & Sirmans, CF 2005, 'Apartment Rent Prediction Using Spatial Modeling.' Journal of Real Estate Research, vol. 27, no. 1, p. 105. Vanichvattana, S 2006, 'Life cycle of Thai Real Estate Business', Government Housing Bank, vol. 47, pp. 38-43. Wiebers, U-C, Metcalfe, M & Zilberman, D 2002, 'The influence of pest management advice on pesticide use in California tomatoes Advances in the Economics of Environmental Resources', in Volume 4 edn, JAI, pp. 81-98. Wilson, R 2004, 'Malthusian Motoring.' Automotive Industries, vol. 184, no. 9, pp. 2004. Winsor, D & Gross, M 2005, The Demographics of housing Demand, January 2005, <www.donahue.umassp.edu/publications/housing/1-demand.html>. Wulff, M 2001, 'Out with the old and in with the new? housing's role in the new social settlement. [Future Directions in Australian Social Policy: New Ways of Preventing Risk. Edited by Hancock, Linda; Howe, Brian; Frere, Marion and O'Donnell, Anthony.]', Growth (Melbourne), vol. 49, pp. 5765. Yates, J 2001, 'The rhetoric and reality of housing choice: the role of urban consolidation', Urban Policy and Research, vol. 19, no. 4, pp. 491-527. Young, R 2003, 'Housing Market Boom in Yuba City, Calif., Area Continues through 2002', Knight Ridder Tribune Business News, p. 1. Zhang, X 1999, A Macro Study of the Housing Demand of the United State, Department of Economics Hankamer School of Business Baylor University, Waco Texas. Zhao, X & Harris, MN 2004, 'Demand for Marijuana, Alcohol and Tobacco: Participation, Levels of Consumption and Cross-equation Correlations', Economic Record, vol. 80, no. 251, pp. 394-410.

287

Appendix A

Terminology

Detached house (Single house)

x a house on its own block of land x containing between two and seven bedrooms (there are only a small number of

properties with more than five bedrooms)

x usually having a fenced backyard and fenced front yards

Semi -detached houses (Twin house)

x also known as half-a-house x one building usually consists of two houses x located on its own block of land and units divided by a common wall x each unit contains one to three bedrooms and x each unit has a fenced front or back yard

Attached house (Town house)

x three or more units built next to each other and divided by common walls x generally two floors x each unit contains between one and four bedrooms and x each unit usually has its own small fenced yard or courtyard

Commercial building and shop house building

x often multi-storey, commercial use on ground floor x less than ten units built next to each other and divided by common walls x generally two to four floors x each unit contains between one and four bedrooms and x each unit has its front face to a road or street

288

Apartment (i.e. flat and condominium)

x contains between one and three bedrooms x usually in a complex of two or more floors x each apartment is on one level x some complexes have a communal laundry and x generally has a communal yard/garden

Many documents and information sources consulted for this research classify housing into three main types. The first type comprises houses, which include detached houses, semi-detached house, and attached houses. The second type consists of commercial buildings and shop-house buildings. The third type comprises apartments, which include flats and condominiums.

This research however uses particular terms in the following manner (A-Z):

Absolute housing demand is the term used to refer to the number of

housing demand that has been counted by using the principle and calculation of this research.
x

Adolescent is the term used to refer to people from 0 to 19 years of

age.
x x

Attached house is the term used to refer to town house.

Condominium is the term used to refer to apartments, flats and condominiums.

Couple head household family is the term used to refer to family

which consists of two parents (in most case it appear as father and mother) who both take care of their family.

289

x x

Detached house is the term used to refer to single house. Elder is the term used to refer to people from 60 and over years of

age.
x

Exiting housing market is the term used to refer to resell housing

market.
x

Extended family is the term used to refer to a family group that

consists of parents, children, and other close relatives, often living in close proximity or a group of relatives, such as those of three generations, who live in close geographic proximity rather than under the same roof.
x

House is the term used to refer to detached houses, semi-detached and attached houses.

Housing affordability actually has no uniform definition. A widely accepted implicit definition states that monthly housing costs for adequate housing should be no more than 30% of household income. As such, housing affordability can be measured as a ratio of housing payments to household income, adjusted for household size. Housing affordability in this research then is the term used to refer to purchasing power of people with willingness to buy housing unit.

Housing demand is the term used to refer to willingness to buy

housing unit with affordability to buy a house or houses of people.

x

Housing demand unit (HDU) is the term used to refer to number of a

person or persons who need to buy a house for living or investing and have affordability to buy.

290

Housing needs unit (HNU) is the term used to refer to number of a

person or persons who need to buy a house for living or investing even have or have no affordability to buy.
x

Housing needs without affordability unit (HNwoA) is the term used

to refer to the number of a person or persons who need to buy a house for living or investing but have no ability to buy due to lack of founds.
x

Housing needs is the term used to refer to needs of people to have a

house for living.

x

Housing sales unit (HSU) is the term used to refer to number of

housing that has been sold to persons in housing demand unit.

x

Housing unit is the term used to refer to all types of housing, counted as the same unit.

x x

Housing is the term used to refer to all types of housing.

Investment housing demand or Speculation housing demand is

the term used to refer to housing demand unit of people who want to buy a house for investment or speculation purpose.
x x

Mature is the term used to refer to people from 20 to 59 years of age. Nuclear family is the term used to refer to a family unit consisting of

a mother and father and their children.

x x

Semi detached house is the term used to refer to twin house.

Shop-house or block is the term used to refer to commercial buildings and shop-house buildings.

Single head household family is the term used to refer to family

which consist of only one person or one parent who take a leadership duty in that family.

291

Bangkok Metropolitan Region or BMR is the term used to refer to

Bangkok and five provinces around Bangkok: Samut Prakarn, Nontaburi and Patum Thanee, Samut Sakorn and Nakorn Patom. Altogether, it covers an area of 7,761.50 km, and with a registered population of 9,988,400 (as of April 4, 2007) it has a population density of 1,286.92 per km.
x

Underlying housing needs units(UHNUs) is the term used to refer to the number of housing needs that has been counted by using the principle and calculation in this research to simulate number of persons who need to buy a house for living or investing even though they have or have no housing affordability. The principle (i.e. from this research) to produce the underlying housing needs units comes from two main factors: demographic and social factors.

292

Appendix B

Table of data used in the model

Table B-1 Population in Thailand from 1975-2003 Population year total 42,391,454 43,213,711 44,272,693 45,221,625 46,113,756 46,961,338 47,875,002 48,846,927 49,515,074 50,583,105 51,795,651 52,969,204 53,873,172 54,960,917 55,888,393 56,303,273 56,961,030 57,788,965 58,336,072 59,095,419 59,460,382 60,116,182 60,816,227 61,466,178 61,661,701 61,878,746 62,308,887 62,799,872 male 21,359,489 21,790,510 22,314,837 22,775,852 23,205,927 23,627,727 24,067,597 24,549,873 24,911,684 25,449,044 26,059,668 26,642,889 27,070,155 27,574,256 28,001,343 28,181,202 28,463,102 29,018,092 29,205,086 29,552,978 29,678,600 29,973,059 30,295,797 30,591,602 30,650,172 30,725,016 30,913,485 31,139,647 female 21,031,965 21,423,201 21,957,856 22,445,773 22,907,829 23,333,611 23,807,405 24,297,054 24,603,390 25,134,061 25,735,983 26,326,315 26,803,017 27,386,661 27,887,050 28,122,071 28,497,928 28,770,873 29,130,986 29,542,441 29,781,782 30,143,123 30,520,430 30,874,576 31,011,529 31,153,730 31,395,402 31,660,225

2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546

1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

density per sq km. 82.62 84.22 86.28 88.13 89.87 91.52 93.30 95.20 96.50 98.58 100.94 103.23 104.99 107.11 108.92 109.73 111.01 112.62 113.69 115.17 115.88 117.16 118.52 119.79 120.17 120.59 121.43 122.39

63,079,765

31,255,350

31,824,415

122.93

From ministry of interior, Thailand

293

Table B-2 Population of Thailand from surveying between1909 and 2000 year total Population male female increased rate

2452 2462 2472 2480 2490 2503 2513 2523 2533 2543

1909 1919 1929 1937 1947 1960 1970 1980 1990 2000

8,149,487 9,207,355 11,506,207 14,464,105 17,442,689 26,257,916 34,397,374 44,824,540 54,548,530 60,916,441

n/a 4,599,667 5,795,065 7,313,584 8,722,155 13,154,149 17,123,862 22,328,607 27,061,733 30,015,233

n/a 4,607,688 5,711,142 7,150,521 8,720,534 13,103,767 17,273,512 22,495,933 27,486,797 30,901,208

n/a 1.22 2.23 2.86 1.87 3.15 2.70 2.65 1.96 1.05

From: the report of Household survey in 1909,1919,1929,1937 and 1947 of ministry of interior The report of household survey in 1960, 1970, 1980, 1990 and 2000 of the statistic bureau

area

Table B-3 Population area density and house in2003 population area density (Sq km.) total male female per sq km.

house

country Bangkok

63,079,765 5,844,607

31,255,350 2,822,171

31,824,415 513,115.0 3,022,436 1,565.2

122.93 17,853,423 3,734.10 2,020,019

From Department of local administration

294

age Population male

20,885,478 1,767,835 1,963,669 1,914,477 3,662,309 3,347,886 3,497,560 973,391 919,392 828,323 576,148 431,404 320,814 578,347 433,060 315,480 1.1 0.7 0.3 0.5 176,330 103,154 75,966 265,667 192,613 137,348 74,082 39,852 26,338 639,463 471,979 355,367 312,680 240,447 178,132 102,248 63,302 49,628 3,673,477 3,513,329 3,110,723 2,504,211 1,944,776 1,608,096 1,392,340 1,096,961 764,623 449,952 237,396 169,361 1,882,378 1,735,005 1,723,601 1,792,088 1,736,866 1,534,008 1,231,726 949,676 777,072 659,538 510,188 349,059 198,052 97,194 63,046

(year)
100.0 7.6 8.4 8.3 8.7 8.6 8.9 9.1 8.6 7.7 6.1 4.7 3.7 3.2 2.5 1.8 1.0 0.6 0.4 0.5 0.9 1.6 1.9 2.3 2.7 3.1 3.4 3.7 3.8 4.6 4.7 903,383 676,181 6.0 6.2 1,215,611 7.6 7.8 1,574,321 8.5 8.7 1,746,028 826,636 745,998 8.9 9.2 1,852,467 879,076 8.8 9.0 1,910,236 907,807 8.8 8.4 1,914,255 917,649 996,606 1,002,429 8.9 8.4 1,624,724 801,256 823,468 8.7 8.0 1,351,669 689,734 661,935 8.8 8.0 1,315,293 675,544 639,749 8.0 7.3 1,205,825 621,789 584,036 3,429,073 3,809,761 3,732,277 100.0 100.0 18,972,330 9,129,755 9,842,575 41,944,111 2,245,274 2,485,841 2,479,735 2,603,399 2,609,487 2,776,388 2,854,780 2,695,855 2,405,038 1,911,948 1,467,079 1,186,391 1,045,482 827,220 593,696 354,148 203,504 155,943

total

male

Table B-4 population by age, sex and municipal limits in 2000 in out of percent municipal municipal limits limits female total male female total male female total

female
21,058,633 1,661,238 1,846,092 1,817,800 1,779,931 1,612,881 1,773,959 1,881,389 1,776,463 1,576,715 1,272,485 995,100 831,024 732,802 586,773 415,564 251,900 140,202 106,315

total

60,916,441

30,015,233

30,901,208

0-4

4,634,898

2,389,624

5-9

5,125,054

2,639,213

10 - 14

5,083,946

2,604,211

15 - 19

5,287,033

2,683,634

20 - 24

5,262,141

2,652,654

25 - 29

5,407,796

2,631,408

30 - 34

5,525,944

2,671,164

35 - 39

5,259,357

2,563,502

40 - 44

4,685,044

2,280,006

45 - 49

3,719,822

1,807,874

50 - 54

2,848,159

1,381,080

55 - 59

2,284,277

1,097,886

60 - 64

1,970,687

925,205

65 - 69

1,530,021

702,801

70 - 74

1,080,103

486,407

75 - 79

626,282

272,134

80 - 84

340,550

137,046

over 85

245,327

89,384

from statistic bureau, office of Prime minister

295

2001 male 30,913,485 31,395,402 62,799,872 2,165,012 2,471,035 2,365,374 2,510,210 2,694,397 2,759,387 2,793,229 2,619,916 2,243,629 1,827,567 1,385,671 1,001,197 857,555 957,546 1,084,247 2,148,128 1,872,448 1,489,173 3,039,054 1,953,811 3,930,215 6.26 4.84 3.42 2.98 2,346,854 4,739,965 7.55 2,721,855 5,413,362 8.62 2,651,000 2,317,495 1,898,060 1,461,853 1,028,648 884,995 2,862,295 5,663,969 9.02 2,797,582 2,754,341 5,479,986 8.73 2,745,551 2,734,435 2,866,387 2,762,362 2,422,470 2,032,155 1,577,201 1,119,480 987,453 2,646,617 5,282,247 8.41 2,668,483 2,613,764 2,417,960 4,780,924 7.61 2,436,368 2,344,556 4,661,217 5,208,192 5,406,128 5,563,479 5,505,556 4,889,452 4,060,211 3,202,542 2,286,141 1,855,785 2,249,055 4,700,093 7.48 2,410,108 2,289,985 4,759,358 7.54 7.39 8.26 8.57 8.82 8.73 7.75 6.44 5.08 3.62 2.94 2,342,579 4,821,676 7.68 2,476,093 2,345,583 4,754,166 7.54 2,042,485 4,042,604 6.44 2,080,555 1,962,049 3,902,496 6.19 2,008,961 2,441,841 2,440,525 2,377,751 2,629,241 2,710,611 2,751,018 2,691,778 2,389,460 1,960,274 1,535,489 1,094,353 875,527 31,139,647 31,660,225 63,079,765 female total % male female total % male female

Table B-5 population by age from 2001-2003 2002 2003

age (year) 31,255,350 31,824,415 1,893,535 2,312,325 2,318,833 2,283,466 2,578,951 2,695,517 2,812,461 2,813,778 2,499,992 2,099,937 1,667,053 1,191,788 980,258

total

total

62,308,887

0-4

4,207,497

6.75

5-9

4,813,614

7.73

10 - 14

4,614,429

7.41

15 - 19

4,928,170

7.91

20 - 24

5,341,014

8.57

25 - 29

5,513,728

8.85

30 - 34

5,655,524

9.08

35 - 39

5,341,771

8.57

40 - 44

4,590,483

7.37

45 - 49

3,781,378

6.07

50 - 54

2,874,844

4.61

55 - 59

2,085,444

3.35

60 - 64

1,815,101

2.91

296

65 - 69 456,531 267,069 138,316 65,916 27,931 13,354 17,058 1,547,770 100.00 1,398,829 2,911,960 4.64 1,535,264 1,376,696 23,811 42,715 0.07 17,855 24,860 34,784 2,875,694 20,295 33,839 0.05 13,425 20,414 30,612 0.05 0.06 4.56 100.00 45,510 74,995 0.12 28,671 46,324 71,214 0.11 104,638 179,575 0.29 69,522 110,053 190,802 0.30 74,208 26,712 11,923 14,521 1,524,953 200,643 354,636 0.56 144,892 209,744 380,257 0.60 155,833 355,262 657,745 1.05 281,305 376,440 715,638 1.13 305,814 409,824 224,424 116,594 44,502 18,689 20,263 1,350,741 571,173 1,106,138 1.76 493,352 612,786 1,145,133 1.82 508,836 636,297

1,491,784

2.39

685,361

806,423

1,523,598

2.43

698,570

825,028

1,580,908

2.51

725,721

855,187

70 - 74

1,027,704

1.65

75 - 79

622,331

1.00

80 - 84

338,959

0.54

85-89

170,554

0.27

90-94

73,441

0.12

95-99

33,649

0.05

over 100

40,869

0.07

unknown

2,946,599

4.73

100.00

from : department of local administration

297

Table B- 6 birth rate by municipal limits in 1985-1986, 1991 and 1995-1996 area birth rate (per 1,000 ) 1985-1986 1991 20.19 1995-1996 17.90

country in municipal limits out of municipal limits

Bangkok

23.87

18.97 24.79
18.92

16.79 20.98
14.65

14.73 18.72
14.24

From : statistic bureau, office of prime minister

298

Table B-7 population over 13 year of age divide by status, age and sex in 2000

total total 28,712,450 2,803,037 405,976 532,144 317,424 243,761 married separate no data be a monk

age (year) and sex status have been marry widow divorce

single

no data of status 170,643

total

48,132,896

14,947,461 32,771,031

13 - 14 406,380 1,759,180 3,419,305 4,440,846 4,590,994 4,256,558 3,453,576 2,686,201 2,175,497 1,885,757 1,468,726 1,038,316 603,957 328,951 236,701 317,680 139,403 80,815 644,650 1,044,288 1,467,849 365,278 385,627 366,003 270,207 179,443 147,708 1,831,660 279,524 2,371,788 229,007 3,148,635 185,237 49,746 33,282 22,434 15,360 9,701 5,485 2,689 1,336 937 3,951,539 147,182 65,815 4,315,412 102,910 65,609 4,188,697 74,478 59,275 77,536 77,682 70,708 54,956 40,089 31,113 25,380 17,904 10,939 5,212 2,368 1,400 3,209,510 45,192 44,403 67,097 1,625,587 20,667 24,308 40,608 361,088 4,257 5,201 8,939 26,895 48,010 53,103 40,860 29,381 21,314 15,002 12,035 10,766 11,890 11,206 11,239 8,169 6,401 5,841

2,060,353

1,991,312

20,086

13,849

317

395

213

5,312

24,037 30,149 34,306 22,459 17,171 14,824 13,574 11,437 10,811 11,621 14,240 14,173 12,286 7,128 3,378 2,167

24,918 18,319 18,844 14,788 17,460 14,858 13,918 9,800 8,004 6,748 6,783 5,711 4,543 2,592 1,802 1,555

15 - 19

5,287,033

4,832,185

20 - 24

5,262,141

3,449,811

25 - 29

5,407,796

1,951,244

30 - 34

5,525,944

1,050,467

35 - 39

5,259,357

638,681

40 - 44

4,685,044

400,994

45 - 49

3,719,822

245,009

50 - 54

2,848,159

143,143

55 - 59

2,284,277

90,411

60 - 64

1,970,687

63,907

65 - 69

1,530,021

41,411

70 - 74

1,080,103

24,958

75 - 79

626,282

12,605

80 - 84

340,550

6,419

over 85

245,327

4,904

From: statistic bureau, office of prime minister

299

total total 13,977,400 617,505 137,386 189,944 142,456 243,761 married separate no data be a monk

age (year) and sex status have been marry widow divorce

single

no data of marry status 107,089

male

23,434,640

8,019,099 15,064,691

13 - 14 101,658 600,942 1,443,363 2,050,311 2,212,556 2,078,860 1,697,407 1,317,702 1,052,894 887,566 672,854 464,444 259,976 131,068 85,100 84,273 46,191 189,439 369,147 575,550 84,588 86,017 64,994 43,596 36,397 792,607 77,537 976,767 54,983 7,007 5,025 3,119 1,884 951 434 299 1,244,438 44,602 10,533 1,619,000 36,415 16,323 1,988,541 31,285 22,638 26,231 19,441 13,686 10,552 8,622 6,283 4,143 2,173 1,018 643 2,121,353 22,985 23,511 29,567 1,959,060 18,216 21,750 29,410 1,364,445 10,606 15,579 24,475 28,258 21,875 15,140 10,165 6,228 4,443 3,585 3,775 3,314 3,253 2,419 1,747 1,570 555,599 4,239 7,009 11,856 22,239 86,128 917 1,129 1,775 11,709

1,052,455

1,006,866

7,990

4,862

128

195

69

2,736

24,037 30,149 34,306 22,459 17,171 14,824 13,574 11,437 10,811 11,621 14,240 14,173 12,286 7,128 3,378 2,167

13,562 10,411 11,673 10,132 11,879 10,695 9,763 6,811 5,204 4,366 4,017 3,306 2,512 1,333 789 636

15 - 19

2,683,634

2,541,416

20 - 24

2,652,654

2,005,733

25 - 29

2,631,408

1,155,454

30 - 34

2,671,164

591,803

35 - 39

2,563,502

325,427

40 - 44

2,280,006

177,809

45 - 49

1,807,874

92,219

50 - 54

1,381,080

47,363

55 - 59

1,097,886

29,005

60 - 64

925,205

19,382

65 - 69

702,801

12,468

70 - 74

486,407

7,165

75 - 79

272,134

3,697

80 - 84

137,046

1,811

over 85

89,384

1,481

From: statistic bureau, office of prime minister

300

age (year) and sex total 17,706,340 12,096 304,722 1,158,238 1,975,942 2,390,535 2,378,438 2,177,698 1,756,169 1,368,499 1,122,603 998,191 795,872 573,872 343,981 197,883 151,601 55,130 34,624 128,241 275,503 468,738 301,039 279,986 205,213 135,847 111,311 675,242 287,741 854,893 224,541 15,427 10,335 6,582 3,601 1,738 902 638 1,127,350 184,405 22,749 1,529,635 148,822 33,423 1,962,998 115,897 43,177 44,477 35,515 26,403 20,561 16,758 11,621 6,796 3,039 1,350 757 2,194,059 79,925 42,098 48,115 2,229,637 56,262 37,525 48,126 1,845,065 34,586 28,824 42,622 24,845 18,985 14,241 11,149 8,774 7,592 7,181 8,115 7,892 7,986 5,750 4,654 4,271 1,069,988 16,428 17,299 28,752 25,771 274,960 3,340 4,072 7,164 15,186 8,987 189 200 144 2,576 14,735,050 2,185,532 268,590 342,200 174,968 married status have been marry widow divorce separate no data be a monk 63,554 11,356 7,908 7,171 4,656 5,581 4,163 4,155 2,989 2,800 2,382 2,766 2,405 2,031 1,259 1,013 919

total

single

no data of marry status

female

24,698,256

6,928,362

13 - 14

1,007,898

984,446

15 - 19

2,603,399

2,290,769

20 - 24

2,609,487

1,444,078

25 - 29

2,776,388

795,790

30 - 34

2,854,780

458,664

35 - 39

2,695,855

313,254

40 - 44

2,405,038

223,185

45 - 49

1,911,948

152,790

50 - 54

1,467,079

95,780

55 - 59

1,186,391

61,406

60 - 64

1,045,482

44,525

65 - 69

827,220

28,943

70 - 74

593,696

17,793

75 - 79

354,148

8,908

80 - 84

203,504

4,608

over 85

155,943

3,423

From: statistic bureau, office of prime minister

301

Table B-8 birth rate death rate and increased rate by aria and region in 1985-1986, 1991 and 1995-1996 1985-1986 death rate birth rate
(per 1,000) (per 1,000) (per 1,000)

1991

1995-1996

area birth rate (per 1,000) 17.90 14.73 18.72

birth rate

(per 1,000)

death rate

death rate
(per 1,000)

increased rate by nature (per 100) increased rate by nature (per 100) 1.43 1.32 1.45 1.74 1.48 1.79 20.98 6.46 16.79 3.62 20.19 5.93

increased rate by nature (per 100) 6.02 4.14 6.50 1.19 1.06 1.22

country 6.44 4.22 6.86

23.87

in municipal limits

18.97

out of municipal limits Bangkok

24.79

18.92 5.76 7.23 6.82 7.03 1.81 22.76 6.33 1.44 17.83 6.70 1.67 17.63 5.55

3.84

1.51

14.65

3.31

1.13 1.21 1.11 1.64

14.24 15.61 14.83 19.96 24.00

3.92 5.78 6.97 6.12 6.69

1.03 0.98 0.79 1.38 1.73

middle region

22.43

northern region

21.60

north east region

24.87

south region

31.17

2.41 25.78 6.61 1.92 From : statistic bureau, office of prime minister

302

Table B-9 the estimate of population and rates in Thailand between 1990-2020 year population midyear (million) increased rate by nature fertilization rate birth rate (per 1000) death rate (per 1000) baby death rate

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

55.84 56.57 57.29 58.01 58.71 59.40 60.00 60.60 61.20 61.81 62.41 62.91 63.43 63.96 64.49 65.03 65.47 65.94 66.19 66.69 67.23 67.71 68.19 68.79 68.96 69.08 69.31 69.58 69.88 70.21 70.50

1.34 1.29 1.24 1.19 1.14 1.08 1.03 0.98 0.95 0.92 0.88 0.85 0.82 0.79 0.76 0.72 0.69 0.66 0.64 0.61 0.59 0.56 0.54 0.51 0.49 0.46 0.44 0.41 0.38 0.36 0.33

2.26 2.22 2.19 2.15 2.11 2.07 2.03 2.00 1.98 1.96 1.93 1.91 1.89 1.88 1.87 1.85 1.84 1.83 1.82 1.81 1.81 1.80 1.79 1.79 1.78 1.78 1.77 1.77 1.77 1.76 1.76

20.26 19.88 19.50 19.12 18.74 18.36 17.98 17.60 17.32 17.04 16.76 16.48 16.20 15.96 15.72 15.48 15.24 15.00 14.80 14.60 14.40 14.20 14.00 13.84 13.68 13.52 13.36 13.20 13.04 12.88 12.72

6.82 6.96 7.10 7.24 7.38 7.52 7.66 7.80 7.84 7.88 7.92 7.96 8.00 8.08 8.16 8.24 8.32 8.40 8.44 8.48 8.52 8.56 8.60 8.70 8.80 8.90 9.00 9.10 9.20 9.30 9.40

35.08 34.47 33.86 33.25 32.64 32.04 31.43 30.82 30.27 29.72 29.17 28.62 28.07 27.59 27.11 26.64 26.16 25.68 25.26 24.83 24.41 23.98 23.56 23.18 22.80 22.41 22.03 21.65 21.27 20.89 20.50

From : statistic bureau, office of prime minister

303

Table B-10 estimation of population by age and sex from 1990-2020

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Unit (1000)
2019 2020

age (year) 1990

1991

1992

1993

1994

1995

1996

total

55,839 56,574 57,294 58,010 58,713 59,401 60,003 60,602 61,201 61,806 62,405 62,914 63,430 63,959 64,492 65,034 65,470 65,938 66,190 66,690 67,230 67,712 68,194 68,789 68,955 69,076 69,314 69,584 69,878 70,213 70,503

male

27,941 28,299 28,650 28,999 29,341 29,675 29,963 30,245 30,531 30,819 31,105 31,348 31,597 31,854 32,110 32,370 32,557 32,772 32,997 33,213 33,422 33,701 33,983 34,385 34,350 34,328 34,445 34,580 34,721 34,893 35,018

0-4

2,788 2,787 2,778 2,764 2,749 2,733 2,730 2,712 2,687 2,658 2,628 2,615 2,597 2,579 2,560 2,542 2,523 2,503 2,485 2,466 2,448 2,465 2,512 2,714 2,515 2,362 2,353 2,342 2,329 2,330 2,277

5-9

2,906 2,872 2,843 2,816 2,789 2,762 2,740 2,727 2,719 2,713 2,703 2,685 2,666 2,645 2,623 2,601 2,566 2,555 2,550 2,539 2,519 2,508 2,493 2,474 2,452 2,429 2,411 2,395 2,378 2,362 2,345

10-14

2,978 2,970 2,957 2,940 2,919 2,895 2,868 2,837 2,805 2,775 2,751 2,733 2,722 2,714 2,705 2,692 2,676 2,656 2,634 2,611 2,591 2,579 2,566 2,551 2,532 2,510 2,493 2,476 2,457 2,439 2,421

15-19

2,936 2,950 2,960 2,966 2,967 2,963 2,955 2,943 2,926 2,906 2,881 2,854 2,823 2,791 2,762 2,739 2,722 2,711 2,703 2,695 2,682 2,671 2,654 2,632 2,607 2,582 2,564 2,548 2,533 2,518 2,501

20-24

2,793 2,823 2,851 2,874 2,895 2,911 2,928 2,939 2,945 2,945 2,941 2,934 2,923 2,907 2,886 2,862 2,835 2,805 2,773 2,745 2,722 2,711 2,704 2,696 2,684 2,666 2,650 2,630 2,608 2,586 2,566

25-29

2,567 2,605 2,643 2,683 2,720 2,753 2,782 2,809 2,834 2,857 2,875 2,894 2,907 2,914 2,916 2,914 2,907 2,896 2,881 2,861 2,838 2,817 2,790 2,759 2,728 2,699 2,682 2,672 2,665 2,656 2,644

30-34

2,311 2,360 2,402 2,440 2,477 2,514 2,545 2,578 2,613 2,648 2,680 2,714 2,748 2,780 2,807 2,830 2,852 2,867 2,876 2,879 2,877 2,878 2,871 2,856 2,834 2,804 2,778 2,748 2,717 2,690 2,668

35-39

1,948 2,023 2,092 2,155 2,212 2,264 2,305 2,339 2,367 2,395 2,425 2,458 2,494 2,533 2,572 2,609 2,647 2,685 2,721 2,753 2,779 2,808 2,829 2,839 2,840 2,833 2,829 2,819 2,804 2,786 2,763

40-44

1,507 1,578 1,658 1,744 1,828 1,907 1,978 2,041 2,098 2,148 2,193 2,232 2,262 2,289 2,315 2,345 2,378 2,415 2,455 2,496 2,535 2,584 2,630 2,670 2,702 2,724 2,749 2,767 2,778 2,783 2,783

45-49

1,260 1,288 1,320 1,360 1,408 1,466 1,534 1,611 1,693 1,773 1,846 1,914 1,976 2,030 2,078 2,120 2,155 2,183 2,206 2,230 2,258 2,301 2,346 2,390 2,432 2,469 2,514 2,557 2,597 2,633 2,662

50-54

1,138 1,154 1,166 1,177 1,191 1,210 1,237 1,268 1,305 1,351 1,406 1,472 1,547 1,625 1,702 1,771 1,837 1,895 1,946 1,990 2,029 2,071 2,102 2,128 2,152 2,179 2,217 2,259 2,303 2,347 2,390

55-59 868 612 391 405 415 427 439 452 400 416 439 469 484 466 636 664 698 742 760 776 504 482 893 925 954 974 968 976

944

977 1,006 1,031 1,052 1,070 1,085 1,097 1,107 1,120 1,139 1,165 1,194 1,229 1,273 1,325 1,387 1,458 1,533 1,605 1,670 1,737 1,796 1,847 1,889 1,924 1,960 1,990 2,015 2,041 2,072 995 1,018 1,040 1,044 1,071 1,111 1,160 1,213 1,250 1,317 1,397 1,474 1,537 1,566 1,616 1,676 1,733 1,777 793 530 500 815 560 518 847 596 539 846 616 566 843 636 593 849 654 620 868 669 646 908 683 673 915 703 703 923 718 732 946 727 759 992 1,065 1,110 1,158 1,211 1,275 1,357 733 784 739 807 742 827 755 848 780 870 820 894 873 919

60-64

682

701

737

780

823

858

65-69

486

494

502

517

545

588

70-74

326

343

357

369

377

386

over75

371

374

378

383

389

396

From: statistic bureau, office of prime minister

304

Table B-11 estimation of population by age and sex from 1990-2020

1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Unit (1000)
2019 2020

age (year) 1990

1991

1992

1993

1994

1995

1996

total

55,839 56,574 57,294 58,010 58,713 59,401 60,003 60,602 61,201 61,806 62,405 62,914 63,430 63,959 64,492 65,034 65,470 65,938 66,190 66,690 67,230 67,712 68,194 68,789 68,955 69,076 69,314 69,584 69,878 70,213 70,503

female 27,898 28,275 28,644 29,011 29,372 29,726 30,040 30,357 30,670 30,987 31,301 31,566 31,833 32,105 32,382 32,664 32,913 33,166 33,193 33,477 33,807 34,011 34,211 34,404 34,605 34,749 34,869 35,004 35,157 35,320 35,486

0-4

2,709 2,709 2,702 2,692 2,679 2,664 2,660 2,643 2,618 2,588 2,557 2,540 2,519 2,499 2,482 2,470 2,474 2,481 2,259 2,293 2,376 2,366 2,355 2,344 2,346 2,291 2,279 2,265 2,250 2,233 2,207

5-9

2,809 2,780 2,756 2,734 2,711 2,688 2,670 2,660 2,654 2,649 2,640 2,622 2,603 2,581 2,558 2,535 2,517 2,499 2,483 2,468 2,452 2,435 2,417 2,398 2,379 2,361 2,344 2,327 2,310 2,294 2,278

10-14

2,872 2,866 2,855 2,841 2,823 2,802 2,779 2,753 2,726 2,701 2,681 2,667 2,658 2,652 2,645 2,633 2,616 2,596 2,573 2,550 2,529 2,510 2,494 2,478 2,463 2,446 2,429 2,411 2,392 2,374 2,356

15-19

2,828 2,843 2,854 2,862 2,865 2,863 2,858 2,848 2,834 2,816 2,795 2,772 2,746 2,719 2,694 2,675 2,661 2,653 2,647 2,639 2,627 2,611 2,590 2,567 2,544 2,524 2,505 2,489 2,474 2,458 2,442

20-24

2,696 2,723 2,749 2,774 2,795 2,813 2,827 2,838 2,846 2,850 2,849 2,844 2,835 2,821 2,804 2,783 2,761 2,735 2,708 2,684 2,665 2,651 2,643 2,638 2,630 2,618 2,602 2,582 2,559 2,536 2,516

25-29

2,512 2,548 2,581 2,613 2,643 2,672 2,695 2,718 2,740 2,761 2,778 2,792 2,804 2,814 2,820 2,820 2,817 2,809 2,797 2,781 2,761 2,740 2,715 2,689 2,666 2,647 2,634 2,626 2,621 2,614 2,603

30-34

2,277 2,324 2,368 2,410 2,450 2,487 2,518 2,547 2,573 2,599 2,623 2,646 2,668 2,690 2,710 2,727 2,744 2,759 2,771 2,779 2,782 2,781 2,775 2,765 2,750 2,732 2,712 2,688 2,663 2,641 2,623

35-39

1,972 2,042 2,102 2,155 2,205 2,253 2,297 2,339 2,377 2,413 2,446 2,475 2,501 2,524 2,547 2,570 2,592 2,614 2,636 2,657 2,676 2,695 2,712 2,727 2,737 2,742 2,743 2,738 2,729 2,716 2,700

40-44

1,541 1,611 1,694 1,783 1,869 1,946 2,014 2,073 2,125 2,172 2,218 2,261 2,301 2,337 2,370 2,401 2,428 2,452 2,473 2,494 2,516 2,539 2,563 2,586 2,608 2,628 2,649 2,668 2,685 2,696 2,702

45-49

1,320 1,346 1,375 1,410 1,456 1,513 1,582 1,664 1,751 1,835 1,910 1,977 2,035 2,085 2,131 2,175 2,217 2,255 2,289 2,321 2,349 2,376 2,400 2,422 2,443 2,465 2,490 2,514 2,538 2,562 2,583

50-54

1,198 1,218 1,234 1,248 1,264 1,285 1,310 1,339 1,373 1,418 1,474 1,542 1,622 1,707 1,789 1,861 1,927 1,983 2,032 2,077 2,119 2,160 2,198 2,232 2,263 2,291 2,319 2,343 2,366 2,387 2,410

55-59 958 686 454 561 571 583 596 612 466 486 515 553 577 630 721 758 799 848 874 898 605 650

996 1,036 1,071 1,102 1,128 1,150 1,171 1,187 1,200 1,216 1,236 1,261 1,289 1,323 1,365 1,420 1,486 1,563 1,646 1,725 1,795 1,860 1,915 1,962 2,005 2,046 2,088 2,125 2,159 2,189 2,217 990 1,026 1,059 1,083 1,086 1,099 1,119 1,143 1,166 1,175 1,201 1,239 1,287 1,342 1,390 1,464 1,550 1,632 1,701 1,739 1,789 1,845 1,899 1,944 923 639 672 950 677 697 984 721 724 990 746 762 994 1,005 1,026 1,064 1,071 1,081 1,106 1,154 1,229 1,285 1,347 1,412 1,483 1,564 772 800 797 838 820 876 841 915 870 956 892 904 910 914 908 917 946 995 1,063 997 1,036 1,075 1,112 1,143 1,175 1,208 1,243 1,279

60-64

723

753

797

846

894

936

65-69

531

542

554

574

607

653

70-74

391

406

419

429

438

447

over 75

524

528

533

538

545

552

From: statistic bureau, office of prime minister

305

Table B-12 estimation of population at midyear by region from 1990-2010 1992

6,495 3,017 2,829 3,613 3,249 10,993 19,867 7,630 58,010 58,713 59,401 60,003 60,602 61,201 61,806 62,405 62,914 7,744 7,854 7,961 8,061 8,157 8,251 8,341 8,427 20,063 20,246 20,406 20,571 20,734 20,906 21,068 21,203 21,327 8,507 63,430 11,056 11,122 11,148 11,172 11,200 11,215 11,253 11,262 11,272 11,283 21,458 8,592 63,959 3,278 3,313 3,343 3,370 3,395 3,422 3,450 3,480 3,495 3,521 3,543 3,659 3,710 3,757 3,805 3,851 3,890 3,942 3,987 4,025 4,071 4,110 4,152 3,573 11,284 21,584 8,677 64,492 2,844 2,858 2,879 2,884 2,891 2,906 2,912 2,920 2,923 2,925 2,928 2,936 2,940 4,196 3,591 11,288 21,719 8,764 65,034 3,104 3,192 3,282 3,368 3,457 3,544 3,634 3,720 3,804 3,890 3,979 4,068 4,160 4,244 2,938 4,229 3,609 11,281 21,801 8,836 65,470 6,635 6,778 6,919 7,061 7,204 7,348 7,496 7,637 7,775 7,917 8,066 8,218 8,375 8,533

unit (1000) 2005 2006 2007

8,677 4,332 2,935 4,271 3,628 11,273 21,914 8,908 65,938

region 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

1990

1991

2008
8,776 4,410 2,933 4,297 3,638 11,268 21,917 8,951 66,190

2009
8,943 4,503 2,931 4,336 3,660 11,259 22,020 9,039 66,690

2010
9,122 4,592 2,927 4,382 3,677 11,250 22,157 9,123 67,230

Bangkok

6,198

6,349

Bangkok boundary area

2,850

2,931

middle region(middle area)

2,791

2,811

middle region(eastern area)

3,517

3,564

middle region(western area)

3,180

3,217

northern region

10,777

10,854

10,924

north eastern region

7,516

19,209

19,435

19,652

southern region total

From: statistic bureau, office of prime minister

7,315

7,413

55,839

56,574

57,294

306

Appendix C

The model formulae documentation listing and instruction

Note * Demographic sector

Total_population (t) = Total_population (t - dt) + (Changing_of__Population_from_1990_to_2006 - Flow_out__of_last_year) * dt INIT Total_population = 4000000

INFLOWS: Changing_of__Population_from_1990_to_2006 = 4E+06*2.7182818284590452353603^(0.033*Year_of_changing_rate_1977_is_th e1st_year)

OUTFLOWS: Flow_out__of_last_year = Total_population

Year_of_changing_rate_1977_is_the1st_year = GRAPH (TIME) (1977, 1.00), (1978, 2.00), (1979, 3.00), (1980, 4.00), (1981, 5.00), (1982, 6.00), (1983, 7.00), (1984, 8.00), (1985, 9.00), (1986, 10.0), (1987, 11.0), (1988, 12.0), (1989, 13.0), (1990, 14.0), (1991, 15.0), (1992, 16.0), (1993, 17.0), (1994, 18.0), (1995, 19.0), (1996, 20.0), (1997, 21.0), (1998, 22.0), (1999, 23.0), (2000, 24.0), (2001, 25.0), (2002, 26.0), (2003, 27.0), (2004, 28.0), (2005, 29.0), (2006, 30.0), (2007, 31.0), (2008, 32.0), (2009, 33.0), (2010, 34.0), (2011, 35.0), (2012, 36.0), (2013, 37.0), (2014, 38.0), (2015, 39.0), (2016, 40.0), (2017, 41.0), (2018, 42.0), (2019, 43.0), (2020, 44.0), (2021, 45.0), (2022, 46.0), (2023, 47.0), (2024, 48.0), (2025, 49.0), (2026, 50.0), (2027, 51.0), (2028, 52.0), (2029, 53.0), (2030, 54.0),

307

(2031, 55.0), (2032, 56.0), (2033, 57.0), (2034, 58.0), (2035, 59.0), (2036, 60.0), (2037, 61.0), (2038, 62.0), (2039, 63.0), (2040, 64.0), (2041, 65.0), (2042, 66.0), (2043, 67.0), (2044, 68.0), (2045, 69.0), (2046, 70.0), (2047, 71.0), (2048, 72.0), (2049, 73.0), (2050, 74.0), (2051, 75.0), (2052, 76.0), (2053, 77.0), (2054, 78.0), (2055, 79.0), (2056, 80.0), (2057, 81.0), (2058, 82.0), (2059, 83.0), (2060, 84.0), (2061, 85.0), (2062, 86.0), (2063, 87.0), (2064, 88.0), (2065, 89.0), (2066, 90.0), (2067, 91.0), (2068, 92.0), (2069, 93.0), (2070, 94.0), (2071, 95.0), (2072, 96.0), (2073, 97.0), (2074, 98.0), (2075, 99.0), (2076, 100)

Adolescent_from_0_to_19_from_equation_trend (t) = Adolescent_from_0_to_19_from_equation_trend (t - dt) + (Adolescent_in_flow Adolescent_out_flow) * dt INIT Adolescent_from_0_to_19_from_equation_trend = 794407

INFLOWS: Adolescent_in_flow = Total_population * ((-0.5676*Year_of_age_group_equation__2001_is_the1st_year_+30.356)/100)

OUTFLOWS: Adolescent_out_flow = Adolescent_from_0_to_19_from_equation_trend

Mature_from_20_to_59_from_equation_trend (t) = Mature_from_20_to_59_from_equation_trend (t - dt) + (Mature_in_flow Mature_out_flow) * dt INIT Mature_from_20_to_59_from_equation_trend = 1467125

308

INFLOWS: Mature_in_flow = Total_population * ((0.3981*Year_of_age_group_equation__2001_is_the1st_year_+56.062)/100)

OUTFLOWS: Mature_out_flow = Mature_from_20_to_59_from_equation_trend

Elder_from_60_and_over_from_equation_trend (t) = Elder_from_60_and_over_from_equation_trend (t - dt) + (Elder_in_flow Elder_out_flow) * dt INIT Elder_from_60_and_over_from_equation_trend = 229515

INFLOWS: Elder_in_flow = Total_population * ((0.2547*Year_of_age_group_equation__2001_is_the1st_year_+8.7703)/100)

OUTFLOWS: Elder_out_flow = Elder_from_60_and_over_from_equation_trend

Year_of_age_group_equation__2001_is_the1st_year_ = GRAPH (TIME) (1977, -23.0), (1978, -22.0), (1979, -21.0), (1980, -20.0), (1981, -19.0), (1982, 18.0), (1983, -17.0), (1984, -16.0), (1985, -15.0), (1986, -14.0), (1987, -13.0), (1988, -12.0), (1989, -11.0), (1990, -10.0), (1991, -9.00), (1992, -8.00), (1993, 7.00), (1994, -6.00), (1995, -5.00), (1996, -4.00), (1997, -3.00), (1998, -2.00), (1999, -1.00), (2000, 0.00), (2001, 1.00), (2002, 2.00), (2003, 3.00), (2004, 4.00), (2005, 5.00), (2006, 6.00), (2007, 7.00), (2008, 8.00), (2009, 9.00), (2010, 10.0),

309

(2011, 11.0), (2012, 12.0), (2013, 13.0), (2014, 14.0), (2015, 15.0), (2016, 16.0), (2017, 17.0), (2018, 18.0), (2019, 19.0), (2020, 20.0), (2021, 21.0), (2022, 22.0), (2023, 23.0), (2024, 24.0), (2025, 25.0), (2026, 26.0), (2027, 27.0), (2028, 28.0), (2029, 29.0), (2030, 30.0), (2031, 31.0), (2032, 32.0), (2033, 33.0), (2034, 34.0), (2035, 35.0), (2036, 36.0), (2037, 37.0), (2038, 38.0), (2039, 39.0), (2040, 40.0), (2041, 41.0), (2042, 42.0), (2043, 43.0), (2044, 44.0), (2045, 45.0), (2046, 46.0), (2047, 47.0), (2048, 48.0), (2049, 49.0), (2050, 50.0), (2051, 51.0), (2052, 52.0), (2053, 53.0), (2054, 54.0), (2055, 55.0), (2056, 56.0), (2057, 57.0), (2058, 58.0), (2059, 59.0), (2060, 60.0), (2061, 61.0), (2062, 62.0), (2063, 63.0), (2064, 64.0), (2065, 65.0), (2066, 66.0), (2067, 67.0), (2068, 68.0), (2069, 69.0), (2070, 70.0), (2071, 71.0), (2072, 72.0), (2073, 73.0), (2074, 74.0), (2075, 75.0), (2076, 76.0)

Note * Social sector

Splitting_of_single_head_household = ((Adolescent_from_0_to_19_from_equation_trend*0.25/100))/19 + ((Mature_from_20_to_59_from_equation_trend*2.49/100))/40 + ((Elder_from_60_and_over_from_equation_trend*1.53/100))/40

Single_head___household__formation = Splitting_of_single_head_household + (Divorce_rate/2)

Marriage_rate = Total_population ((1.68*(1/(year_of_marriage_&_divorce_equation^0.1486)))/100)

310

Divorce_rate = Marriage_number__in_couple_unit *2 *((23.031*year_of_marriage_&_divorce_equation^0.2263)/100)

year_of_marriage_&_divorce_equation = GRAPH (TIME) (1996, 1.00), (1997, 2.00), (1998, 3.00), (1999, 4.00), (2000, 5.00), (2001, 6.00), (2002, 7.00), (2003, 8.00), (2004, 9.00), (2005, 10.0), (2006, 11.0), (2007, 12.0), (2008, 13.0), (2009, 14.0), (2010, 15.0), (2011, 16.0), (2012, 17.0), (2013, 18.0), (2014, 19.0), (2015, 20.0), (2016, 21.0), (2017, 22.0), (2018, 23.0), (2019, 24.0), (2020, 25.0), (2021, 26.0)

Marriage_number__in_couple_unit = Marriage_rate/2

Single_head___household__formation = Splitting_of_single_head_household + (Divorce_rate/2)

Couple_head__household__formation = Marriage_number__in_couple_unit

sHS_needs(t) = sHS_needs(t - dt) + (Housing__needs__rate sHS__needs_rate_out_flow) * dt INIT sHS_needs = 60000

INFLOWS: Housing__needs__rate = Single_head___household__formation+Couple_head__household__formation OUTFLOWS: sHS__needs_rate_out_flow = sHS_needs

311

The_rest_of_sHS_needs_each_year = sHS_needs-sHS_demand

312

Note * Economic sector

GDP_growth_index = GRAPH (TIME) (1981, 6.30), (1982, 4.10), (1983, 7.30), (1984, 7.10), (1985, 3.50), (1986, 4.90), (1987, 9.50), (1988, 13.2), (1989, 12.0), (1990, 11.2), (1991, 8.60), (1992, 8.10), (1993, 8.30), (1994, 9.00), (1995, 9.20), (1996, 5.90), (1997, -1.40), (1998, -10.5), (1999, 4.50), (2000, 4.60), (2001, 1.90), (2002, 5.40), (2003, 6.90), (2004, 6.20), (2005, 4.50), (2006, 5.10), (2007, 4.50), (2008, -10.5), (2009, 4.50), (2010, 4.60), (2011, 1.90), (2012, 5.40), (2013, 6.90), (2014, 6.20), (2015, 4.50), (2016, 5.10), (2017, 4.50), (2018, -10.5), (2019, 4.50), (2020, 4.60), (2021, 1.90), (2022, 5.40), (2023, 6.90), (2024, 6.20), (2025, 4.50), (2026, 5.10)

CO_of_GDP_on_Affordability__&__Investment = GRAPH(GDP_growth_index) (0.00, 0.12), (2.00, 0.6), (4.00, 1.50), (6.00, 3.50), (8.00, 6.55), (10.0, 8.63), (12.0, 9.82), (14.0, 10.5), (16.0, 11.1), (18.0, 11.9), (20.0, 12.5)

sHS_demand(t) = sHS_demand(t - dt) + (Housing__demand__rate sHS__demand_rate_out_flow) * dt INIT sHS_demand = 60000

INFLOWS: Housing__demand__rate = (sHS_needs+The_rest_of_sHS_needs_each_year) *CO_of_GDP_on_Affordability__&__Investment

313

OUTFLOWS: sHS__demand_rate_out_flow = sHS_demand

the_rest_of_sHS_demand_each_year = sHS_demand-sHS_sales

CO_of_GDP_on___Consumption__Decision = GRAPH(GDP_growth_index) (0.00, 0.075), (2.00, 0.09), (4.00, 0.15), (6.00, 0.24), (8.00, 0.45), (10.0, 0.675), (12.0, 0.9), (14.0, 1.11), (16.0, 1.26), (18.0, 1.35), (20.0, 1.41)

Housing_loan = GRAPH (TIME) (1978, 29000), (1979, 36000), (1980, 41000), (1981, 45000), (1982, 40000), (1983, 35000), (1984, 30000), (1985, 35000), (1986, 35000), (1987, 30000), (1988, 40110), (1989, 53736), (1990, 59368), (1991, 78306), (1992, 111346), (1993, 147829), (1994, 209811), (1995, 223408), (1996, 242000), (1997, 202720), (1998, 103733), (1999, 64301), (2000, 108886), (2001, 112611), (2002, 164851), (2003, 296661), (2004, 296025), (2005, 297407), (2006, 279392), (2007, 262638), (2008, 103733), (2009, 64301), (2010, 108886), (2011, 112611), (2012, 164851), (2013, 296661), (2014, 296025), (2015, 297407), (2016, 297407), (2017, 297407), (2018, 103733), (2019, 64301), (2020, 108886), (2021, 112611), (2022, 164851), (2023, 296661), (2024, 296025), (2025, 297407), (2026, 297407)

housing_price_index = GRAPH (TIME) (1977, 53.9), (1978, 54.2), (1979, 56.2), (1980, 63.4), (1981, 68.2), (1982, 78.5), (1983, 83.3), (1984, 88.9), (1985, 94.6), (1986, 100), (1987, 107), (1988, 112), (1989, 117), (1990, 121), (1991, 124), (1992, 124), (1993, 126), (1994, 127),

314

(1995, 129), (1996, 131), (1997, 132), (1998, 133), (1999, 135), (2000, 131), (2001, 131), (2002, 132), (2003, 136), (2004, 143), (2005, 156), (2006, 142), (2007, 143), (2008, 144), (2009, 144), (2010, 145), (2011, 146), (2012, 147), (2013, 147), (2014, 148), (2015, 149), (2016, 149), (2017, 150), (2018, 150), (2019, 151), (2020, 152), (2021, 152), (2022, 153), (2023, 153), (2024, 154), (2025, 154), (2026, 155)

Fraction_of__housing_loan_over_housing_price_index = Housing_loan/housing_price_index

Co_of_Loan_per__Price_Index_on_Consumption_Decision_in_BMR = GRAPH(Fraction_of__housing_loan_over_housing_price_index) (0.00, 0.931), (300, 1.13), (600, 1.31), (900, 1.43), (1200, 1.53), (1500, 1.61), (1800, 1.69), (2100, 1.76), (2400, 1.80), (2700, 1.83), (3000, 1.88)

sHS_sales(t) = sHS_sales(t - dt) + (Housing_buying__rate sHS__sales_rate__out_flow) * dt INIT sHS_sales = 60000

INFLOWS: Housing_buying__rate = DELAY((sHS_demand+the_rest_of_sHS_demand_each_year) *CO_of_GDP_on___Consumption__Decision *Co_of_Loan_per__Price_Index_on_Consumption_Decision_in_BMR,4/12)

315

OUTFLOWS: sHS__sales_rate__out_flow = sHS_sales

316

Note * Validation sector

Accumulated_sHS_Vacancy(t) = Accumulated_sHS_Vacancy(t - dt) + (Housing__vacancy__rate) * dt INIT Accumulated_sHS_Vacancy = 0

INFLOWS: Housing__vacancy__rate = sHS_sales - sHS_needs

Housing_sales = GRAPH (TIME) (1977, 63163), (1978, 48119), (1979, 63775), (1980, 63652), (1981, 45892), (1982, 41881), (1983, 26297), (1984, 50638), (1985, 50086), (1986, 33722), (1987, 53353), (1988, 67451), (1989, 80031), (1990, 102335), (1991, 129688), (1992, 108001), (1993, 134086), (1994, 171254), (1995, 172419), (1996, 166785), (1997, 145355), (1998, 63864), (1999, 33382), (2000, 32028), (2001, 34023), (2002, 37182), (2003, 56085), (2004, 69050), (2005, 70072), (2006, 75000)

Coefficient_of__HS_Sale_on_HS_Supply = GRAPH(Housing_sales) (0.00, 0.2), (20000, 0.3), (40000, 0.4), (60000, 0.51), (80000, 0.61), (100000, 0.75), (120000, 1.49), (140000, 1.75), (160000, 1.88), (180000, 1.93), (200000, 1.97)

sHS_supply(t) = sHS_supply(t - dt) + (Housing_supply_rate sHS__supply_rate__out_flow) * dt INIT sHS_supply = 60000

317

INFLOWS: Housing_supply_rate = sHS_sales* Coefficient_of__HS_Sale_on_HS_Supply

OUTFLOWS: sHS__supply_rate__out_flow = sHS_supply

318

Appendix D

Mathematical prove for the evidence supported the principle in Chapter 7

A feature of linear trend line -which can equally divide the data to its above-area and under-area- is unintentionally found in this research process. Since it is not the aim of this research, all contents of this issue is described in this appendix. Purpose of explanation in this appendix is to make a smooth link of contents in conclusion and suggestion principle about housing consumption in Chapter 7.

To prove this pre-mentioned feature, this research adopts a mathematical function (i.e. linear trend-line) to make an equilibrium line. The equilibrium line -in this research- means the line that housing sales equal to housing needs in a market. The evidence for proving that the linear function calculates and plots the line to divide the graphs area into two parts (i.e. upper part and lower part) equally is explained as follows.

The principle to calculate the linear function line is that the linear function (by itself) tries to minimize all of distances or R2 (error in statistical terminology) between itself and the data plotting on the graph. As a result, the summation value of distances between every data points above the linear line and the linear line (i.e. Blue or dashed line) is equal to the summation value of distances between every data point below the linear line and the linear line (i.e. red line). The detail is described in the figure below.

319

Data of mathematical prove of linear funtion 10 9 8 7 6 Y unit 5 4 3 2 1 0 0 2 4 X unit 6 8 10 Data Linear (Data)

Figure in appendix D-1: Mathematical prove of linear function for distance issue

Figure D-2 shows that linear line divides the data into two groups. The next Figure shows further step to application of the linear line regards to its ability to equally divide the summation of both groups area simultaneously. This principle is true only the X- intervals between points are equal. This principle will be proven mathematically later.

320

Data of mathematical prove of linear funtion 10 9 8 7 6 Y unit 5 4 3 2 1 0 0 1 2 3 4 5 X unit 6 7 8 9 10 Data Linear (Data)

Figure in appendix D-2: Mathematical prove of linear function for area issue

Therefore the summation of the area above the linear line and the summation of the area under the linear line are also be divided equally. In Figure D-2 the dark blue represents the area above-linear line and the dark red represents the area under-linear line. The principles about triangle area calculation which used to calculate the summation of those areas are not explained here.

Data of mathematical prove of linear funtion 10 9 8 7 6 Y unit 5 4 3 2 1 0 0 1 2 3 4 5 X unit 6 7 8 9 10 Data Linear (Data)

Figure in appendix D-3: Mathematical prove of linear function for triangle area issue

321

The mathematical equation to prove this principle is provided below.

Min ei2 ei # 0

Equation in appendix D-1

If ei # 0, And Intervals of independent variables are equal ei+ - ei ei+ = ei ei+ * x - ei- * x A+ = Athen, above area = below area
= 0

Where ei x ei+ eiA+

= error (difference between observed value and the linear line) = intervals of X-axis in graphs = error above linear line = error below linear line = area above linear line = area below linear line

A-

From this logical reason, the principle is applied to develop the housing consumption principle mentioned in Chapter 7. The application used for housing demand field is shown in Figure D-4 below. The linear line and the underlyinghousing-need line are compared and calculated to find out the surplus result. This number is deduced to be the number of vacant house left in a housing market. All details are described in Chapter 7.

322

y = 936.66x + 62202 Housing sale data and linear trend 2 R = 0.0334 compare with underlying housing needs number 200,000 180,000 160,000 140,000 housing unit 120,000 100,000 80,000 60,000 40,000 20,000 0 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 year Housing sale Linear (Housing sale) Simulated housing needs number

Source: simulation program

Figure in appendix D-4: Housing sales data and the linear trend line compare with the underlying housing-need line

The above-area number and below-area number are displayed in Table in appendix D-1 below. The table shows similarity of the summation of above-area number and below-area number with a small missing number (i.e. 9 units). Thus, this line is claimed that it is the line represented equilibrium of housing sales data as it divides two accumulated areas (above and below) equally.

The minus numbers in the shaded cells represent the distance between the linear equation line and the housing sales line for the lower area. The plus numbers represent the distance between the linear equation line and the housing sales line for the upper area.

323

Table in appendix D-1: Comparison between linear line value and housing sales data
*Distance of linear line and housing sales data to prove theory of linear regression equation (housing unit) 24 -15,956 -1,237 -2,297 -20,993 -25,941 -42,462 -19,057 -20,546 -37,847 -19,152 -5,991 5,652 27,020 53,436 30,812 55,961 92,192 92,420 95,681 66,759 -15,755 -54,716 -51,608 -51,639 -49,373 -31,407 -19,378 -19,293 -15,302 9

year

linear line value

housing sales data (housing unit) 63,163 48,119 63,775 63,652 45,892 41,881 26,297 50,638 50,086 33,722 53,353 67,451 80,031 102,335 129,688 108,001 134,086 171,254 172,419 176,616 148,631 67,054 29,029 33,074 33,980 37,182 56,085 69,050 70,072 75,000 2,301,616

1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 total

63,139 64,075 65,012 65,949 66,885 67,822 68,759 69,695 70,632 71,569 72,505 73,442 74,379 75,315 76,252 77,189 78,125 79,062 79,999 80,935 81,872 82,809 83,745 84,682 85,619 86,555 87,492 88,428 89,365 90,302 -

* Minus sign (-) is the deficit number between housing sales data and the linear line value

The equilibrium of housing sales line -in this research- means the accumulated number of housing sales is equal to the accumulated number of housing needs in a housing market. This assumption brings the idea of housing demand modelling to make an appropriate supply control. Fluctuations of housing supply can seriously harm economy in holistic view. Therefore the suitable number of housing

324

production against housing needs will lead to an efficient market cycle between demand and supply.