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
B
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
29 th September 2008 C
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 lo Ze. D
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 E
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 Thailand’s Emerging Economic Policies and Strategies ...... 3-95 F
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 G
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 H
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 I
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
J
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 K
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 L
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 M
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 N
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 O
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 P
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 Q
Equation 5-3……………..…..………………………………………………5-209 Equation 6-1……………..…..………………………………………………6-223 Equation 6-2……………..…..………………………………………………6-224 Equation in appendix D-1……………..…..………….……………………...... 321 R
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 country’s 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 provinces- forms the case study in this research. The importance of the real estate sector was S 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.
T
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 1996–1997.
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 housing- demand 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 Thailand’s 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:
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 housing- demand 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 researcher’s 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:
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 housing- demand 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 self-
contained 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 privacy…the 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 real- world 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 housing- demand 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:
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.
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.
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
P N = Equation 1-1 n * h
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.
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)
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 Change household Age Group per age group head (thousand) (thousand)
Total 1,949.8 0-14 3.7 3.7 15-19 21.1 17.4 20-24 43.1 22.0 25-29 62.7 19.6 30-34 104.1 41.4 35-39 142.9 38.8 40-44 217.5 74.6 45-49 230.7 13.2 50-54 231.1 0.4 55-59 210.0 -21.1 60-69 382.2 172.3 70-79 228.1 -154.1 80-89 65.4 -162.7 1 90 7.4 -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 housing- demand 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 people’s 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 model’s 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 (R 2) 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 model- simulation 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:
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
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.
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.
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:
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 single- head household formation—calculated 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 2-
Literature 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 3-
Background 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 4-
Overview 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 5-
Simulation 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 6-
Model 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 model’s 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 7-
Conclusions 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)
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)
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)
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
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)
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.
‘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.
‘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.
‘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.
‘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 people’s financial ability to buy a house
or houses.
‘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.
‘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.
‘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.
‘Housing needs’ is the term used to refer to needs of people to have a
house to live in.
‘Housing sales unit (HSU)’ is the term used to refer to number of
houses that have been sold to persons in housing demand unit.
‘Housing unit ’ is the term used to refer to all types of housing, counted
as the same unit.
‘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.
‘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.
‘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.
‘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 km², and with a registered population of
9,988,400 (as of April 4, 2007) it has a population density of 1,286.92
per km².
‘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 sector’s 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, P h, P o, T), Equation 2-1
Where,
Q = housing consumption
Y = household income
Ph = the relative price of housing
Po = a vector of price of other goods and services
T = a vector of taste factors
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 T = t (H), Equation 2-2
Where,
T = a vector of taste factors
H = a vector of household characteristics
Then Q = q (Y, P h, P o, H), Equation 2-3
2-45
Where,
Q = housing” consumption
Y = household income
Ph = the relative price of housing
Po = a vector of price of other goods and services
H = 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
Q o = qo (Z o) + u o Equation 2-5
Q r = qr (Z r) + u r Equation 2-6
The tenure choice equations above, describes:
The unobservable index, I, the household’s likelihood of owning versus
renting, as a function of the independent variables, X.
The owner’s demand for housing services, Q o is function of variable Z o. 2-47
The renter’s demand for housing services, Q r is function of the variable
Zr.
The U’s 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 housing- service 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 housing- demand 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 research’s model. However, renter’s 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 neoclassical- economic 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 model’s 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 housing- demand 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.
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 Kong’s 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 Thailand’s 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, cross- sectional 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 $ " 0 # "1 pop # " 2 social # " 3econ # !. Equation 2-7
Where,
HD = Housing demand
" 0 = Constant value
"1 = Coefficient of population factors
" 2 = Coefficient of social factors
" 3 = Coefficient of economic factors
! = Error value
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 model’s methodology, using System Dynamics processes, is as follows:
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 11 th 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+bX 1+ cX 2+dX 3+ e Equation 2-8
Where, 2-61
Y = dependent variable
X1-3 = independent variable
A = constant value b, c and d = coefficient e = error value
Then, this case can be set all factors influent to the quantity of orange as the equation below:
O = a(T a)+b(T y )+c(R)+d(B)+e Equation 2-9
(In this case) where,
O = number of orange
Ta = number of all orange trees
Ty = number of young (mature) orange trees
R = quantity of rain
B = number of bees a, b and c = coefficient e = error value
From the equation above, it can be found that there are some problems in this equation regarding the following points:
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 research’s 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
Give empirical content to Generate understanding of causes economic theory and effects creating a problem Purpose Empirically verify economic Identify solutions, in the form of theory policies, to problems.
Reductionist perspective; focus on Holistic perspective; include all a small number of variables and factors that have significant impact keep the remaining variables on how the problem develops over constant time
Long-term (relatively speaking); Short-term perspective in study both short- and long-term forecasting consequences
Scope Closed system view (few or no exogenous variables); both source Often disconnected from time of problem and solution are found (in the case of equilibrium models) within the modelled system itself
Aggregated system view; details Open system view are removed through averaging, (many exogenous variables) and sequences of events are grouped into continuous flows High level of aggregation
Build on economic theory about Behaviour can be reproduced from the nature (e.g., linear) of given initial state and system structure, economic relationships without the need for exogenous input. Results are based on correlations Assumptions identified within data. 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 2-65
Quantitative prediction Qualitative (as opposed to quantitative); patterns of behaviour Determination of the equilibrium rather than predictions, forecasts Results state or prognosis
Main focus is on the product Focus is on the process (learning) (i.e. estimates and hypotheses rather than on the product tests). (model/results). Source: conclude from Myrtveit( 2005)
Housing-demand study dilemma
In all markets -including housing- “demand” is an economic term quantifying people’s 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 study’s 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 model’s
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 (R 2), 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 model’s 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 model’s 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 researcher’s 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 demand’s 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 research’s 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. Forrester’s 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).
Forrester’s 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:
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 world’s 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 world’s 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-to- understand 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:
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 research’s 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:
Income data - people’s average income at the year ( \)
Expense data – people’s average expenses at the year ( \)
Savings data - people’s average savings at the year or income -
expense ( \)
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 research’s model; however, the model’s 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 model’s 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). Thailand’s 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 km² and the registered population in 2007 was 10,061,726. It therefore has, a population density of 1,296.36 per km².
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 km² (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 37. Phayao 2. Buriram 38. Phetchabun 3. Chachoengsao 39. Phetchaburi 4. Chai Nat 40. Phichit 5. Chaiyaphum 41. Phitsanulok 6. Chanthaburi 42. Phra Nakhon Si Ayuthaya 7. Chang Mai 43. Phrae 8. Chang Rai 44. Phuket 9. Chon Buri 45. Prachin Buri 10. Chumphon 46. Prachuap Khiri Khan 11. Kalasin 47. Ranong 12. Kamphaeng Phet 48. Ratchaburi 13. Kanchanaburi 49. Rayong 14. Khon Kaen 50. Roi Et 15. Krabi 51. Sakon Nakhon 16. Krung Thep Mahanakhon (Bangkok) 52. Samut Prakan 17. Lampang 53. Samut Sakhon 18. Lampang 54. Samut Songkhram 19. Loei 55. Sara Buri 20. Lop Buri 56. Satun 21. Mae Hong Son 57. Sing Buri 22. Maha Sarakham 58. Sisaket 23. Nakhon Nayok 59. Songkhia 24. Nakhon Pathom 60. Sukhothai 25. Nakhon Phanom 61. Suphan Buri 26. Nakhon Ratchasima 62. Surat Thani 27. Nakhon Sawan 63. Surin 28. Nakhin Si Thammarat 64. Tak 29. Nan 65. Trang 30. Narathiwat 66. Trat 31. Nong Khai 67. Ubon Ratchatani 32. Nonthaburi 68. Udon Thani 33. Pathum Thani 69. Uthai Thani 34. Pattani 70. Uttaradit 35. Phangnga 71. Yala 36. Phatthalung 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.
Thailand’s Emerging Economic Policies and Strategies
The economy and politics of Thailand from the 20 th to 21 st 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 1996-
1997 as follows:
Prior to 1997, Thailand had been one of the world’s 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.
Thailand’s 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 4 th 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 7 th 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 Ratio between Bangkok Metropolitan (with out Bangkok housing loans Region Year Metropolitan Region) in Bangkok Metropolitan Housing Rate of Rate of Housing loans Region and all loans growth growth (M Baht) other region (M Baht) (%) (%) (%) 1995 94,719 - 44,825 - 67.9 1996 122,337 29.2 73,439 63.8 62.5 1997 165,963 35.7 109,841 49.6 60.2 1998 175,296 5.6 118,715 8.1 59.6 1999 164,706 -6.0 115,450 -2.8 58.8 2000 161,871 -1.7 112,892 -2.2 58.9 2001 162,115 0.2 111,365 -1.4 59.3 2002 167,465 3.3 126,656 13.7 56.9 3-99
2003 184,737 10.3 147,428 16.4 55.6 2004 214,132 15.9 178,907 21.4 54.5 average 59.4 Source: Government housing Bank (2005)
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 industry’s 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 Region’s 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 vis-
a-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
0.00
1 5 7 9 7 9 1 8 8 8 8 9 9 0 05 9 9 9 9 9 0 growth growth rate 19 1983 1 1 1 1991 1993 1995 1 1 20 2003 2 -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. 3-104
vacant house unit
400000
350000
300000
250000
200000
150000
vacanthouse unit 100000
50000
0 1996 1997 1998 1999 2000 2001 2002 2003 year
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 market’s 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. 1977 The period from 1984 to 1985 was when the first condominiums were 1978 completed and sold.
1979
1980 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 1981 1987 and 72% in 1989.
1982 From 1988 to 1989, the condominium market expanded again.18,776 units 1983 were completed across Thailand. Half were built in the BMR. In the meantime, an increasing number of low price condominiums were constructed 1984 in the BMR.
1985 In 1989, 80,031 housing units were completed - an all-time record for the 1986 BMR. There was transformation in financing, with expanded capital loans from 1987 Bangkok International Banking Facilities.
1988 By 1991, land prices had increased to 21.5 times the price in 1985. 1989 In 1993, 23 % of people who bought houses or units and 31 % of people who 1990 bought condominiums at the Housing and Condominium Exhibition did so for
1991 investment and speculation.
1992 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 1993 investment and speculation.
1994
1995 In 1995 there were about 300,000 units vacant in the Bangkok Metropolitan Region (rose to 350,000 units in 1998). 1996
1997 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. 1998 In 1997 there was an economic crisis stemming from devaluation of the Thai 1999 Baht. Banks stopped giving loans for housing projects and faced non performing loan problem. There was shrinkage of housing demand consequent 2000 of the economic problems. The rest of South East Asia was also dramatically 2001 affected.
2002 From 2001 to 2003, the government subsidised people buying houses by 2003 reducing interest rates and fees and taxes on housing purchase. 2004 From 2004 to 2006, the political situation was unstable. The government was 2005 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 2006 election. year Figure 3-8 : time line of housing market in the Bangkok Metropolitan Region 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 1988 1989 year
Housing project
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
100000 unit 72,072 67,054 69,050 56,085 50000 33,880 37,182 29,029 33,074
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
142,023 150000 113,150 100000 64,909 41,300 50000 16,000 1,071 1,357 3,040 6,157 0 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 year
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) Units People (%) <500,000 1,432 6.8 500,001 - 1,000,000 4,797 22.6 1,000,001 -1,500,000 3,855 18.2 About 71 % is the result from the price between 1,500,001 - 2,000,000 3,163 14.9 500,000 to 3,000,000 2,000,001 -3,000,000 3,294 15.5 3,000,001 -4,000,000 1,618 7.6 4,000,001 - 5,000,000 1,042 4.9 5,000,001 - 7,000,000 1,014 4.8 >7,000,000 418 2 Non price expected 556 2.6 Total 21,189 100
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% detached house 45% condominium 30%
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 67.2 64.6 60.2 61.7 64.1 60
50 46.2 42.3 40 40
29.3 35.8 31.2 30 34.2 27.8 23.1 22.3 25.4 18.9 19.2 20 20.2 17.7 14.9 13.2 12.2 8 15.1 10 8.2 13.7 13.8 0 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 year
house block 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% ) 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 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 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 Ratio between Bangkok (with out Bangkok housing loan in Metropolitan Region Year Metropolitan Region) Bangkok Rate Metropolitan Housing Rate of Housing loan Region and all loan of growth (M Baht) growth other region (M Baht) (%) (%) (%) 1995 94,719 - 44,825 - 67.9 1996 122,337 29.2 73,439 63.8 62.5 1997 165,963 35.7 109,841 49.6 60.2 1998 175,296 5.6 118,715 8.1 59.6 1999 164,706 -6.0 115,450 -2.8 58.8 2000 161,871 -1.7 112,892 -2.2 58.9 2001 162,115 0.2 111,365 -1.4 59.3 2002 167,465 3.3 126,656 13.7 56.9 2003 184,737 10.3 147,428 16.4 55.6 2004 214,132 15.9 178,907 21.4 54.5 average 59.4 Source: Government housing Bank (2005)
Graph of new housing loan from 1997 to 2006 M Baht
350,000 296,661 294,403 300,000 279,392 241,172 262,638 250,000 223,408 202,720 200,000 164,851 150,000 112,611 103,733 108,886 100,000 64,301 50,000
0 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
year
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
Thailand’s 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 1991 1995-1996 Increase Increase Increase Area Birth Death rate Birth Death rate Birth Death rate rate rate rate rate rate rate (per 1,000) (per 1,000) by (per 1,000) (per 1,000) by (per 1,000) (per 1,000) by nature nature nature (per 100) (per 100) (per 100) Thailand 23.87 6.44 1.74 20.19 5.93 1.43 17.90 6.02 1.19 Bangkok Metropolitan 18.92 3.84 1.51 14.65 3.31 1.13 14.24 3.92 1.03 Region 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 region
1996 1997 1998 1999 2000 2001 2002
Thailand 3.7 3.7 3.7 3.6 3.6 3.5 3.4
Bangkok
Metropolitan 3.2 3.4 3.3 3.2 3.2 3.3 3.2
Region
Source: The National Statistical Office of Thailand 3-122
Summary
The Bangkok Metropolitan Region covers an area of 7,761.50 km² and has a registered population of 9,988,400 (as at April 4, 2007). Its population density is
1,286.92 per km², 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 Population Pop. Density Provinces km² in 2007 /km²
Bangkok 1,568.20 5,704,723 3,637.75
Nonthaburi 622.30 1,006,660 1,617.64
Samut Prakan 1,004.50 1,115,026 1,110.03
Pathum Thani 1,525.90 871,627 571.22
Samut Sakhon 872.30 465,450 533.59
Nakhon Pathom 2,168.30 824,447 380.44
Total 7,761.50 9,988,400 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.
The 1 st box represents the number of people in the total population
The 2 nd box represents the number of people in the mature
population
The 3 rd box represents the number of people in the mature population
who have housing needs 4-124
The 4 th 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 group’s 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:
An increase in GDP represents an average increase in people’s income.
A decrease in GDP represents an average decrease in people’s 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 model’s 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 unit’s 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 (R 2) at 0.9295, as shown in equations 4-1, 4-2 and figure 4-3 below.
y = 4E+06e 0.033x Equation 4-1
Or,
y = 4,000,000*e 0.033x Equation 4-2
4-136
Where, y = number of population x = number of year e = 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.
y = 4E+06e 0.033x population Population data in BMR R2 = 0.9295 12,000,000
10,000,000
8,000,000
6,000,000
4,000,000
2,000,000
0
80 82 84 86 88 90 92 94 96 98 1976 1978 19 19 19 19 19 19 19 19 19 19 2000 2002 2004 2006 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
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 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 ^ Simulation data 1976 n/a (4,000,000) 1977 n/a 4,134,202 1978 n/a 4,272,907 1979 n/a 4,416,265 1980 n/a 4,564,433 1981 n/a 4,717,572 1982 n/a 4,875,850 1983 n/a 5,039,437 1984 n/a 5,208,513 1985 n/a 5,383,261 1986 n/a 5,563,873 1987 n/a 5,750,543 1988 n/a 5,943,477 1989 n/a 6,142,884 1990 6,198,000 6,348,981 1991 6,349,000 6,561,993 1992 6,495,000 6,782,151 1993 6,635,000 7,009,696 1994 6,778,000 7,244,875 1995 6,919,000 7,487,945 1996 7,061,000 7,739,169 1997 7,204,000 7,998,823 1998 7,348,000 8,267,187 1999 7,496,000 8,544,556 2000 7,637,000 8,831,231 2001 8,737,000 9,127,523 2002 9,677,000 9,433,756 2003 9,812,000 9,750,264 2004 9,638,000 10,077,391 2005 9,792,000 10,415,492 2006 9,943,000 10,764,938 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 (R 2). 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 2002 2003 % Total % total % Total
(Total) 62,308,887 100 100 62,799,872 100 100 63,079,765 100 100 Age 0 - 4 4,207,497 6.75 4,042,604 6.44 3,902,496 6.19 5 - 9 4,813,614 7.73 ^29.79 4,821,676 7.68 ^29.21 4,754,166 7.54 ^28.66 10 - 14 4,614,429 7.41 (31.19) 4,700,093 7.48 (30.56) 4,759,358 7.54 (29.93) 15 - 19 4,928,170 7.91 4,780,924 7.61 4,661,217 7.39
Average of ^ 30.577349 % adolescent 20 - 24 5,341,014 8.57 5,282,247 8.41 5,208,192 8.26 25 - 29 5,513,728 8.85 5,479,986 8.73 5,406,128 8.57 30 - 34 5,655,524 9.08 5,663,969 9.02 5,563,479 8.82 35 - 39 5,341,771 8.57 ^56.47 5,413,362 8.62 ^56.84 5,505,556 8.73 ^57.26 40 - 44 4,590,483 7.37 (59.14) 4,739,965 7.55 (59.47) 4,889,452 7.75 (59.87) 45 - 49 3,781,378 6.07 3,930,215 6.26 4,060,211 6.44 50 - 54 2,874,844 4.61 3,039,054 4.84 3,202,542 5.08 55 - 59 2,085,444 3.35 2,148,128 3.42 2,286,141 3.62
Average ^ 59.496846 % of mature 60 - 64 1,815,101 2.91 1,872,448 2.98 1,855,785 2.94 65 - 69 1,491,784 2.39 1,523,598 2.43 1,580,908 2.51 70 - 74 1,027,704 1.65 1,106,138 1.76 1,145,133 1.82 75 - 79 622,331 1.00 657,745 1.05 715,638 1.13 ^9.01 ^9.31 ^9.52 338,959 0.54 354,636 0.56 380,257 0.60 80 - 84 (9.43) (9.74) (9.95) 85 - 89 170,554 0.27 179,575 0.29 190,802 0.30 90 - 94 73,441 0.12 74,995 0.12 71,214 0.11 95 - 99 33,649 0.05 33,839 0.05 30,612 0.05 over 100 40,869 0.07 42,715 0.07 34,784 0.06
Average ^ 9.7103632 % of elder Unknown 2,946,599 4.73 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 y = -0.5676x + 30.356 R2 = 0.9998 35
30
25
20
15
10
5 percentage of adolescent in of adolescent BMR percentage 0 2001 2002 2003 2004 2005 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 Equation 4-3
2 R = 0.9998
Where,
Y = percentage of adolescent population at time x = year (set 2001 as the 1 st year) 4-142
Mature proportion y = 0.3981x + 56.062 R2 = 0.9989 60 59 58 57 56 55 54 53 52 51 percentage of mature in percentage BMR 50 2001 2002 2003 2004 2005 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 Equation 4-4
2 R = 0.9989
Where,
Y = percentage of mature population at time x = year (set 2001 as the 1 st year) 4-143
Elder proportion y = 0.2547x + 8.7703 R2 = 0.9905 12
10
8
6
4
2 percentage of elder in percentage BMR 0 2001 2002 2003 2004 2005 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 Equation 4-5
2 R = 0.9905
Where,
Y = percentage of elderly population at time x = year (set 2001 as the 1 st year)
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:
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:
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 = number of single head household
NTH = number of total household
NCHH = 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
Marriage (couple) Total population in Total Group of head municipal area household ages household (thousand) (thousand) (thousand) Total 19,376.7 5,288.8 3,338.9 0-14 4,021.5 5.6 1.9 15-19 1,423.6 55.8 34.7 20-24 1,640.4 217.5 174.5 25-29 1,856.2 404.9 342.2 30-34 1,861.7 558.0 453.9 35-39 1,760.6 629.6 486.7 40-44 1,606.9 667.5 450.0 45-49 1,410.0 657.8 427.1 50-54 1,092.8 561.9 330.8 55-59 807.8 445.6 235.7 60-69 1,124.6 662.2 280.0 70-79 591.8 337.2 109.0 80-89 152.4 76.7 11.3 1 90 26.2 8.5 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 Group of head Order of ages household cohort (thousand) Total 1,949.8 1 0-14 3.7 2 15-19 21.1 3 20-24 43.1 4 25-29 62.7 5 30-34 104.1 6 35-39 142.9 7 40-44 217.5 8 45-49 230.7 9 50-54 231.1 10 55-59 210.0 11 60-69 382.2 12 70-79 228.1 13 80-89 65.4 14 1 90 7.4 Source: * 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 = Number of splitting of single head household
NSHH = Number of single head household
And,
NSHH = NSHH (2) - NSHH (1) Equation 4-8
Where,
NSHH = Number of single head household
NSHH (2) = Number of single head household of older cohort
NSHH (1) = 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 Number of
head growth (change) of Group of household the formation of ages Order of (thousand) single head cohort household Total 1,949.8 per group of ages (thousand) 1 0-14 3.7 3.7 2 15-19 21.1 17.4 3 20-24 43.1 22.0 4 25-29 62.7 19.6 5 30-34 104.1 41.4 6 35-39 142.9 38.8 7 40-44 217.5 74.6 8 45-49 230.7 13.2 9 50-54 231.1 0.4 10 55-59 210.0 -21.1 11 60-69 382.2 172.3 12 70-79 228.1 -154.1 13 80-89 65.4 -162.7 14 1 90 7.4 -58.0 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
Number of Number of Single change change Order of Group of household per group of per each cohort cohort ages head ages group (thousand) (thousand) (thousand)
Total 1,949.8 1 0-14 3.7 3.7 0.26 2 15-19 21.1 17.4 3.48 3 20-24 43.1 22.0 4.39 4 25-29 62.7 19.6 3.92 5 30-34 104.1 41.4 8.28 6 35-39 142.9 38.8 7.76 7 40-44 217.5 74.6 14.92 8 45-49 230.7 13.2 2.64 9 50-54 231.1 0.4 0.08 10 55-59 210.0 -21.1 -4.22 11 60-69 382.2 172.3 17.23 12 70-79 228.1 -154.1 -30.81 13 80-89 65.4 -162.7 -32.54 14 1 90 7.4 -58.0 -11.60 Source: calculation of Excel program
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 Marriage Single of Number Number Number Ratio number of of head head of of of the ages household household population head (thousand) change change splitting in (thousand) household per per (formation) municipal group each of area (thousand) (thousand) cohort single (thousand) head group (thousand) household per total population Total 19,376.7 5,288.8 3,338.9 1,949.8 (percent)
0-14 4,021.5 5.6 1.9 3.7 3.7 0.26 0.01
15-19 1,423.6 55.8 34.7 21.1 17.4 3.48 0.24
20-24 1,640.4 217.5 174.5 43.1 22.0 4.39 0.27
25-29 1,856.2 404.9 342.2 62.7 19.6 3.92 0.21
30-34 1,861.7 558.0 453.9 104.1 41.4 8.28 0.44
35-39 1,760.6 629.6 486.7 142.9 38.8 7.76 0.44
40-44 1,606.9 667.5 450.0 217.5 74.6 14.92 0.93
45-49 1,410.0 657.8 427.1 230.7 13.2 2.64 0.19
50-54 1,092.8 561.9 330.8 231.1 0.4 0.08 0.01
55-59 807.8 445.6 235.7 210.0 -21.1 -4.22 -0.52
60-69 1,124.6 662.2 280.0 382.2 172.3 17.23 1.53
70-79 591.8 337.2 109.0 228.1 -154.1 -30.81 -5.21
80-89 152.4 76.7 11.3 65.4 -162.7 -32.54 -21.35 1 90 26.2 8.5 1.0 7.4 -58.0 -11.60 -44.20 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
%(NHH - NMHH) !!$!"## ! RSHHpTP = Equation 4-9 RofGOA *TP
%(NHH - NMHH) !!! Or, RSHHpTP = RofGOA * 100 Equation 4-10 TP
Where,
RSHHpTP = ratio of splitting of single head household per total population
NHH = number of total head household
NMHH = number of married head household (couple head household)
Rof GOA = range of each group of ages
TP = total population in municipal area
And
(NHH – NMHH) = NHH – NMHH Equation 4-11
Where,
NHH = NHH 2 – NHH 1
NMHH = NMHH 2 – NMHH 1
And
NHH 2 = number of total head household at time 2
NHH 1 = number of total head household at time 1
NMHH 2 = number of married head household (couple head household) at
time 2
NMHH 1 = 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 Total Single Number Number Ratio Ratio of population head of of of the of the in household ages municipal (thousand) change change splitting splitting area per per each of of (thousand) group cohort single single (thousand) group head head (thousand) household household per total by age population group (percent) (percent) Total 19,376.7 1,949.8 4.27 0-14 4,021.5 3.7 3.7 0.26 0.01 0.25 15-19 1,423.6 21.1 17.4 3.48 0.24 20-24 1,640.4 43.1 22.0 4.39 0.27 25-29 1,856.2 62.7 19.6 3.92 0.21 30-34 1,861.7 104.1 41.4 8.28 0.44 35-39 1,760.6 142.9 38.8 7.76 0.44 2.49 40-44 1,606.9 217.5 74.6 14.92 0.93 45-49 1,410.0 230.7 13.2 2.64 0.19 50-54 1,092.8 231.1 0.4 0.08 0.01 55-59 807.8 210.0 -21.1 -4.22 -0.52 60-69 1,124.6 382.2 172.3 17.23 1.53 70-79 591.8 228.1 -154.1 -30.81 -5.21 1.53 80-89 152.4 65.4 -162.7 -32.54 -21.35 1 90 26.2 7.4 -58.0 -11.60 -44.20 Source: *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.
* X * (0.29/100) Y * (2.49/100) Z * (1.53/100) NSSHHpY ' & & Equation 4-12 19 40 40
Where,
NSSHHpY = number of splitting single head households per year
X = number of adolescents from 0 to 19
(0.29/100) = proportion of adolescent creating a single household per year
19 = age range for the adolescent group (from 0 to 19)
Y = number of mature people from 20 to 59
(2.49/100) = proportion of mature people creating a single household per
year
40 = age range for the mature group (from 20 to 59)
Z = number of elders from 60 to 100
(1.53/100) = proportion of elders creating a single household per year
40 = 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 (R 2) from regression equations in these cases do not provide a high level of confidence.
However, the trend for both factors follows the data for Thailand’s 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 * Number ^ Percent in Bangkok * Number ^ Percent Year of of Metropolitan of divorce of divorce marriage marriage Region 1996 7,348,000.00 64,179 1.65 14,742 22.97 1997 7,496,000.00 64,229 1.62 16,483 25.66 1998 7,637,000.00 52,626 1.30 17,699 33.63 1999 8,737,000.00 58,499 1.42 17,042 29.13 Source: * The National Statistical Office of Thailand (* data from historical data, ^data from calculation of Excel program)
-0.1486 % of population Marriage proportion in BMR y = 1.68x R2 = 0.6215 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00 1996 1997 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 (R 2) at 0.6215. Then this method provides the equation as shown below (see Figure 4-7).
TP * 1.68 * Y -0.1486 MN = Equation 4-13 100
Where,
MN = marriage number
TP = total population
Y = year of marriage & divorce equation (set 1996 as the 1 st year)
Divorce ratio of marriage People in BMR y = 23.031x 0.2263 % of marriage R2 = 0.6855 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 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 (R 2) at 0.6855. Then this method provides the equation below (see Figure 4-8).
TP * 23.031 * Y 0.2263 DN = Equation 4-14 100
Where,
DN = divorce number
TP = total population
Y = year of marriage & divorce equation (set 1996 as the 1 st 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
Marriage Divorce year percent of divorce (people) (people)
1993 484,569 46,953 9.69 1994 435,425 46,903 10.77 1995 470,751 53,560 11.38 1996 436,831 56,718 12.98 1997 396,928 62,379 15.72 4-161
1998 324,262 67,551 20.83 1999 354,198 61,377 17.33 2000 337,140 70,882 21.02 2001 324,661 76,037 23.42 2002 291,734 77,735 26.65 2003 328,356 80,886 24.63 2004 365,721 86,982 23.78 2005 345,234 90,688 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 model’s 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 people’s 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 Equation 4-15
Where,
RNHN = the remaining number of housing needs
NHN = number of housing needs
NHD = number of housing demand
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.
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).
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.
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
y = -1.0436Ln(x) + 8.4333 GDP GDP in Thailand R2 = 0.0344 15.00
10.00
5.00
0.00
0 2 4 6 0 2 6 8 2 4 6 8 8 8 9 9 9 0 0 98 988 99 00 -5.0019 1 19 19 1 19 19 1994 1 19 2000 2 20 20
-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 (R 2) at 0.0344 as shown below.
Y = -1.0436 Ln(x) + 8.4333 Equation 4-16
Where,
Y = GDP growth index x = year (set 1980 as the first year)
Ln = mathematical value (Formally, Ln (a) may be defined as the area
under the graph of 1/x from 1 to a, as the integral) 4-166
Loan y = 39295x 0.6 (M Bath) Housing loan in Thailand R2 = 0.5772 350,000
300,000
250,000
200,000
150,000
100,000
50,000
0
01 1985 1987 1989 1991 1993 1995 1997 1999 20 2003 2005 2007 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 (R 2) at 0.5772 as shown below.
Y = 39295 *x 0.6 Equation 4-17
Where,
Y = housing loan x = year (set 1985 as the first year) 4-167
percent of index y = 93.081x 0.1362 (1986 as 100% ) Housing Price Index in Thailand 2 R = 0.9249 180
160 140
120
100
80 60
40
20 0
9 5 9 1 5 9 1 5 7 7 8 8 9 9 9 0 0 0 9 9 9 9 0 0 1977 1 1981 1983 1 1987 19 1 1993 1 1997 19 2 2003 20 2 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 (R 2) at 0.9249 as shown below.
Y = 93.081 *x 0.1362 Equation 4-18
Where,
Y = housing price index x = year (set 1985 as the first year) 4-168
The assumptions relating to the impact of economic data on housing demand in this model are defined as follows:-
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 ^ Data from equation for price year index index 1977 53.90 - 1978 54.20 - 1979 56.20 - 1980 63.40 - 1981 68.20 - 1982 78.50 - 1983 83.30 - 1984 88.90 - 1985 94.60 - 1986 100.00 - 1987 106.90 - 1988 111.80 - 1989 117.10 - 1990 121.10 - 1991 124.00 - 1992 n/a 123.56 1993 n/a 125.55 1994 n/a 127.37 1995 n/a 129.03 1996 n/a 130.57 1997 n/a 132.00 1998 n/a 133.34 1999 n/a 134.60 2000 131.20 - 2001 130.90 - 2002 131.80 - 2003 135.70 - 2004 143.00 - 2005 155.80 - 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 Maximum co Purpose value efficient value efficient GDP on affordability & growth (1998) (1988) investment C0.12 C10.3 index -10.5 13.2 (from housing needs to (%) housing demand) 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 factor’s 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 co Purpose value efficient value efficient GDP on consumption growth (1998) (1988) decision C0.075 C10.1 index -10.5 13.2 (from housing demand (%) to housing sales) 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 Housing ^Ratio of housing loan on year price loan housing price index index 1977 ^29,000 53.90 538.03 1978 ^29,000 54.20 535.06 1979 ^36,000 56.20 640.57 1980 ^41,000 63.40 646.69 1981 ^45,000 68.20 659.82 1982 ^40,000 78.50 509.55 1983 ^35,000 83.30 420.17 1984 ^30,000 88.90 337.46 1985 ^35,000 94.60 369.98 1986 ^35,000 100.00 350.00 (minimum value) 1987 ^30,000 106.90 280.64 1988 ^40,110 111.80 358.77 1989 53,736 117.10 458.89 1990 59,368 121.10 490.24 1991 78,306 124.00 631.50 1992 111,346 ^123.56 901.15 1993 147,829 ^125.55 1,177.45 1994 209,811 ^127.37 1,647.26 1995 223,408 ^129.03 1,731.44 1996 241,172 ^130.57 1,853.41 1997 202,720 ^132.00 1,535.76 1998 103,733 ^133.34 777.96 1999 64,301 ^134.60 477.72 2000 108,886 131.20 829.92 2001 112,611 130.90 860.22 2002 164,851 131.80 1,250.77 2003 296,661 135.70 2,186.15 (maximum value) 2004 296,025 143.00 2,070.10 2005 279,407 155.80 1,908.90 2006 279,392 141.81 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 Purpose value efficient value efficient Housing loan over (1986) (2003) on consumption housing C1.164 C1.771 decision 350 2186.15 price (from housing demand index to housing sales) (ratio) 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 4-
16, below. 4-177
Table 4-16 : housing sales and housing supply
*Housing *Housing year sales supply 1977 63,163 n/a 1978 48,119 n/a 1979 63,775 n/a 1980 63,652 n/a 1981 45,892 n/a 1982 41,881 n/a 1983 26,297 n/a 1984 50,638 n/a 1985 50,086 n/a 1986 33,722 n/a 1987 53,353 n/a 1988 67,451 n/a 1989 80,031 n/a 1990 102,335 n/a 1991 129,688 81,657 1992 108,001 58,497 1993 134,086 113,150 1994 171,254 253,028 1995 172,419 155,522 1996 166,785 120,671 1997 145,355 36,537 1998 63,864 2,407 1999 33,382 4,068 2000 32,028 6,789 2001 34,023 13,126 2002 37,182 29,217 2003 56,085 55,539 2004 69,050 68,579 2005 70,072 63,579 2006 75,000 66,118 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 housing- supply 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 Purpose value efficient value efficient Housing on housing supply (1983) (1995) sales (from housing sales to 26297 C0.33 172419 C1.91 (unit) housing supply) 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 model’s accuracy.
These are listed below (see Figure 4-17). Each equation set is then shown in Table
4-18 below.
Demographic stocks
total population
adolescents from 0 to 19 year of age
mature people from 20 to 59 year of age
elders from 60 and over year of age
Social stock
housing needs (sHS needs)
Economic stock
housing demand (sHS demand)
Check point stock
housing sales (sHS sales)
housing supply (sHS supply)
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 Stock values (1976) Total population 4,000,000 Adolescents from 0 to 19 794,407 Mature people from 20 to 59 1,467,125 Elders from 60 and over 229,515 Housing needs 60,000 Housing demand 60,000 Housing sales 60,000 Housing supply 60,000 Housing vacancy (accumulated value) 0 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 model’s 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 model’s 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:
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 Equation 5-1
Where,
NVH = number of vacant house
NHS = number of housing sales
NHN = number of housing needs
This equation is used to demonstrate and verify the comparison between the simulated housing vacancy line and historical housing vacancy line (see Figure 5-
14).
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 = number of housing supply
NHS = number of housing sales
C of NHS on NHSP = 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 5-
13: 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.
Verification of the logical assumptions which create the equations used in
the model
Verification of the logical assumptions which create the graphical
functions used in the model
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 model’s 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:
Comparison between simulation of the housing needs number and the
historical housing sales number (Figure 5-7)
Comparison between simulation of the housing demand number and the
historical housing sales number (Figure 5-10)
Comparison between simulation of the housing sales number and the
historical housing sales number (Figure 5-12)
Comparison between simulation of the housing supply number and the
historical housing supply number (Figure 5-13)
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).
Housing sale data and linear trend y = 936.66x + 62202 compare with underlying housing needs number R2 = 0.0334 200,000
180,000
160,000
140,000
120,000
100,000 80,000 housing unit 60,000 40,000
20,000
0 19771979 1981 1983 19851987 1989 1991 19931995 1997 19992001 2003 2005 year Housing sale Simulated housing needs number Linear (Housin g sale )
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 housing housing surplus or total of surplus sales data needs deficit and deficit
1977 63,163 43,258 19,905 1978 48,119 44,849 3,270 56,192 1979 63,775 46,412 17,363 1980 63,652 47,998 15,654 1981 45,892 49,627 -3,735 1982 41,881 51,306 -9,425 1983 26,297 53,042 -26,745 1984 50,638 54,835 -4,197 -82,820 1985 50,086 56,689 -6,603 1986 33,722 58,605 -24,883 1987 53,353 60,586 -7,233 1988 67,451 62,634 4,817 1989 80,031 64,751 15,280 1990 102,335 66,940 35,395 1991 129,688 69,203 60,485 1992 108,001 71,542 36,459 570,047 1993 134,086 73,960 60,126 1994 171,254 76,460 94,794 1995 172,419 79,045 93,374 1996 176,616 78,663 97,953 1997 148,631 77,267 71,364 1998 67,054 77,381 -10,327 1999 29,029 78,187 -49,158 2000 33,074 79,407 -46,333 2001 33,980 80,912 -46,932 2002 37,182 82,638 -45,456 -279,120 2003 56,085 84,545 -28,460 2004 69,050 86,609 -17,559 2005 70,072 88,815 -18,743 2006 75,000 91,153 -16,153 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 = number of housing vacancy
NHS = number of housing sales
NHN = 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 simulat ed 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 5-
12). 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 research’s forecasting adopted in two scenarios:
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.
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
(R 2) 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 y = -0.5676x + 30.356 R2 = 0.9998 35.00
30.00
25.00
20.00
15.00
10.00
5.00
0.00 percentage of adolescent in of adolescent BMR percentage 2 3 4 5 6 7 8 9 0 1 2 4 13 15 00 00 00 01 01 01 0 01 0 2001 200 200 200 200 200 2 2 2 2 2 2 2 2 2 2016 2017 year(set 2001 as a 1ST year)
Adolescent proportion Linear (Adolescent proportion)
Source: Excel program
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 (R 2) 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. 6-219
Mature People in the Bangkok Metropolitan Region projection
Mature proportion y = 0.3981x + 56.062 R2 = 0.9989 64
62
60
58
56
54
percentage of mature in percentage BMR 52
2 3 4 5 6 9 0 1 5 6 7 00 00 00 01 01 01 01 2001 200 200 2 2 2 2007 2008 200 2 2 2012 2013 2014 201 2 2 year(set 2001 as a 1ST year)
Mature proportion Linear (Mature proportion)
Source: Excel program
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 (R 2) 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. 6-220
Elderly Popele in the Bangkok Metropolitan Region projection
Elder proportion y = 0.2547x + 8.7703 R2 = 0.9905 14
12
10
8
6
4
2 percentage of elder in percentage BMR 0
2 3 4 5 6 7 3 5 6 7 001 00 00 00 00 00 00 012 01 014 01 01 01 2 2 2 2 2 2 2 2008 2009 2010 2011 2 2 2 2 2 2 year(set 2001 as a 1ST year)
Elder proportion Linear (Elder proportion)
Source: Excel program
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 (R 2) 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 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 Number of Married Single Number of Number of ratio of population total household household change change of ages in household head head per group per each splitting of municipal head (thousand) (thousand) (thousand) cohort single area (thousand) group head (thousand) (thousand) household per total population (percent)
Total 19,376.7 5,288.8 3,338.9 1,949.8
0-14 4,021.5 5.6 1.9 3.7 3.7 0.26 0.01
15-19 1,423.6 55.8 34.7 21.1 17.4 3.48 0.24
20-24 1,640.4 217.5 174.5 43.1 22.0 4.39 0.27
25-29 1,856.2 404.9 342.2 62.7 19.6 3.92 0.21
30-34 1,861.7 558.0 453.9 104.1 41.4 8.28 0.44
35-39 1,760.6 629.6 486.7 142.9 38.8 7.76 0.44
40-44 1,606.9 667.5 450.0 217.5 74.6 14.92 0.93
45-49 1,410.0 657.8 427.1 230.7 13.2 2.64 0.19
50-54 1,092.8 561.9 330.8 231.1 0.4 0.08 0.01
55-59 807.8 445.6 235.7 210.0 -21.1 -4.22 -0.52
60-69 1,124.6 662.2 280.0 382.2 172.3 17.23 1.53
70-79 591.8 337.2 109.0 228.1 -154.1 -30.81 -5.21
80-89 152.4 76.7 11.3 65.4 -162.7 -32.54 -21.35
1 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 (i.e. equation 4-3)
0.25/100 = 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 (i.e. equation 4-4)
2.49/100 = 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.
! (NHH - NMHH) SHH = Equation 6-2 NRAG
Where,
SHH = percentage of the single-head households formed for each age group
NHH = 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 Single head Households Total of each of household formed group age (household) (percent) (percent) Total 1,949,800 2.77 4.27 0 – 14 3,700 0.01 Adolescent 15 – 19 21,100 0.28 0.25 20 – 24 43,100 0.27 25 – 29 62,700 0.21
30 – 34 104,100 0.41 Mature 35 – 39 142,900 0.47 40 – 44 217,500 0.93 45 – 49 230,700 0.19 2.49 50 – 54 231,100 0.01
55 – 59 210,000 Negative value 60 – 69 382,200 1.53 70 – 79 228,100 Negative value Elder 80 – 89 65,400 Negative value 1.53 90 and over 7,400 Negative value
Source: The National Statistical Office (* Data from historical data, ^ Data from calculation) 6-226
-0.1486 % of population Marriage proportion in BMR y = 1.68x R2 = 0.6215 1.80 1.60 1.40 1.20 1.00 0.80 0.60 0.40 0.20 0.00
8 6 96 04 12 14 002 010 19 199 2000 2 20 200 2008 2 20 20 2016 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 (R 2) 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. 6-227
Divorce ratio of marriage People in BMR y = 23.031x 0.2263 % of marriage R2 = 0.6855 50.00 45.00 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00
98 02 06 10 14 0 1996 19 2000 20 2004 20 2008 20 2012 2 2016 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 (R 2) 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 Marriages Divorces % of marriages ending in divorce 1993 484,569 46,953 9.69
1994 435,425 46,903 10.77
1995 470,751 53,560 11.38
1996 436,831 56,718 12.98
1997 396,928 62,379 15.72
1998 324,262 67,551 20.83
1999 354,198 61,377 17.33
2000 337,140 70,882 21.02
2001 324,661 76,037 23.42
2002 291,734 77,735 26.65
2003 328,356 80,886 24.63
2004 365,721 86,982 23.78
2005 345,234 90,688 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
y = -1.0436Ln(x) + 8.4333 GDP GDP in Thailand R2 = 0.0344 15.00
10.00
5.00
0.00
4 2 6 0 4 0 4 98 988 99 99 0 00 01 01 -5.001980 1982 1 1986 1 1990 1 1994 1 1998 20 2002 2 2006 2008 2 2012 2 2016
-10.00
-15.00 year (set 1980 as 1st year)
GDP data Log. (GDP data)
Source: Excel program
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 (R 2) at = 0.0344 (see Figure 6-7).
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 (R 2) 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 model’s forecast would vary. 6-231
Projected Housing Loans in Thailand
y = 39295x 0.6 Loan Housing loan in Thailand R2 = 0.5772 (M Bath) 350,000
300,000
250,000
200,000
150,000
100,000
50,000
0
9 01 003 1985 1987 1989 1991 1993 1995 1997 1999 20 2 2005 2007 200 2011 2013 2015 2017 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 (R 2) 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 0.1362 (1986 as 100% ) Housing Price Index in Thailand y = 93.081x R2 = 0.9249 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 2009 2011 2013 2015 2017 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 (R 2) 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.
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 graph’s 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 6-
16) (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 Bangkok’s
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 buyer’s 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:
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.
The second objective was to verify the model’s 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.”
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.
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 Model’s 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:
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 10-
40% 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 model’s 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 this calculation are described in
Appendix D.
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 housing- need 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 Ove r-sur plus area
Deficit area
Over-deficit area Surplus area
Time Underlying housing-need line
Housing-sale line Spring box and its forces
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
Thailand’s 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.
Housing sale data and linear trend y = 936.66x + 62202 compare with underlying housing needs number R2 = 0.0334 200,000
180,000
160,000
140,000
120,000
100,000 80,000 housing unit 60,000 40,000
20,000
0 19771979 1981 1983 19851987 1989 1991 19931995 1997 19992001 2003 2005 year Housing sale Simulated housing needs number Linear (Housin g sale )
Source: simulation program
Figure 7-5 : Housing sales and underlying housing-needs
Interestingly, the different accumulated numbers between the underlying housing- need 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
Accumulated ^housing Over- needs *Accumulated of simulated (Equilibrium ) year needs housing vacant housing housing linear line consumption vacancy 1977 n/a 43,258 - 19,880 63,139 1978 n/a 44,849 - 19,226 64,075 1979 n/a 46,412 - 18,600 65,012 1980 n/a 47,998 - 17,951 65,949 1981 n/a 49,627 - 17,259 66,885 1982 n/a 51,306 - 16,516 67,822 1983 n/a 53,042 - 15,717 68,759 1984 n/a 54,835 - 14,860 69,695 1985 n/a 56,689 - 13,943 70,632 1986 n/a 58,605 - 12,964 71,569 1987 n/a 60,586 - 11,919 72,505 1988 n/a 62,634 - 10,808 73,442 1989 n/a 64,751 - 9,627 74,379 1990 n/a 66,940 - 8,375 75,315 1991 n/a 69,203 - 7,049 76,252 1992 n/a 71,542 - 5,647 77,189 1993 n/a 73,960 - 4,165 78,125 1994 n/a 76,460 - 2,601 79,062 1995 n/a 79,045 - 954 79,999 1996 275,039 78,663 230,334 2,272 80,935 1997 336,084 77,267 234,940 4,605 81,872 1998 341,925 77,381 240,367 5,428 82,809 1999 348,308 78,187 245,925 5,558 83,745 2000 347,810 79,407 251,201 5,275 84,682 2001 353,800 80,912 255,907 4,706 85,619 2002 350,259 82,638 259,824 3,917 86,555 2003 352,645 84,545 262,771 2,947 87,492 2004 n/a 86,609 264,591 1,820 88,428 2005 n/a 88,815 265,142 551 89,365 2006 n/a 91,153 264,290 -851 90,302 total 352,645 2,037,317 264,290 264,290 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 housing-
need 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:
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 world’s
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
ABS. (2007). "Measuring wellbeing: frameworks for Australian social statistics."from http://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 modeling." from http://web.aces.uiuc.edu/aim/model/day1text1.htm .
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: 357- 82.
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): 381- 393.
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, N- 5020 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
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 ,
---- 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 planning—A 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 Society , 6 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 high- risk 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 Low- Cost Refinancing.' Real Estate Economics , vol. 32, no. 4, p. 589.
Ross, MW, Rosser, BRS & Stanton, J 2004, Beliefs about cybersex and Internet- mediated 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. 57- 65.
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)
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 yards
Semi -detached houses (Twin house)
also known as half-a-house
one building usually consists of 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)
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
each unit usually has its own small fenced yard or courtyard
Commercial building and shop house building
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 288
Apartment (i.e. flat and condominium)
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 number of
housing demand that has been counted by using the principle and
calculation of this research.
‘Adolescent’ is the term used to refer to people from 0 to 19 years of
age.
‘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
‘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.
‘Exiting housing market’ is the term used to refer to resell housing
market.
‘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.
‘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.
‘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.
‘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.
‘Housing needs’ is the term used to refer to needs of people to have a
house for living.
‘Housing sales unit (HSU)’ is the term used to refer to number of
housing that has been sold to persons in housing demand unit.
‘Housing unit ’ is the term used to refer to all types of housing, counted
as the same unit.
‘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.
‘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.
‘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 km², and with a registered
population of 9,988,400 (as of April 4, 2007) it has a population density
of 1,286.92 per km².
‘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 density year per sq total male female km. 2518 1975 42,391,454 21,359,489 21,031,965 82.62 2519 1976 43,213,711 21,790,510 21,423,201 84.22 2520 1977 44,272,693 22,314,837 21,957,856 86.28 2521 1978 45,221,625 22,775,852 22,445,773 88.13 2522 1979 46,113,756 23,205,927 22,907,829 89.87 2523 1980 46,961,338 23,627,727 23,333,611 91.52 2524 1981 47,875,002 24,067,597 23,807,405 93.30 2525 1982 48,846,927 24,549,873 24,297,054 95.20 2526 1983 49,515,074 24,911,684 24,603,390 96.50 2527 1984 50,583,105 25,449,044 25,134,061 98.58 2528 1985 51,795,651 26,059,668 25,735,983 100.94 2529 1986 52,969,204 26,642,889 26,326,315 103.23 2530 1987 53,873,172 27,070,155 26,803,017 104.99 2531 1988 54,960,917 27,574,256 27,386,661 107.11 2532 1989 55,888,393 28,001,343 27,887,050 108.92 2533 1990 56,303,273 28,181,202 28,122,071 109.73 2534 1991 56,961,030 28,463,102 28,497,928 111.01 2535 1992 57,788,965 29,018,092 28,770,873 112.62 2536 1993 58,336,072 29,205,086 29,130,986 113.69 2537 1994 59,095,419 29,552,978 29,542,441 115.17 2538 1995 59,460,382 29,678,600 29,781,782 115.88 2539 1996 60,116,182 29,973,059 30,143,123 117.16 2540 1997 60,816,227 30,295,797 30,520,430 118.52 2541 1998 61,466,178 30,591,602 30,874,576 119.79 2542 1999 61,661,701 30,650,172 31,011,529 120.17 2543 2000 61,878,746 30,725,016 31,153,730 120.59 2544 2001 62,308,887 30,913,485 31,395,402 121.43 2545 2002 62,799,872 31,139,647 31,660,225 122.39 2546 2003 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 Population year increased rate total male female
2452 1909 8,149,487 n/a n/a n/a
2462 1919 9,207,355 4,599,667 4,607,688 1.22
2472 1929 11,506,207 5,795,065 5,711,142 2.23
2480 1937 14,464,105 7,313,584 7,150,521 2.86
2490 1947 17,442,689 8,722,155 8,720,534 1.87
2503 1960 26,257,916 13,154,149 13,103,767 3.15
2513 1970 34,397,374 17,123,862 17,273,512 2.70
2523 1980 44,824,540 22,328,607 22,495,933 2.65
2533 1990 54,548,530 27,061,733 27,486,797 1.96
2543 2000 60,916,441 30,015,233 30,901,208 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
Table B-3 Population area density and house in2003 population area density area house total male female (Sq km.) per sq km. country 63,079,765 31,255,350 31,824,415 513,115.0 122.93 17,853,423
Bangkok 5,844,607 2,822,171 3,022,436 1,565.2 3,734.10 2,020,019 From Department of local administration 294
Table B-4 population by age, sex and municipal limits in 2000 in out of age Population percent municipal municipal limits limits (year) total male female total male female total male female total male female total 60,916,441 30,015,233 30,901,208 100.0 100.0 100.0 18,972,330 9,129,755 9,842,575 41,944,111 20,885,478 21,058,633 0 - 4 4,634,898 2,389,624 2,245,274 7.6 8.0 7.3 1,205,825 621,789 584,036 3,429,073 1,767,835 1,661,238 5 - 9 5,125,054 2,639,213 2,485,841 8.4 8.8 8.0 1,315,293 675,544 639,749 3,809,761 1,963,669 1,846,092 10 - 14 5,083,946 2,604,211 2,479,735 8.3 8.7 8.0 1,351,669 689,734 661,935 3,732,277 1,914,477 1,817,800 15 - 19 5,287,033 2,683,634 2,603,399 8.7 8.9 8.4 1,624,724 801,256 823,468 3,662,309 1,882,378 1,779,931 20 - 24 5,262,141 2,652,654 2,609,487 8.6 8.8 8.4 1,914,255 917,649 996,606 3,347,886 1,735,005 1,612,881 25 - 29 5,407,796 2,631,408 2,776,388 8.9 8.8 9.0 1,910,236 907,807 1,002,429 3,497,560 1,723,601 1,773,959 30 - 34 5,525,944 2,671,164 2,854,780 9.1 8.9 9.2 1,852,467 879,076 973,391 3,673,477 1,792,088 1,881,389 35 - 39 5,259,357 2,563,502 2,695,855 8.6 8.5 8.7 1,746,028 826,636 919,392 3,513,329 1,736,866 1,776,463 40 - 44 4,685,044 2,280,006 2,405,038 7.7 7.6 7.8 1,574,321 745,998 828,323 3,110,723 1,534,008 1,576,715 45 - 49 3,719,822 1,807,874 1,911,948 6.1 6.0 6.2 1,215,611 576,148 639,463 2,504,211 1,231,726 1,272,485 50 - 54 2,848,159 1,381,080 1,467,079 4.7 4.6 4.7 903,383 431,404 471,979 1,944,776 949,676 995,100 55 - 59 2,284,277 1,097,886 1,186,391 3.7 3.7 3.8 676,181 320,814 355,367 1,608,096 777,072 831,024 60 - 64 1,970,687 925,205 1,045,482 3.2 3.1 3.4 578,347 265,667 312,680 1,392,340 659,538 732,802 65 - 69 1,530,021 702,801 827,220 2.5 2.3 2.7 433,060 192,613 240,447 1,096,961 510,188 586,773 70 - 74 1,080,103 486,407 593,696 1.8 1.6 1.9 315,480 137,348 178,132 764,623 349,059 415,564 75 - 79 626,282 272,134 354,148 1.0 0.9 1.1 176,330 74,082 102,248 449,952 198,052 251,900 80 - 84 340,550 137,046 203,504 0.6 0.5 0.7 103,154 39,852 63,302 237,396 97,194 140,202 over 85 245,327 89,384 155,943 0.4 0.3 0.5 75,966 26,338 49,628 169,361 63,046 106,315 from statistic bureau, office of Prime minister 295
Table B-5 population by age from 2001-2003 2001 2002 2003 age (year) total % male female total % male female total % male female total 62,308,887 30,913,485 31,395,402 62,799,872 31,139,647 31,660,225 63,079,765 31,255,350 31,824,415 0 - 4 4,207,497 6.75 2,165,012 2,042,485 4,042,604 6.44 2,080,555 1,962,049 3,902,496 6.19 2,008,961 1,893,535 5 - 9 4,813,614 7.73 2,471,035 2,342,579 4,821,676 7.68 2,476,093 2,345,583 4,754,166 7.54 2,441,841 2,312,325 10 - 14 4,614,429 7.41 2,365,374 2,249,055 4,700,093 7.48 2,410,108 2,289,985 4,759,358 7.54 2,440,525 2,318,833 15 - 19 4,928,170 7.91 2,510,210 2,417,960 4,780,924 7.61 2,436,368 2,344,556 4,661,217 7.39 2,377,751 2,283,466 20 - 24 5,341,014 8.57 2,694,397 2,646,617 5,282,247 8.41 2,668,483 2,613,764 5,208,192 8.26 2,629,241 2,578,951 25 - 29 5,513,728 8.85 2,759,387 2,754,341 5,479,986 8.73 2,745,551 2,734,435 5,406,128 8.57 2,710,611 2,695,517 30 - 34 5,655,524 9.08 2,793,229 2,862,295 5,663,969 9.02 2,797,582 2,866,387 5,563,479 8.82 2,751,018 2,812,461 35 - 39 5,341,771 8.57 2,619,916 2,721,855 5,413,362 8.62 2,651,000 2,762,362 5,505,556 8.73 2,691,778 2,813,778 40 - 44 4,590,483 7.37 2,243,629 2,346,854 4,739,965 7.55 2,317,495 2,422,470 4,889,452 7.75 2,389,460 2,499,992 45 - 49 3,781,378 6.07 1,827,567 1,953,811 3,930,215 6.26 1,898,060 2,032,155 4,060,211 6.44 1,960,274 2,099,937 50 - 54 2,874,844 4.61 1,385,671 1,489,173 3,039,054 4.84 1,461,853 1,577,201 3,202,542 5.08 1,535,489 1,667,053 55 - 59 2,085,444 3.35 1,001,197 1,084,247 2,148,128 3.42 1,028,648 1,119,480 2,286,141 3.62 1,094,353 1,191,788 60 - 64 1,815,101 2.91 857,555 957,546 1,872,448 2.98 884,995 987,453 1,855,785 2.94 875,527 980,258 296
65 - 69 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 1.65 456,531 571,173 1,106,138 1.76 493,352 612,786 1,145,133 1.82 508,836 636,297 75 - 79 622,331 1.00 267,069 355,262 657,745 1.05 281,305 376,440 715,638 1.13 305,814 409,824 80 - 84 338,959 0.54 138,316 200,643 354,636 0.56 144,892 209,744 380,257 0.60 155,833 224,424 85-89 170,554 0.27 65,916 104,638 179,575 0.29 69,522 110,053 190,802 0.30 74,208 116,594 90-94 73,441 0.12 27,931 45,510 74,995 0.12 28,671 46,324 71,214 0.11 26,712 44,502 95-99 33,649 0.05 13,354 20,295 33,839 0.05 13,425 20,414 30,612 0.05 11,923 18,689 over 100 40,869 0.07 17,058 23,811 42,715 0.07 17,855 24,860 34,784 0.06 14,521 20,263 unknown 2,946,599 4.73 1,547,770 1,398,829 2,911,960 4.64 1,535,264 1,376,696 2,875,694 4.56 1,524,953 1,350,741 100.00 100.00 100.00
from : department of local administration 297
Table B- 6 birth rate by municipal limits in 1985-1986, 1991 and 1995-1996 birth rate (per 1,000 ) area 1985-1986 1991 1995-1996 country 23.87 20.19 17.90 in municipal limits 18.97 16.79 14.73 out of municipal limits 24.79 20.98 18.72 Bangkok 18.92 14.65 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 age status (year) total have been marry be a no data of single and sex total married widow divorce separate no data monk status total 48,132,896 14,947,461 32,771,031 28,712,450 2,803,037 405,976 532,144 317,424 243,761 170,643 13 - 14 2,060,353 1,991,312 20,086 13,849 317 395 213 5,312 24,037 24,918 15 - 19 5,287,033 4,832,185 406,380 361,088 4,257 5,201 8,939 26,895 30,149 18,319 20 - 24 5,262,141 3,449,811 1,759,180 1,625,587 20,667 24,308 40,608 48,010 34,306 18,844 25 - 29 5,407,796 1,951,244 3,419,305 3,209,510 45,192 44,403 67,097 53,103 22,459 14,788 30 - 34 5,525,944 1,050,467 4,440,846 4,188,697 74,478 59,275 77,536 40,860 17,171 17,460 35 - 39 5,259,357 638,681 4,590,994 4,315,412 102,910 65,609 77,682 29,381 14,824 14,858 40 - 44 4,685,044 400,994 4,256,558 3,951,539 147,182 65,815 70,708 21,314 13,574 13,918 45 - 49 3,719,822 245,009 3,453,576 3,148,635 185,237 49,746 54,956 15,002 11,437 9,800 50 - 54 2,848,159 143,143 2,686,201 2,371,788 229,007 33,282 40,089 12,035 10,811 8,004 55 - 59 2,284,277 90,411 2,175,497 1,831,660 279,524 22,434 31,113 10,766 11,621 6,748 60 - 64 1,970,687 63,907 1,885,757 1,467,849 365,278 15,360 25,380 11,890 14,240 6,783 65 - 69 1,530,021 41,411 1,468,726 1,044,288 385,627 9,701 17,904 11,206 14,173 5,711 70 - 74 1,080,103 24,958 1,038,316 644,650 366,003 5,485 10,939 11,239 12,286 4,543 75 - 79 626,282 12,605 603,957 317,680 270,207 2,689 5,212 8,169 7,128 2,592 80 - 84 340,550 6,419 328,951 139,403 179,443 1,336 2,368 6,401 3,378 1,802 over 85 245,327 4,904 236,701 80,815 147,708 937 1,400 5,841 2,167 1,555
From: statistic bureau, office of prime minister 299
age status (year) total have been marry be a no data of single and sex total married widow divorce separate no data monk marry status male 23,434,640 8,019,099 15,064,691 13,977,400 617,505 137,386 189,944 142,456 243,761 107,089 13 - 14 1,052,455 1,006,866 7,990 4,862 128 195 69 2,736 24,037 13,562 15 - 19 2,683,634 2,541,416 101,658 86,128 917 1,129 1,775 11,709 30,149 10,411 20 - 24 2,652,654 2,005,733 600,942 555,599 4,239 7,009 11,856 22,239 34,306 11,673 25 - 29 2,631,408 1,155,454 1,443,363 1,364,445 10,606 15,579 24,475 28,258 22,459 10,132 30 - 34 2,671,164 591,803 2,050,311 1,959,060 18,216 21,750 29,410 21,875 17,171 11,879 35 - 39 2,563,502 325,427 2,212,556 2,121,353 22,985 23,511 29,567 15,140 14,824 10,695 40 - 44 2,280,006 177,809 2,078,860 1,988,541 31,285 22,638 26,231 10,165 13,574 9,763 45 - 49 1,807,874 92,219 1,697,407 1,619,000 36,415 16,323 19,441 6,228 11,437 6,811 50 - 54 1,381,080 47,363 1,317,702 1,244,438 44,602 10,533 13,686 4,443 10,811 5,204 55 - 59 1,097,886 29,005 1,052,894 976,767 54,983 7,007 10,552 3,585 11,621 4,366 60 - 64 925,205 19,382 887,566 792,607 77,537 5,025 8,622 3,775 14,240 4,017 65 - 69 702,801 12,468 672,854 575,550 84,588 3,119 6,283 3,314 14,173 3,306 70 - 74 486,407 7,165 464,444 369,147 86,017 1,884 4,143 3,253 12,286 2,512 75 - 79 272,134 3,697 259,976 189,439 64,994 951 2,173 2,419 7,128 1,333 80 - 84 137,046 1,811 131,068 84,273 43,596 434 1,018 1,747 3,378 789 over 85 89,384 1,481 85,100 46,191 36,397 299 643 1,570 2,167 636 From: statistic bureau, office of prime minister 300
age status (year) total have been marry be a no data of and sex single total married widow divorce separate no data monk marry status female 24,698,256 6,928,362 17,706,340 14,735,050 2,185,532 268,590 342,200 174,968 - 63,554 13 - 14 1,007,898 984,446 12,096 8,987 189 200 144 2,576 - 11,356 15 - 19 2,603,399 2,290,769 304,722 274,960 3,340 4,072 7,164 15,186 - 7,908 20 - 24 2,609,487 1,444,078 1,158,238 1,069,988 16,428 17,299 28,752 25,771 - 7,171 25 - 29 2,776,388 795,790 1,975,942 1,845,065 34,586 28,824 42,622 24,845 - 4,656 30 - 34 2,854,780 458,664 2,390,535 2,229,637 56,262 37,525 48,126 18,985 - 5,581 35 - 39 2,695,855 313,254 2,378,438 2,194,059 79,925 42,098 48,115 14,241 - 4,163 40 - 44 2,405,038 223,185 2,177,698 1,962,998 115,897 43,177 44,477 11,149 - 4,155 45 - 49 1,911,948 152,790 1,756,169 1,529,635 148,822 33,423 35,515 8,774 - 2,989 50 - 54 1,467,079 95,780 1,368,499 1,127,350 184,405 22,749 26,403 7,592 - 2,800 55 - 59 1,186,391 61,406 1,122,603 854,893 224,541 15,427 20,561 7,181 - 2,382 60 - 64 1,045,482 44,525 998,191 675,242 287,741 10,335 16,758 8,115 - 2,766 65 - 69 827,220 28,943 795,872 468,738 301,039 6,582 11,621 7,892 - 2,405 70 - 74 593,696 17,793 573,872 275,503 279,986 3,601 6,796 7,986 - 2,031 75 - 79 354,148 8,908 343,981 128,241 205,213 1,738 3,039 5,750 - 1,259 80 - 84 203,504 4,608 197,883 55,130 135,847 902 1,350 4,654 - 1,013
over 85 155,943 3,423 151,601 34,624 111,311 638 757 4,271 - 919 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 1991 1995-1996 death increased increased increased birth rate death death area rate rate birth rate rate birth rate rate (per 1,000) rate rate (per 1,000) by nature (per 1,000) by nature (per 1,000) by nature (per 1,000) (per 1,000) (per 100) (per 100) (per 100) country 23.87 6.44 1.74 20.19 5.93 1.43 17.90 6.02 1.19 in municipal limits 18.97 4.22 1.48 16.79 3.62 1.32 14.73 4.14 1.06 out of municipal 24.79 6.86 1.79 20.98 6.46 1.45 18.72 6.50 1.22 limits Bangkok 18.92 3.84 1.51 14.65 3.31 1.13 14.24 3.92 1.03 middle region 22.43 5.76 1.67 17.63 5.55 1.21 15.61 5.78 0.98 northern region 21.60 7.23 1.44 17.83 6.70 1.11 14.83 6.97 0.79 north east region 24.87 6.82 1.81 22.76 6.33 1.64 19.96 6.12 1.38 south region 31.17 7.03 2.41 25.78 6.61 1.92 24.00 6.69 1.73 From : statistic bureau, office of prime minister
302
Table B-9 the estimate of population and rates in Thailand between 1990-2020 population increased baby fertilization birth rate death rate year mid year rate by death rate (per 1000) (per 1000) (million) nature rate
1990 55.84 1.34 2.26 20.26 6.82 35.08 1991 56.57 1.29 2.22 19.88 6.96 34.47 1992 57.29 1.24 2.19 19.50 7.10 33.86 1993 58.01 1.19 2.15 19.12 7.24 33.25 1994 58.71 1.14 2.11 18.74 7.38 32.64 1995 59.40 1.08 2.07 18.36 7.52 32.04 1996 60.00 1.03 2.03 17.98 7.66 31.43 1997 60.60 0.98 2.00 17.60 7.80 30.82 1998 61.20 0.95 1.98 17.32 7.84 30.27 1999 61.81 0.92 1.96 17.04 7.88 29.72 2000 62.41 0.88 1.93 16.76 7.92 29.17 2001 62.91 0.85 1.91 16.48 7.96 28.62 2002 63.43 0.82 1.89 16.20 8.00 28.07 2003 63.96 0.79 1.88 15.96 8.08 27.59 2004 64.49 0.76 1.87 15.72 8.16 27.11 2005 65.03 0.72 1.85 15.48 8.24 26.64 2006 65.47 0.69 1.84 15.24 8.32 26.16 2007 65.94 0.66 1.83 15.00 8.40 25.68 2008 66.19 0.64 1.82 14.80 8.44 25.26 2009 66.69 0.61 1.81 14.60 8.48 24.83 2010 67.23 0.59 1.81 14.40 8.52 24.41 2011 67.71 0.56 1.80 14.20 8.56 23.98 2012 68.19 0.54 1.79 14.00 8.60 23.56 2013 68.79 0.51 1.79 13.84 8.70 23.18 2014 68.96 0.49 1.78 13.68 8.80 22.80 2015 69.08 0.46 1.78 13.52 8.90 22.41 2016 69.31 0.44 1.77 13.36 9.00 22.03 2017 69.58 0.41 1.77 13.20 9.10 21.65 2018 69.88 0.38 1.77 13.04 9.20 21.27 2019 70.21 0.36 1.76 12.88 9.30 20.89 2020 70.50 0.33 1.76 12.72 9.40 20.50 From : statistic bureau, office of prime minister 303
Table B-10 estimation of population by age and sex from 1990-2020 Unit (1000 ) age (year) 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
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 70,50 3
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 35,01 8
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 2,27 7
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 2,34 5
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 2,56 6
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 2,64 4
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 2,66 8
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 2,76 3
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 2,78 3
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 2,66 2
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 2,39 0
55-59 944 977 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 2,07 2
60-64 682 701 737 780 823 858 868 893 925 954 974 968 976 995 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 1,77 7
65-69 486 494 502 517 545 588 612 636 664 698 742 760 776 793 815 847 846 843 849 868 908 915 923 946 992 1,065 1,110 1,158 1,211 1,275 1,35 7
70-74 326 343 357 369 377 386 391 400 416 439 469 484 504 530 560 596 616 636 654 669 683 703 718 727 733 739 742 755 780 820 87 3
over75 371 374 378 383 389 396 405 415 427 439 452 466 482 500 518 539 566 593 620 646 673 703 732 759 784 807 827 848 870 894 91 9
From: statistic bureau, office of prime minister
304
Table B-11 estimation of population by age and sex from 1990-2020 Unit (1000 ) age (year) 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
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 70,50 3
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 35,48 6
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 2,20 7
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 2,27 8
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 2,35 6
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 2,44 2
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 2,51 6
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 2,60 3
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 2,62 3
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 2,70 0
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 2,70 2
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 2,58 3
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 2,41 0
55-59 996 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 2,21 7
60-64 723 753 797 846 894 936 958 990 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 1,94 4
65-69 531 542 554 574 607 653 686 721 758 799 848 874 898 923 950 984 990 994 1,005 1,026 1,064 1,071 1,081 1,106 1,154 1,229 1,285 1,347 1,412 1,483 1,56 4
70-74 391 406 419 429 438 447 454 466 486 515 553 577 605 639 677 721 746 772 797 820 841 870 892 904 910 914 908 917 946 995 1,06 3
over 75 524 528 533 538 545 552 561 571 583 596 612 630 650 672 697 724 762 800 838 876 915 956 997 1,036 1,075 1,112 1,143 1,175 1,208 1,243 1,27 9 From: statistic bureau, office of prime minister
305
unit Table B-12 estimation of population at midyear by region from 1990-2010 (1000) region 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Bangkok 6,198 6,349 6,495 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 8,677 8,776 8,943 9,122
Bangkok boundary area 2,850 2,931 3,017 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 4,332 4,410 4,503 4,592 middle region(middle area) 2,791 2,811 2,829 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 2,938 2,935 2,933 2,931 2,927 middle region(eastern area) 3,517 3,564 3,613 3,659 3,710 3,757 3,805 3,851 3,890 3,942 3,987 4,025 4,071 4,110 4,152 4,196 4,229 4,271 4,297 4,336 4,382 middle region(western area) 3,180 3,217 3,249 3,278 3,313 3,343 3,370 3,395 3,422 3,450 3,480 3,495 3,521 3,543 3,573 3,591 3,609 3,628 3,638 3,660 3,677 northern region 10,777 10,854 10,924 10,993 11,056 11,122 11,148 11,172 11,200 11,215 11,253 11,262 11,272 11,283 11,284 11,288 11,281 11,273 11,268 11,259 11,250 north eastern region 19,209 19,435 19,652 19,867 20,063 20,246 20,406 20,571 20,734 20,906 21,068 21,203 21,327 21,458 21,584 21,719 21,801 21,914 21,917 22,020 22,157 southern region 7,315 7,413 7,516 7,630 7,744 7,854 7,961 8,061 8,157 8,251 8,341 8,427 8,507 8,592 8,677 8,764 8,836 8,908 8,951 9,039 9,123 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 From: statistic bureau, office of prime minister
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 graph’s 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 R 2 (“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 Data 5 Linear (Data) Y unit 4
3 2
1 0 0 2 4 6 8 10 X unit
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 group’s 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 Data 5 Linear (Data) Y unit 4
3
2
1 0 0 1 2 3 4 5 6 7 8 910 X unit
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 Data 5 Linear (Data) Y unit 4
3
2
1 0 0 1 2 3 4 5 6 7 8 910 X unit
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.
2 Min 5ei ; 5ei " 0 Equation in appendix D-1
If 5ei " 0,
And ; 5ei+ - 5ei- = 0
Intervals of independent variables are equal ; 5ei+ = 5ei-
; 5ei+ * x - 5ei- * x
; 5A+ = 5A-
then, above area = below area
Where
ei = error (difference between observed value and the linear line)
x = intervals of X-axis in graphs
ei+ = error above linear line
ei- = error below linear line
A+ = area above linear line
A- = area below linear line
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 underlying-
housing-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
Housing sale data and linear trend y = 936.66x + 62202 compare with underlying housing needs number R2 = 0.0334 200,000
180,000
160,000
140,000
120,000
100,000 80,000 housing unit 60,000 40,000
20,000
0 19771979 1981 1983 19851987 1989 1991 19931995 1997 19992001 2003 2005 year Housing sale Simulated housing needs number Linear (Housin g sale )
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 housing sales linear line data to prove theory year data value of linear regression (housing unit) equation (housing unit) 1977 63,139 24 63,163 1978 64,075 -15,956 48,119 1979 65,012 -1,237 63,775 1980 65,949 -2,297 63,652 1981 66,885 -20,993 45,892 1982 67,822 -25,941 41,881 1983 68,759 -42,462 26,297 1984 69,695 -19,057 50,638 1985 70,632 -20,546 50,086 1986 71,569 -37,847 33,722 1987 72,505 -19,152 53,353 1988 73,442 -5,991 67,451 1989 74,379 5,652 80,031 1990 75,315 27,020 102,335 1991 76,252 53,436 129,688 1992 77,189 30,812 108,001 1993 78,125 55,961 134,086 1994 79,062 92,192 171,254 1995 79,999 92,420 172,419 1996 80,935 95,681 176,616 1997 81,872 66,759 148,631 1998 82,809 -15,755 67,054 1999 83,745 -54,716 29,029 2000 84,682 -51,608 33,074 2001 85,619 -51,639 33,980 2002 86,555 -49,373 37,182 2003 87,492 -31,407 56,085 2004 88,428 -19,378 69,050 2005 89,365 -19,293 70,072 2006 90,302 -15,302 75,000 total - 9 2,301,616 * 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.