An Asian International Real Estate Review

By Kim Hin David HO

Direct real estate market analysis is a rigorous investigative approach for academic research, and for direct real estate investment research in practice. ‘An Asian International Real Estate Review’ considers the subject in the context of economic theory pertaining to market disequilibria, utilizing data from major cities in Asia as case studies. Such an approach makes it possible to determine what really defines an Asian direct real estate sector. What is being measured? How does it behave (in terms of price and non-price factors)? How is it structured? How effectively does it achieve sustainable total returns? And how does it manage direct real estate market uncertainty?

Direct real estate market uncertainty originates from both the demand-side and the supply-side of the market. The market responds to structural macroeconomic and microeconomic factors that in turn are affected by related public policies. Such factors and policies interact to affect Asian direct real estate in unique ways since the Asian currency crisis of 1997. ‘An Asian International Real Estate Review’ shows that while the details of direct real estate market analysis are different for the various Asian cities (and their direct real estate sectors) owing to their different stages of maturity, underlying principles nevertheless apply. ‘An Asian International Real Estate Review’ also looks at managing direct real estate market uncertainty at the portfolio level via the analytical techniques of direct real estate asset allocation, direct real estate value-at-risk (VaR), real option analysis and pricing.

• Language: English
• Publisher: Partridge Publishing Singapore
• Release date: Oct 20, 2021
• ISBN: 9781482879278

Kim Hin David HO

Dr HO Kim Hin / David is Honorary Professor in Development Economics & Land Economy, awarded by the UK public university, the University of Hertfordshire. He retired end-May 2019 as Professor (Associate) (Tenured) from the National University of Singapore. Professor HO spent the last thirty-one years across several sectors, which include the military, oil refining, aerospace engineering, public housing, resettlement, land acquisition, land reclamation, real estate investment , development and international real estate investing. He spent six years in the real estate career as part of the executive management group of Singapore Technologies at Pidemco Land Limited, and as part of the senior management team of the Government of Singapore Investment Corporation’s GIC Real Estate Private Limited. Seventeen years are spent in the National University of Singapore at the then School of Building and Estate Management, the Department of Real Estate, School of Design and Environment, where his research expertise is in two areas. First is international real estate in the area of risk-return behavior behind international real estate investing in direct and indirect real estate. Secondly, is urban and public policy analysis involving real estate, sea transport, public housing, land and land use. Schooled in development economics and in land economy at the University of Cambridge, England, he has effectively extended these disciplines to examine his two expertise areas. Apart from being well versed in econometrics, his quantitative interests include real estate demand and supply, investment and finance, artificial intelligent modeling in real estate and system dynamics modeling for real estate market analysis and public policy analysis. He is the Member of the Royal Economics Society (U.K.), Academic Member of the National Council of Real Estate Investment Fiduciaries (U.S.), Fellow of the American Real Estate Society (U.S.), member of the American Economic Association (U.S.) and member of the Economic Society of Singapore and the Singapore Institute of Management. He holds the degrees of Master of Philosophy (1st Class Honors with Distinction), Honorary Doctor of Letters and the Doctor of Philosophy from the University of Cambridge, U.K. He has published widely in top international journals and conferences, in chapters of international academic book publishers. Dr Ho has written 11 major books (including this book), undertaken many consultancies and funded research projects. He has written a total of about 275 published works (with 91 in peer reviewed, reputable international journals). He is an editorial board member of the Journal of Economics & Public Finance, Real Estate Economics journal, Journal of Property Research, Journal of Property Investment & Finance, Journal of Real Estate Finance & Economics, the Property Management journal and the International Journal of Strategic Property Management. He has published widely in conferences, Finance, chapters of international academic book publishers, undertaken many consultancies and funded research projects. He is an immediate past Governor of the St Gabriel's Foundation that oversees nine schools in Singapore; and a District Judge equivalent member of the Valuation Review Board, Ministry of Finance, Singapore, and the Singapore Courts.

Book preview

Copyright © 2021 by Kim Hin David HO (Dr) (Professor).

All rights reserved. No part of this book may be used or reproduced by any means, graphic, electronic, or mechanical, including photocopying, recording, taping or by any information storage retrieval system without the written permission of the author except in the case of brief quotations embodied in critical articles and reviews.

Because of the dynamic nature of the Internet, any web addresses or links contained in this book may have changed since publication and may no longer be valid. The views expressed in this work are solely those of the author and do not necessarily reflect the views of the publisher, and the publisher hereby disclaims any responsibility for them.

www.partridgepublishing.com/singapore

Foreword

This book is dedicated to real estate scholastic work, in advancing the greater understanding of real estate market analysis and real estate finance rigorous disciplines. This is because there has been limited research in bringing out clearly the uncertainty or risk, which is quantifiable uncertainty in real estate market analysis. Even real estate market research, which is carried out as an industry practice among private real estate research, is no exception. Another reason is that it has been widely accepted that while the financial revolution has substantially changed many sectors of the financial industry, it has made little impact on real estate development and investment practice as well as scholastic work. Furthermore, while it is readily acknowledged that despite its huge share in the world wealth, real estate research is overall still a poorly researched subject area.

As a result, the industry tends to be dominated by traditional real estate analysts with little understanding of real estate market uncertainty and capital markets. These commentators are widely regarded to spend too much time worrying about local space supply and demand conditions, while totally losing sight of the ever-changing real estate market and capital market conditions.

The theme of this book is real estate market analysis and real estate market uncertainty, which in turn can be optimally managed under the theoretical conceptions of real estate finance, provided the uncertainty is quantifiable. The book thus defines what real estate market analysis is broadly about from the economic framework of market disequilibria and deploys case studies involving major cities in Asia. This framework enables real estate market analysis to attempt what defines an Asian real estate sector; what is being measured; how it behaves (in terms of price and non-price factors); how it is structured and how it effectively achieves its objectives of sustainable total returns and manageable real estate market uncertainty. Real estate market analysis would be different for the various Asian cities (and their real estate sectors), due to their different stages of maturity. Hence, this book divides the discussion of the cities into several main groups, to highlight the differentiation and the extent of real estate market uncertainty:

(NB. RE = real estate)

The corresponding real estate market analyses would point to real estate market uncertainty being derived from internal and external demand and supply sources. These conform to the structural macroeconomic and microeconomic factors that uniquely affect an Asian real estate sector. Managing real estate market uncertainty optimally is achieved at the portfolio level through real estate asset allocation. This is important because the real estate portfolio is able to virtually eliminate the unique (i.e. specific) uncertainties among the various Asian real estate sectors; thus, retaining within the portfolio only the systemic (i.e. market-wide) uncertainty. Apart from real estate asset allocation, the alternative and modern approach to risk management at the portfolio level, is the value-at-risk (VaR) approach. Another modern and important alternative to coping with uncertainty is real option analysis and pricing that help to better define real estate market uncertainty in extent and time. Real option analysis and pricing also represent uncertainty via a decision tree and the risk-neutral probability conception, to comprehend how uncertainty impacts on the value of real estate investment decisions. The pricing of uncertainty is based on the risk-free hedge security conception. These are best examined at the micro level of the investment in a real estate development opportunity on vacant land. They can tell us about the value of waiting to invest when the net present value (NPV) approach would suggest refraining from investment.

Nevertheless, the real estate sectors (markets) in Asia offer promising prospects since the Asian currency crisis of 1997. It is now timely to take stock and assess how the sectors would pan out for the future, well into at least rest the next century.

I am very pleased to present my thinking and research in international real estate with particular emphasis on Asia. The region’s vast potential for real estate is itself a large incentive for international real estate research and education that has inspired me to document the significant work I have done over the years. Much of the work is based on my experience of some twenty-five years.

My passion in international real estate is enkindled during an extended stint in private industry from October 1996 to May 2002. During this period, I had been given the rare opportunity as an academician to spend a very meaningful experience with Temasek Holding’s Pidemco Land Ltd (now known as CapitaLand Ltd), and with the Government of Singapore Investment Corporation’s GIC Real Estate Pte Ltd.

I wish all readers a pleasurable reading of this book, and I thank you sincerely for your support without which the publication of this book would be made even more difficult.

Kim Hin David HO

Ph.D. (Cambridge), M.Phil. (1st Class Honors with Distinction) (Cambridge),

S.E. (USNPS), F.A.R.E.S., A.M.N.C.R.E.I.F., M.A.E.A., M.E.S.S., M.S.I.M

Honorary Professor

(Development Economics & Land Economy)

University of Hertfordshire

Hatfield, UK.

National University of Singapore

School of Design & Environment

2021

Ad Majorem Dei Gloriam

Acknowledgements

The Author wishes to extend his most sincere appreciation to the School of Design & Environment, under the highly able Deanship of the Provost & Chair Professor (Dr) LAM Khee Poh, of the National University of Singapore. The same wish is extended to the University of Cambridge and the University of Hertfordshire in Hatfield, UK. These three tertiary institutions of higher learning and research are globally leading Universities, inspiring and encouraging both modern and contemporary urban and real estate studies.

About the Author

image1.jpg

Professor (Dr) Kim Hin David HO

PhD (Development Economics) (Cambridge), MPhil (1st Cl Hons and a Star for Distinction) (Development Studies & Land Economy) (Cambridge); Honorary Professor (Development Economics & Land Economy) (Uni of Hertfordshire); Honorary Doctorate of Letters (International Biographical Centre) (Cambridge); Systems Engineering (US Naval Postgraduate School), MRES (UK), AM NCREIF (US), FARES (US), MAEA (US), MESS, MSIM. Retired Professor (Associate) (Tenured) (International Real Estate) (Department of Real Estate) (School of Design and Environment) (National University of Singapore, NUS). Home Address: Block 220 Ang Mo Kio Avenue 1 #02-807, Singapore 560220; email address: [e-mail].

Professor HO Kim Hin / David spent 31 years across several sectors, including the military, oil refining, aerospace engineering, public housing, resettlement, land acquisition, reclamation and international real estate investing. 6 years were in Pidemco Land Ltd (now CapitaLand Ltd) and GIC Real Estate Pte Ltd. 17 years were in the NUS School of Design and Environment at the Department of Real Estate. He holds the Master of Philosophy (First Class Honours with the Star For Distinction), Doctor of Philosophy from the University of Cambridge; and the Honorary Professor from the University of Hertfordshire. He has published widely in 275 articles (inclusive of 91 articles in top peer reviewed, international journals; pertaining to real estate investment, real estate development, urban policy, consultancies, public cum private funded research projects and so also published 15 major books. He was governor of the St Gabriel’s Foundation and member (District Judge equivalent) of the Valuation Review Board under the Singapore Ministry of Finance and the Singapore Courts.

Contents

Foreword

Acknowledgements

About the Author

Chapter 1     Real Estate Market Disequilibria Framework

Chapter 2     Real Estate Market Measures

Chapter 3     Asia Real Estate Performance Analysis

Chapter 4     Preamble to the Real Estate Market Analysis of Key Cities in Asia

EMERGING REAL ESTATE SECTORS

Chapter 5     Malaysia Real Estate Market

Chapter 6     Bangkok Real Estate Market

DEVELOPING REAL ESTATE SECTORS

Chapter 7     China Office Sectors – Shanghai and Beijing CBDs

Chapter 8     The Beijing Commodity Housing Sector

Chapter 9     The Beijing Serviced Apartment Market

Chapter 10   Shanghai Commodity Housing Sector

Chapter 11   Shanghai Retail Real Estate Sector

Chapter 12   Tianjin Real Estate Sectors

MATURING REAL ESTATE SECTORS

Chapter 13   Seoul Real Estate Sectors

Chapter 14   Tokyo Real Estate Sectors

Chapter 15   The Hong Kong Office Sector

Chapter 16   The Hong Kong Mass Residential Market

Chapter 17   Hong Kong Hotel Sector

Chapter 18   The Singapore Office Sector

Chapter 19   The Singapore Residential Sector

MANAGING REAL ESTATE MARKET UNCERTAINTY

Chapter 20   Real Estate Asset Allocation

Chapter 21   Real Options and Some Closing Questions

153329.png

Chapter 1

Real Estate Market

Disequilibria Framework

There are quite a number of models that are available to analyze economic sectors at the local, regional and national levels, and the associated policy implications. They usually adopt a general equilibrium framework that works on the notion that a set of fragmented markets exists and operates smoothly, except for whatever policy restrictions are imposed, for e.g. price ceilings, floors and rationing. In other words, the general equilibrium framework operates solely on the mutual adjustment of quantities and prices at each point in time. Now, a modest start is taken in this chapter to develop a conceptual framework that moves away from that of the general equilibrium. The conceptual framework of analysis in this chapter seeks to link the different sectors of the national economy. Sloman (1991) explains that general equilibrium is where equilibrium exists in all markets, i.e. a situation where all the millions of markets throughout the economy are in a simultaneous state of equilibrium. The process of price and quantity adjustment continues to restore general equilibrium as a result of any change in the conditions of demand and supply. If there is perfect competition and an absence of externalities (the latter referring to third-party’ utilities or disutilities) then the economy, when in general equilibrium, would be socially efficient. A state of general Pareto optimality would exist. In this perfect market economy, should any change in the conditions of demand or supply occurs (tastes change and technology changes for instance), general equilibrium would be temporarily destroyed and Pareto optimality would no longer exist.

Pareto optimality is then restored not by government intervention but rather by the individual actions of producers, consumers and factor owners all seeking their own self-interest. The entire process of adjustment continues, being merely confined to the individual actions of producers, consumers and factor owners all seeking their own self-interest, until general equilibrium and Pareto optimality are restored. The process of restoring equilibrium does not need to resort to any form of government intervention. However, the restoration of a socially-efficient general equilibrium is an elusive one in the real world. Part of the problem is that markets tend to take a long time to adjust to disequilibria, given the short-run immobility of factors. Part of the problem is also the lacking of perfect competition, and the presence of externalities. The problem is further compounded unless there is government intervention via adopting an appropriate bundle of public (urban) policies. Because of these inadequacies, the general equilibrium framework is not considered for further scrutiny in developing the conceptual framework of this chapter.

The conceptual framework of analysis in this chapter therefore investigates the dynamic interaction between real estate markets (sectors), the domestic economy and public policy. It goes beyond a mere description of the economic structure of a domestic economy in the way that an econometric model attempts to do so. The elements that make up an econometric model are usually represented by ordinary least-square equations that include behavioral ones, definitions and identities. The elements can also be represented by an input-output model, which is based on single-point observations. In general, however, model building is conducted for various purposes - the four main ones seem to be pure economic science, economic forecasting, government revenue forecasting, and policy analysis (including impact analysis). Often, one purpose dominates but the model is prone to highly uneven treatment, owing to the special interest of the modeler. This helps to explain why a model’s ability to track the economy varies so greatly from one variable to another. Hence, the conceptual framework is to be designed to incorporate the dynamics of micro-, macro- economic forces and, where possible and subject to information being readily available, the policy handles that operate as a whole to capture the special features of a particular country’s (or city’s in the case of Singapore for instance) national economic development. Internal and external demand can be represented in the Keynesian pattern of output, expenditure, income and employment. It is equally important to consider the adjustments on the supply-side of the domestic economy in response to demand. Conceptualized under demand-side and supply-side considerations, the dynamic interaction between the real estate sectors, the domestic economy and public policycan thus be better understood.

Real Estate and Public Policy

Prices in the form of land prices, real estate prices, transportation costs and wages affect location rent to which investment responds. Location rent as a real estate market outcome or output is determined by real estate demand and real estate supply regardless of speed of adjustment. Real estate demand itself depends on personal income and government policies such as income tax policy, consumption tax policy, property tax policy, corporate tax policy, residential construction expenditure policy, etc. Real estate supply consists of units completed and under construction for the office, retail, residential, industrial warehousing sectors. These real estate sectors constitute the wider real estate market.

Public policy in turn also affects non-earnings income comprising fringe benefits (normally incorporated into the wage rate), unemployment insurance, other transfer payments and real estate income (ie. dividends, rent, interest). Transfer payments would comprise selective payments based on stringent or relaxed criteria set by the Government, for targeted beneficiaries in society that usually include the destitute, pensioners, unemployed single parents, the mentally retarded, the physically handicapped and the aged. Such payments can also include regular payments to subsidize residential rents, mortgage payments, sale prices or public utilities charges (for water, electricity and gas). The effect of public policy in enhancing wealth accumulation can be readily observed when government offers to qualifying households, a subsidized interest rate for personal residential loans in order to promote affordable home ownership on a large scale (as against renting), in either private or public housing or both. Another observation of this wealth-accumulation outcome is in low-interest government loans extended to a local or national housing authority for its public housing development programme. Wealth accumulation is reinforced and enhanced by the quality of physical infrastructure provision, financial capital for funding requirements and a competitive labour market.

Physical Infrastructure Provision

This section is concerned with physical infrastructure, which can be provided by the public and private sectors. The extent of public infrastructure provision is determined by expenditure policy under the government budget. The provision would cover port infrastructure; inland transportation network such as roads, rail and bridges that enhances public housing infrastructure; public utilities; health infrastructure; education infrastructure; parks and leisure infrastructure; seaport and airport infrastructure. These infrastructure projects need not be under the sole purview of the public sector. At the same time, private sector investments can take root in them through joint ventures or full equity participation if there is business profit to be made and the cost of capital is adequately attractive.

Financial Capital

Financial capital has its source from savings under different forms. There is the business saving charged by business firms for the capital consumption allowances. When new output is produced, a part of the nation’s capital equipment is worn out and business firms deduct a capital consumption allowance from their income to provide funds to replace worn-out equipment. The net business saving refers to that part of a corporation’s income withheld as undistributed profit. Personal income after allowing for personal taxes and personal consumption expenditures, produce personal saving that is another financial capital source. Finally, the government surplus represents the difference between government tax revenue and government spending. The financial capital so obtained provides the working and venture capital to spend on more capital goods (ie. investment, indigenous or inward).

Labour Market

Industry output, lagged or otherwise, stimulates desired labour use (or labour demand) which is also affected by investment demand, indigenous or inward investment in new capacity, wage rates which may be industry specific, possible skill requirements which may be industry-specific, and the level of employment. Wage rates, employment, domestic prices, land and real estate prices affect the competitiveness of all the enterprises in the various economic sectors. Wage rates and employment also determine the wage bill, the major part of personal income that in turn affects the demands for all of the five industrial categories. In response to labour demand, the labour supply adjustment mechanisms are the migration rate, the commuting rate and the rate of natural increase. Migration and commuting can be expected to change fast enough to affect wage rates in the short run. Migration may depend on wage rates, personal income, level of employment and domestic prices (consumer price index as a proxy). The labour supply would normally be represented by the labour force participation rate (or activity rate).

As a result of the foregoing discussion, it can be readily inferred that while a key purpose is to develop a disequilibrium model of the real estate market in order to explain and forecast house price movements, it may well be meaningful to look at other structural models. Many real estate markets in Asia can be understood within the context of the disequilibrium model, owing partly to the existence of illiquid markets and partly to market non-transparency, given a persistent vacancy problem and the corresponding uncertainty of rental income. Price and quantity adjustments are inclined to dampen to differing extent in order to clear the market disequilibrium. Volatility is introduced in the real estate market in the short run. A long-run equilibrium model facilitates better understanding of Asian real estate markets as it is less centred on the short-run market dynamics. Nevertheless, disequilibrium models set the path for further research with regard to several aspects of the price discovery process by which real estate market performance is formed. These aspects, apart from the speeds of price-quantity adjustment, are primarily due to the real estate market informational inefficiency in the form of liquidity variation, market timing or other institutional factors. This can therefore explain why the real estate markets in Asia react differently among themselves.

Structural Models

Many models, particularly, of the housing market sector originated from large-scale econometric models of the entire economy (Alberts, 1962; Maisel, 1963; Brady, 1967). These models, developed in the 1960s, were usually formulated within the Keynesian framework. Housing demand was modelled under Keynesian investment demand albeit from a supply perspective, with a central focus on housing starts and completions. Although partially complete, these models are rudimentary. The competitive theory of housing was subsequently developed (Muth, 1960; Olsen, 1969). The heterogeneity of housing is defined using the concept of housing services where one standard unit (i.e. stock) of housing would correspond to one standard unit of housing service flow per unit time. The durability issue is not a consideration while tenure choice is defined in terms of the different estimates for rents and owner-occupiers. The competitive theory has been used in the estimation of demand and supply elasticities, with estimates varying widely (DeLeeuw, 1971). However, no clear relationship is made between tenure choice and the consumption decision (Henderson and Ioannides, 1986). Nevertheless, the stock adjustment models (Muth, 1960) takes the durability issue into consideration through modelling a two-period adjustment instead of a one-period equilibrium models.

For simplification in calculus form for ease of understanding, the desired stock at time t, S*t, may well be expressed as a function of several variables Xt, e.g. price, income, credit and demographic variables. Thus,

152755.png

where a is a vector of parameters and ut is the error term. Actual stock St is envisaged to adjust towards desired stock at the rate 162971.png :

152748.png

where 163004.png (0< 163019.png <1) is a speed of adjustment coefficient while actual stock is envisaged to be the sum of previous stock, appropriately depreciated at constant rate 163044.png , and the new completions q(t) so that

152741.png

Equation 1-3 above can be solved for the new completions q(t) via substituting eq (1) into eq (2) second and the resulting equation substituted into eq (3), after placing qt onto the right hand side.

An improvement is to let the speed of adjustment coefficient 163067.png to depend on some other variables instead of being a fixed constant value. The solution then becomes more complex (Rowley and Trivedi, 1975). In the same manner, the depreciation rate can be allowed to vary in more complex solutions. The stock-flow model is an improvement of the stock adjustment model in which a flow equation is to be incorporated. It thus enables a price (i.e. capital value) equation to be estimated. It means that residential completions qt can be expressed as a function of previous housing starts, HSt-i, of the functional form qt = 163440.png [ 163472.png i HSt-i] in which i runs from zero to n years. By substituting this functional form into the a stock supply expression, then setting demand to be equal to supply at time t, Dt = St, a price equation P(t) is expressed in terms of all other observable variables.

Disequilibria models

A main limitation of the stock-flow models is the assumption of equilibrium (from Dt = St). Markets are expected to clear in the short run through price or quantity adjustments. In the 1950s, Blank and Winnick (1953) argued that residential markets may well be in disequilibria. Disequilibria occur whenever there is uncertainty, be it a price change to be perceived as transient or permanent, an information search, given the imperfections of the housing market, is a time-consuming process. On the whole, real estate markets are characterised by disequilibria. In the case of excess demand (or low vacancy rate), stock supply tends to be inelastic) and an upward shift in demand would result in a substantial rise in prices. When vacancy rates rise, there is considerable excess supply. Quantity adjustment may not be enough owing to production lags (typically two to four years for project completion). There are then two possible outcomes.

The first outcome is that the market is in a short-run disequilibrium with excess supply. Thus, the short-run adjustment towards long-run equilibrium is through prices. However, price adjustment tends to be slow but once excess supply builds up and oversupply is expected to be no longer temporary, then price adjustment tends to be rapid. Real estate developers cut prices steeply to clear excess housing units. The second outcome is that neither quantities nor prices adjust so that the real estate market is in long-run disequilibrium. The expectation is that in adjusting towards long-run equilibrium, the real estate market is continually disturbed by external shocks. In practice, most real estate markets are in between the two possible outcomes. Hence, real estate markets are inclined to adjust towards equilibrium if demand is changing at a relatively constant rate and supply adjusts to demand. The effect of external shocks can be substantial and adjustment towards a stationary equilibrium never occurs. One possibility is to model the price adjustment as a function of excess demand, that is 163478.png Pt = f(Dt-St), with the linear functional form being the most common (Fair, 1971). More complex functions forms e.g. the cubic equation, have been suggested (Hendry, 1984).

Long-Run Equilibrium Model

In their book entitled ‘Urban Economics and Real Estate Markets’, D. DiPasquale and W. Wheaton (1996), presented a framework (see Fig 1.1) depicting the long-run equilibrium of the real estate market. The four-quadrant framework seeks to take account of the various impacts of the broader economy on the real estate market, and its corresponding movement towards long-run equilibrium.

Fig 1.1 A Long-Run Equilibrium Framework

83739.png

(Source: DiPasquale and Wheaton, 1996)

Starting from the northeast quadrant, rent is determined in the property market when the demand for space, D is equal to the stock of space, S. Demand is a function of rent [R] and the conditions in the economy, D[R, Economy] = S.

With rent determined in the property market, investors can subsequently establish the corresponding capitalization rate (really the interest rate [i] and returns in the broader capital market for all assets), to determine the price [P]. Thus, the northwest quadrant represents the asset market and is where the price in the asset market is determined, ie. P=R/i. The price line P in the Asset Market represents the capitalization rate for real estate assets, ie. the ratio of rent-to-price. This is the current yield that investors demand in order to hold real estate assets. Generally, the capitalization rate is made up of the long-term interest rate, the expected growth in rent, the risks associated with rent and the treatment of taxes. A higher capitalization rate is represented by a clockwise rotation of the price line P, while a decline in the capitalization rate is represented by a counter clockwise rotation.

Having determined rent in the property market and price in the asset market, the next step is to determine the associated level of new construction. The southwest quadrant is that portion of the asset market in which the creation of new assets is determined. Here, the curve f(C) represents the replacement cost of real estate and therefore the latter’s price. The cost of replacement through new construction is assumed to increase with greater building activity (C), and so the curve proceeds along a southwesterly direction. It intersects the price axis at that minimum dollar value (per unit of space) required to get some level of new development underway.

Given a certain price level for real estate assets from the northwest quadrant, a line projected down to the replacement cost curve and then crossing over horizontally to the vertical axis determines the level of new construction in which replacement costs equal asset prices. Lower levels of construction would lead to excess profits, whereas higher levels would be unprofitable. Hence, new construction occurs at that level, C, at which the asset price P is equal to the replacement cost f(C) ie. P=f(C).

Finally, in the southeast quadrant, the annual flow of new construction, C, is converted into a long-run stock of real estate space. The change in the stock, 163483.png S, in a given period is equal to new construction minus looses from the stock measured by the depreciation rate, d, ie. 163510.png S=C- 163543.png S. The curve that proceeds away from the origin along a south easterly direction represents that level of stock that requires an annual level of construction for replacement just equal to that value on the vertical axis. At that level of stock and the corresponding level of new construction, the stock of space will be constant over time because the depreciation rate will equal new net construction. Hence 163555.png S=0 and S=C/ 163563.png .

In effect, the DiPasquale-Wheaton framework shows that starting with a certain level of stock, the real estate market determines rent that is then translated into a corresponding property price by the asset market. These asset prices, in turn, generate new construction, which in the property market, eventually yields a new level of stock. The combined property and asset markets are in equilibrium when the starting and ending levels of stock are the same.

The Macroeconomy Relationship

From the DiPasquale-Wheaton framework depicting how the real estate market functions, it is readily observed that any movement in Gross Domestic Product (GDP), prime lending rate (PLR) and the supply of new private housing are bound to have an impact on the private residential market, and in particular, private housing prices. These cause-effect relationships are illustrated in Fig 1.2 that presents the conceptual framework.

Fig 1.2 Conceptual Framework

83963.png

(Source: Ho and Cuervo, 1999)

With regard to Fig 1.2 it can be readily observed that as an economy expands, national income rises and the outcome should be a greater demand for housing space. For a given level of housing space, rents must therefore rise if the demand to use space is to be equal to the available space. These higher rents then lead to greater house prices, which in turn generate a higher level of new construction. Eventually, this leads to a greater stock of space associated to a new market equilibrium. Demand for housing has been widely studied, and it has been shown that income is an important determinant of residential price movements, which in turn depends on the economic well being of the country (Ong and Teck, 1996).

Economic theory suggests that interest rates and house prices are inversely related. Generally, reduced interest rates tends to increase housing demand, and therefore pushes up house prices. This effect is, however, softened by a similar increase in the supply of housing in response to higher house prices and lower construction financing costs, ie. interest rates. Thus, interest rates influence house prices through the demand for, and supply of private housing.

The development of new housing is expected to be positively related to private housing prices since developers, in general, undertake a commitment to start a project based on an estimate of what demand will be at the projects’ completion (McCue and Kling, 1994). Increases in residential prices serve as a signal to developers that the demand for housing is good and, therefore, the developer increases the supply of housing. There is however a delay - a time to build problem - inherent in the development of new housing units because of the nature of the industry requiring detailed plans, approvals from the competent authorities, financing, and the structures have to be constructed. Overall, new construction should have a positive effect on returns (on prices) in the short run (ibid.). Thus, adjustments to the stock of housing (i.e. new construction), occurs only slowly over time and often with lags. Such housing stock adjustments respond to the prices determined by the market’s short-run equilibrium (DiPasquale, D. and W. Wheaton, 1996).

Given the dynamics of private house prices and the corresponding adjustments expected of certain macroeconomic and property specific variables, namely, real gross domestic product (GDP), the prime lending rate (PLR) and the number of private housing starts (PST), the overall contemporaneous relationship among these variables will have to incorporate an Error Correction Mechanism (ECM) in the Vector Autoregressive (VAR) model. The ECM is necessary to account for the short-run dynamics of all the variables included in the cointegrating regression, ie. the VAR model.

Concept of Stationarity

In developing time series models it is essential to know whether the underlying stochastic process that generated the series can be assumed to be invariant with respect to time, that is, if it is stationary (Pindyck and Rubinfeld, 1991).

Stationarity of the time series is a desired characteristic in order to use standard estimation techniques, such as ordinary least squares (OLS) for the estimation of regression relationships. Otherwise, the use of non-stationary time series variables may give rise to so called spurious regression (Granger and Newbold, 1974), from which no appropriate inferences can be drawn (Hataiseree, R., 1993).

A spurious regression has a high coefficient of determination (R-square) and significant t-statistics but the relationship contained in the regression itself is not underpinned by economic theory. Phillips (1986) mentions that it has been shown in a number of theoretical works in which the statistical properties of regression analysis using non-stationary time series (i.e. a series with a unit root) are dubious (Charemza W. and D. Deadman, 1993).

In practice, however, most economic time series exhibit a trend and are therefore non-stationary. To overcome this problem, the non-stationary time series is differenced d number of times until it is transformed to a stationary time series. A non-stationary series which can be transformed to a stationary series by differencing d times is said to be integrated of order d (Engle and Granger, 1987). Thus, a series Xt integrated of order d is conventionally denoted as Xt ~ I(d). If Xt is stationary in levels, then no differencing is necessary, and Xt ~ I(0). This practical solution to unit root problems was first proposed by Box and Jenkins (1970).

Fortunately, most economic time series data can usually be transformed to a stationary series after differencing once or twice. Although differencing removes the problem of spurious regressions, it holds the inherent problem of removing valuable information concerning long-run relationships among the variables (Davidson, et. al., 1978). This concern, however, will be dealt with by incorporating an error correction mechanism in a time series model.

Autocorrelation Function

An informal test of stationarity is carried out by inspecting the sample autocorrelation function ( 163576.png k) over lagged periods (k) plotted in a correlogram. The 163587.png k indicates how much correlation there is between neighboring data points in a time series and is defined as

152708.png

Where the numerator is the covariance between Yt, a variable at time t, and Yt+k and the denominator is the variance of the stochastic process in Yt.

If the 163595.png k falls off rather quickly toward zero as the number of lagged periods (k) increases, the time series is typically stationary. On the other hand, if 163603.png k does not fall off quickly as k increases, the series is nonstationary (Pindyck and Rubinfeld, 1991).

Unit Root Tests

Due to the imprecise nature of using the correlogram of the autocorrelation function ( 163619.png k) to detect the possible presence of a unit root, more formal tests are necessary to determine whether a time series is stationary (i.e. it does not contain a unit root).

For example, consider a series that is generated from the following first-order autoregressive model,

152437.png

Where { 163641.png t} is generated from a white-noise process. An error sequence { 163670.png t} is a white-noise process if each value in the sequence has a zero mean, a constant variance and is serially uncorrelated.

Testing the null hypotesis that a1= 0, eq (5) can be estimated using ordinary least squares (OLS). The condition that { 163697.png t} is a white-noise process and |a1| < 1 ensures that the {Yt} sequence is stationary and the estimate of a1 is efficient. By calculating the standard error of the estimate of a1 a t-test can be used to determine whether a1 is significantly different from zero.

However, if one wants to test the null hypothesis of a unit root process, such that a1=1, against the alternative hypothesis |a1|< 1, the {Yt} sequence is seen to be generated by the nonstationary process:

116507.png

Thus, if a1=1 then the variance becomes infinitely large as t increases, and it is not appropriate to use OLS to estimate and perform the test on the coefficient a1. Hence, the usual t-test cannot be used to test the null hypothesis that a1=1.

Fortunately, David Dickey and Wayne Fuller have introduced the more formal unit root test, hereafter known as the Dickey-Fuller (DF) test for time series data (Dickey, D. and W. Fuller, 1979; 1981).

The DF test considers three regression equations to test for the presence of a unit root (Enders, 1995):

152426.png

Eq [6.1] is a pure random walk, eq [6.2] is the random walk with an intercept and eq [6.3] includes both an intercept and the linear time trend (T) apart from the random walk.

The coefficient of interest in all the above equations is 163699.png , such that if 163705.png =0 the {Yt} sequence contains a unit root, and is therefore non-stationary. The test involves estimating one (or more) of the equations using OLS in order to obtain the estimated value of 163707.png and the corresponding standard error.

Comparing the computed t-statistic with the appropriate value reported in the DF tables, it can be known whether to accept or reject the null hypothesis that 163733.png =0. For the sake of obtaining a stationary time series it is necessary to reject the null hypothesis of a unit root. This methodology is the same irregardless of whether the equations to be tested contains an intercept and/or a time trend as in eqs (6.2) and (6.3) respectively.

Dickey and Fuller (1981) provides three additional F-statistics (called 163755.png 1, 163769.png 2 and 163802.png 3) to test for the joint hypotheses on the intercept (a0), time trend (a2) and unit root ( 163813.png ) for eqs (6.1), (6.2) and (6.3) respectively.

The complete set of test statistics and their critical values for a sample size, say of 50, is summarized in Table 1.

Table 1. Table for DF statistics for joint F-test

167624.png

Notes: The critical values are for the 95% and 99% confidence interval.

            This table was adapted from Enders (1995), p. 223.

A weakness of the original DF test is that it does not take account of possible autocorrelation in the error process { 163884.png t}. If 163886.png t is autocorrelated (that is, it is not white noise), then the OLS estimate of eqs (6.1), (6.2) and (6.3) are not efficient. Dickey and Fuller (1981) provides a simple solution by using lagged left-hand side variables as additional explanatory variables to approximate the autocorrelation. This test is known as the Augmented Dickey-Fuller test, hereafter known as the ADF test, which is widely regarded as being the most efficient test for stationarity and hence, cointegration (Charemza and Deadman, p. 135, 1993). Thus eqs (6.1), (6.2) and (6.3) are transformed to the ADF equivalent as:

152418.png

Where 163888.png is the differencing operator, 163890.png t is the error term that is independently and identically distributed (i.i.d.) and ao, a2, 163892.png i and 163894.png i, i = 1,2....k are the regression parameters which can be estimated using the OLS method. k represents the number of lags, and sufficient lags should be included in the autoregressive (AR) process such that the error term will approach white noise in relatively short lags.

The practical rule-of-thumb for establishing the value of k (the number of lags for 163896.png Yt-i ) is that it should be relatively small in order to save on the degrees of freedom, but large enough to allow for the existence of autocorelation in 163898.png t. One can select the appropriate lag length by starting with a relatively long lag length and paring down the period of k for the model by the usual t-test and/or F-tests.

The t-ratio of the lagged level term is compared with the critical value to test the null hypothesis of a unit root. Take note however that the t-ratio statistic used in the ADF test is not the ordinary student t distribution but rather a Monte Carlo generated simulation of the distribution. Thus, if the computed t ratio is greater than the critical value, the null hypothesis of a unit root is rejected, and the series is deemed to be stationary.

Testing for the Order of Integration

Besides testing whether a data time series is stationary, the unit root tests are also used to determine the order of integration of a time series.

The first step in this unit root test is to check whether the time series in its levels, Xt is stationary. If the null hypothesis of a unit root cannot be rejected, the times series Xt is differenced once (i.e. 163900.png Xt = Xt ‒ Xt‒1 ) and then tested for the presence of a unit root again. This process is continued by increasing the order of differencing until the ADF test shows that the unit root series has been transformed into a stationary time series. Thus, if a time series Xt achieves stationarity after differencing once, it is said to be integrated of the order of one, i.e. Xt 163902.png I(1). Most economic time series are either integrated of an order of one or two (Nelson and Plosser, 1982).

Dickey and Pantula (1987) suggest a simple extension of the basic procedure if more than one unit root is suspected (Enders, 1995). If two roots are suspected, the following equation is estimated:

152410.png

Using the appropriate t-statistic, one has to determine whether 163904.png 1 is significantly different from zero. If the null hypothesis that 163906.png 1 is equal to zero cannot be rejected, then conclude that the time series {Yt} is I(2). If 163908.png 1 differs from zero then go on to determine whether there is a single root by estimating the following equation:

152400.png

Since there are not two unit roots, one should find that 163910.png 1 and/or 163912.png 2 differ from zero. Under the null hypothesis of a single unit root, 163951.png 1 < 0 and 163953.png 2=0; under the alternative hypothesis, {Yt} is stationary, so that 163955.png 1 and 163957.png 2 are both negative. Thus, estimate eq (11) and use the DF critical values to test the null hypothesis 163959.png 1=0. If you reject this null hypothesis, conclude that {Yt} is stationary.

Cointegration and the Error Correction Mechanism

The formal concept of cointegration was introduced by Engle and Granger (1987). Cointegration refers to a linear combination of nonstationary variables as proposed by economic theory, to form a stationary variable. Thus, for example, economic theory tells us that consumption expenditures and disposable income should move together over the long run, such that a linear combination of the two random walk variables should be stationary when combined.

Take the simple case of two non-stationary time series, xt and yt. If there exists a non-zero constant 163961.png such that zt = xt‒ 163963.png yt is a stationary or I(0) process. Then, xt and yt are said to be cointegrated with a cointegrating parameter 163965.png .

For the general case (Enders, 1995) considers a set of economic variables in long-run equilibrium when:

116715.png

If we let 163967.png and Xt denote the vectors ( 163969.png 1, 163971.png 2,... 163973.png n )

and (X1t, X2t,... Xnt ) respectively, the system is in long-run equilibrium when 163975.png Xt =0. The deviations from long-run equilibrium - called the equilibrium error - is 163977.png t, so that 163979.png t 163981.png Xt. If the equilibrium is meaningful then it must be the case that the equilibrium error process is stationary, and any deviations from long-run equilibrium is temporary in nature.

Engle and Granger (1987) provide the following definition of cointegration:

The components of a vector 116737.png are said to be cointegrated of order d, b denoted as Xt 163983.png CI(d, b), if: (i) all components of Xt are I(d); and (ii) there exists a vector 163985.png 0, where 116752.png such that the linear combination results in a relationship,

116769.png

that has an order of integration of I(d-b), where b>0. The vector 163987.png is called the co-integrating vector.

The relationship between error correction models and cointegration was first demonstrated by Granger (1981) in what is widely known as the Granger representation theorem. The theorem states that if a set of I(1) variables is cointegrated, there always exists an error correction mechanism (ECM) representation among the variables concerned (Engle and Granger 1987). Thus, one can estimate an ECM that takes into account the short-run dynamics of all variables included in the cointegrating regression.

As shown by Charemza and Deadman (1993), suppose that Xt, Yt are both I(1) and that the long run relationship between them is given by Yt 163989.png 163991.png 163993.png Xt. An important situation occurs if the variables Xt, Yt are CI(1,1) and have the cointegrating vector [ 163995.png , -1], so that the deviations of Yt from its long run path Yt 163997.png are I(0). If this is the case then a model in first differences incorporating an ECM can be developed. Specifically, for the equation:

152379.png

Here both the dependent variable 163999.png Yt and the regressors 164001.png Xt and (Yt‒1 - 164003.png Xt‒1 ) are I(0). This model incorporates both a long run solution and also an ECM when 164005.png 2 is negative. Consequently, in a long run relationship between two variables both must be integrated of the same order if the error term is to be I(0). Stationarity of the error term is important if one is to examine models incorporating an ECM as eq (12).

Hataiseree (1993) presents the simple case of two variables, Xt and Yt, where the ECMs can be expressed as:

152370.png

where Zt = Xt ‒ 164007.png Yt is I(0) and represents the error correction term which displays the size of the preceeding error derived from the long run equation. 164009.png 1t and 164011.png 2t are white noise and at least one of the speed of adjustment parameters, 164013.png 1 or 164015.png 2, is not equal to zero. In this setting, Xt and Yt are functions of distributed lags of first differences of Xt and Yt as well as the one-period lag of the error-correction term.

The error correction eqs (13) and (14) may be interpreted as the disequilibrium Xt = 164017.png Yt in the long run. It will therefore be noted that the ECM requires that both Xt and Yt be cointegrated, so that eqs (13) and (14) are in equilibrium with all the components

(i.e. 164019.png Xt 164021.png Yt 164023.png and Zt ) being I(0). Thus, as Engle and Granger (1987) has shown, all terms in the error correction models are I(0); and the converse is also true such that if Xt 164025.png I(1) are generated by an error correction model, then Xt is necessarily co-integrated.

Testing for Cointegration: Engle-Granger Methodology

The following steps are proposed by Engle and Granger (1987) to determine whether there exists an equilibrium relationship betwen two variables, Yt and Zt which are integrated of order 1 (Enders, 1995):

Step 1: The first step is to pretest each variable, Yt and Zt to determine its order of integration. The DF and ADF tests can then be used to know the number of unit roots present in each of the variables to be tested. The variables have to be integrated of the same order in order to conclude that they are co-integrated.

Step 2: If both Yt and Zt are I(1), the next step is to estimate the long-run equilibrium relationship given by:

152355.png

In order to determine if the variables Yt and Zt are cointegrated of order (1,1), one has to test if the estimated residuals 116912.png are found to be stationary. Using a DF test on the autoregression of the residuals, one can conclude whether the residual is stationary. Thus:

152348.png

If one can reject the null hypothesis Ho: a1 = 0, then one can conclude that the residuals does not contain a unit root (i.e. it is stationary), and therefore conclude that Yt and Zt are cointegrated. Engle and Granger provide test statistics to test the hypothesis that a1 = 0.

If the 164027.png t series exhibits serial correlation then an ADF test will have to be used. Thus, instead of using the results from eq (16), one will have to estimate the following autoregression equation:

152340.png

And if -2<a1<0, one can conclude that the residuals are stationary and the time series Yt and Zt are cointegrated of order CI(1,1).

Step 3. Estimate the error correction model.

Since the variables Yt and Zt are cointegrated, the residuals from the equilibrium regression can be used to estimate the error correction model in the following form:

152317.png

The value of the residual 116992.png estimates the deviation from long-run equilibrium in period (t-1). Hence, it is possible to use the saved residuals 117003.png obtained in STEP 2 as an instrument for the expression Yt ‒1 ‒ 164029.png t ‒1 in eqs (18) and (19). Thus, using the saved residuals from the estimation of the long-run equilibrium relationship one can estimate the error-correcting models as follows:

152310.png

From the above equations, one would note that:

(i) The OLS is an efficient estimation strategy since each equation contains the same set of regressions.

(ii) Since all the terms in eqs (20) and (21) are stationary, the test statistic used in traditional vector autoregressive (VAR) analysis is appropriate for these equations. Thus, the restrictions that all 164031.png jk 164033.png i 164035.png 0 can be checked using an F-test. If there is a single cointegrating vector then restrictions concerning 164037.png y or 164039.png zcan be conducted using a t-test.

Step 4: Assess model adequacy.

The following procedures will help one determine whether the estimated error-correction model is appropriate:

(i) One should be able to determine whether the residuals of the VAR approximate white noise. If the residuals are serially correlated, one should re-estimate the model using longer lag lengths.

(ii) One should be able to determine that the value of the speed of adjustment coefficients 164041.png y and 164043.png z lead to direct convergence. Direct convergence necessitates that 164045.png y be negative and 164047.png z be positive for eqs (20) and (21) respectively. If say 164049.png z in equation [3.19] is zero then the change in 164051.png t does not at all respond to the deviations from long-run equilibrium in (t-1). If 164061.png z is zero and all 164063.png 21 (i) = 0, then it can be said that {∆Yt } does not Granger cause { 164065.png t}. It is interesting to note that one or both of these coefficients should be significantly different from zero if the variables are co-integrated.

Testing for Co-integration - The Johansen Methodology

Although the two-step Engle and Granger methodology for co-integration testing is rather straightforward, it does have several shortcomings (Enders, 1995). First, the method does not have a systematic procedure for the separate estimation of the multiple co-integrating vectors. In practice, it is possible to find that one regression indicates the variables to be co-integrated, whereas reversing the order indicates no co-integration. This feature should not be so since the test for co-integration should be invariant to the choice of the variable selected for normalization.

The second shortcoming of the Engle and Granger methodology is that it relies on a two-step estimation process. Thus any error generated in the first step to generate the error series is carried over to the second step.

Fortunately, Johansen (1988) and Johansen and Juselius (1990) have introduced the Johansen methodology to do away with the two-step estimation process used in the Engle and Granger methodology. Moreover, here there is no need to specify endogenous and exogenous variables.

The Johansen methodology is based on a maximum likelihood procedure that determines the number of cointegrating vectors in a vector autoregressive (VAR) process to be estimated.

The first requirement in this method is for all the variables to be integrated of the same order. Having established this, the second step is to determine the appropriate lag length to be used in the VAR model. The method used in this study is to start with a relatively large lag length and pare down the model by the usual t-test and/or F-test. For example, one could estimate an equation using a lag length of k*. If the t-statistic on lag k* is insignificant at some specified critical value, reestimate the regression using a lag length of k*-1. The process is repeated until the lag is significantly different from zero. This procedure will yield the true lag length provided that the initial choice of lag length includes the true length.

A brief discussion of the Johansen methodology is presented by Hataiseree (1993):

Given that the vector of n variables, Xt = (X1t,..., Xnt), is generated from the following Vector AutoRegressive (VAR) process of order k:

152293.png

Expressing the VAR model in the error-correction form:

152282.png

To detect the number of cointegrating vectors, Johansen and Juselius (1990) suggest estimating the rank, r, of the 164067.png matrix. The rank indicates the order of cointegration and should not be more than n, the number of variables of the vector Xt. For a (n x n) 164069.png matrix, the rank r may take the following forms: (i) r=n, the matrix 164071.png has full rank implying that the Xt is stationary; (ii) r=0, the matrix 164073.png is zero and a first difference VAR is appropriate; and (iii) 0

Johansen has derived two likelihood ratio statistics for testing the number of cointegrating vectors, r. The first statistic is known as the Trace Test. It is used to test the null hypothesis that there are at most r cointegrating vectors against the alternative hypothesis of r or more such vectors. The second statistic is labelled the Maximal Eigenvalue Test and is employed to test the null hypothesis of at most r cointegrating vector against the alternative hypothesis of r+1 cointegrating vectors.

Johansen and Juselius (1990) provide the critical values of the Trace and Maximal Eigenvalue statistics obtained using simulation studies. In establishing the number of cointegrating vectors, the results of the the Trace and Maximal Eigenvalue Test can conflict. In such situations, the Maximal Eigenvalue Test is preferred as this has a sharper alternative hypothesis. It is usually preferred in determining the final number of cointegrating vectors (Enders, 1995).

After determining the number of cointegrating vectors, r, and knowing the 164089.png matrix, a Vector AutoRegressive structural model which incorporates an error correction vector is estimated in the following form:

ie. regressing

152272.png

The estimates for the coefficients of the variables for eq (24) - a0, 117177.png , and ut - will be used to construct the VECM (i.e. eq 25) which will be used for making the short-term forecast.

152262.png

Of high relevance at this juncture is the observation by Johansen (1995) with regard to the appropriate lag length of up to typically two lags for macroeconomic data:

Our experience with analysing macro economic data is that the models quite often provide an adequate fit of the data.....It is an empirical finding that we have usually been able to do so with two lags for seasonally unadjusted data.

Faced with the pragmatic choice of at most a two-lag length for the VAR model, the next step is to test for the presence of co-integrating relationships.

Finally, the Vector Autoregressive Error Correction Mechanism (VECM) under the Johansen methodology is estimated to make short-term forecasts of changes in private housing prices. The results of the estimated VECM showed that changes in private housing prices is significantly explained by real GDP, changes in prices of private housing prices one quarter back, the price level of private housing two quarters’ back, and the prime lending rate. The forecast of price changes using the VECM is seen to be sub-optimal. However, the forecasted (i.e. ex ante) price changes do represent the price changes in private housing that move in line with the overall growth of economy, after incorporating the error correction mechanism for short-term movements in prices.

In summary, the various econometric models that have been discussed in this chapter offer a comprehensive theoretical underpinning for analysing the real estate markets in the developing countries of Asia. As an integral part of the market analysis, the price discovery process would be able to highlight the market fundamentals, their explanatory factors in the short- and long- run as well as the cyclical behaviour of the predominantly inefficient real estate markets of Asia. It is therefore appropriate to measure the performance of these various real estate markets as presented in Chapter 2. Then it would be meaningful analyse the cyclical pattern and the stationarity inherent to these measures of the real estate markets in Asia.

153329.png

Chapter 2

Real Estate Market Measures

This chapter presents a holistic approach to facilitate real estate market analysis of the countries in Asia that involves several key real estate market measures. While it serves to offer a consistent way to analyse and interpret fundamental real estate market factors, it also serves to uncover some aspects of the price discovery process for the real estate markets in Asia. The measures are typically encountered in the real estate investment advisory industry, by industry players and in scholarly real estate research. In reality, these real estate market measures form the primary research data on which primary empirical analysis would be based, and subsequent real estate investment decisions are made. They are acceptable by researchers in urban economics, real estate economics, real estate finance and public policy.

For the purpose of enhancing the clarity of the real estate market measures, a list is discussed below while furnishing the basis and assumptions deployed in the measurement whenever it is needed.

Rents

Rental definitions vary to some extent but do not present a serious problem on the whole. Within the Jones Land Lasalle family, all the rental definitions ignore unusual lease terms and vacancy (i.e. full occupancy assumed).

JLL REIS (Real Estate Intelligence Service) Asia

JLL REIS Asia mainly measures ‘Net Effective Rent’ in the following manner:

Gross rent is the sum of net rent and outgoings (i.e. management fees, utilities, government rates/taxes) payable by the tenant and based on net floor area (NFA).

Net rent is the market rent stated on the lease on a net floor and annual basis, exclusive of outgoings and landlord incentives. For retail, it is the rent achievable for a prime unit (good location on the best floor or level of the retail centre).

Net Effective Rent is the current market rental income receivable by the landlord, net of landlord incentives i.e. rent-free period, fitting out contributions.

Non-JLL-REIS-Asia Local Convention

Beijing