price indicees eurostat ks ra 12 022 en

5/24/2018 Price Indicees Eurostat KS RA 12 022 En

1/186

Handbook on ResidentialProperty Prices Indices (RPPIs)

2013 edition

M e t h o d o l o g i e s &

W o r k i n g p a p e r s


2/186


3/186

M e t h o d o l o g i e s &

W o r k i n g p a p e r s

2013 edition

Handbook on ResidentialProperty Prices Indices (RPPIs)


4/186

Europe Direct is a service to help you find answers toyour questions about the European Union.

Freephone number (*):

00 800 6 7 8 9 10 11(*) Certain mobile telephone operators do not allow accessto 00 800 numbers or these calls may be billed.

More information on the European Union is available on the Internet (http://europa.eu).

Cataloguing data can be found at the end of this publication.

Luxembourg: Publications Office of the European Union, 2013

ISBN 978-92-79-25984-5doi:10.2785/34007

Cat. No KS-RA-12-022-EN-N

Theme: Economy and finance

Collection: Methodologies & Working papers

European Union, International Labour Organization, International Monetary Fund, Organisation for EconomicCo-operation and Development, United Nations Economic Commission for Europe, The World Bank, 2013Reproduction is authorised provided the source is acknowledged.

Photo credits: PhovoirReproduction of photos is allowed for non-commercial purposes and within the sole context of this publication.

Printed in Belgium

PRINTED ON ELE MENTAL CHLORINEFREE BLEACHED PAPER(ECF)

Also available under the title Handbook on Residential Property Prices Indices (RPPIs),

ILOISBN 978-92-2-127359-2 (paperback)ISBN 978-92-2-127360-8 (PDF)

OECDISBN 978-92-6-419718-3 (PDF)

The opinions expressed and arguments employed herein do not necessarily reflect the official views of the ILO, IMF, OECD,UNECE, the World Bank or of the governments of their member countries or those of Eurostat or the European Commission.
http://europa.eu/http://europa.eu/


5/186

3

Table of contents

Handbook on Residential Property Prices Indices (RPPIs)

Table of contents

Foreword RPPI 7

Preface 9

1. Introduction 11

2. Uses of Residential Property Price Indices 15

3. Elements for a Conceptual Framework 21

4. Stratification or Mix Adjustment Methods 37

5. Hedonic Regression Methods 49

6. Repeat Sales Methods 65

7. Appraisal-Based Methods 73

8. Decomposing an RPPI into Land and Structures Components 81

9. Data Sources 101

10. Methods Currently Used 113

11. Empirical Examples 139

12. Recommendations 155

Glossary 161Bibliography 167

Index 177


6/186

4

Table of contents


List of tables

3.1. Estimated Rent to Value Ratios as Percentages (Capitalization Ratios)........................................................35

4.1. Sample Probability of a Sale in Each Cell .................................................................................................................43

4.2. Matched Model Fisher Chained and Fixed Base Price Indices, Mean, Median

and Representative Model Price Indices ..................................................................................................................45

4.3. Rolling Year Fixed Base Fisher, Fisher Chained Moving Average and Fisher Fixed Base Moving

Average Price Indices ......................................................................................................................................................47

5.1. Log-Linear Time Dummy Price Indices and the Chained Stratified Sample Mean Fisher

Price Index ...........................................................................................................................................................................60

5.2. Linear Time Dummy Price Indices, the Log Log Time Dummy Price Index and

the Chained Stratified Sample Mean Fisher Price Index ....................................................................................62

5.3. Chained Laspeyres, Paasche and Fisher Hedonic Imputation Price Indices............................................... 64

6.1. Repeat Sales Price Index, Chained Stratified Sample Mean Fisher Price Index and Hedonic

Imputation Fisher Price Index ......................................................................................................................................71

7.1. SPAR Index, Hedonic Imputation Fisher Price Index and Repeat Sales Index............................................ 79

8.1. The Price of Land (PL1

), the Price of Quality Adjusted Structures (PS1

), the Overall Cost

of Production House Price Index (P1) and the Fisher Hedonic Imputation House Price Index............. 86


), the Price of Structures (PS2

), the Overall Price Index Using Splines

on Land (P2) and the Fisher Hedonic Imputation Price Index ..........................................................................88



), the Overall House

Price Index with Monotonicity Restrictions on Structures (P3)

and the Overall House Price Index Using Splines on Land (P2) ........................................................................90



)

and the Overall House Price Index using Exogenous Information on the Price of Structures (P4)..... 92

8.5. House Price Indices Using Exogenous Information (P4) and Using Monotonicity Restrictions

(P3), the Chained Fisher Hedonic Imputation Index and the Chained Fisher Stratified

Sample Index .....................................................................................................................................................................93




Price Index using Exogenous Information on the Price of Structures (P4) and their Rolling

Window Counterparts (PRWL

) and (PRW

) ......................................................................................................................95

8.7. Approximate Stock Price Indices and Based on Hedonic Imputation (PStock1

)

and Stratification (PStock2

) and the Fisher Hedonic Imputation Sales Price Index.......................................97

8.8. Approximate Price Indices for the Stock of Houses (PStock

), the Stock of Land (PLStock

),

the Stock of Structures (PSStock

) and the Corresponding Sales Indices (PL4

and P4)....................................99

10.1. Indices of Property Prices Published in Japan......................................................................................................123

10.2. Indices of Residential Property Prices Published in the UK .........................................................................128 10.3. Tenure Status All Housing in South Africa (According to Census 2001)..................................................134

10.4. Distribution of Number of Rooms

in Informal Dwellings ....................................................................................................................................................135

10.5. Price Determinants ........................................................................................................................................................136

10.6. Percentage of Materials Used in the Construction of Informal and Traditional Dwellings

in South Africa .................................................................................................................................................................136

10.7. Evaluation of Barriers ....................................................................................................................................................137

11.1. Means, Medians, Percent Changes, Standard Deviations, and Skewness..................................................141

11.2. Regional Expenditures, Prices and Volumes (Implicit Quantities) Using Median Prices

as the Regional Prices ...................................................................................................................................................143

11.3. Overall House Price Indices using Median Prices and Alternative Formulae to Aggregateover Regions A, B and C ...............................................................................................................................................144


7/186

5

Table of contents


11.4. Regional Expenditures, Prices and Volumes (Implicit Quantities) Using Mean Prices

as the Regional Prices ...................................................................................................................................................144

11.5. Overall House Price Indices using Mean Prices and Alternative Formulae to Aggregate

over Regions A, B and C .............................................................................................................................................145 11.6. Log-linear Regression Results for a Simple Example .........................................................................................146

11.7. Results from a Pooled Regression for Years 2006 and 2007 ............................................................................148

11.8. Results from a Pooled Regression for Years 2006 to 2008 ...............................................................................148

11.9. Results from a Pooled Regression for Years 2007 and 2008 ............................................................................149

11.10. Results from a Regression for 2006 ..........................................................................................................................150

11.11. Results from a Regression for 2007 ..........................................................................................................................150

11.12. Mean Values of the Characteristics for the Base Period (2006)......................................................................151

11.13. Repeat Sales Data ...........................................................................................................................................................151

11.14. Dummy Variables for Repeat Sales...........................................................................................................................152

11.15. Unweighted Repeat Sales Regression ....................................................................................................................153

11.16. Weighted Repeat Sales Regression ..........................................................................................................................153 11.17. Repeat Sales Price Indices (2002 = 100) .................................................................................................................154

11.18. Growth Rates in Percent for the Various House Price Indices (2007)...........................................................154


8/186

6

Table of contents


List of figures

4.1. Matched Model Fisher Chained and Fixed Base Price Indices, Mean, Median

and Representative Model Price Indices ..................................................................................................................45

4.2. Rolling Year Fixed Base Fisher, Fisher Chained Moving Average and Fisher Fixed

Base Moving Average Price Indices ...........................................................................................................................47

5.1. Log-Linear Time Dummy Price Indices and the Chained Stratified Sample

Mean Fisher Price Index .................................................................................................................................................59

5.2. Linear Time Dummy Price Indices, the Log Log Time Dummy Price Index

and the Chained Stratified Sample Mean Fisher Price Index ............................................................................61

5.3. Chained Laspeyres, Paasche and Fisher Hedonic Imputation Price Indices............................................... 63

5.4. The Fisher Imputation Price Index, the Chained Stratified Sample Mean Fisher

Price Index, the Linear Time Dummy Price Index and the Log Log Time Dummy

Price Index ...........................................................................................................................................................................64

6.1. Repeat Sales Price Index, Chained Stratified Sample Fisher Price Index and Hedonic

Imputation Fisher Price Index ......................................................................................................................................70

7.1. SPAR Index, Hedonic Imputation Fisher Price Index and Repeat Sales Index............................................ 79



),

the Overall Cost of Production House Price Index (P1) and the Fisher Hedonic

Imputation House Price Index .....................................................................................................................................86


), the Price of Structures (PS2

), the Overall Price Index

Using Splines on Land (P2) and the Fisher Hedonic Imputation Price Index.............................................. 88




Price Index with Monotonicity Restrictions on Structures (P3) and the Overall House

Price Index Using Splines on Land (P2) .....................................................................................................................89



) and the Overall

House Price Index using Exogenous Information on the Price of Structures (P4).....................................91

8.5. House Price Indices Using Exogenous Information (P4) and Using Monotonicity

Restrictions (P3), the Chained Fisher Hedonic Imputation Index and the Chained

Fisher Stratified Sample Index .....................................................................................................................................93

8.6. Approximate Stock Price Indices and Based on Hedonic Imputation (PStock1

)

and Stratification (PStock2

) and the Fisher Hedonic Imputation Sales Price Index.......................................96

8.7. Approximate Price Indices for the Stock of Houses (PStock

), the Stock of Land (PLStock

),

the Stock of Structures (PSStock

) and the Corresponding Sales Indices (PL4

and P4)....................................98

Diagram: House purchase timeline and house price indices ..........................................................................................103

10.1. Four Residential Property Price Indices for Canada ..........................................................................................120

10.2. Property Information Flow ..........................................................................................................................................125 10.3. Four Residential Price Indices for Japan (January 1999=100)........................................................................126

10.4. House Purchase Time-line ...........................................................................................................................................127

10.5. NHB RESIDEX Indices India Citywise index ........................................................................................................129

10.6. Quarterly National House Price Index for Existing Units Nominal and Real..........................................131

10.7. Quarterly National Real House Price Index for Existing Units Annual Percentage

Changes .............................................................................................................................................................................131

10.8. Annual National House Price Index for Existing Units (1) ..................................................................................132

10.9. Annual Real House Price Index for Existing Units Principal metropolitan areas .................................132

10.10. Annual Real House Price Index for Existing units: Houses with Subsidies (VIS)

and Houses without (NOVIS) ......................................................................................................................................133

11.1: Distribution of House Prices in 2008 .......................................................................................................................141


9/186

7

Foreword - RPPI


Foreword RPPITis Handbook on Residential Property Prices Indices(RPPIs) represents the first comprehensive overview o conceptual

and practical issues related to the compilation o price indexes or residential properties.Te development o the RPPI Handbookhas been co-ordinated by the Statistical Office o the European Union (Eurostat),under the joint responsibility o six organizations - International Labour Organization (ILO), International Monetary Fund(IMF), Organisation or Economic Co-operation and Development (OECD), Statistical Office o the European Union(Eurostat), United Nations Economic Commission or Europe (UNECE) and World Bank - through the mechanism o anInter-Secretariat Working Group on Price Statistics (IWGPS). Te Handbook is published jointly by these organizations.

Te aim o the RPPI Handbookis to give practical guidance on the compilation o house price indexes and to increaseinternational comparability o residential property price indexes. Te Handbookoutlines the different user needs, givesdetails on data needs and methods, and provides recommendations. Te primary purpose o the Handbookis to assistproducers o residential property price indexes, particularly in countries that are revising or setting up their RPPIs. TeHandbookdraws on a wide range o experience and expertise in an attempt to describe practical and suitable measurementmethods. It should also help countries to produce their RPPIs in a more comparable manner. As it brings together a large

body o knowledge on the subject, the Handbookmay be used or sel-learning, or as a teaching tool or training courseson residential property price indexes.

Other RPPI users, such as businesses, policy-makers or researchers may also find the Handbookuseul as a source o in-ormation, not only about the different methods that are employed in collecting data and compiling such indexes, but alsoabout their limitations. In this respect, it may acilitate the interpretation o the results.

Te drafing and revision have entailed many meetings over a three-year period, in which RPPI experts rom national sta-tistical offices, international and regional organisations, universities and research institutes have participated. Teir collec-tive advice and wisdom were indispensable or the compilation o this Handbook. An electronic version o the Handbookis available on the Internet at http://epp.eurostat.ec.europa.eu . Te IWGPS views the Handbookas a living document thatwill be amended and updated to address particular points in more detail.

Comments on the Handbook are welcomed by the IWGPS, and should be sent to Eurostat (e-mail: ESA-hicp-methods@

ec.europa.eu). Tey will be taken into account in any uture revisions.

Walter RadermacherChief Statistician of the European UnionDirector GeneralEurostat - Statistical Office of the European Union

Raael Diez de MedinaChief StatisticianDirector of the Department of StatisticsInternational Labour Organisation

Alredo M. LeoneActing DirectorStatistics DepartmentInternational Monetary Fund

Martine DurandChief StatisticianDirector of Statistics DirectorateOrganisation for Economic Co-operation and Development

Lidia BratanovaDirector of Statistical DivisionUnited Nations Economic Commission for Europe

Shaida BadieeDirector, Development Data GroupWorld Bank
http://epp.eurostat.ec.europa.eu/mailto:[email protected]:[email protected]:[email protected]:[email protected]://epp.eurostat.ec.europa.eu/


10/186


11/186

9

Preface


Preface

IntroductionTe aim o this Handbook is to acilitate the setting-up o residential property price indices in countries where these arestill missing and the improvement o existing price indices where this is deemed necessary. It is designed to give practicalguidance on the compilation o house price indices, both in developed and less developed countries, and to increase inter-national comparability o residential property price indices. It explains the different user needs, gives details on data andmethods that can be used to compile residential property price indices and provides recommendations. Te production othe Handbook was unded and supported by Eurostat.

Background

Te need or property price indices that are fit-or-purpose was recognised at a conerence organised jointly by theInternational Monetary Fund (IMF) and the Bank or International Settlements (BIS) in Washington DC, October 2003.

As a result, a chapter on residential property price indices was added to the IMFs Compilation Guide o FinancialSoundness Indicators. Te idea o a more detailed Handbook dates back to a workshop organised by the Organisation orEconomic Co-operation and Development (OECD) and the IMF on Real Estate Price Indices in Paris, November 2006.Te Handbook would complement the existing international manuals on consumer price indices, producer price indicesand import-export price indices that were produced under the auspices o the Inter-Secretariat Working Group on PriceStatistics.

Eurostat agreed to take this initiative orward by supporting and unding the preparation o the Handbook, given thestrong links to its ongoing work on the inclusion o owner-occupied housing in the Harmonized Index o Consumer Prices(HICP) and the role that house price indices have in the set o Principal European Economic Indicators.

At the Eurostat-IAOS-IFC conerence on residential property price indices, held in Basel, 11-12 November 2009, theHandbook plan was discussed. Preliminary versions o the Handbook were presented and discussed at several occasions,in particular at the UNECE-ILO Meeting on Consumer Price Indices in Geneva, 10-12 May 2010, a workshop held in TeHague, 10-11 February 2011, and the twelfh Ottawa Group meeting in Wellington, 4-6 May 2011.

A Guide to Readers

Although not all o the chapters are sel-contained, the Handbook is not designed to be read rom cover to cover. For ex-ample, some o the chapters can easily be skipped by compilers who are particularly interested in methodological issues.Further details on the contents o the Handbook are given in Chapter 1.

Te Handbook cannot be too prescriptive or two reasons. Firstly, it is not always possible to give practical guidance assome o the solutions to conceptual problems are not always clear-cut and there are choices to be made about precisely howa practical solution is implemented. Secondly, what is applicable and what can be achieved will depend on the data andresources available to the individual national statistical institute (or other compiling institute).

Acknowledgements

Te writing o the Handbook was led by Statistics Netherlands; Bert M. Balk co-ordinated the project activities. Jan deHaan and W. Erwin Diewert acted as main editors. Te authors o the individual chapters are as ollows:

PrefaceBert Balk, Jan de Haan and David Fenwick

1. IntroductionBert Balk

2. Uses of Residential Property Price IndicesDavid Fenwick

3. Elements for a Conceptual FrameworkErwin Diewert

4. Stratification or Mix Adjustment MethodsJan de Haan and Erwin Diewert


12/186

10

Preface


5. Hedonic Regression MethodsJan de Haan and Erwin Diewert

6. Repeat Sales MethodsJan de Haan

7.Appraisal-Based MethodsJan de Haan

8. Decomposing an RPPI into Land and Structures ComponentsErwin Diewert

9. Data SourcesDavid Fenwick

10.Methods Currently UsedDavid Fenwick

11. Empirical ExamplesMarc Prudhomme and Erwin Diewert

12. Recommendations David Fenwick, Erwin Diewert and Jan de Haan

GlossaryJan de Haan

Te quality o the Handbook was increased by the valuable contributions o many individuals and organisations, includinginput rom both compilers and users o residential property price indices in different parts o the world. Te number o

contributors is, o course, too great to mention them all by name.Te BIS (and in particular Paul Van den Bergh) have been excellent hosts or the Basel workshop in 2009. Many thanks goto UNECE (and in particular Carsten Boldsen) who were also heavily involved in the organisation o the Basel workshop,and o the special session on the RPPI Handbook during the joint UNECE/ILO CPI meeting in 2010.

Special thanks are due to Irmtraud Beuerlein, Simon Cot, Lee Everts, Gregory Klump, Jose Vicente Romero, PatrickSabourin, A.P. Saxena, and Chihiro Shimizu or contributing to the country-based case studies and to Emily Carless,Preechaya Chavalittumrony, Ali Hepen, Marissa Gonzalez Guzman and Hector Zarate, who provided other backgroundinormation on published indices. Useul comments on preliminary drafs o the Handbook were received rom CarlosBrs, Morris Davis, Martin Eiglsperger, imothy Erickson, Rui Evangelista, Dennis Fixler, John Greenlees, Brian Gra,Vanda Guerreiro, Ronald Johnson, Marcel van Kints, Andrew Leventis, Bogdan Marola, Daniel Santos, Mick Silver, LeoSveidkauskas, Randall Verbrugge, David Wasshausen, and participants at the workshop in Te Hague, in particular MarcFrancke and Jan Walschots. Eurostat, the BIS, the IMF and the ECB also provided helpul comments. Tanks are also dueto Rens Hendriks and Ning Huang or comments and computational assistance.


13/186

1Introduction


14/186

12

Introduction1


1.6 Broadly speaking, two separate types o RPPI canbe distinguished: a constant quality price index or thestocko residential housing at a particular moment in timeand a constant quality price index or residential property

salesthat took place during a particular period o time. Teconstruction o these two types o index will be different;most particularly, the weighting associated with the twotypes will differ.

1.7 Chapter 3 continues by summarizing the ourmain approaches to constructing an RPPI. In the final sec-tions a number o miscellaneous topics are addressed, suchas the requency o an RPPI, the consistency o monthlywith quarterly estimates and the consistency o quarterlywith annual estimates, revision policies, and seasonaladjustment.

1.8 Chapters 4-7 review in depth the main methodsor compiling RPPIs. Te simplest methods are based onsome measure o central tendency o the distribution otransaction prices in a period, in particular the mean or themedian. Since house price distributions are generally posi-tively skewed (predominantly reflecting the heterogeneousnature o housing, the positive skew in income distribu-tions, and the zero lower bound on transaction prices), themedian rather than the mean is ofen used. As no data onhousing characteristics are required to calculate the medi-an, a price index that tracks changes in the price o the me-dian house sold rom one period to the next can be easilyconstructed. Another attraction o median indices is that

they are easy to understand.

1.9 One major drawback o simple median basedindices is that they provide very noisy estimates o pricechange. Te set o houses actually traded in a period, or asample thereo, is typically small and not necessarily rep-resentative o the total stock o houses. Changes in the mixo properties sold will thereore affect the sample medianprice more than the median price o the housing stock. Aperhaps bigger problem than short-term noise is system-atic error, or bias. A median index will be subject to biaswhen the quality o the housing stock changes over time.Bias can also arise i certain types o houses are sold more

requently than other types o houses and at the same timeexhibit different price changes.

1.10 A general technique or reducing sample selectionbias is (post-) stratification. Tis technique, which is alsoknown as mix adjustment, is discussed in Chapter 4.

1.11 Chapter 5 reviews the hedonic regression ap-proach. Tis approach recognizes that heterogeneousgoods can be described by their attributes or characteris-tics. Tat is, each good is essentially a bundle o peror-mance characteristics. In the housing context, this bundlemay contain attributes o both the structure and the lo-cation o the properties. Although there is no market or

1.1 Residential property is both a source o wealthand, insoar as property owners live in or on their proper-ty, an important determining actor in their cost o living.Te price o a house is something different rom the cost o

dwelling services it provides, though the two concepts areobviously interlinked.

1.2 Monitoring the development o house prices isconsidered important, especially in times o economicturbulence. Yet the way house price development is meas-ured varies per country, and even within a country thereare sometimes two or more competing methods in use.Tis situation is o course not avourable or the design oconsistent policy measures based on solid internationalcomparisons.

1.3 Against this background it is understandable that

it was proposed that a handbook be prepared on housing,or broader residential property, price indices. (1) Te pri-mary goals o the handbook are

to provide guidance or those wishing to set up residen-tial property price indices or modiy existing indices inview o international harmonisation;

to provide a discussion and comparison o the varioustargets and their corresponding conceptual rameworks;

to provide an inventory o existing practices.

Te contents o the handbook are briefly outlined below.

1.4 Chapter 2reviews a number o areas where resi-

dential property price indices (RPPIs) play a role. Te ol-lowing applications are considered:

as a macro-economic indicator o economic activity;

or use in monetary policy and inflation targeting;

as a tool or estimating the value o a component o realwealth;

as a financial stability or soundness indicator to measurerisk exposure;

as a deflator in the National Accounts;

as an input into citizens decision making on whether tobuy or sell residential property;

as an input into the Consumer Price Index; and

or use in making inter-area and international compari-sons.

1.5 In Chapter 3on the uses o an RPPI, the ocus willbe to fill in gaps in the System o National Accounts and inthe compilation o a Consumer Price Index. It is likely thati appropriate RPPIs can be constructed to fill in these gaps,then the resulting amily o RPPIs will meet the needs omost users.

(1) Actually, this was one of the conclusions of the OECD-IMF Workshop on Real EstatePrice Indices (Paris, 6-7 November 2006).


15/186

13

1Introduction


become clear in Chapters 4-7, most methods are unable todecompose an RPPI into a land and a structures compo-nent. Chapter 8discusses how hedonic regression methodscan be used to obtain such a decomposition and considers

how to construct an RPPI or the stocko housing when he-donic regression methods are used. Using the actual data,this chapter also suggests ways to overcome several practi-cal problems that are ofen encountered in empirical worko this nature, such as a high correlation between the sizeo the structure and the size o the land.

1.17 In practice, because o the high cost o undertakingpurpose-designed surveys o house prices, the approachesadopted by statistical agencies and others to constructRPPIs have been mainly a unction o the house price datasets generated by the legal and other processes associatedwith buying a house. Te indices so constructed can varyaccording to the point in the house purchasing process atwhich the price is measured, or instance whether the finaltransaction price or the earlier valuation used or secur-ing a loan is taken. Also, the amount o detailed inorma-tion available on the characteristics o the properties soldwill affect index compilation methods, ofen acting as aconstraint on the techniques available to quality adjust orhouses o different sizes and locations. Tus, data availabil-ity has historically been a constraint on the approach usedor index construction.

1.18 Chapter 9qualitatively examines the different data

sources that can be used or constructing RPPIs, such asprinted news media, real estate agents, mortgage compa-nies, property registers and tax offices. In the final section,attention is paid to the situation in many developing coun-tries where data are scarce and the issue o property owner-ship is ambiguous.

1.19 Chapter 10 catalogues the availability o RPPIs indifferent countries and also presents some case studies.It relies on meta-data gathered by various organisations,including the European Central Bank and the Bank orInternational Settlements, and more recently a act-findingexercise conducted by Eurostat in connection with the in-

clusion o owner-occupied housing costs in the EuropeanUnions Harmonised Index o Consumer Prices, which wasextended to cover some non-EU countries.

1.20 Chapter 11provides additional practical guidanceby demonstrating the working o the RPPI constructionmethods (excluding the SPAR method) that were outlinedin Chapters 4, 5 and 6 on simple examples using small datasets.

1.21 Chapter 12concludes by providing recommenda-tions.

characteristics, since they cannot be sold separately, thedemand and supply or the properties implicitly determinethe characteristics marginal contributions to the prices othe properties. Regression techniques can be used to esti-

mate those marginal contributions or implicit prices.

1.12 Tis chapter discusses, in a non-technical way, themain models used as well as the methods to orm RPPIsrom estimation o such models. Te overall evaluation othe hedonic regression method is that it is probably thebest method that could be used in order to construct con-stant quality RPPIs or various types o residential prop-erty. However, it is also the most data-intensive method.

1.13 Te repeat sales method, reviewed in Chapter 6,utilizes inormation on the same properties which havebeen sold more than once. Because only matched mod-

els are used, there is no change in the quality mix to con-trol or. In its basic orm, the only inormation required isprice, sales date and address o the property. So the repeatsales method is much less data- intensive than hedonicmethods. Also, the repeat sales method will automaticallycontrol or micro location (address), something which he-donic methods are unable to do.

1.14 Te matched model methodology, where prices oexactly the same item are compared over time, is the natu-ral starting point or the construction o any price index.Because o the low incidence o transactions, and becausethe quality o houses continually changes, the standard

matched model methodology cannot be applied straight-orwardly. Te repeat sales method attempts to deal withthis issue by looking only at properties that have been soldmore than once over a sample period. Tis, however, canlead to a relatively low number o observations and to sam-ple selection bias. o overcome such problems, assessedvalues o the properties could be used.

1.15 In many countries, official government assess-ments are available or all properties, because such dataare needed or taxation. I the assessments pertain to somereerence date, an RPPI can be constructed by relating ac-tual sale prices to assessed values. Tis constitutes a variant

o the matched model methodology, the distinct eaturebeing that compositional change is accounted or. In thiscase, there is no need to use econometric techniques. Tevarious assessment-based methods, and in particular thesale-price appraisal ratio (SPAR) method, are reviewed inChapter 7.

1.16 Chapters 4-7 all end with empirical examplestested on actual data in order to illustrate the methodsdiscussed and to provide additional background mate-rial. Te data set covers 14 quarters o residential propertysales or a relatively small town in the Netherlands. As will


16/186


17/186

2Uses of ResidentialProperty Price Indices


18/186

16

Uses of Residential Property Price Indices2


monitoring house inflation, as experienced by purchasers,may best be estimated by collecting inormation on currenttransaction prices and using this inormation to constructa price index for the sales of housing units. In contrast, to

estimate an economys (real) stock o wealth, inormationon the sample o transacted dwellings must ideally be sup-plemented by inormation on the stock o non-transacteddwellings in order to construct aprice index for the housingstock. Tis may be done by re-weighting to reflect the di-erent mix o houses in the housing stock compared withtransactions but the adequacy o this method depends onwhether the dwellings that are actually transacted can act asa proxy or the ones that have not been subject to a changeo ownership. I the price o houses that have not changedownership is not available and inormation on their num-bers and characteristics is limited or even non-existent, the

user needs to be assured that the profile o the transactionsis representative o the overall housing stock. In practice,the latter condition may not be ully met as different sec-tors o the housing market can be influenced by differentactors and the limited number o transactions may lead tounreliable or even non-existent data on prices or some othese different strata.

2.5 Te (price determining) attributes o individualhouses ofen change over time. Tese changes include im-provements to the dwelling in the orm o renovations tokitchens and bathrooms, replacement windows with insu-lated glazing, or the installation o energy efficient heating

or air-conditioning systems, and also extensions o thestructure which reflect the recent trend in many coun-tries towards larger houses. Improvements and extensionswill be partially offset by depreciation o the structures.Irrespective o the purpose o the index, an ideal RPPIshould be adjusted or all o those changes. o put it differ-ently, the index should represent changes in the prices oproperties that are comparable in quality over time.

2.6 Te need or quality adjustment extends beyondcontrolling or home improvements and depreciation,however. Te mix o dwellings that are sold in one period islikely to be different rom that in the next period when, say,the sample o houses sold consists o more larger housescompared to the previous period. Such compositional ormix changes may have a cyclical pattern because sales olarger houses will typically decline as an economy enters arecession. Compositional changes o the sample over time,just like quality changes o the individual dwellings, shouldnot be interpreted as price changes measurement tech-niques are required to adjust the price changes or qualitymix changes. A short overview o the various methods thatare available to solve the problems o quality (mix) changewill be provided in Chapter 3. A detailed discussion othese methods will ollow in Chapters 4-7.

Introduction2.1 Tere are many areas o society where individu-

als or organisations use residential property price indices(RPPIs) directly or indirectly either to influence practicaldecision making or to inorm the ormulation and conducto economic policy. Different uses can have a significantimpact on the preerred coverage o the index and also onthe appropriate methodology applied or its construction.

2.2 From an individual households perspective, realestate ofen represents the single largest investment in theirportolio. It also accounts or the largest share o wealth inmost nations balance sheets. Changes in house prices canhave ar-reaching implications or individuals. For exam-ple, changes in housing equity and household debt levelscan permeate through to the overall economy. In act, con-sumer spending is ofen affected by changes in house pricesas a result o wealth effects and its effect on consumer con-fidence. House prices influence home improvement andrenovations expenditures, which in many countries arehigher than overall spending on new house construction.House prices play a major role in the measurement o theaffordability o home-ownership, a key housing policy ob-jective in some countries. House price changes also influ-ence the decision to build new houses (the supply side) aswell as the decision to become a homeowner (the demandside). (1) Investors turn to house price indices to not onlymeasure wealth but also to help in assessing current and

uture rates o return. (2)2.3 From a broader perspective, analysts, policymak-

ers, and financial institutions ollow trends in house pricesto expand their understanding o real estate and creditmarket conditions as well as to monitor the impact on eco-nomic activity, and financial stability and soundness. (3)For instance, mortgage lenders will use inormation onhouse price inflation to gauge deault risk. Central banksofen rely on movements in house price indices to monitorhouseholds borrowing capacity and debt burden (4) andtheir effects on aggregate consumption. (5)

2.4 In this context it should be emphasised that thedifferent uses o residential property price indices mayrequire different conceptual bases and methodology, al-though in practice, other actors sometimes come intoplay, such as data availability. (6) In general, no single in-dicator o house price change can satisy every purpose.For instance, the price dynamics o the housing market or

(1) See Duffy (2009).(2) Residential construction investment accounts for about 5% of GDP in the euro area.(3) See Case and Wachter (2005).(4) See Finocchiaro and von Heideken (2007).(5) See Case et al. (2001), Phang (2004) and Belsky and Prakken (2004).(6) See Fenwick (2006) and also Chapter 9.


19/186

17

2Uses of Residential Property Price Indices


employment and higher incomes or a wide range oworkers involved in the housing market, such as realestate agents, construction workers and proessionals inthe financial and the legal proessions. Expectations o

higher uture returns on property investment lead build-ers to start new construction and this is accompanied byhigher market demand in property-related sectors romowner-occupiers and property investors. (9) In addition,building activity will tend to increase rom more homerenovations.

Higher house prices tend to lead to increased sales ofexisting housing unitsand this in turn can lead to addi-tional tax revenues in the orm o property transer taxesgenerated rom the higher volume and value o propertysales. Tese increased tax revenuescan lead to increasedgovernment spending which in turn provides additional

economic stimulus. Rising real estate prices will lead to improvements in the

household sectors balance sheet (the wealth effect) andthis in turn will generally lead to increased householdspending on consumption and investment. (10) Accord-ing to a report by the U.S. Congressional Budget O-fice (2007), when house prices surged in the 1990s and2000s, consumer spending grew aster than incomes.Tis household wealth effect generally leads to increasesin spending by consumers on home renovations and re-pairs in addition to increased spending on other goodsand services.

2.10 O course, the above stimulative effects o increas-ing house prices go into reverse when (real) house pricesall. It is thereore important that the public and economicpolicy makers have at their disposal accurate and timelyinormation on movements in real estate prices.

2.11 Asset prices, including real estate prices, are a keyindicator or more ully understanding the dynamics othe economy. (11) According to Plosser (2007), asset pricescontain important inormation about the current and u-ture state o the economy and can play an important role inthe deliberations o central bankers as they seek to achievetheir objectives o price stability and sustainable output

growth.

For Use in Monetary Policyand Inflation Targeting

2.12 In addition to the above general interest in moni-toring property prices, many central banks have inflationtargets which can directly involve indices o propertyprices. For instance, central banks in some countries uti-lize a Monetary Conditions Index (MCI) as a day-to-day

(9

) See Zhu (2005).(10) See Campbell and Cocco (2007).(11) See Turvey (1989) and Goodhart (2001).

A Review of the DifferentUses of Residential Property

Price Indices2.7 Residential property price indices have a number

o important uses:

as a macro-economic indicator o economic growth;

or use in monetary policy and inflation targeting;

as an input into estimating the value o housing as a com-ponent o wealth;

as a financial stability or soundness indicator to measurerisk exposure;

as a deflator in the national accounts;

as an input into an individual citizens decision makingon whether to buy (or sell) a residential property;

as an input into the consumer price index, which in turnis used or wage bargaining and indexation purposes; (7)

or use in making inter-area and internationalcomparisons.

Each use is considered in turn.

As a Macro-Economic Indicatorof Economic Growth

2.8 Rising house prices are ofen associated with pe-riods o economic expansion while alling house pricesofen correspond with a slowing economy. Goodhart andHomann (2006) show that or 16 industrialised countriesthere exists a strong correlation between house prices andeconomic activity. In act the six major banking crises inadvanced countries since the mid 1970s were all associ-ated with the bursting o a housing bubble (Reinhart andRogoff, 2009). (8) In the main, house prices are treated asa leading indicator although there is some debate aboutwhether house price change is a leading, lagging or coinci-dent economic indicator.

2.9 What is clear is that rising house prices are ofenassociated with economic growth through at least threechannels:

Higher (relative) house prices tend to stimulate increasedconstruction activity, which in turn leads to higher

(7) The inclusion of a house price index in the calculation of a CPI depends on theobjectives of the CPI and, in particular, whether an acquisitions, payments or user costapproach is adopted. Further discussion of these issues is given in the Consumer PriceIndex Manual (ILO et al., 2004) and the Practical Guide to Producing Consumer PriceIndices (United Nations, 2009).

(8) Claessens, Kose and Terrones (2008; 25) find that ... recessions associated with houseprice busts are on average over a quarter longer than those without busts. Moreover,output declines (and corresponding cumulative losses) are typically much larger in

recessions with busts, 2.2 (3.7) percent versus 1.5 (2.3) percent in those without busts.These sizeable differences also extend to the other macroeconomic variables, includingconsumption, investment and the unemployment rate.


20/186

18



wealth effect that can lead to increases in consumption andincreased household borrowing.

2.15 More generally, individuals will have an indirect

stake in real estate asset prices, including residential prop-erty, through pension unds and other direct investmentsin real estate.

As a Financial Stability or SoundnessIndicator to Measure Risk Exposure

2.16 Financial Soundness Indicators (FSIs) are indica-tors o the current health and soundness o the financialsystem and institutions o a country and o their corporateand household components. Tey include both aggregatedindividual institution data and indicators that are repre-

sentative o the markets in which the financial institutionsoperate, including statistics on real estate prices. FSIs arecalculated and disseminated or the purpose o supportingnational and international surveillance o financial systems.Te IMF developed FSIs with a view to monitoring andstrengthening the global financial system and to increas-ing stability ollowing the financial market crises o the late1990s, and as a way o combating the subsequent growingnumber o banking crises that have occurred globally. Tecompilation guide or financial soundness indicators pro-vides some advice on compiling house price indices whilstacknowledging the relative absence o international expe-rience and guidance and the absence o a comprehensive

ramework or constructing such indices. More recently,the October 2009 Report to the G-20 Finance Ministersand Central Bank Governors on the Financial Crisis andInformation Gaps(13) mentions that inormation on dwell-ings and their associated price changes are critical ingredi-ents or financial stability policy analysis.

2.17 Sharp alls in real estate prices have a detrimentalimpact on the health and soundness o the financial sectorand on the financial situation o individuals and o indi-vidual households, by affecting credit ratings, the value ocollateral, and the debt to equity ratio.

2.18 It should come as no surprise that the relation-ship between real estate cycles and economic cycles is welldocumented and that the role o real estate prices in debtfinance and financial crises has long been recognised. Tishas led to the use o residential property price indices asindicators o financial stability, particularly in countrieswhere real estate accounts or a significant proportion onational and household wealth, and where the propensityo home ownership is relatively high.

2.19 Te use o trends in residential property prices, andreal estate prices more generally, as an indicator o finan-cial soundness, has been supported by in-depth analytical

(13) Available at: http://www.imf.org/external/np/seminars/eng/2010/infogaps/index.htm.

operating target or the conduct o monetary policy. Inan expanded version o this index, as that suggested byJarociski and Smets (2008) and Goodhart and Homann(2007), the MCI would include some measure o house

prices because o the important role this variable plays inthe inflationary process and or economic perormance.Other central banks who have an inflation target based onthe Consumer Price Index (CPI) will indirectly take intoaccount the movement in house prices when setting in-terest rates, depending in part on the treatment o OwnerOccupied Housing (OOH) in their countrys CPI. Tis is-sue is discussed urther in Chapter 3.

2.13 It can be argued that in the uture, residentialproperty prices are likely to play an increasing role in theconduct o monetary policy. Over recent years an inflationtarget has been used by a growing number o countries

to define and operate their monetary policy rameworks.Te IMF (2007) provides a list o 28 countries classifiedas inflation targeters according to their exchange ratearrangements (without speciying the target or inflationmeasure). Carare and Stone (2003) extend this analysisurther by classiying countries that use an inflation targetor monetary policy, into ully-fledged inflation targeters,eclectic targeters and inflation targeting lite regimes, us-ing the clarity and credibility (12) o the commitment tothe inflation target to classiy individual countries. Teauthors then identiy 42 medium and large country cen-tral banks who have some orm o floating exchange rate

mechanism (i.e. not adopting a fixed exchange rate) leav-ing their degree o commitment to an inflation target asthe defining monetary objective. Tey estimated that by2001 some 7 industrial and 11 emerging markets operatedully-fledged inflation targeting, that is have a medium tohigh level o credibility, clearly commit to their inflationtarget and institutionalize this commitment in the orm oa transparent monetary ramework that osters account-ability o the central bank to the target. Te number ocountries operating ully-fledged inflation targeting hasbeen increasing over the years.

As an Input for Estimating the Valueof Housing as a Component of Wealth

2.14 House prices are an input into the measurement oaggregate wealth in the economy. Existing dwelling unitsare part o the balance sheet accounts in the System oNational Accounts (SNA). Tus it is necessary to have aprice index or this asset class in order to orm estimateso real household wealth. As was mentioned in the intro-duction to this chapter, rising house prices will generate a

(12) Clarity is gauged by the public announcement of the inflation target and by the

institutional arrangements for accountability. Credibility is measured indirectly using asa proxy the actual inflation outturn and by market ratings of long-term local currencygovernment debt.
http://www.imf.org/external/np/seminars/eng/2010/infogaps/index.htmhttp://www.imf.org/external/np/seminars/eng/2010/infogaps/index.htm


21/186

19

2Uses of Residential Property Price Indices


that two o the most recent and widely available reerenceson the compilation and use o national accounts deflators,SNA (1993) and the Eurostat (2001) Handbook on Priceand Volume Measures in National Accounts, pre-date the

CPI Manual (2004) and PPI Manual (2004).

2.23 Te CPI and PPI Manuals were developed in par-allel and take advantage o the latest research into indexnumber theory and practice, which is not ully reflectedin the official literature on national accounts. (16) Te twomanuals are essentially based on the same underlying eco-nomic principles and statistical theory. Tey provide acomprehensive and coherent overview o the conceptualand theoretical issues associated with consumer and pro-ducer price indices and translate these into available op-tions or practical measurement. Te CPI Manual also act-ed as a catalyst or the new ILO Resolution on Consumer

Price Indices, which was passed in 2003.

As an Input into an Individual CitizensDecision Making on Whether to Buy(or Sell) a Residential Property

2.24 Te buying or selling o a dwelling is typically thelargest financial transaction a household will enter intoduring his or her lie. Changes in house prices are there-ore likely to influence substantially whether a householdpurchases a property and also the budget plans and savings

decisions o the prospective house buyers and sellers. Tepurchase o a house is considered by many owner-occupiersboth as a means o obtaining shelter services and as acapital investment, the latter potentially providing an op-portunity or significant capital gains in the longer-term.Current price levels and trends, together with expectationsabout uture trends in house prices and mortgage interestrates, (17) will influence an individuals decision on whetherto purchase now or postpone the purchase. Te opportu-nity cost associated with the sums o money involved willalso come into play as prospective purchasers evaluate thealternative choices available to them. For instance, pro-spective purchasers will ofen take into account the impact

o changes in house prices on market rents.

2.25 More generally, individuals also have an indirectstake in real estate asset prices through pension unds andother investments or which house prices will likely havean effect. For instance, the portolios o some pensionunds include apartment blocks whose rents provide an

most attention in the literature is devoted to price i ndices. Once somehow estimated,price indices are in fact used, if at all, primarily to deflate nominal or monetary totals inorder to arrive at estimates of underlying real magnitudes.

(16) The CPI and PPI Manuals are consistent with the material in Chapter 16 of SNA (1993)and also with the 2008 System of National Accounts but delve deeper into theproblems associated with the construction of price indexes, particularly at lower levels

of aggregation.(17) Interest rate policy will have an impact both on inflation and on net disposable income

after the payment of i nterest.

studies. Included amongst the vast amount o materialpublished on this subject is a paper by Nabarro and Key(2003) who present a model or real estate and lending cy-cles, supported by case studies. Teir paper traces the cycli-

cal evolution rom initial indicators provided by the rentalmarket, to property prices and through to balance sheetso borrowers and lenders, and draws attention to a numbero relevant indicators o the real estate market. It describeswhat the authors call the dangerous interdependence be-tween real estate cycles and financial systems. Whilst theauthors acknowledge the highly unpredictable nature othe real estate cycle and its different characteristics andproperties rom one cycle to the next, they discuss the link-ages between real estate cycles and debt finance to identiyareas where improved inormation could support effectivecounteracting strategies and policies. Tey explain how a

reliable and cost-effective system o perormance measure-ment and monitoring can be developed and implementedand suggest how such a system can provide a mechanismor analytical decision making, designed to impact uponthe behaviour o the real estate sector.

2.20 Inormation on residential property and otherproperty prices needs to be supplemented by relevant andtimely detailed analyses, and by other inormation such asthe proportion o houses being purchased with cash ratherthan being financed through a loan. Te average ratio oloan to property price, and how this is distributed, pro-vides an indication o the exposure o the borrower and

the lender, as does the price to earnings ratio and, to acertain extent, the volume o transactions. (14) Similarly, amore detailed analysis o the types o houses being sold byregion will show whether activity in the housing marketis concentrated in particular segments o the market suchas high-end properties or in certain geographical locationssuch as the capital city or large urban areas.

As a Deflator in the National Accounts

2.21 National statistical agencies use house price indi-ces in at least two ways. First, the structures component oa price index or newly-built houses is ofen used to deflate

current price values or residential construction in the na-tional accounts; see Bover and Izquierdo (2003). Second,house price indices may be included in the construction othe CPI, depending on the choice o its conceptual basis.Tis second use is considered below and in more detail inChapter 3.

2.22 Price indices and deflators are seemingly differ-ent entities within a wider group o statistics relating toprices. (15) It is pertinent to note against this background

(14) Past observation suggests that when price-to-earnings ratios get to an unsustainablehigh level, the adjustment is initially seen in a reduction in the volume of housing

turnover rather than in transaction prices.(15) However, the underlying theory of deflators and (direct) price indexes is the same; see

Chapter 16 in SNA (1993). Samuelson and Swamy (1974) note the following: Although


22/186

20



than making, say, national comparisons over time becauseinter-area/international comparisons require comparabletypes o housing across the regions/countries being com-pared (or comparable inormation on the characteristics

o housing units across the regions i a hedonic regressiontechnique is used) in order to construct a constant qualityprice index.

2.29 Te European Central Bank (ECB) in co-operationwith the central banks o the individual countries o theeuro-zone and the European Union has an interest incomparative measurement o changes in residential prop-erty prices across different euro-area countries and or theeuro-area as a whole. Te raw data used here come romvarious national sources and have primarily been collectedand documented by the Bank or International Settlements(BIS). (18) Since 2001, the ECB has compiled an aggregate

index or the euro-area by weighting together changes inprices or houses and flats or the euro-area countries. (19)Te national methodologies associated with the figuresavailable or each individual country and or the euro-areaaggregate, have improved over recent years but perhaps allshort o the standards applied to other economic statisticsand price indicators or the euro-area. (20) Te BIS has alsobrought together residential property price statistics or thenon-euro area countries o the European Union and has inmany cases been conronted with even more pronouncedissues concerning data comparability and quality.

2.30 Such comparisons can be conounded by meth-

odological and coverage differences and also by differencesin the requency and timeliness o the data. Some o thesedifferences arise rom the different sources o data used tocompile national indices. Chapter 9 explores these datasources in more detail and Chapter 10 gives an inventoryo the different methods used by countries to compile theirindices o residential property prices. It can be observedthat a notable proportion o countries, including some de-veloped countries, do not have reliable residential propertyprice indices.

(18) The BIS data set of residential property price statistics is available at: http://www.bis.org/statistics/pp.htm.

(19) See box Preliminary evidence on developments in euro area residential property pricesin the October-2001 issue of the ECB Monthly Bulletin.

(20) See Eiglsperger (2010), page 233.

income and where a capital gain is expected to materialiserom an increase in the property value.

As an Input into the Constructionof a Consumer Price Index (CPI)

2.26 House prices will directly affect measured infla-tion when the CPI includes owner-occupier housing costsand the method o measurement draws on house prices asone o the inputs. Measured inflation is indirectly affectedi house prices influence market rents, which constituteanother element o a CPI, and where additionally imput-ed rents are used as a proxy or owner-occupied housingcosts. Renting and buying can be substitutes and the levelo house prices will have an impact on the rate o returnobtained by a landlord on his or her investment and alsoon the rent charged.

2.27 Te treatment o Owner Occupied Housing(OOH) is one o the most difficult challenges aced bycompilers o consumer price indices. Tere are a numbero alternative conceptual treatments and the choice be-tween them can have a significant impact on the overallindex, affecting both the weight attributed to OOH (andby implication to an RPPI) and the measured rate o infla-tion. In essence there are our possible main approachesto including OOH in a CPI: the acquisitions approach, thepayments or money outlays approach, the user cost ap-proach and the rental equivalence approach. Te first threeapproaches require the construction o a housing price in-dex. Tese various approaches to the treatment o OOH arereviewed in more detail in Chapter 3.

For Use in Making Internationaland Inter-area Comparisons

2.28 House price indices are also used in conjunctionwith (comparable) benchmark data on house price levelsacross regions or countries to generate inter-area or in-ternational comparisons o living cost differentials. Teproblems that arise in attempting to price the services oOOH in a national context also arise in the context o in-ter-area and international comparisons. In the latter con-text, however, the problems are somewhat more difficult
http://www.bis.org/statistics/pp.htmhttp://www.bis.org/statistics/pp.htmhttp://www.bis.org/statistics/pp.htmhttp://www.bis.org/statistics/pp.htm


23/186

3Elements for a ConceptualFramework


24/186

22

Elements for a Conceptual Framework3


Such statements indicate that the construction o an RPPIwill be much more difficult than the construction o a nor-mal price index based on a matched model methodology.It should be recognized at the outset that, because o the

difficulties resulting rom the uniqueness o each dwellingunit, it would not be possible to construct a perect RPPI;it will only be possible to construct an approximation tothe theoretically ideal index or each purpose.

3.3 Te question o what is the purpose o an RPPIhas been addressed in Chapter 2, where the many uses oan RPPI were considered. Tis chapter ocuses on the useso RPPIs to fill in gaps in the System o National Accountsand in the construction o a CPI. It is likely that i appropri-ate RPPIs can be constructed to fill in these gaps, then theresulting amily o RPPIs will meet the needs o most users.

3.4 Broadly speaking, two separate RPPIs can be dis-tinguished: 1) a constant quality price index or the stockoresidential housing at a particular moment in time; and 2)a constant quality price index or residential property salesthat took place during a particular period o time. Te con-struction o these two types o index will be different; e.g.,the weighting associated with the two types will differ. Inthis chapter, the main approaches to constructing an RPPIwill be briefly discussed. Details on these methods will bepresented in Chapters 4 to 7.

3.5 A variety o miscellaneous topics will be ad-dressed in the final our sections o this chapter. Tese top-

ics include the requency o the RPPI and user needs, theconsistency o monthly with quarterly estimates and theconsistency o quarterly with annual estimates, revisionpolicies, and seasonal adjustment.

Residential Property PriceIndices and the Systemof National Accounts

3.6 Te System of National Accounts (SNA) 1993andits recent updating, the System of National Accounts2008, (1) provide a comprehensive accounting rameworkor an economy. Te SNA partitions the value flows in theeconomy into various meaningul categories and providesa reconciliation o the flow accounts with the correspond-ing stock accounts. It is urthermore recommended to de-compose the values in the cells o these accounts into priceand volume (or quantity) components.

(1) See Eurostat, IMF, OECD, UN and the World Bank (1993) and (2009).

Introduction3.1 What makes the construction o a residential

property price index (RPPI) so challenging? Tis questionwas addressed in Chapter 1 o this Handbook but it will beuseul to remind readers about the main problems, whichare as ollows:

Te compilation o price indices typically relies onmatching the prices or identical items over time. How-ever, in the housing context, each property has a uniquelocation and usually a unique set o structural charac-teristics. Tus, the matched model methodology will bedifficult or impossible to apply.

ransactions are sporadic.

Te desired index number concept may not be clear, or

put another way, there are several distinct purposes orwhich an RPPI is required and, broadly speaking, differ-ent purposes require different indices.

For some purposes, notably the construction o nationalbalance sheets and the estimation o user costs o owneroccupied housing, a decomposition o a property priceinto land and structures components is required but it isunclear how best to accomplish such a decomposition.Tis issue will be discussed in more detail in Chapter 8below.

3.2 Te first two difficulties are well recognized in thehousing measurement literature as the ollowing quota-

tions indicate:

Te price o housing is harder to measure than that o most

other goods and assets because o three key distinguishing

characteristics. First, and most importantly, dwellings are

heterogeneous. No two dwellings are identical, i only

because they cannot occupy quite the same location. Tis

means that sampled house prices may be a poor indicator

o all house prices because we cannot always reliably

predict the sales price o a given dwelling rom the price o

another. Robert Wood (2005; 213).

Te fundamental problem that price statisticians facewhen attempting to construct a real estate price indexis that exact matching of propertiesover time is not pos-sible for two reasons: (i) the property depreciates overtime (thedepreciation problem) and (ii) the property mayhave had major repairs, additions or remodeling doneto it between the two time periods under consideration(the renovations problem). Because of the above two prob-lems, constructing constant quality real estate price in-dices cannot be a straightforward matter; some form ofimputation or indirect estimation will be required. ErwinDiewert (2009b; 92).


25/186


26/186

24



but the CPI manual suggests our possible approaches. (7)Tese approaches treat the unique character o OOH,which involves both the acquisition o a house and the con-sumption over time o the flow o services o the house, in

a different manner.

Te money outlays or payments approach. In this ap-proach, the out o pocket expenses o home ownershipare simply added up. Tese costs include expenditureson maintenance and repair, mortgage interest costs, in-surance premiums, property taxes and condominiumcharges (i the housing unit is a condominium). woimportant types o implicit cost and one important im-plicit benefit o home ownership are not included. Tetwo omitted costs are depreciation and the opportunitycost o the unds that are tied up in the homeowners eq-uity in the house; the implicit omitted benefit is any (net)

capital gains that may accrue to the owner during thetime period under consideration. (8) Te money outlaysapproach is useul i an analyst wishes to ocus on thedisposable income o households. However, it is not par-ticularly useul as a measure o household consumptionservices (because o the omission o the costs and bene-fits mentioned above).

Te (net)acquisitions approach. (9) In this approach, theservices o OOH are ignored in the CPI except when anew housing unit is introduced into the market place.Te purchase price o the new dwelling unit is chargedto the period o purchase so that a purchase o a new

house is treated in the same manner as the purchase oa nondurable good or service, i.e. the purchase is treatedin the same way as the purchase o other durable goods.A variant o this approach is to decompose the sellingprice o the newly built residential property into landand structures components and to use just the structurescomponent as the price which will enter into the CPI.

Te rental equivalence approach. In this approach, a priceis imputed or the shelter cost o the owner occupiedhousing unit (both or new and existing units), which isequal to the price at which the unit could be rented. (10)

Te user cost approach. In this approach, the financial

opportunity cost o owning the house and using its ser-vices during the reerence period is calculated.

(7) Diewert (2002) (2009a) (2009b) provides more discussion of alternative methods.(8) The money outlays concept is explained in some detail in Baldwin, Nakamura and

Prudhomme (2010).(9) For a comprehensive practical treatment of the net acquisitions approach, see Eurostats

(2012) Technical Manual on Owner Occupied Housing.(10) This approach is consistent with the treatment of OOH in National Accounts. In the

SNA, OOH is considered a fixed asset, unlike other durables (such as washing machines,furniture, cars etc). The purchase of a house is considered an i nvestment and includedin gross fixed capital formation and thus excluded from household final consumptionexpenditure; the same goes for ex tensions of the house and major repairs. However, theownership of a house provides a service which is consumed over time by the ownerand the value of this service is included in household final consumption expenditure.

complexities in the system o national accounts that havenot yet been definitively resolved. From the viewpoint othe construction industry, these transactions costs are notpart o the revenues that accrue to the construction sec-

tor, so these costs should not be included in the value othe output o the construction industry. However, romthe viewpoint o the sector that purchases the new hous-ing unit, these transactions costs are a real cost and theymust be accounted or. Tere are a number o ways thattransactions costs associated with the purchase o a newhousing unit could be treated (rom the viewpoint o thepurchaser):

simply attribute all o the costs to the period o purchaseand treat the transactions costs as an expenditure by thepurchaser (4) (which is an acquisitions approach to thesecosts);

include transactions costs as part o the structures compo-nent o the value o the purchase so that these costs wouldbe amortized over time using the same depreciation ratethat was being used to depreciate the structure; or

separately amortize the transactions costs according tothe average length o time a residential property o thetype under consideration is being held beore it is resold.

Conceptually, the last treatment seems preerable (5) butthe first and second treatments will lead to a simpler seto accounts. Tese issues need to be studied urther by na-tional accountants with input rom the broader economics

community.

Residential Property PriceIndices and the ConsumerPrice Index

3.15 Pricing the services o Owner Occupied Housing(OOH) in a Consumer Price Index (CPI) is extensivelydealt with in the Consumer Price Index Manual. (6) Tere is

no universal consensus on the treatment o OOH in a CPI

(4) The price index that could be used to convert the nominal value of transactioncharges into a real amount (or volume) is a composite purchase price index for thetype of property under consideration which includes both the land and structurescomponents.

(5) This is the treatment used by the Australian Bureau of Statistics. An unresolved issue isthe choice of price deflator in order to form real amortization charges. That is, shoulda structures price index be used or should a composite structures and land price beused? In the case of real estate the commissions are generally proportional to theoverall price of the property (the sum of the land and structures components) so itwould be appropriate to use a composite property price index for the deflation ofthis component of transactions costs. Government transactions taxes or stamp dutiesmay impose different rates on the land and structures components of the sale and soworking out an appropriate real price for this component of transactions costs may berather complicated. Again, it may be acceptable to avoid all of these complexities and

just use a composite purchase price index to do the deflation.(6) See ILO, IMF, OECD, Eurostat, UN and World Bank (2004), Chapter 23.


27/186

25

3Elements for a Conceptual Framework


a price index or the structures component of newly-builtresidential properties that were sold during a given pe-riod o time, which is needed or a narrowly defined netacquisitions approach where only the structures compo-

nent would be included in the purchase.

Main Methods3.19 o measure pure price change, real estate prices

must be adjusted or quality change. In other words, tocompile a constant quality RPPI, it will be necessary tosomehow control or any variations in the amounts o theprice determining characteristics o the properties. Temost important characteristics are:

the area of the structure (in squared eet or in meterssquared);

the area of the landthat the structure sits on (in squaredeet or meters squared);

the locationo the property;

the ageo the structure;

the type of structure; the structure can sit entirely in thelot without sharing any walls with an adjacent struc-ture (detached dwelling unit) or share one wall with aneighbouring unit (semi-detached dwelling unit), orthe dwelling unit can be a single apartment or unit in

a multiamily residence (apartment or condominiumbuilding);

the materials usedin the construction o the house (pri-marily wood, brick, concrete or traditional materials; i.e.,a shack or shanty), and

other price determining characteristicssuch as the num-ber o bedrooms, the number o bathrooms, a garage, aswimming pool, air conditioning, distance to amenities,etc.

3.20 Four main methods have been suggested in the lit-erature to control or changes in the amounts o the prop-erty characteristics: stratification or mix adjustment, repeat

sales methods, hedonic regression methods, and the use oproperty assessment inormation. Below, a brie overviewo the our methods is provided. More details can be oundin Chapters 4-7.

3.21 Stratificationo transactions according to some othe price determining characteristics is a straightorwardand computational simple way to adjust or changes in thequality mix o the samples in different time periods. By de-fining a number o reasonably homogeneous strata or cells,the average selling price within each cell can be used as a(proxy to a) constant quality price or that type o property.Regular index number theory can then be applied to aggre-

gate up the average prices by cell into an overall index. Suchstratification methods are also known as mix adjustment

Since the CPI Manual (2004) was written, a fifh conceptor pricing the services o OOH has been suggested: (11)

Te opportunity cost approach. In this approach, the price

for the services of an owned dwelling unit is set equal tothe maximumof its rental equivalence and user cost prices.

3.16 Te conceptual differences between these ap-proaches should be underlined. Te rental equivalence ap-proachand the user cost approachprice the services o anowner occupied dwelling. Te payments approachmeas-ures the out o pocket expenses o home ownership. Tenet acquisitions approachtakes a completely different per-spective, implicitly allocating all the services o the newlypurchased dwelling to the period o purchase.

3.17 In the above approaches except thepayments andrental equivalenceapproaches, there is a need or constant

quality price indices or either newly-built dwelling unitsor or the existing stock o dwelling units. Te user costandopportunity cost approachesto pricing the services o a resi-dential housing unit are not entirely straightorward. TeAppendix to this chapter outlines the mechanics o theseapproaches.

3.18 o summarize, RPPIs are needed in the construc-tion o a CPI and to deflate several value flows and stockholdings in the national accounts. For both CPI and na-tional accounts purposes, it will be useul or necessary tohave a decomposition o the price indices into structuresand land components. More specifically, it would be useul

to be able to construct the ollowing set o RPPIs: (

12

) a price index or the total stocko residential housing at

a particular moment in time, which is needed or esti-mating real changes o the economys stock o residentialhousing, a component o a nations real wealth;

a price index or the owner occupied stock o residen-tial housing (a subindex o the index in the bullet pointabove), which is needed to construct estimates or thevalue o OOH services based on user cost or opportunitycost principles;

a price index or residential property sales(both newly-built and existing dwelling units) that took place during a

given period o time, which is needed or estimating thereal output o the residential real estate services sector;

a price index or the sales of newly-built residential prop-ertiesduring a given period o time, which is required i abroadly defined net acquisitions approach is used whereboth the structures and land components would be in-cluded in the purchase;

(11) See Diewert (2009b), Diewert and Nakamura (2009) and Diewert, Nakamura andNakamura (2009).

(12) Fenwick (2005) (2006) argued that i t would be useful to develop a coherent conceptualframework for a family of real estate price indexes. It can be seen that user needswill vary and that in some instances, more than one measure of house price or realestate inflation may be required. It can also be seen that coherence between different

measures and with other economic statistics is important and that achieving this willbe especially difficult as statisticians are unlikely to have an ideal set of price indicatorsavailable to them. David Fenwick (2006; 8).


28/186

26



another variant o the method known as the hedonic impu-tation method.

3.24 Many countries tax real estate property and are

likely to have an official property valuation office that pro-vides periodic appraisals o all taxable real estate proper-ties.Assessment-based methodscombine selling prices withappraisals to compute price relatives (sale price appraisalratios) and control or quality mix changes. Te Sale PriceAppraisal Ratio (SPAR) method is based on the matchedmodel methodology. In contrast to the repeat sales meth-od, it relies on all (single and repeat) sales data, and thereis no revision o previously estimated indices. O coursethe method can only be applied in countries where reliableassessed values o the properties are available.

3.25 I the reerence period is a year, all methods will

tend to generate similar estimates o the trend in residentialproperty price changes or an entire country. However, aswill be seen in the examples presented in Chapters 4-7 andChapter 11, different methods do generate small but sig-nificant differences in trends while or shorter periods theycan lead to rather different estimates o price change. Tevarious methods could also produce different signals oturning points.

3.26 As hedonic methods assume that inormation onthe characteristics o the properties sold is known, thesamples can be stratified and, i a sufficient number o ob-servations is available, separate indices can be estimated

or the strata. In other words, hedonic regression meth-ods can provide a set o constant quality price indices orvarious types o property. Obviously, i data on some pricedetermining characteristics are available, then repeat salesand assessment-based methods can also be combined withstratification.

3.27 Stratification can also be used to approximate astock based RPPI. In this case the stratum weights will bebased on census data pertaining to the value o the owneroccupied housing stock. Te stratum price indices will stillbe based on sample data o properties sold. Within eachstratum, the properties traded are now treated as a (ran-dom) sample rom the stock. Since long time intervals be-tween two censuses is the norm, stock value weights canusually only be updated very inrequently.

3.28 As was discussed previously, or various purposesit is necessary to decompose the overall price o a prop-erty into (additive) components that reflect the price o thestructure and the price o the land the structure is locatedon. In Chapter 8 it is shown how hedonic regression tech-niques can be used to accomplish this decomposition.

methods. Wood (2005) describes this method in the ol-lowing way:

House price observations are grouped into sets or

cells o observations on houses with similar location andphysical attributes. [..] Te mean prices in each cell areweighted together to give a mix adjusted price. A changein the composition o the sample will alter the number oobservations in each cell. But i the cells are defined suffi-ciently precisely, so that all elements o the cell have similarprices and price trends, then such compositional changeswill not systematically affect the mix adjusted house price.Robert Wood (2005; 214).

3.22 Te repeat sales methodaddresses the quality mixproblem by comparing properties that have sold more thanonce over the sample period. Restricting the comparison to

units that have sold repeatedly ensures that the price rela-tives compare like with like, provided that the quality othe houses remained unchanged. Te standard repeat salesmethod is based on a regression model where the repeatsales data pertaining to all periods are pooled. A poten-tial drawback o this approach is the issue o revisions:when new periods are added to the sample and the modelis re-estimated, the previously estimated price indices willchange. An advantage o repeat sales methods is that, be-cause properties are matched at the address level, loca-tion, an important actor affecting real estate prices, is heldconstant.

3.23 One other potential drawback o the repeat salesmethod is that it does not account or quality changes othe sampled houses; over time a dwelling unit can undergorenovations and be subject to depreciation. Consequently,the quality o the property can vary with time. Hedon

price indicees eurostat ks ra 12 022 en

Documents

o g i e s w o r

fixed base price indices

i n g p

european commission

papers european union

p e r s2013 editionhandbook

sales methods

mix adjustment methods