11285-1

Embed Size (px)

Citation preview

  • 8/13/2019 11285-1

    1/11

    Inflation Hedging Characteristics of Chinese Real Estate Market1

    Chu Yongqiang

    (National University of inga!ore 11"#$$%

    &'stractNowadays Chinese real estate market is developing very fast, more and moreindividuals, as well as institutional investor are considering investing in real estate. Forthis purpose, we must know the return characteristics of real estate, as compared withthe inflation rate. In this article, we study the inflation hedging characteristics ofChinese real estate market following the OLS methodology developed y Fama andSchwert!"#$$%. &nd we use &'I(& model to first decompose the inflation rate toe)pected and une)pected inflation rate. *he results of OLS do not provide evidence tosupport the hypothesis that real estate acts as effective inflation hedge. &nd we also usecointegration test to study the long term relationship of real estate market and inflation,then the +ranger causality test is carried out to see the causality relationship. &gain wefind no evidence of long term hedge aility of Chinese real estate. owever thecausality test do show that the inflation leads real estate return.)ey *ordsChinese 'eal -state (arket, Inflation edging, -)pected Inflation 'ate

    ection I Introduction

    *he relationship etween real estate return and inflation has een suected to e)tensiveempirical research since the seminal paper y Fama and Schwert !"#$$%. *he Chinese (arket is ofinterest ecause of the following reasons,. First Chinese real estate market is an emerging, highlyspeculative market, which is e)pected to show much different characteristics compared with themature market. Second, the inflation hedging characteristics in China has not een studied. *hird,more and more attention has een made aout the direct investment on real estate, so it isimportant to study the investment return with respect to the corresponding inflation rate.

    In this article, we will firstly use the conventional OLS methodology to test the inflationhedging aility of real estate return ased on the /uarterly data from "##012332. &nd then acointegration test ased on the e)tended yearly data will e done to study the long termrelationship and +ranger Causality test will also e employed to study the causality relationshipetween real estate return and inflation rate.

    *he rest of this article will e organi4ed as follows5 Section II will provide a rief literatureview of inflation hedging characteristics of various assets, especially those of real estate6 thetheoretical framework will e given in Section III6 *he methodology of various model mentionedaove will e discussed in Section I76 section 7 will riefly descrie the data for this article6 theresult of various models will e given out in section 7I6 section 7II concludes the article.

    ection II +iterature ,ie-

    *he empirical study of the relationship of asset return and inflation was firstly done y Famaand Schwert !"#$$%. In their seminal paper, the return of 8S treasury ills, long term treasuryonds, private residential real estate, human capital and common stocks are studied. *hey usedconventional OLS method to study the inflation hedging characteristics of these assets, and thenominal return of *reasury ill will used as the appro)imation of the e)pected return. *hey haveshown that only private residential real estate is a complete hedge against oth e)pected and

    "C-N-*9Correspondence &ddress5 :epartment or 'eal -state, School of :esign and -nvironment, National 8niversity ofsingapore, ; &rchitecture :rive, Singapore, ""$

  • 8/13/2019 11285-1

    2/11

    une)pected inflation, while government det instruments are only complete hedge againste)pected inflation, and human capital is at est a partial hedge against e)pected and une)pectedinflation, and common stocks are shown to e negatively related to oth e)pected and une)pectedinflation.

    >arkham, ?ard and enry !"##0% e)amined inflation hedging characteristics of 8@ property.*hey used an alternative method to decompose the inflation rate, which is &'I(& !", ", A%. *hey

    also used cointegration and causality analysis to test the long1term relationship. *hey have shownthat in the short run, changes in e)pected and actual inflation affects returns to direct real estate,and that the property return and inflation are cointegrated in the long run.

    +at4laff !"##;% used another decomposition method of &'I(&!",3,A% and studied therelationship etween inflation and e)cess return.

    Stevenson and (urray !"###% e)amined the inflation hedging aility of Irish real estate,using oth the conventional OLS and cointegration analysis. is results show that Irish real estatedoes not provide a good hedge against inflation which differs from the maority of empiricalevidence from other real estate markets.

    Stevenson !"###% showed that although the conventional OLS models provides littleevidence of a consistent and stale relationship, the cointegration results, especially those otainedfrom the -ngle1+ranger procedure, provide strong evidence to support the hypothesis that thehousing market return and inflation are cointegrated.

    Stevenson !2333% focused on the long term relationship of inflation and housing market,which shows that the cointegration test provides strong evidence that housing and inflation share acommon long term trend, while *arert !"##0% focused on the inflation hedging characteristics ofvarious property types.

    ection III .heoretical /ra0e-ork

    Irving Fisher !"#A3% pointed out that the nominal interest rate can e e)pressed as the sum of ane)pected real return and an e)pected inflation rate. ?hen this is applied to asset return, it statesthat the e)pected nominal return contain market assessment of e)pected inflation rate. So in anefficient market, the price of any asset will e determined such that the e)pected nominal return on

    the asset from "t

    tot

    is the sum of the appropriate e/uilirium e)pected real return and the estpossile assessment of the e)pected inflation rate from "t to t, which is5

    " " "! % ! % ! %

    t t tE R E r E I

    = +

    R1nominal returnr- real rate of returnI1 inflation rate

    &nd again Fisher suggests that the e)pected return is determined y real factors as the realand monetary sectors of the economy are largely independent. *his is made it convenient for us tostudy the return1inflation relationship without introducing a complete general e/uilirium modelfor e)pected real return.*hus we otain another form of !"%5

    "! %

    t t t t R E I

    = + +

    !2% can e estimated y OLS. &side from the e)pected return, we are also interested in theunanticipated component of inflation, which can e otained y5

    " "! % ! ! %%

    t t t t t t t R E I I E I

    = + + +

    &s ased on FisherBs model, all assets should have a coefficient ".3 = , and when ".3 = , we

    say that the asset provides complete hedge against e)pected inflation, and ".3 = suggests that itprovides complete hedge against une)pected inflation.

    Section I75 (ethodology

    1 2+ Regression Model

  • 8/13/2019 11285-1

    3/11

    >ased on the analysis aove, we will use OLS to estimate !A%, which is also suggested y

    Fama and Schwert !"#$$%. I will test whether or is statistically indistinguishale from unity.If oth coefficients are e/ual to ", then the real estate is a complete hedge against inflation. &nd in!A%, the constant can e interpreted as the e/uilirium real rate of return.

    ?hen testing the inflation hedging aility, the pro)y of inflation rate is an important factor. InFama and Schwert, they used *reasury ill return a pro)y. ?hile in China, the *reasury ill ishighly regulated y the government, and is illi/uid, so we cannot use the return of *reasury ill asappro)imation of e)pected inflation. Instead I use &'I(& model to otain the e)pected inflationas +at4laff !"##;% and >arkham, ?ard and enry !"##0% did.

    &s mentioned y many literatures, this methodology suggested y Fama and Schwert has aig prolem in that it does not consider the stationarity of time series data, which may result inspurious regression. !Sing 2333 !a%, Sing 2333!%,*arert "##0%. In this article, the stationarity ofthe time series data of real estate return and inflation will e tested to insure that there is nospurious regression.

    3 Cointegration

    ?hile OLS has een widely utili4ed as the classical method in the e)amination of inflationhedging aility of asset, more and more literature argue that the OLS regression may not e

    suitale when the asset is illi/uid, especially real estate. Cointegration method is useful in thiscase to study the long term return1inflation relationship even though the short term relationshipmay vary. *his study follows the two step procedure proposed y -ngle and +ranger !"#$%.Initially real estate return and inflation series are tested for stationarity using augmented :ickey1Fuller test !&:F%5

    "

    "

    J

    t t j t j

    j

    y y y

    =

    = +

    &nd the Cointegration test is ased on that the series at level is I!"%, the residual of theregression then is again tested for stationarity y &:F, if the residual is stationary, then the twoseries are cointegrated which means that they have long term relationship.

    t t ty x = + +

    &nd t is suect to the &:F test.

    4 5ranger Causality .est

    &fter testing the cointegration, +ranger causality test will e used to test causality etweeninflation and real estate return to determine which one causes the other. If the variales are notcointgrated in the levels, then5

    " "

    " "

    J J

    t i t i i t i t

    i i

    y y x u

    = =

    = + + +

    2 2

    " "

    J J

    t i t i i t i t

    i i

    x y x v

    = =

    = + + +

    If the variales are cointegrated, then the +ranger causality1-rror Correction model !+C1-C(% must e used5

    " " ", "

    " "

    J J

    t i t i i t i t t

    i i

    y y x z u

    = =

    = + + + +

    2 2 ", "

    " "

    J J

    t i t i i t i t t

    i i

    x y x z v

    = =

    = + + + +

    *he term "tz is added to reflect the long1term relationship. *o test the causality, the F test

    will e used to test the oint significance of i .

    ection , 6ata 6escri!tion

  • 8/13/2019 11285-1

    4/11

    *he actual inflation rates of various cities are calculated from the CDI, which is availalefrom :atastream. *o decompose the inflation rate to e)pected and une)pected inflation, &'I(&!",3,A% model was used as suggested y +at4laff !"##;%. *he statistical properties of inflation rateof these cities are shown from tale " to tale ;.

    Table 1Inflation Rate of Beijing

    8I' -I' &I' (ean 13.3330$3 13.33

  • 8/13/2019 11285-1

    5/11

    Table 2Inflation Rate of Shanghai

    8I' -I' &I'

    (ean 1;.0;-13< 13.33

  • 8/13/2019 11285-1

    6/11

    From these tales we can see that over the studied years, these cities e)perienced a lowinflation period or even deflation, which is accompanied y the high growth rate of Chinese macroeconomy. &nd more the inflation rate is of little volatility which can e seen from the standarddeviation.

    *he data of real estate return is otained from Chinese 'eal -state Inde) !C'-I%. *his inde)contains the /uarterly real estate price inde) of maor Chinese cities since "##eiing, Shen4hen, Chengdu, which can represent the north, south, eastand west part of China. In each city, oth residential property, commercial property and officeproperty are e)amined. &s mentioned earlier, the &:F test must e done to test the stationarity.

    From the tale < to tale , we can see that most of the series are stationary at the level ofut as the short termdevelopment of Chinese real estate market and the availaility of data, this is what we can do est.!In the tales,'D1return of real estate, ''D1return of residential property, 'OD1return of officeproperty, 'CD1return of commercial Droperty, -I'1 the e)pected inflation rate, 8I'1une)pectedinflation rate.GG significant at ", G significant at

  • 8/13/2019 11285-1

    7/11

    -I' 1

  • 8/13/2019 11285-1

    8/11

    Table 1,. The )*S Result of Shanghai Real +state Ma!et

    Intercept -I' 8I' &'!"% F1statistic '2!&dusted%

    'D 3.333$ 13.32#A$ 13.3A0 3.#0$GG A

  • 8/13/2019 11285-1

    9/11

    oth e)pected and une)pected inflation rate in those developed countries.&nother finding of the result is that almost in all cases the lagged term is significant at

  • 8/13/2019 11285-1

    10/11

    Table 1%. Unit Root Test of the Regession Residual

    Level

    :F!no lag% 1".#&:F !" lag% 1"."0

    &:F !; lag% 1"."2

    *his shows that the two series are not cointegrated ecause the unit root test show elowcritical value statistic even for :F test. *herefore it can e concluded that there is no evidence thatthe Chinese real estate market is a long term hedge against inflation. &nd the result is the samewith *arert !"##0% and Stenvenson and (urray !"###%, however it is contrary to the findings of>arkham, ?ard and enry !"##0%, (atysiak et al !"##0% and Stevenson !2333%.

    &s there is no evidence of cointegration of real estate return and inflation rate nationwide, anerror1correction term is not re/uired in the +ranger causality model. *he result of the +rangercausality test is listed elow5

    Table 1&. ange Causalit Test of Real +state Retun and Inflation

    Null ypothesis5F1Statistic Droaility

    I' does not +ranger Cause '-*8'N 3.#$#A2 3.;"0;