Housing Prices and Business Cycle in China: A DSGE Analysis

Transcription

1 MPRA Munich Personal RePEc Archive Housing Prices and Business Cycle in China: A DSGE Analysis Qing He and Fangge Liu and Zhongxin Qian and Terence Tai Leung Chong Renmin Universiy of China, The Chinese Universiy of Hong Kong, Nanjing Universiy 21 July 2016 Online a hps://mpra.ub.uni-muenchen.de/75921/ MPRA Paper No , posed 1 January :48 UTC

2 Housing Prices and Business Cycle in China: A DSGE Analysis Qing He School of Finance & China Financial Policy Research Cener, Renmin Universiy of China Fangge Liu Academic Journal Press, Renmin Universiy of China Zongxin Qian School of Finance, Renmin Universiy of China 1 Terence Tai Leung Chong Deparmen of Economics and Lau Chor Tak Insiue of Global Economics and Finance, The Chinese Universiy of Hong Kong and Deparmen of Inernaional Economics and Trade, Nanjing Universiy Absrac We invesigae he ineracion beween he real esae marke and he business cycle volailiy in China over he pas wo decades. A Bayesian dynamic sochasic general equilibrium (DSGE) model wih nominal sickiness and collaeral consrains is esimaed. I is found ha shocks from he housing marke (e.g., loan-o-value raio and housing preference shocks) affec he macroeconomy of China. The ineracive feedback beween credi consrains and housing prices amplifies he impac of various economic shocks, which plays an imporan role in explaining he business cycle volailiy in China. Keywords: housing prices; business cycle; DSGE; China 1 We hank Donald Lien, Taoxiong Liu, Ke Tang and seminar paricipans a he Second Inernaional Conference on The Chinese Economy: Pas, Presen and Fuure, held by Insiue of Economics, Tsinghua Universiy for heir suggesive and helpful commens. This paper is suppored by he Fundamenal Research Funds for he Cenral Universiies, and he Research Funds of Renmin Universiy of China (13XNJ003). Corresponding auhor: Zongxin Qian, School of Finance, Renmin Universiy of China. qianzx@ruc.edu.cn. All remaining errors are ours. 1

3 JEL Classificaions: E32, E42, R31 2

4 1. Inroducion Over recen decades, he boom bus cycles in he real esae marke have been widely documened and discussed (Hiraa e al., 2012). Considerable aenion is paid by policymakers and researchers o he ineracion beween housing prices and business cycle flucuaions. For insance, Iacoviello (2005) analyses he collaeral channel: rising housing prices lead o rising collaeral value and increasing borrowing capaciy, which promoes consumpion and economic growh. Benio (2006) suggess ha credi consrains play an imporan role in he housing marke and significanly affec business cycle volailiy. Iacoviello and Neri (2010) use a Bayesian approach o quanify he conribuion of he housing marke o business cycle flucuaions. Previous sudies have focused on developed counries, and lile is known abou emerging marke economies. How real esae asses inerac wih business cycle volailiy is an imporan issue ye o be fully undersood. In his paper, we add o he lieraure by invesigaing how he housing marke ineracs wih he shocks ha drive economic flucuaions in China. The case of China is of ineres for several reasons. Firs, China exhibis almos all of he noable feaures ha characerize emerging marke counries. For example, China s financial sysem is characerized by he repressive financial policies ha are prevalen in many emerging counries. Small and medium-sized firms rely heavily on informal finance, a ypical feaure for mos emerging marke economies (He e al., 2016). Second, despie he financial repression, China s housing marke has grown rapidly since The propery prices in China have risen rapidly over he las decade. Home prices are 10 imes he median household income and in Beijing he raio is close o 22 imes. Undersanding he housing price dynamics and associaed governmen policies in China provides valuable experiences for oher emerging marke economies. Finally, as China has played a significan role in he global marke, he collapse of he housing marke in China will have a devasaing impac on he world economy. Therefore, how he Chinese governmen should preven excessive speculaion in he housing marke and a subsequen housing marke collapse has received increasing inernaional aenion. To shed ligh on his issue, we develop a DSGE model based on Iacoviello (2005) o invesigae he channels hrough which he housing marke affecs he economy in 3

5 China. An imporan aribue of our model is ha i akes ino accoun he nominal price sickiness and collaeral consrains, which allows us o sudy he ineracions beween he housing marke and macroeconomic flucuaions. Specifically, we exend he model of Iacoviello (2005) in wo aspecs: firs, we incorporae fiscal policy ino he model so ha he relaive conribuion of fiscal policy and moneary policy o macroeconomic flucuaions can be evaluaed. Second, we allow a sochasic shock o loan-o-value raio (LTV) so ha impacs of shocks caused by cenral banks moneary policy can be invesigaed. The main finding of his paper is ha he ineracive feedback beween credi consrains and housing prices magnifies he impac of various economic shocks. Because of he ineracive effec, he shocks from he housing marke, e.g., loan-o-value raio and housing preference shocks, are he imporan deerminans of business cycle volailiy in China. Housing preference shocks are he mos imporan drivers of housing price flucuaions. Policies which can affec he effecive housing demand are herefore imporan for smoohing housing price flucuaions. However, such polices also lead o a decline in oupu and employmen rae. The res of his paper is organized as follows: Secion 2 presens basic facs on China s housing secor. Secion 3 describes he DSGE model. Secion 4 and 5 presen he model esimaion and inference resuls. Secion 6 concludes his paper. 2. Housing Marke in China Over he las hree decades, China has mainained a rapid economic growh rae. However, is privae housing marke did no exis unil On July 18, 1994, The Sae Council issued he Decision on Deepening Urban Housing Reform and iniiaed he commercializaion of urban houses. The governmen began o change he urban employee s compensaion sysem by disribuing money insead of houses according o heir working performance. This reform has seadily promoed he privaizaion of housing in China by selling former sae-owned properies o urban households. However, in he iniial sage, mos houses were sill conrolled by he governmen, and households sill relied on he governmen house disribuion sysem o obain a propery. On July 3, 1998, he Sae Council issued he Noice on Furher Deepening Ur- 4

6 ban Housing Sysem Reform and Speeding up he Housing Consrucion. To remove he barrier for he commercial housing marke, his reform abolished he house disribuion sysem. Urban households had o ener he commercial housing marke o purchase and sell heir real esae asses. To encourage furher developmen of residenial houses for low-and-medium-income ciizens, he Sae Council issued he Noice on Promoing he Susainable and Healhy Developmen of he Real Esae Marke. Several policies were implemened o solve he housing problem of urban, low-income families. Afer a series of reforms, he privae housing marke was esablished, and i has experienced a rapid expansion over he pas decade. Panel A and B of Figure 1 show he annual changes in floor space of newly sared commercial residenial housing and he invesmen in residenial housing respecively beween 1998 and Noe ha here were only million square meers of residenial houses in 1998, while million square meers were buil in During he same period, invesmen in residenial housing increased 19.9 imes, and accouns for more han 10 per cen of GDP a he end of [Inser Figure 1 here] Wih he expansion of he housing marke, bank credi plays an increasing role in financing he residenial housing invesmen. Figure 2 shows ha he raio of bank loans o residenial housing over GDP increases from less han 1% o more han 17.23%. The loan ousanding o residenial housing was less han 100 billion RMB in 1998, and reached 9800 billion RMB a he end of The growh rae of residenial housing loans is much higher han he growh rae of he money supply. 2 This suggess ha an increasing percenage of money supply is channeled ino he real esae marke. [Inser Figure 2 abou here] Figure 3 shows he annual changes in he nominal and real housing prices from 1998 o 2013, where he real housing prices are measured as he nominal housing prices deflaed by he consumer price index. I can be seen ha residenial propery prices rose rapidly, boh in nominal and real erms, from 1998 o The average 2 The annual average growh rae of he money supply from 1998 o 2013 was 16.8%. 5

7 housing price almos ripled from 1998 o During he global financial crisis from , housing prices fell by 14%, bu recovered rapidly owing o a number of simulus policies, e.g., a lower ineres rae and bank reserve raio. 3 Housing prices have risen rapidly since [Inser Figure 3 abou here] Due o he fear of asse bubbles, he Chinese governmen has adoped a series of policies, e.g., a igher moneary policy, higher ransacion axes, a reducion in morgage lending and increases in down paymens, 4 o cool down he demand for real esae and reduce speculaions in he housing marke. Table 1 summarizes he primary policies adoped by he Chinese governmen o cool down he housing marke. [Inser Table 1] The Chinese governmen relies on adminisraive measures o conrol housing prices. One such measure is o conrol he down-paymen raio (loan-o-value raio). Despie he effors made by he governmen, he housing price level in China relaive o household income is sill very high compared o he res of he world. 3. The Model The model feaures four ypes of economic agens, namely enrepreneurs, reailers, paien and impaien households. The paien households are lenders while he impaien households and enrepreneurs are borrowers. The paien households have lower discoun raes han impaien ones (Iacoviello, 2005). Reailers are direcly owned by he lenders. Following Iacoviello (2005), he supply side is divided ino a producion secor and a reailing secor o simplify aggregaion. The flexible-price producion secor, managed by he enrepreneurs, faces a borrowing consrain due o he same ype of conrac incompleeness as he one specified in Kiyoaki and Moore (1997). In he reailing secor, here is no conrac incompleeness and borrowing con- 3 To offse adverse global economic condiions, he Chinese governmen launched a CNY 4-rillion simulus plan on Nov. 9, 2008, o boos domesic demand. Due o he relaionship beween lending condiions and a lower ineres rae, boh developers and buyers were able o easily ge loans o inves in he housing marke. 4 In March, 2013, he Sae Council made new policies o conrol house prices: (1) major ciies should esablish an annual housing price conrol arge. (2) Banks should no give loans o developers who hoard land. 6

8 srain. The nominal price is sicky. 3.1 Households secors The paien households maximize heir expeced lifeime uiliy given by ' L ' ' ' E (ln c j ln h ), 0 (1) subjec o he consrain: R b c q h b wl F ' ' ' ' 1 ' ' ' ' V h,, (2) ' where is he subjecive discoun facor; c ' is he consumpion a ime, h ' is he real esae sock; j is he weigh on he real esae sock which is assumed o be ime varying o allow for a preference shock; ' L are he number of working hours; is he real housing price; R 1 is he nominal gross ineres rae; inflaion rae; P Q q P is he gross P 1 ' b is he real sock of lending provided by he household; w ' is he ' real wage rae; F is he lump-sum profi from he reailers; is ax paid o he gov- ' ' ' 2 ' ernmen; h h( V h / h ) qh /2 is he housing adjusmen cos; h, are con-, 1 1 san parameers. Equaion (2) saes ha in each period, he households receive loan repaymens, working wages, profis from he reailing secor and use hem o buy consumpion goods, purchase houses, issue new loans and pay axes. The impaien households have o borrow o purchase houses since heir discoun '' ' facor is. If a borrower breaches he conrac, he lender can obain he ne value of he real esae by paying ransacion coss. Therefore, he firm's borrowing canno exceed he ne presen value of he real esae imes a loan-o-value raio. The gross ineres rae is nominally specified in he loan conrac. Therefore, inflaion can reduce he real value of he deb. More specifically, he impaien households maximize '' L '' '' '' E (ln c j ln h ), 0 7

9 (3) subjec o he budge consrain: and he borrowing consrain R b c q h b w L -, '' '' '' '' 1 '' '' '' '' V h, b m E ( q h / R ) '' '' '' 1 1 LTV, (4) (5) where '' b is he borrowing of he impaien households; 8 '' w is heir wage rae; '' is ax paid o he governmen; '' ( ' / ' ) 2 ' h h V h h qh /2 is he housing adjusmen cos; '' m is he loan-o-value raio; LTV,, 1 1 is a shock o he loan-o-value raio. In each period, he impaien households receive loans and wage paymens and use he money o buy consumpion goods, purchase houses, pay axes and repay he deb. Before we carry on, i is imporan o have some discussion on he housing preference parameer j. Firs, i is assumed o be he same across boh household secor, so a shock o i is an aggregae shock. Second, since we have used he represenaive agen approach o aggregae boh he paien and impaien household secors, j no only reflecs unexpeced changes in individual households' preferences, bu also reflecs anyhing ha affec he ransformaion of each individual household's preferences ino he aggregae effecive housing demand. An imporan Chinese macroprudenial policy which can affec he effecive housing demand is resricions on housing purchase (see Table 1). Even if individual households have very srong preference for housing, adminisraive resricions will preven ha srong preference from forming a srong effecive demand for housing. Therefore, by sudying he effecs of he housing preference shocks, we can analyze he effecs of his kind of macroprudenial policies. Anoher shock which is relaed o governmen policies is he loan-o-value shock. '' The loan-o-value raio m measures he amoun of loan a represenaive house buyer can obain wih a given marke value of he house. Obviously, a leas wo governmen credi policies can affec he aggregae loan-o-value raio. The firs policy is he minimum down paymen requiremen (see Table 1). The second is he adminisraive policy which direcly affecs aggregae credi supply. Suppose he loan-o-value raio se by an individual bank for he bank's own risk managemen purpose is 0.8, i does

10 no mean a poenial borrower wih a collaeral valued a 6 million RMB can borrow 4.8 million. If he down paymen is se o be 40 percen, a leas 2.4 million has o be paid by he buyer herself, so he acual loan-o-value raio will be lower han 0.8. In pracice, he Chinese cenral bank usually applies adminisraive policies o influence he commercial banks aggregae credi supply. Suppose a bank's loan-o-value raio is 0.8 and here is no down paymen requiremen, i could be ha a house buyer's arge house has a collaeral value of 6 million bu she canno borrow up o 4.8 million. This is because here may be oo many borrowers whose loan demand adds up o a oal amoun higher han he oal supply he bank can offer. The credi policies work hrough changing he gap beween he demand and supply. 3.2 Producion and reailing secors price Enrepreneurs produce wholesale goods. Reailers buy he wholesale goods a w P and re-sell hem a price P. Enrepreneurs choose consumpion c, invesmen I K (1 ) K 1, housing h, '' L o maximize heir expeced lifeime uil- employmen of wo ypes of labors iy subjec o he producion echnology ' L and E 0 ln c, (6) ' (1 ) (1 )(1 ) k 1 1 Y A K h L L, (7) he budge consrain Y / X b c qvh R b / wl w L I + ' ' '' '' 1 1 e, K,, (8) and he borrowing consrain b me ( q h / R ), 1 1 LTV, (9) 9

11 ' where is he discoun facor; is he capial depreciaion rae; K is he capial sock; A is he echnology parameer; X P is he markup of reail price over P w he whole sale price; is ax paid o he governmen; ( V h / h ) qh /2 is 2 e, e he housing adjusmen cos of enrepreneurs; ( I / K ) K /(2 ) is he K, 1 1 capial adjusmen cos; m is he loan-o-value raio faced by enrepreneurs; α, μ, v, e, and ϕ are consan parameers. The enrepreneurs receive sales revenue and loans and use hem o buy consumpion goods, purchase houses, repay deb, pay salaries and axes and make invesmen. The reailing secor feaures monopolisic compeiion. The mass of reailers is 1. 1/ / 1 Indexing he reailers by z, he final goods are Y ( Y( z) dz), where 1. Assuming ha he reail price can change in every period only wih probabiliy of 1-θ and denoing he rese price by f 1 0 P * ( z ), he reailer's opimizaion problem is o choose P * ( z ) o maximize he ne presen value of is lifeime profi flows. The firs order condiion o his opimizaion problem is as follows: * k *, K K k 0 Pk XK P ( z) X E{ [ ] Y ( z)} 0, (10) where c is he sochasic discoun facor, ' K, '' c X 1 is he seady-sae markup. Combining his resul wih he aggregae price index P [ P (1 ) P ] and log-linearizing, we obain he usual New Keynesian e * 1/1 1 Phillips curve ^ ^ ^ 1 X, where (1 )(1 ) /. We add a cos-push erm u o he righ hand side of he Phillips curve as an addiional source of business cycle flucuaions in our model. 3.3 Equilibrium, moneary policy and shock processes The marke clearing condiions for he labor marke, housing marke, final goods marke, and he loan marke are respecively: 10

12 L L L, ' '' h h h H, ' '' c c c I G Y ' '' (11) (12) (13) ' '' b b b 0. (14) The moneary policy on ineres rae follows r R 1 r r Y 1 r R R, R ( R ) [ rr ( Y / Y) ] e, (15) where H is he fixed supply of houses; rr and Y are he seady-sae real ineres rae and oupu respecively; e R, is he moneary policy shock wih a variance 2 R. The governmen budge consrain is G. ' '' The governmen expendiure evolves as follows: ln G (1 g) ln Gg ln G 1 eg,, (16) where G is he seady-sae governmen expendiure, e g, is a i.i.d. shock wih vari- 2 ance g. Oher shock processes follow: ^ j j e j ^ 1 j, (17) (18) (19) ^ e ^ LTV, LTV LTV, 1 lv, (20) 11

13 where ha denoes percenage deviaion from he seady-sae; variances of e j,, e, e, e lv, are 2 j, 2, 2 a, 2 lv, respecively., a, 4. Calibraion and Esimaion Table 2 liss he calibraed parameers of he model. For sandard parameers, he values are se wihin he range considered in he real business cycle lieraure. The discoun facors (β, β ) of he paien and impaien households are se o 0.99 and 0.985, respecively, o mach China s ineres raes on savings and personal housing morgages. The discoun facor for enrepreneurs is se as 0.95 o mach he Chinese firms housing morgage rae, 21%, which is he average housing morgage rae of he informal finance secor hrough our sample period. 5 As he official lending rae is highly regulaed in China, mos small and medium-sized firms find i difficul o access exernal finance from he formal financial secor in China (Allen e al., 2005). The official lending rae is less likely o reflec he acual borrowing cos of he firms. 6 δ is se o be 0.025, corresponding o an annual capial depreciaion rae of 0.1 (Bai e al., 2006). Housing share (v) and capial share (μ) are 0.11 and 0.39 respecively. The sum of hose wo parameers equals 0.5, he same as he value esimaed by Bai e al. (2006). We se he housing share o be he value esimaed by Huang e al. (2008). The values of oher calibraed parameers follow Iacoviello (2005). [Inser Table 2 abou here] Oher parameers are esimaed by employing he Meropolis Hasings Bayesian mehod wih 10,000 burn-in draws and 100,000 ieraions. Due o he sylized naure of DSGE models, maximum likelihood esimaion ofen generaes absurd parameers which are a odds wih prior informaion. The Bayesian mehod can avoid his problem. Moreover, prior informaion adds curvaure ino he poserior densiy surface so ha numerical maximizaion can beer idenify model parameers han he maximum likelihood esimaion when he likelihood funcion is fla around some parameer values. We employ quarerly daa of five macroeconomic variables from 1998Q1 o 5 Daa obained from he CEIC. 6 We also se he discoun facor for enrepreneurs o mach he official lending rae. I urns ou ha our primary resuls remain qualiaively unchanged. 12

14 2013Q4: real GDP (Y), consumer price inflaion (π), real governmen expendiure (G), invesmen (I), and he ineres rae (R). Our daa are drawn from he CEIC daabase. Real variables are a consan prices in All variables are seasonally adjused by he US Census X12 mehod. The HP filer is employed o derend he daa. The summary saisics of he prior and poserior disribuions are summarized in Table 3. The prior mean of he housing preference parameer j is se o be 0.2, much higher han he US level. We se a relaively large prior sandard error of his parameer o reflec he uncerainy of is rue value. Priors of he moneary policy parameers are se o be he OLS esimaes. Prior means of housing adjusmen cos parameers are se o 1, higher han he US level (0), o reflec he heavy regulaions and higher ransacion coss on housing sock adjusmen in China. However, he prior sandard deviaions of hese adjusmen parameers are se relaively large o provide room for he daa o adjus o he prior. [Inser Table 3 abou here] 5. Empirical Resuls 5.1 Impulse response Figure 4 repors he economic response o various srucural shocks. Noe ha afer a posiive echnology shock, inflaion goes down while oupu goes up. The ineres rae goes down as a reacion o falling inflaion. Oupu expansion pushes up housing prices. Afer a posiive housing preference shock, housing prices rise. The housing boom leads o oupu expansion and inflaion, and a subsequen increase in ineres raes. As we have discussed above, he governmen's resricions on housing purchase can cause a negaive housing preference shock, so our resuls sugges ha hose resricions can effecively slow down he housing price increase, bu also reduce oupu. Therefore, i is imporan o pay aenion o he coordinaion beween he macroprudenial policy which aims a smoohing house price flucuaions and regular macroeconomic policies which aims a fighing recession and mainaining full employmen. A posiive loan-o-value shock leads o more lending and aggregae demand. The 13

15 increasing demand leads o higher oupu level. The oupu expansion raises he nominal ineres rae, and lowers he consumpion and housing demand of he paien households. Therefore, he housing price drops, alhough oupu expansion leads o a sronger indusrial demand for housing. Inflaion gradually climbs o a level higher han he seady-sae level and says above he seady-sae level for a long ime. As he indusrial demand for housing increases, i finally offses he decline in he housing demand of he savers, and he housing price increases. As we discussed above, boh a higher down paymen requiremen and a igher adminisraive credi supply policy cause a negaive loan-o-value shock. Therefore, our resuls sugges ha if he aim is o slow down he housing price increase, such policies only become effecive wih lags. In he firs few quarers afer he inroducion of he policies, housing price will be even higher. Moreover, hese policies will have a negaive impac on oupu and employmen. A conracionary moneary policy shock raises he nominal ineres rae, which leads o lower oupu level and inflaion. The housing price also declines. I urns ou ha he impac of he unexpeced moneary ighening is more persisen on oupu han on inflaion. Inflaion soon sars o increase again afer a emporary decline. The rising inflaion reduces he real deb value of he borrowers and encourages invesmen in capial goods and housing. As a resul, boh housing price and oupu gradually recover. A cos-push shock leads o higher inflaion, oupu and housing price. On he one hand, he increase in cos lowers profi, prevens invesmen and indusrial housing demand. On he oher hand, inflaion lowers he real value of nominal deb, which loosens he exernal financing consrain, encourages invesmen and increases indusrial housing demand. The deb devaluaion effec also increases he housing demand of impaien households. When he cos-push shock hi he economy, he laer effec dominaes, so aggregae demand increase, which pushes up boh oupu and housing price. Bu as ime goes by, he negaive impac of he cos increase on profi and invesmen becomes more and more imporan. Therefore, boh oupu and housing price decrease owards he seady-sae values. An increase in governmen spending increases boh oupu and ineres rae. The increase in he ineres rae lowers inflaion rae. Because he governmen spending increases axaion, i lowers consumpion and housing demand. As a resul, he oupu 14

16 expansion effec is limied and he housing price declines. [Inser Figure 4 here] To compare our findings wih exising lieraures, we also repor he impulse responses o various srucural shocks of he US model calibraed by using he parameers in Iacoviello (2005). Noe ha Iacoviello (2005) does no include he governmen spending shock and LTV shock in his model. No resuls relaed o he wo shocks are repored in he US model. Figure 5 shows ha he impulse responses o all shocks excep he cos-push shock are qualiaively similar in boh our Chinese model and he US model of Iacoviello (2005). The cos-push shock increases inflaion in boh counries. However, he impulse responses of oupu, and housing price differ qualiaively in he wo counries. While oupu and housing price iniially increase in response o he cos-push shock in China, he iniial impulse responses of hose wo variables are negaive in he US model. I may indicae ha he size of deb devaluaion effec of inflaion was limied in he US. [Inser Figure 5 here] 5.2 Variance decomposiion Table 4 presens resuls from variance decomposiion. Panel A repors he resuls for China. I shows ha he moneary policy shock based on ineres rae adjusmen has limied impac on China s business cycle flucuaions. By conras, he loan-ovalue (LTV) shock plays a noable role. The LTV shock parly reflecs changes in he PBOC s credi policy. Therefore, his resul suggess ha he credi policy has more influence on China's business cycles han he ineres rae policy. The cos-push shock explains nearly half of China s business cycle flucuaions. Such shocks are relaed o various adminisraive policies, which direcly affec he price level. The housing preference shock also plays an imporan role. As we have discussed above, his shock is also relaed o governmen regulaions. Puing ogeher, hose resuls sugges ha adminisraive policies of he Chinese governmen have conribued a lo o China's business cycle flucuaions. In he fuure, he governmen needs o ransform is policy o make i more marke-oriened which can help beer manage he expecaions of economic agens and smooh he business cycles. In addiion, we also observe ha he housing preference shock is he mos im- 15

17 poran source of flucuaions in boh he housing price and invesmen. Recall ha a posiive housing preference shock increases boh housing price and oupu. Therefore, he Chinese governmen faces a ough policy radeoff: a policy aimed a smoohing he housing price migh dampen invesmen, cause oupu fall and unemploymen. Panel B of Table 4 repors he variance decomposiion resuls aken from Iacoviello and Neri (2010). Similarly, i shows ha cos-push shocks are imporan drivers of he business cycle flucuaions in boh China and he US. However, we also find noable differences across hese wo counries. Compared o he US, he cenral bank's ineres rae policy shocks do no play a major role in China's business cycle flucuaions. This is consisen wih he general observaion ha adminisraive measures are more effecive in adjusing he real economy and price level in China (He e al., 2013). As a proxy of he PBOC's adminisraive measures, he LTV shock explains a large par of China's business cycle flucuaions. The housing preference shock plays a much larger role in China han in he US wih regards o invesmen and house price flucuaions. There are wo possible explanaions. Firs, compared o he US, he Chinese governmen relies more on adminisraive policies o figh economic flucuaions (He e al., 2013). Housing preference is likely o be influenced by adminisraive policies. Second, house secor has a larger weigh in he Chinese households' uiliy funcion. Thus, he shocks o housing preference play an imporan role on driving house price and business cycle volailiies. [Inser Table 4 abou here] 6. Conclusion This paper invesigaes he ineracion beween he housing marke and he real economy in China. I is found ha he shocks relaed o he housing marke, e.g., loano-value raio and housing preference shocks, are imporan deerminans of business cycle flucuaions in he counry. The housing preference shock is also he mos imporan driver of housing price flucuaions. These resuls have valuable policy implicaions for he Chinese economy. Alhough China has adoped a series of governmen policies o avoid overheaing is housing marke, mos of hem are heavily relaed wih governmen adminisraive inervenions. These policies migh be effecive in slow down he housing price increase 16

18 and preven excessive speculaions in he housing marke from creaing a sysemic financial risk, hey also conribue o oupu flucuaions. Paricularly, due o he economy's heavy dependence on he housing secor, downward changes of he housing price dampen invesmen and cause oupu falls. Alhough his is no a problem, someime even desirable, when he economy is overheaed and face he hrea of inflaion, i will pose a ough policy choice for he governmen when he economy is in recession and here is he hrea of unemploymen. In his sense, he enire economy is capured by he housing secor. To relax he policy consrain for he governmen, fuure policies should encourage innovaion and echnological improvemens which help he economy grow wihou a heavy dependence on he housing secor. Moreover, compared wih he adminisraive policies, marke-oriened ineres rae policy has limied impac on he business cycle flucuaions in China. Fuure reforms are necessary o increase he effeciveness of he marke-oriened policies. I should be noed ha our analysis is a counry level. China s large geographical scale leads o grea diversiy in is regional economies. Therefore, fuure sudies along his line should focus on he provincial level analysis and urban rural comparison. Reference Allen, F., Qian, J. and M. Qian (2005). Law, Finance, and Economic Growh in China, Journal of Financial Economics, 77(1), Bai, C. E., Hsieh, C. T. and Y. Qian (2006). The Reurn o Capial in China, Brookings Papers on Economic Aciviy, 2006(2), Benio, A. (2006). The Down-Paymen Consrain and UK Housing Marke: Does he Theory fi he Facs?, Journal of Housing Economics, 15 (1), He, Q., Leung, P. K. and T. T. L. Chong (2013). Facor-Augmened VAR Analsis of he Moneary Policy in China, China Economic Review, 25, He, Q. Xue, C., & Zhu, C. (2016). Financial Developmen and he Paerns of Indusrial Specializaion: he Evidence from China, Review of Finance, forhcoming Hiraa, H., Kose, M. A., Orok, C. and M. E. Terrones (2012). Global House Price Flucuaions: Synchronizaion and Deerminans, Naional Bureau of Economic Research Working paper No

19 Huang, H., Wu, C. and X. Du (2008). Real Esae Invesmen and Economic Growh, Finance and Trade Economics (in Chinese), 8, Iacoviello, M. (2005). House Prices, Borrowing Consrains and Moneary Policy in he Business Cycle, American Economic Review, 95(3), Iacoviello, M. and S. Neri (2010). Housing Marke Spillovers: Evidence from an Esimaed DSGE Model, American Economic Journal: Macroeconomics, 2(2), Kiyoaki, N. and Moore, J. (1997). Credi Cycles, Journal of Poliical Economy, 105, Table 1 Governmen policies Time Policy 2005 Eigh real esae marke regulaion measures Major Conen (1) sabilizing housing price; make governmens responsible for housing price sabiliy; (3) reform he supply srucure; (4) conrol he passive housing demand, paricularly housing deconsrucion; (5) opimizing household housing demand; (6) monioring developmens of real esae marke; (7)enforcing housing policies on boh demand and supply; monioring housing price sabilizaion policies Six real esae marke regulaion measures 2009 Four real esae marke regulaion measures (1) adjus supply srucure; (2) make beer use of ax, credi and land policies; (3) conrol he size and progress of housing desrucion, slow down he growh of passive housing demand; (4) resore marke order in he real esae secor; (5) speed up he developmen of cheap rening housing, improve regulaions on he housing wih regulaed prices, develop he secondary marke and rening marke; (6) more ransparency. (1) increase effecive supply; (2) suppor self-occupancy, suppress speculaion, differeniae credi policies; (3) more marke discipline; (4) suppor he developmen of welfare housing 2010 Eleven real esae marke regulaion measures Sric regulaion on second house purchase, he down paymen has o be a leas 40%, window guidance on credi exended o he real esae secor. 18

20 2012 New Eigh real esae marke regulaion measures 2013 Five real esae marke regulaion measures (1)Furher implemen local governmen responsibiliy; (2) Increase of indemnificaory comforable living engineering consrucion force (3)Adjus and improve he relaed ax policy, and srenghening ax collecion and adminisraion; (4) Srenghening decreases. i housing credi policy. (5) Sric housing land supply managemen (6) Reasonable guide housing demand (7) Implemen housing safeguard and sable prices work inerviews accounabiliy mechanism;(8) Persisence and srenghen public opinion guide (1) Improve he responsibiliy o sabilize he housing prices, e.g., major ciies should esablish an annual housing price conrol arge;(2) Resoluely curb speculaive invesmen buyers e.g. In ciies, where house price is higher han he conrol arge price, People s Bank of China (PBOC) branches will increase down paymen requiremens ; (3) To increase he supply of ordinary commodiy housing and land.; (4) Accelerae he planning and consrucion of affordable housing projecs;(5) Srench marke supervision 19

21 Table 2 Calibraed Parameers Descripion Parameer Value Discoun facor, paien households 0.99 Discoun facor, impaien households Discoun facor, enrepreneurs 0.95 Capial depreciaion rae Seady-sae markup 1.05 Housing share 0.11 Capial share 0.99 Capial adjusmen cos 2 20

22 Table 3 Prior and poserior disribuions of he srucural parameers Prior Poserior Parameers Disribuion s.d. mean Mean 10% 90% Housing preference j Bea LTV raio(enrepreneurs) m Bea LTV raio(paien households) m '' Bea Wage share of paien households Bea Housing adjusmen cos(enrepreneurs) e Inverse-Gamma Housing adjusmen cos(impaien households) h Inverse-Gamma Calvo sickiness parameer Bea Moneary policy coefficien (oupu gap) Y Normal Moneary policy coefficien(inflaion) Normal Moneary policy coefficien(lagged ineres rae) R Bea Auoregressive coefficien of he preference shock j Bea Auoregressive coefficien of he cos-push shock u Bea Auoregressive coefficien of he echnology shock a Bea Auoregressive coefficien of he LTV shock lv Bea S.d. of he moneary policy R Inverse-Gamma S.d. of he preference shock j Inverse-Gamma S.d. of he cos-push shock u Inverse-Gamma S.d. of he echnology shock a Inverse-Gamma S.d. of he LTV shock lv Inverse-Gamma

23 Table 4 Variance decomposiion Moneary Policy Housing preference Cospush Technology LTV Governmen spending Panel A: China Oupu Invesmen Inflaion Housing price Panel B: US Oupu Invesmen Inflaion Housing price Noes: repored numbers are percenages. US resuls are aken from Table 6 of Iacoviello and Neri (2010). They have a more disaggregaed decomposiion on differen ypes of echnology shocks. The LTV shock and governmen spending shock are no considered in heir paper. 22

24 Figure 1 Panel A Floor space of newly sared of commercial residenial housing (Million square meers) Floor space Daa source: CEIC Panel B: Invesmen in residenial housing (Billion RMB) GDP Invesmen in residenial housing Daa Source: CEIC 23

25 Figure 2 Personal Loan ousanding (Billion RMB) 22.50% 18.00% 13.50% 9.00% 4.50% 0.00% Daa Source: CEIC Loan Ousanding(Billion RMB) Loan Ousanding/GDP 24

26 Figure 3 Residenial house prices (RMB per square) Daa Source: CEIC Nominal Real 25

27 Figure 4 Impulse response (China) 26

28 Figure 5 Impulse response (US) 27