A Study On The Price Transmission: Experience From Chinese Market

Transcription

1 Cambridge Journal of China Sudies 17 A Sudy On The Price Transmission: Experience From Chinese Marke Dr Kang LIU 1 Cenre of Developmen Sudies, Universiy of Cambridge kl387@cam.ac.uk Absrac: This aricle examines he flucuaion characerisics of price level in China, and sudies he ransmission mechanism of price hrough Granger Causaliy es. I shows he ransmission relaionship beween CPI and PPI in China is no sable from 2000 o 2013: a bidirecional ransmission was observed a he iniial sage of he es, no ransmission a he inerim sage and bidirecional ransmission a he las sage. The flucuaions of he food price are he main reason of he flucuaions of price level in China, and are also he main reason of backward ransmission beween CPI and PPI. The money supply in China had a relaively more significan impac on CPI, bu a very resric impac on he level of PPI. Key words: Price Transmission, Granger Causaliy es 1 Dr Kang LIU, Pos-doc researcher, Cenre of Developmen Sudies, Universiy of Cambridge & Insiue of Finance and Banking, Chinese Academy of Social Sciences.

2 Volume 9, No INTRODUCTION Some new characerisics of price ransmissions were observed in China in he las 10 years. For example, he flucuaions of PPI are much more obvious han ha of CPI, he flucuaions of PPI canno ransmi o CPI in ime. Those phenomena no only conor he ransmission funcion of he price signal, bu also imposed difficulies on macro-conrol and he seing of moneary policies in China. Therefore, undersanding he process of price ransmission and masering he law of price ransmission are significanly meaningful for he regulaion of economics and finance. Relevan lieraures provide heoreical analysis and empirical ess abou he ransmissions beween PPI and CPI from several differen aspecs. The radiional price ransmission heory mainly concenraes on supply shock, which claims ha he flucuaions of price usually ransmi from PPI-measured upsream price index o CPI-measured downsream price index. I means ha PPI could be he leading indicaor of CPI. Having sudied he law of ransmissions in Unied Saes, Silver and Wallace (1980) confirmed ha PPI could lead o he changes of CPI. Kyrsou and Labys (2006) examined he monhly daa in Unied Saes from 1970 o 2002, and found ou ha all commodiies index in PPI have a dynamic and non-linear influence on CPI, which showed ha PPI could be he leading indicaor of inflaion. Whereas Jianqi Chen (2008) esed he causaliy beween PPI and CPI in China from 1999 o 2008, and discovered ha he relaionship beween PPI and CPI is of no apparen saisical significance. He hinks he reverse relaionship beween PPI and CPI in China is mainly due o he synheic acion of invesmen-driven economy and mainenance of gloomy consumpion. Demand deerminaion heory claims ha he demand of consumers for finished goods deermines he demand for inermediae goods. Hence, he price changes of he finished good should have influences on he inpu price of inermediae goods. For example, Colclough and Lange (1982) suggesed ha he changes of he commodiy price would influence he price of inpu goods in producion process. Liping He e al. (2008) suggesed ha CPI Granger-Causes PPI by using Granger es o examine he causaliy beween PPI and CPI in China from 2001 o 2008, which indicaes ha, among all sors of facors, demand facors have a more significan influence on price level han supply facors. Afer examined he price level of China from 1997 o 2008, Zhiqing Dong e al (2009) discovered ha, quaniaively, CPI is a leading indicaor of PPI and he leading period is 3 monhs. I can be seen ha from above lieraures, wheher he ransmissions beween CPI and PPI are forward or backward, here are always suppors from eiher heory analysis or empirical es.

3 Cambridge Journal of China Sudies 19 By using error correcion model, Jingqi Song and Xiaohui Shu (2008) discovered ha here exiss muual causaliy beween PPI and CPI in China in long erm, bu only unidirecional causaliy exiss in shor erm period. Yanqun Zhang (2007) suggesed a differen conclusion by using CVAR model. He verified ha he changes of raw maerials price index can be he leading indicaor for boh CPI and PPI in shor ime period, bu in long erm he endency of CPI deermines PPI. However, Weikang Xu (2010) found ou ha he ransmissions beween CPI and PPI are muual wihin his es period. Why do differen papers give differen conclusions, or even opposie? Firs of all, he differen consiuens of price index in differen counries are he main reason for inconsisen conclusions of he ransmissions beween PPI and CPI derived from differen markes. Secondly, as an acceleraed developing counry, i is difficul for he saic es o be insrucive in solving pracical problems, which is due o he exreme unseadiness of he es conclusion caused by exensions of he sample period. Given he above, his aricle sudies he characerisics of price changes and ransmission characerisics in China from wo aspecs. On one hand, analyzing he consiuens and he characerisics of flucuaions of Chinese PPI and CPI, and esing he main reason of flucuaions of PPI and CPI in differen ime periods. On he oher hand, by using dynamic Ganger Causaliy es o exam he ime-varying characerisics of he ransmission beween PPI and CPI. Also, on he basis of dynamic Granger Causaliy es, Granger - F es (Fei, 2012) is conduced o examine he facors ha cause he ransmissions of PPI and CPI from a quaniaive perspecive. 2. THE FLUCTUATIONS AND CONSTITUENTS OF PPI AND CPI 1. Index Analysis of CPI Consiuion This secion firs calculaes he weigh of each sub index in CPI by using regression mehod, and hen used he weigh o calculae he conribuion raio of each sub index. The formula is following: GXCPI w / CPI i, i * CPI i, (1) GXCPI i, denoes he conribuion raio of he i ype CPI (e.g. food) wihin ime. CPI i, denoes he year-on-year growh rae of i ype CPI wihin ime. CPI denoes he year-on-year growh of he overall level of CPI wihin ime

4 Volume 9, No w i denoes he weigh of he i ype index from January 2001 o December 2013 When using regression mehod o calculae he weigh of each sub index in CPI, in order o ensure he indices are saionary ime series, HP mehod is conduced o calculae he gap of indices. The smoohing parameer of HP is In order o reflec he conribuion raios more direcly, we calculae average conribuion raio of every index in confidence inerval in his secion. The formula is following: GXCPI i 1 T T 1 GXCPI i, (2) The curren rise of CPI was mainly driven by he rise of food price, and he cos of housing also played an imporan role. Table 1 shows ha during he period from January 2001 o December 2013, he mean value of he conribuion raio of he rise in he price of food and housing price accouned for 59.6% and25.6% respecively, followed by educaion and enerainmen (7.2%) and healh care (5.9%). Table 1 Esimaed Weigh Value of he Indices in he CPI and he Conribuion Raio of he Variables o he CPI ( ) Variables Weigh Coefficien Weigh Saisic Mean Value of he Conribuion Raio Food Alcohol & Tobacco Clohing Family Equipmen Medical Care Communicaions Enerainmen & Educaion Housing

5 Cambridge Journal of China Sudies 21 Noe: The weighs of all kinds of variables in he CPI may be slighly adjused annually, bu he esimae of weigh coefficiens in he able can broadly reflec he average level of all kinds of indices weighs in he CPI. The same in Table 2 and Table 3. Afer December 2002, a susained rising rend of CPI showed in China, from 0.4% in January 2003 o 5.2% in Sepember A ha ime, four of eigh caegories of consumpion price indices (food, Clohing, educaion and housing) rose in varying degrees, of which he Food CPI showed he maximum growh rae, rising from -0.1 % o 14.6% from November2002 o July From a pracical perspecive, China enered he fourh consecuive year of grain reducion in 2003.and he grain price increased due o he shorage of supply. This growh leads he direcion of he rend of CPI o a posiive way. In. 2004, he prices rise of grain and financial subsidies lead he price of Grain o increase obviously. In he case of he ever-increasing grain supply, grain price fell gradually. As a resul, he Food CPI sared o fall afer i reached 14.6%. The index fell down o 4% in January Influenced by he change of he Food CPI, oal CPI began o fall when i reached 5.3% in July The index fell down o 1.9% in January Figure 2 confirmed he absolue influence on he oal CPI caused by he Food CPI of which he conribuion raio flucuaed around 100% and even reached up o 295% in February CPI showed he increase rend again from Ocober 2006, which rose up o 8.7% from 1.4% in February Clohing CPI rose up o 2.4% from 0.8%, Family Equipmen & Service CPI rose up o 2.1% from 1.7%, Medical Care CPI rose up o 3.2% from 1.3%, Housing CPI rose up o 6.6% from 4.6%, and Food CPI rose enormously up o 23.3% from 2.2%. From a pracical perspecive, his rise of Food CPI is mainly due o he increase of pork and mea prices. Compared wih 2006, he pork producion of China in 2007 declined by 7.8%, which caused a shorage of pork supply. As a resul, he price of pork rose up and pushed Food CPI o increase. Figure 2 confirmed ha, from Ocober 2006 o February 2008, he conribuion raio of Food CPI sill mainained a high level of 80%. From February o Ocober in 2009, he negaive growh of CPI was mainly caused by he decrease in he Housing CPI. The conribuion raio of Housing CPI was abou 63%(according o equaion (2)). During his period, Housing price fell down o -3.8% from -2.9% and he oher prices also fell down in varying degrees. The Food CPI increased slighly of which he conribuion raio ended o zero. Therefore, he increasing srucural price leads o he rising of CPI. In addiion, he rising CPI due o cos-push price increase raher han demand-driven price increase. The direc cause of

6 Volume 9, No he significan growh rend of food prices was he decrease of supply caused by cos-push. Acually, he seasonal reason leads o shorage supply. Following he sharp rise in food prices, marke supply will gradually increase. A he same ime, he prices of he producion will gradually increase o a sabilized level / / / / / / / / / / / / / / / / / / / /09 food CPI Housing Cpi Figure 1 he conribuion raio of Food CPI and Housing CPI 2. Index Analysis of PPI Consiuion 1) PPI-means of subsisence As he same as he analysis of CPI, firsly, his secion measures each sub - index weighs of means of subsisence by adoping he regression mehod. The nex sep is o calculae he conribuion raio of all kinds of indices on he means of subsisence. The price of he means of subsisence rose sharply hree imes in he confidence inerval, which was mainly due o he rise of Food PPI: (1) The firs rise was in he period from February 2002 o Ocober 2004, Subsisence PPI rose from -2.9% o 1.7%. A ha ime, he price of food, clohing and daily necessiies increased in varying degrees. The Food PPI showed he maximum growh rae, which increased from -0.4% o 5.6%; Clohing PPI and Daily Necessiies PPI increased by 3.3% and 5.6% respecively. However, he price of consumer durables had been in a negaive growh. Thus, in his period, he increase of he PPI was he resul of he general rise in he prices of all kinds of means of subsisence of which he food price increased a he highes rae. (2) The second rise sared from Augus 2006, he Subsisence PPI rose up o 3.7% from 0.3% in he period from Augus 2006 o November However, his growh was mainly driven by he rise of food price. Food PPI raised by 9%, reached up o 9.1% from 0.1% during July 2006 o November 2007; Daily Necessiies PPI raised by 2.3%; Clohing PPI basically mainained sably a 1.2%; he increase range of

7 Cambridge Journal of China Sudies 23 Durable PPI shrank bu sill mainained a negaive growh. According o equaion (2) from Augus 2006 o November 2007, he conribuion raio of Food PPI on means of Subsisence PPI reached up o 88%, which confirmed ha he price of means of subsisence was mainly driven by he price of food. (3) The hird price rise of means of subsisence was from January 2010 o Sepember The conribuion raios of food, clohing, housing and durable consumpion were 69.1%,23.6%,16.7% and -10.2% respecively. Therefore, food was sill he main cause of he price growh of means of subsisence. Table 2 Esimaed Weigh Values of he Indices in he means of Subsisence PPI (and he Conribuion Raio of he Variables o he means of Subsisence PPI)( ) Variables Weigh Coefficien Weigh Saisic Mean Value of he Conribuion Raio Food Clohing Daily Necessaries Durable Consumpion ) PPI-producion secion The conribuion raios in Table 3 and he endency of he raios in Figure 3 shows ha, in he confidence inerval, he rise of Producion PPI was mainly driven by he rise of he price of raw maerials. Since he economic growh of China resumed he endency of increase, he Producion PPI rose hree imes: (1) The firs rise period sared from January 2002 o November 2004, he index raised gradually up o 6.7% from -5% in March Afer a sligh decline, he index resumed o increase o 11.7% in November 2004; he rend of Mining PPI and Raw Maerials PPI were similar o he rend of he Producion PPI, which increased by 3.3% and 5.6% respecively; he Processing Indusry mainained a rise endency hroughou he period, he Processing PPI rose up o 4.6% from -3.7%. According o equaion (2), in his period, he proporion of he conribuion raio of raw maerials, mining and processing on he rise of Producion PPI were 55%, 19% and 30%. From a pracical perspecive, he invesmen expansion in China caused ensions on domesic resources supply since According o saisics, in 2003, he consumpion of crude oil, coal, iron ore, seel, aluminum and cemen in China separaely accouned for 7.4 %, 31%, 30%, 27%, 25% and 40 % of he whole

8 Volume 9, No consumpion of he world. As a resul, he price of energy, raw maerials and some producion prices rose rapidly. Afer November 2004, he rise of he price of mining and raw maerials producion declined sharply, he rise of he Processing price was relaively sable. The rise of means of producion also coninued o fall down sharply and reached down o 1.84% in Sepember (2) In he following period, he price increased again and reached up o 10.83% in Sepember From Sepember 2007 o Sepember 2008, he oal conribuion raio of he rise of he price of mining and raw maerials on he rise of he price of means of producion is 69% and he conribuion raio of processing indusry was abou 30%. (3) The hird rise occurred from Augus 2009(-10.10%) o May 2010(8.79%). This rise was also pushed by mining indusry, raw maerials indusry and processing indusry and he oal conribuion raio of hese hree indusries was 80%. Table 2 Esimaed Weigh Values of he Indices in he Subsisence PPI (and he Conribuion Raio of he Variables o he Subsisence PPI)( ) Variables Weigh Coefficien Weigh Saisic Mean Value of he Conribuion Raio Mining Indusry Raw Maerials Indusry Processing Indusry / / / / / / / / / / / / / / / / / / / /09 Mining Raw Maerials Processing Figure 2 he conribuion raio of mining, raw maerials and processing indusry on means of producion

9 Cambridge Journal of China Sudies STUDY OF THE TRANSMISSION RELATIONSHIP BETWEEN PPI AND CPI Afer 2000, he price ransmission relaionship in China shows many new characerisics:he flucuaions of upsream produc price are far greaer han he flucuaions in downsream produc price.; he flucuaions in upsream producs canno be ransmied o downsream promply and he change in price of means of producion canno be ransmied o he household consumpion price. Therefore, many scholars applied he Granger causaliy es o conduc an empirical sudy of he ransmission relaionship beween PPI and CPI (Liping He, e al, 2008; Weikang Xu, e al, 2010). The Granger causaliy es is proposed by Granger (1969) o sudy he causal relaionship beween variables, which has been widely used in research as well as in he indusry. This mehod gives a qualiaive conclusion on a se significance level: for insance, in his hesis, he PPI Granger-causes he CPI or he PPI does no Granger-cause he CPI. However, wheher he ransmission relaionship beween he CPI and he PPI of China changes during esing or no? If here are any relaionship changes, wha is he incenive leading o he change F? To answer he quesions above, on he basis of he dynamic Granger es, his hesis sudies he ransmission relaionship beween he PPI and he CPI by consrucing a Granger-F es. 1. Saionariy Tes of Daa The Granger causaliy es requires ha he ime series of PPI and CPI should be saionary, bu according o he uni roo es, i is found ha he PPI from January2001 o December 2013 is saionary while he CPI is no (as in Table 4). For his reason, in his hesis, he HP mehod is used o obain he gap value of PPI and CPI, denoed as PPI gap and CPI gap wih smoohing parameer of HP being 14400; i has been verified ha boh PPI gap and CPI gap are saionary ime series. Hereinafer, for convenience, PPI and CPI denoe he gap value forms of heir respecive ime series. Table 4 Saionariy Tes of Time Series (January2001 o December2013) Variable Tes form (c,,k) ADF saisics P value Resul CPI (c,0,2) Non-saionary PPI (c,0,2) Saionary CPI gap (0,0,3) Saionary PPI gap (c,0,4) Saionary

10 Volume 9, No Descripion: (1) c and in he es form denoe he consan and rend erm, and k denoes he lag order (2) he choice of k is based on he minimum of AIC and SC. 2. PPI And CPI: Who Drives Who? The definiion of Granger causaliy is: if he lag value of a variable can help predic anoher variable, hen he variable Granger-causes he oher variable. The formula can be expressed as: CPI a PPI a a a a CPI 12 1 a PPI 22 If and only if he coefficiens 1 1 a... a p a 12 p 11 p 21 p a CPI 12 p a PPI 22 p p 1 2 (3) in he coefficien marices are all zero, variable PPI does no Granger-cause CPI which is equivalen o ha CPI is exogenous o PPI. The null hypohesis for his es H is: all 0 p a 12 are 0, i.e. PPI does no Granger-cause CPI. The way o conduc a recursive Granger es is: suppose he confidence inerval is from appropriae sub-inerval from o 0 o ; a firs, selec an 0 T (n<t) and conduc a Granger es; and hen, as he n number of new values observed increases, expand he sample inerval and repea he Granger es unil reaching he end of he sample inerval ( T ). For he convenience of descripion, his hesis will give he name GF o he saisics measuring he causaliy beween PPI and CPI, referencing Fei (2012) mehod, wih GF saisics= disribuion. S F c, in which RSS is he residual sum of square of he equaion 1, he residual sum of square of he lagged variable excluding final ime value of he confidence inerval. F c, ( RSS0, RSS1, )/ p S RSS /( 2p 1) 1 CPI in equaliy (3); T follows F RSS is 0, PPI ; P is he lag period; is he T is he criical value of F disribuion a ime (10% level of significance). Therefore, if S > F (GF c, 1), PPI Granger-causes CPI ; oherwise, PPI does no Granger-cause CPI. On he basis of dynamic recursive Granger es, he causes of he flucuaions of GF saisics can be sudied hrough he consruc of a Granger-F es, i.e. wha are he facors causing he forward or backward ransmission beween PPI and CPI. Figure 4 depics he rend char of GF saisics based on he recursive Granger es. The black doed line represens he null hypohesis ha PPI does no Granger-cause CPI, and when he line is above "1", i indicaes ha he null hypohesis is rejeced, i.e. PPI Granger-cause CPI. The black solid line represens he null hypohesis ha CPI does no Granger-cause PPI, and when he line is above "1", i indicaes ha he null hypohesis is rejeced, i.e. CPI

11 Cambridge Journal of China Sudies 27 Granger-cause PPI. Moreover, when GF saisics increases, i indicaes ha he saionariy of he Granger causaliy beween CPI and PPI increases lag lag 2002M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M07 CPI cause PPI PPI cause CPI 10% criical value CPI cause PPI PPI cause CPI 10% criical value lag lag 2002M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M M07 CPI cause PPI PPI cause CPI 10% criical value CPI cause PPI PPI cause CPI 10% criical value Figure 3 GF Saisics Trend Char As can be seen from he GF saisics in Figure 4, afer 2000, he price ransmission of China has been hrough 3 sages: bidirecional ransmission beween PPI and CPI during he early sage of he confidence inerval; no ransmission relaionships over he inerim of he confidence inerval; bidirecional ransmission again during he final sage. The rend char shows wo disinc characerisics: (1) he saionariy of he Granger causaliy for he backward ransmission from CPI o PPI is significanly higher han ha of he forward ransmission (GF saisics of backward ransmission is higher han ha of he forward ransmission); (2) he forward ransmission apparenly lags behind he backward ransmission. 3. Empirical Analysis From a heoreical sandpoin, here are wo main forms of price ransmission, namely cos-push and demand-pull. Cos-push ransmission is ha he changes in upsream price cause price changes in midsream and downsream producs. I can also be explained from he

12 Volume 9, No perspecive of price index as a variaion in PPI, a resul of he price changes in purchasing price indices of raw maerials, fuels and power for indusrial enerprises, which leads o a variaion in CPI. Conversely, demand-pull ransmission is ha he changes in downsream price give rise o changes in midsream price, and hen cause changes in upsream price level. I is he variaions of CPI ha cause he variaions of PPI, and lead o relevan changes in he purchasing price of raw maerials, fuels and power. However, wheher he ransmissions beween PPI and CPI can be achieved or no is depend on he supply-demand relaionship beween upsream and downsream produc markes. The upsream producs mainly includes he indusries of excavaion, producion and machining of raw maerials, fuels and power, while he downsream producs mainly includes consumer daa indusry. Under he condiion of shorage economy, he rising coss lead direcly o he increase of commodiy prices, which proved by he hisory of he high inflaion period in China in he 1980s and 1990s. Since he lae 1990s, due o he oversupply in he marke of downsream producs in China, a huge producion and processing capaciy was formed in he downsream of he indusrial chain, hus resuled in a relaively insufficien final consumpion demand. The increase of invesmen demand drives up he price level when he supply of upsream producs has limied elasiciy, while he producion capaciy of downsream producs has a large number of relaive surpluses a he same ime, hus he rise in price of consumer erminal would lead o a loss of he corresponding marke share. As a resul, he price increase of upsream raw maerials canno ransmi o downsream marke smoohly, namely, here is a lack of necessary marke environmen o achieve he ransmissions beween PPI and CPI. By conras, he over-demand of upsream markes leads o an inense compeiion among he buyers. Therefore, people will observe a price soar in upsream marke as a resul of a surge in demand caused by he expansion of invesmen. Meanwhile, he serious decline in supply of some food (e.g. grain and pork) disequilibraes he balance beween supply and demand, causing a sharply jump in he price of hese food. A combinaions of hese facors discussed above conribue o a rapid growh in CPI, and resul in a sharply rising in PPI evenually. Tha is, compared o forward ransmission, CPI can backward ransmi o PPI in a relaively smooher way. Hence, he demand-supply relaionship of upsream and downsream markes causes he wo obvious characerisics of he rend of GF saisics. From 2002 o 2003, PPI and CPI ransmi muually. The ransmission from PPI o CPI is mainly because of he srained supply of domesic resource caused by invesmen expansion in China afer 2000, which leads o a rapid increase in he price of upsream producs, such as energy, raw maerials and some of he means of producion, and hen causes rises in price of

13 Cambridge Journal of China Sudies 29 household consumpion and service secor. Afer November 2004, he price level of means of producion in PPI coninued o fall sharply, which is due o he coninuous falling of he year-on-year growh rae of excavaion and raw maerial indusrial producs, hus have los is driving force of pulling up he level of CPI. Therefore, he GF saisics of lag 1, lag 2, lag3 and lag 4 all saw rapid declined by he end of 2004, and hen flucuaed around he zero value. From he Granger causaliy F es in Table 5, i can be seen ha he null hypohesis ha he purchasing price of raw maerials, from 2002 and 2003, fuels and power indices did no Granger-causes GF saisics is rejeced, which proves ha he rise in price of he former Granger-causes he ransmission from PPI o CPI over his ime period. From July 2002 o January 2004, he main reason of observing a ransmission from CPI o PPI is ha he reducion in food producion causes he increase of food price. Is ransmission pahway is ha he increase in food prices drives up he level of CPI, hen ransmis o he means of producion in PPI, and evenually leads he means of producion and he overall level of PPI o increase. From he Granger causaliy F es in Table 6, i can be seen ha in he year of 2002 and 2003, he null hypohesis ha he food CPI flucuaions did no Granger-cause GF saisics is rejeced, which proves ha he increase in food price Granger-causes he backward ransmission from CPI o PPI over his ime period. From he GF saisics in Table 4, i can be observed ha afer experienced an unrelaed period beween 2004 and 2007, PPI and CPI sared o achieve a backward ransmission in he second half of 2007, and hen ransmied muually by he end of The main reason of backward ransmission is ha he supply of pork and mea produc in food CPI has experienced a severe decline beween he year of 2007 and 2008, which disequilibraed he original supply-demand balance, and led o a rapid growh in he price of hese food, and hen caused he ransmission from CPI o PPI. Wih he spreading and deepening of he world financial crisis in he second half of 2008, he food price dropped back and ended owards sabiliy, which caused a decline in CPI and dragged down he level of PPI. I can be seen from he GF saisics rend diagram ha differen lagged values of GF saisics (backward ransmissions from CPI o PPI) all reached he levels above 1 in succession from he las half of 2007 o he year of 2008, during which he food caegory of CPI was experiencing a high-speed growh. I can be concluded ha PPI and CPI achieved backward ransmission over his ime period. From he Granger Causaliy F es in Table 6, i can be seen ha he food caegory of CPI Granger-causes he GF saisics over his period, which proves ha he shor erm flucuaions of food price is a significan cause of he backward ransmission of PPI and CPI. In addiion,

14 Volume 9, No Table 5 Granger - F es proves ha he ransmission beween PPI and CPI is caused by he coninuous increase of he price of food in means of subsisence in PPI. Table 5 Granger-causaliy F es: Who caused he forward ransmission from PPI o CPI Tes period Null Hypohesis F saisics P-value Conclusion The purchasing price of raw maerials, fuels and power did no Granger-cause GF saisics Food PPI did no Granger-cause GF saisics The purchasing price of raw maerials, fuels and power did no Granger-cause GF saisics Food PPI did no Granger-cause GF saisics Rejec Accep Rejec Rejec Noe: The bes lagged value was calculaed according o AIC and SC principle; GF saisics describes he dynamic causaliy of he ransmission from PPI o CPI. The purchasing price index of raw maerials, fuels and power and he food caegory CPI and GF are all non-saionary process over he es period discussed above. Therefore HP mehod is conduced o obain he gap value of he daa lised above. Afer examining he purchasing price index of raw maerials, fuels and power,he gap values of CPI and GF saisics are all saionary process, herefore Granger Causaliy F es can be conduced. Table 6 Granger-causaliy F es: Who caused he backward ransmission from CPI o PPI Tes period Null Hypohesis F saisics P-value Conclusion Food CPI did no Granger-cause GF saisics Rejec

15 Food CPI did no Granger-cause GF saisics Cambridge Journal of China Sudies Rejec Noe: The bes lagged value was calculaed according o AIC and SC principle; GF saisics describes he dynamic causaliy of he ransmission from PPI o CPI. The daa lised above are all non-saionary process over he es period. Therefore he Granger - F es is conduced afer using he same mehod in Table 5 o obain he gap values of relevan daa. IV. GRANGER CAUSALITY TEST BETWEEN M2 AND SUB-INDICES OF CPI This secion sudies wheher he changes of he moneary policy iself (eg. The growh rae of money supply) Granger-causes he consiuions of CPI and PPI. Given ha he conrol of he amoun of money in circulaion is sill he main policy insrumen and inermediae arge of People s Bank of China, we use he growh rae of M2 o represen he changes of moneary policy. Table 7 demonsraes he oucome of he Granger Causaliy es. From he P-value lised below i can be seen ha he money supply Granger-caused no only he overall level of CPI, bu also he CPI level of food, obacco and alcohol, family equipmen, enerainmen and educaion and housing. From he respec of PPI index, here was no enough evidence o prove ha money supply Granger-caused he overall PPI index and he index of means of producion; however, he money supply, under a significan level of 10%, Granger-caused he PPI index of he means of subsisence. This explained ha he changes of money supply could provide relaively sufficien predicive informaion for he flucuaions of he overall level of CPI and is sub-indices, bu was no sufficien enough o predic he flucuaions of he overall level of PPI and he level of means of producions. Table 7 Wheher M2 caused he changes of price indices Growh Rae of M2 did no Granger-cause he variables lised below F saisics P-value Conclusion CPI Rejec Food Rejec Tobacco & Alcohol Rejec Clohing Accep Family Equipmen Rejec

16 Volume 9, No Medical Care Rejec Communicaion Accep Enerainmen & Educaion Rejec Housing Rejec PPI Accep Means of Producion Accep Mining Indusry Accep Raw Maerial Indusry Accep Processing Indusry Accep Means of Subsisence Rejec Food Accep Clohing Rejec Daily Necessaries Rejec Durable Consumpion Accep Noe: The bes lagged value was calculaed according o AIC and SC principle; The growh rae of M2 and every price index lised above are all non-saionary process over he es period. Therefore HP mehod is conduced o obain he gap value of he daa lised above. The gap values are all saionary process afer examinaion. Therefore Granger Causaliy F es can be conduced. V. CONCLUSIONS This hesis conducs empirical sudies on he flucuaion characerisics, he influenial facors and he ransmission relaionship CPI and PPI, The main mehods and conclusions are as follows: 1. By he ime-varying analysis of conribuion raio of he componens wihin CPI, i is found ha wihin he confidence inerval (January2001-Augus2013), rise in CPI was mainly srucural price rise caused by he change in food prices raher han overall price rise. In addiion, he rise in CPI is he growh of cos-push price rise raher han demand-pull price. The former leads o reducion of supply, hen food prices increase dramaically. This shorage

17 Cambridge Journal of China Sudies 33 supply is seasonal raher han regular. As he food prices drasically increase, marke supply gradually increases, and he corresponding prices will gradually sable. 2. By he ime-varying analysis of conribuion raio of he componens wihin PPI, i is found ha wihin he confidence inerval, he price of means of subsisence of PPI shows several sharp increases, mainly because i is pulled by he rise of food price index; while he increase in price of means of producion of PPI is mainly because of he pull by he rise in prices of raw maerial indusrial producs. 3. This hesis applies a recursive Granger causaliy es o sudy he dynamic ransmission relaionship beween PPI and CPI. The dynamic research mehod no only differeniaes if PPI and CPI achieve forward, backward or bidirecional ransmission a differen imes, bu also esimae he saionariy of boh when ransmiing a differen imes. According o he flucuaions of GF saisics: (1) Afer 2000, PPI and CPI are observed o have hree sages: a bidirecional ransmission a he iniial sage of he es; no ransmission a he inerim sage and bidirecional ransmission a he las sage. (2) GF saisics show wo disinc characerisics: one is ha he saionariy of he Granger causaliy for he backward ransmission from CPI o PPI is significanly higher han ha of he forward ransmission (GF saisics of backward ransmission is higher han ha of he forward ransmission); he oher is ha he forward ransmission apparenly lags behind he backward ransmission. 4. This hesis sudies he causes of he change of ransmission relaionship beween PPI and CPI by conducing a Granger-F es. The sudy confirms ha: (1) he wo backward ransmissions of CPI and PPI in he confidence inerval are boh caused by he flucuaions of food price; (2) he forward ransmission beween PPI and CPI from 2002 o 2003 is caused by he flucuaions of he purchasing price of raw maerials, fuels and power; while he forward ransmission beween PPI and CPI afer 2007 is caused by he flucuaions in food prices of means subsisence of PPI. 5. Afer he sudy of Granger Causaliy es, we found ha he flucuaions of he money supply (year-on-year growh rae of M2) can induce changes of he overall level of CPI and majoriy of he sub-indices of CPI, bu does no have significan impac on he level of PPI and he index of means of producion. Therefore, i is inappropriae o use moneary policy o manage he flucuaions of he PPI level of means of producion.

18 Volume 9, No REFERENCES [1] Chen Jianqi (2008): Inflaion Regulaion and he Reserve Relaionship beween CPI and PPI, China Indusrial Economics, 11, (China) [2] Dong, Zhiqing, Cai, Yucheng, and Xie Jiazhen ((2009): The Cycle Co-flucuaion Effec of CPI and PPI, China Indusrial Economics, 10, (China) [3] Fei Zhaoqi (2012): Sock Marke Co-movemen: Economic Fundamenals and Conagion, Chinese Review of Financial Sudies", 5, (China) [4] He Liping, Fan Gang, and Hu Jiani (2008): CPI v. s. PPI: Which Drives Which?, Economic Research Journal, 11, (China). [5] Song Jinqi and Shu Xiaohui (2008): The Relaionship beween PPI and CPI, Price: Theory & Pracice", 10, (China) [6] Xu Weikang (2010): A Query o he Paper CPI v.s. PPI: Which Drives Which, Economic Research Journal, 5, (China). [7] Zhang Yanque (2007): Are Commodiy Prices he Forward Indicaors of he Consumer Price in China?, The Journal of Quaniaive & Technical Economics, 12, (China) [8] Colclough, W. and M. Lange (1982): Empirical Evidence of Causaliy from Consumer o Whole Sale Prices, Journal of Economirics, 19(2), [9] Granger.C. W. J. (1969): Invesing Causal Relaions by Economeric Models and Cross-Specral Mehods, Economerica, 37, [10] Kyrsou, C. and W. Labys (2006): Evidence for chaoic dependence beween US inflaion and commodiy prices, Journal of Macroeconomics, 28, [11] Silver,J.and T.Wallace (1980): The Lag Relaionship beween Wholesale and Consumer Pricey -An applicaion of he Haanaka-Wallace Procedure, Journal of Economerics, 12,