HOW RELATIVE PRICE VARIABILITY IS RELATED TO UNANTICIPATED INFLATION AND REAL INCOME?

Transcription

1 45 Paksan Economc and Socal Revew Volume 5, No. 1 (Summer 014), pp HOW RELATIVE PRICE VARIABILITY IS RELATED TO UNANTICIPATED INFLATION AND REAL INCOME? SAGHIR PERVAIZ GHAURI, ABDUL QAYYUM and MUHAMMAD FAROOQ ARBY* Absrac. The paper examnes he behavor of prce seng agens as refleced n relave prce changes n response o demand and supply facors. We have followed Parks (1978) model on monhly daa of consumer prce ndex of Paksan. Resuls show ha relave prce varably s deermned manly by supply facors, as refleced by unancpaed nflaon. The demand facors (.e., changes n real ncome) are found nsgnfcan n deermnng relave prce varably. Keywords: Ancpaed nflaon, Unancpaed nflaon, Relave prce varably, Macroeconomc framework JEL classfcaon: E10, E31, E64 I. INTRODUCTION The relave prce varably (RPV) and s relaonshp wh nflaon, ncome, moneary expanson, ec. have go consderable aenon n economc research, boh heorecal and emprcal, as he subjec has mporan lessons for welfare cos of nflaon and neuraly of moneary *The auhors are, respecvely, Jon Drecor a Sae Bank of Paksan, Karach; Jon Drecor a Paksan Insue of Developmen Economcs, Islamabad; and Senor Economs a Sae Bank of Paksan, Karach (Paksan). Correspondng auhor e-mal: saghr.pervaz@sbp.org.pk Noe: Ths paper s a par of Ph.D. dsseraon beng wren by Saghr Pervaz Ghaur.

2 46 Paksan Economc and Socal Revew polcy. Whle heorecal models, lke menu-cos and asymmerc nformaon, predc a posve assocaon beween nflaon and RPV, he emprcal research has found a range of resuls, ncludng nsgnfcan, posve sgnfcan, lnear and non-lnear relaonshps beween he wo. Cukerman (1983) has presened a comprehensve analycal survey of he subjec. A ypcal explanaon o suppor he relaonshp beween RPV and nflaon s ha ndusres and secors may dffer n her speed of adjusmen o nomnal shocks. As fxed coss are assocaed wh changng prces, ndvdual commody prces change only a dscree nervals, whch creae dvergence n relave prces. A second approach s based on dfferences n shor-run supply elasces across ndusres. If he shor-run supply elascy n one ndusry s smaller han ha n anoher bu he long-run elasces are smlar, hen a demand shf beween he ndusres may resul n a change n boh aggregae and relave prces even f he shf s ulmaely neural n s relave-prce effec. Anoher possble explanaon for he relaonshp s he hypohess of mperfec nformaon, accordng o whch: as ndvdual frms lack full nformaon regardng general prce level, hey may confuse changes n he general prce level wh relave prce changes. 1 Apar from he naure of relaonshp beween RPV and nflaon, he emprcal sudes also explore whch componen of nflaon s more relevan for explanng RPV,.e. expeced nflaon, unexpeced nflaon, or nflaon uncerany. Parks (1978), for example, fnds a posve relaonshp beween RPV and unexpeced nflaon, whle Lach and Tsddon (199) fnd ha expeced nflaon has a sronger effec on prce varably han unexpeced nflaon. Grer and Perry (1996), on he oher hand, show ha only ex ane nflaon uncerany ncreases relave prce varably. A prelmnary examnaon of prce daa n Paksan also shows a lnk beween he nflaon and RPV. As shown n Fgure 1, overall CPI nflaon n a gven monh and range of ndvdual commody prce changes move ogeher, n general, over me. For a formal enqury of he subjec on Paksan daa, we have followed Parks (1978) who has nroduced a mehodology o explore he mpac of unexpeced nflaon and real ncome on relave prce changes. As a frs sep, he proposed a measure of changes n relave prces whch was found o move n lne wh he general nflaonary rends. And hen, he esmaed a se of regresson equaons o 1 See Caraballo and Dabus (008) for an accoun of varous heorecal explanaons on he relaonshp and a revew of emprcal sudes.

3 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income 47 esmae he quanave relaonshp of relave prce varably wh demand and supply facors, whch he proposed on he bass of a sandard macroeconomc framework. He used hs framework o analyze he movemens n consumer prces n Uned Saes durng and Neherlands durng FIGURE 1 Range of Prce Changes of Indvdual Commodes and Overall Inflaon In he nex secon of he presen sudy, a bref revew of some sudes havng smlary wh Parks work for dfferen counres and n dfferen me perods s gven. Afer ha, he mehodology used n hs sudy s presened followed by he resuls. The las secon concludes he paper. II. LITERATURE REVIEW Parks sudy promped he debae on relaonshp beween relave prce changes and nflaon, hough a number of auhors addressed hs ssue even before Parks sudy n For example, Glejser (1965) found nflaon an mporan deermnan of relave prce changes (measured by weghed sandard devaon) for 15 OECD counres durng A smlar sudy was done by Okun (1971) ha compared 17 OECD counres for he perod and found a posve assocan beween a counry s average nflaon and sandard devaon of GDP deflaor. Vnng and Elwerosk

4 48 Paksan Economc and Socal Revew (1976) concluded ha varance of relave prce changes was assocaed wh general nflaon varably (durng ) n US. They found such assocaon for boh wholesale and consumer prce ndces and calculaed varance of relave prce changes by akng every sub-ndex of he man seres and calculang a varance for each pon n me. However, her sudy was crczed for weaknesses n measures of boh general prce nflaon and relave prce varably. Movaed by Parks sudy, Ashley (1981) used Granger causaly ess o conclude ha flucuaons n he nflaon help cause flucuaons n relave prces, bu no vce-versa. Cukerman and Wachel (198) used a framework ha was also bul on Parks work excep ha hey allowed nflaon expecaons o vary across markes. Snce equlbrum prces (and her raes of change) n dfferen markes may dffer, nflaon expecaons across markes may also vary. They showed ha here mgh be a posve relaonshp beween relave prce varably and he varance of nflaon expecaons. They also showed ha changes n he varance of eher aggregae demand or supply shocks would cause ncreased relave prce varably. Fscher (198) suded he naure and drecon of causaly beween relave prce varably and nflaon for Germany and US economes n order o undersand he underlyng process behnd nflaon and s socal welfare cos. He found, whle he lnk beween unancpaed nflaon and relave prce varably was srong n boh he economes, was weak n case of ancpaed componen of nflaon and relave prce varably. By esmang a small VAR model, he also concluded ha he relaonshp beween relave varably and nflaon semmed manly from polcy responses o supply shocks. Blejer (1983) examned he experence of Argenna havng hgh nflaon coupled wh rade lberalzaon polces; and suded he response of relave commody prces o nflaonary pressures. He underook a dealed analyss of monhly prces of 61 compones of consumer prce ndex beween 1977 and 1981 and concluded ha ndvdual commody prces had flucuaed over a much wder range han he overall CPI possbley mplyng menu-cos of nflaon. He also noed a clear upward rend n he relave prce of servces, wh a consequen reducon n he relave prce of goods and food producs. Among he facors affecng relave prce varably, hs sudy found ha only unxpeced componens of nflaon and moneary growh had sgnfcan mpacs, whle he expeced pars of hese varables had nsgnfcan coeffcens.

5 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income 49 Lach and Tsddon (199) analyzed he effecs of nflaon on he dsperson of food prces n Isreal by usng dsaggregaed daa of Conrary o oher sudes, hey found ha he effec of expeced nflaon on nramarke prce varably was sronger han ha of unexpeced nflaon. A smlar resul was obaned by Loy and Wеavеr (1998) n case of Russan food markes who showed ha was ancpaed nflaon whch nduced dsorons n rеlavе prcеs nsead of unancpaеd nflaon or nflaon uncеrany. Chang and Cheng (000) examned a dsaggregaed daa se of US prces n pos-war perod o explore he lnk beween nflaon and relave prce varably. As a frs sep, hey used a model o esmae nflaon varably condonal on pas unexpeced nflaon and pas nflaon varably. Then hey relaed o relave prce varably; and found a posve relaonshp of RPV wh boh he nflaon rae and nflaon varably. They also concluded ha he relaon remaned robus o ol-prce shocks. Ukoha (007) esmaed hs relaonshp n case of Ngera wh a focused sudy of relave prce volaly of agrculure commodes durng 1970 o 003. He also found a posve sgnfcan mpac of overall nflaon on relave prce varably of agrculural commodes boh n he shor run and he long run. On he bass of hs resuls, he also suggesed polces o preven agrculure secor from adverse mplcaons of nflaon. Recenly, Cho (010) presened new heorecal and emprcal nsghs relang o he relaonshp beween nflaon and RPV on he bass of dsaggregaed CPI daa for US and Japan. He found a non-lnear and U-shape relaonshp beween nflaon and RPV. However, he relaonshp was no sable over me; nsead vared que sgnfcanly wh he changes n nflaonary epsodes or moneary polcy regmes. Thus hs fndngs are agans he popular heorecal models of prce seng lke menu cos or mperfec nformaon models whch ypcally predc a posve assocaon beween nflaon and RPV. Noneheless, hs resul can be n lne wh an alernave heorecal explanaon,.e. Calvo scky prce model ha ncorporaes secoral heerogeney n prce rgdy. A smlar U-shape relaonshp beween RPV and nflaon was also found by Akmal (011) n case of Paksan. He also found ha hreshold level of nflaon n erms of RPV vared wh general nflaonary phases,.e., n perod of hgh nflaon, he hreshold nflaon s also hgh and vce versa. There s hardly any oher sudy on hs subjec n case of Paksan. We have re-examned hs relaonshp; however, hs sudy s dfferen from

6 50 Paksan Economc and Socal Revew Akmal (011) n hree aspecs: (a) we have used more dealed daa,.e., 9 compose ems of monhly consumer prce ndex compared wh 1 broad groups used n Akmal (011); (b) we have examned he mpac of unancpaed nflaon as well as overall nflaon on RPV; and (c) we have also suded he mpac of real ncome, as a demand facor, under macroeconomc framework as suggesed by Parks (1978). III. DATA AND METHODOLOGY The prce daa have been obaned from Paksan Bureau of Sascs (PBS), whch releases wo ses of prce daa,.e. (a) prces of 374 commodes n consumer baske ( base); and (b) 9 compose ndces whereby smlar commodes are grouped ogeher. PBS has publshed prces of 374 commodes by cy level, bu weghs of hese commodes are no n publc doman, whereas weghs of 9 compose commodes are publshed. Snce for our work, we need boh prce ndces as well as her weghs, so we have used 9 compose ndces for overall, food and non-food groups. The daa used n hs sudy s for he perod from July 001 o June 011. We have used LSM ndex as a proxy of real ncome. By defnng relave prce as a rao of ndvdual prce ndces of 9 compose commodes o he overall prce ndex (.e. p / P ), he rae of change n h commody s relave prce s worked ou as: Dp DP ; where p s he ndex of he h commody n me perod, P s overall consumer prce ndex, whch s a weghed average of ndvdual ndces (.e., P 9 1 w p ), w s weghs of ndvdual commodes n he CPI baske whch sum o one, and D represens he frs dfference of naural logarhm of he ndces. Whle he average of raes of change n relave prces s zero by defnon, we ake he varance of hese changes as a measure of he degree of relave prce varably, followng Parks (1978). I s calculaed as a weghed sum of he squared devaons of he ndvdual raes of prce change around he average, ha s: GDP or oher componens of GDP are no avalable n monhly frequency n Paksan. However, LSM has close proxmy as has srong backward lnkages wh agrculure secor and forward lnkages wh servces secor ncludng rade, ranspor and fnancal servces.

7 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income V w Dp DP (1) As argued by Parks, V (.e., varance of relave prce changes) s also a measure of non-proporonaly of he prce movemens as: (a) f all prces change by he same rae hen he varance of relave prce changes wll be zero; and (b) wll be hgher he more non-proporonal he prce changes are across commodes. We calculae boh he DP (overall nflaon) and V from he daa se of monhly ndces of 9 compose commodes. As he daa se covers a en-year perod (wh 10 observaons), he relaonshp beween he relave prce varably and overall nflaon accouns for epsodes of boh nflaon and deflaon. Parks suggess wo ypes of specfcaons o capure hs relaonshp, as gven below: V a b DP u () Where V a b DP (or DP b DP u DP ) represens posve (or negave) prce change. The second specfcaon allows us o dfferenae he degrees of response o nflaon (posve prce changes) and deflaon (negave prce changes). The above models presen prelmnary evdence of he exsence of ceran knd of assocaon beween relave prce varably and nflaon. Parks (1978) developed a rgorous model o undersand he naure of hs relaonshp, parcularly n he conex of un-ancpaed nflaon and ncome. For he convenence of readers, key feaures of hs framework whch we used o derve fnal equaon for he deermnans of relave prce varably, esmaed for Paksan, are presened here. Parks framework s bul upon a supply-demand apparaus wh focus on he effecs of prce expecaons. Le s he quany of he h commody suppled n perod, he supply funcon s as follows: p ln q a b ln * ct (4) P * Where P s ancpaed level of overall consumer prce ndex, and T represens rend varable. The supply funcon parameers are represened by a, b and c. For a posvely sloped supply funcon, b > 0. The demand funcon s specfed as below: (3)

8 5 Paksan Economc and Socal Revew ln q p f lnm d e ln (5) o Where m s nomnal ncome, e s he own-prce elascy, and f o he ncome elascy of demand. For smplcy, we gnore cross-prce elasces and assume he sum of ncome elascy and own prce elascy s zero o manan homogeney. For negavely sloped demand funcon, e < 0. By akng he frs dfferences of he logarhmc forms of he above supply and demand funcons and solvng for reduced form equaons under marke clearng assumpon, we oban he followng: Dp Dq k e e Dm b DP k * c b Dm b DP * 1 Where k. The above model predcs ha ancpaed overall ( b e ) nflaon affecs prce changes of ndvdual commodes posvely and quanes negavely, whle ncome has posve effec on boh prce and quany. However, he exen of effecs depends on he sze of supply and demand elasces. From equaon (6), we can oban he relave prce changes for ndvdual commodes by subracng DP (overall nflaon) from boh sdes of he equaon;.e. Dp DP k c * e Dm b DP c DP Wh a lle manpulaon, becomes: Dp Where g ( Dm DP ) unancpaed nflaon. DP k e g b n c (8a) s real ncome growh and n DP DP * (6) (7) (8) s By combnng equaons (1) and (8a), we can decompose he deermnans of relave prce varance no supply and demand parameers nvolvng real growh and unancpaed nflaon, as gven below: Dp DP w k e g b n c V w (9) The equaon (9) gves us he followng quadrac equaon (lnear n parameers) whch can be esmaed hrough ordnary leas square mehod: V a0 a1 g an a3gn a4g a5 n (10)

9 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income 53 The coeffcens of he above equaon (.e., a 0, a 1,, a, a 3, a 4, a 5 ) are dfferen combnaons of weghs of ndvdual commodes (w 1 ) and parameers of he model (e and b ). Whle we have dealed prce daa for compung varance of relave prces (V ) and real ncome as dscussed above, we need some operaonal defnon of unancpaed nflaon (n ) n order o esmae he above equaon. Parks uses a smple me seres model of he form DP DP 1 o ge a measure of unancpaed nflaon. We have developed a unvarae ARIMA model on acual prce level and used s fed values as a measure of ancpaed nflaon. 3 Dfference beween he acual and he fed values s unancpaed nflaon. We have aken care of me seres properes of he varables ncluded n he model and found hem approprae o be used n ordnary leas square regressons. IV. RESULTS Our prelmnary ess for a relaonshp beween relave prce varably and nflaon based on models () and (3) show ha here exss such assocaon as gven n Table 1. Wha surprsed Parks for Neherlands (Parks, 1978, p. 84) s also rue n case of Paksan,.e., he assocaon beween he wo varables s sronger n he case deflaon (prce declnes) han for nflaon (prce ncreases). I s neresng o noe ha prce decreases are less common n our sample only 17 nsances of negave prce changes ou of 119 observaons whch ndcaes downward prce rgdy. However, her mpac on relave prce varance s hgh. Possble explanaon of hs resul s: as mos prce decreases are relaed o food ems whch have srong seasonaly n prces, hey are unable o brng proporonal prce declnes n oher ems; as a resul, relave prce varance becomes hgher. 4 Ths explanaon s subsanaed by esmang he same equaons ( and 3) for food and non-food groups separaely. The mpac of prce deflaon on he varance reduces sharply (he coeffcen reduces from for overall baske o for food group), whch ndcaes seasonal prce declnes n food ems brng proporonal change n prces of he whole food group and hus have lesser mpac on he group s relave prce varance. In a sharp 3 We have seleced he parsmonous models (on he bass of AIC) as ARIMA (3,1,3) for overall prce ndex; (3,1,) for food prce ndex; and (4,1,4) for non-food prce ndex. 4 There were 30 nsances of negave prce changes n food ndex whle only 5 n non-food ndex.

10 54 Paksan Economc and Socal Revew conras o, he coeffcen of prce deflaon n case of non-food group ncreases sharply (from o ), whch ndcaes lackluser proporonal declnes n prces of such ems. Consan 0.00 (11.181) (DP)^.70 (3.35) TABLE 1 Regresson Resuls: Dependan Varable = VP (Equaons and 3) For Overall For Food Group For Non-food Group 0.00 (10.00) (DP+)^ (4.333) (DP )^ (5.11) 0.00 (10.06) (3.945) (9.36) (3.99) (4.794) (0.633) 6.61 (9.59) (0.395) 6.69 (9.346) (1.16) DW R Noe: Suden -values are n parenheses. On he oher hand, prce ncreases (as well as overall nflaon) have more profound mpac on relave prce varance n case of food group compared wh non-food. I mples ncrease n ceran food prces cause relavely less proporonal changes n prces of oher ems whle ncreases n non-food prces drve oher prces upward wh a hgher proporon. However, as menoned above, hese resuls do no help us fgurng ou he effecs of unancpaed nflaon and real varables n a sandard demandsupply framework. However, esmaon resuls of equaon (10) gve mporan nsghs n hs regard, as repored n Table. Ineresngly, he coeffcens of real ncome are sascally nsgnfcan for all he hree equaons,.e., for varance of relave prces of overall CPI baske, food group and non-food group. Noneheless, he unancpaed nflaon srongly affecs relave prce varaons. Incdenally, he naure of relaonshp beween RPV and unexpeced nflaon s he same U-shape as found by Akmal (011) n case of acual nflaon and RPV. In case of overall CPI, unancpaed nflaon, lower han 0.6 percen affecs RPV negavely whle

11 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income 55 ha hgher han affecs RPV posvely. Such urnng pons for food and non-food groups are 0.3 percen and 0.7 percen respecvely. TABLE Regresson Resuls; Dependan Varable = VP (Equaon 10) Consan (10.996) g ( 0.076) n (3.744) g n ( 0.865) g (0.319) n (.519) Overall Food Group Non-food Group (9.731) (0.05) (5.161) ( 0.373) ( 0.908) (3.30) (7.031) ( 0.678) (4.30) (0.58) (1.031) (8.818) DW R Noe: Suden -values are n parenheses. Moreover, he supply sde facors, represened by he nercep, also have sgnfcan posve effec on relave prce varably, hough he mpac s no very srong. V. CONCLUSION The sudy of relave prce varably has been he subjec of consderable neres for pas several decades as no only gves nsghs o prce seng mechansm n an economy bu also s consdered an ndcaor of supply shocks. Moreover, he naure of relaonshp beween nflaon and relave prce varably has useful polcy mplcaons. In case of Paksan, a sgnfcan assocaon beween overall nflaon and RPV s already documened by Akmal (011); however, we have exended he work o

12 56 Paksan Economc and Socal Revew examne he mpac of unancpaed nflaon and real ncome on RPV on he bass of a larger se of dsaggregaed prce daa. In hs paper, we have followed Parks (1978) mehodology whch has been developed over a macroeconomc framework. We have found ha changes n real ncome have nsgnfcan mpac on relave prce varably. The resuls make sense as changes n ncome (wh gven preferences) almos evenly affec demand for all consumer ems, whch may lead o relavely proporonal changes n her prces. I can be a case parcularly n a developng economy lke Paksan, havng a large nformal secor, where response of frms s less consraned by wage conracs; and where capacy ssues are less heerogeneous. On he oher hand, unancpaed nflaon, whch usually comes from em-specfc supply facors, may affec prces of dfferen ems unevenly before s fully ransmed o general nflaon. The resuls sugges a careful macroeconomc polcy for prce sably, as he mpac of demand managemen on relave prce varably s no found sgnfcan. However, he resuls do no exclude he possble nfluence of demand facors on rae of nflaon as such. Ths research can be exended furher o esmae he mpac of specfc supply facors, lke admnsered prces and exchange rae movemens, along wh demand facors on relave prce varably.

13 GHAURI e al.: Relave Prce Varably, Inflaon and Real Income 57 REFERENCES Akmal, Muhammad (011), Inflaon and relave prce varably. SBP Research Bullen, Volume 7, No., pp Ashley, Rchard (1981), Inflaon and he dsrbuon of prce changes across marke: A causal analyss. Economc Inqury, Volume 19, Issue 4, pp hp://dx.do.org/ /j b00344.x Blejer, M. I. (1983), On he anaomy of nflaon: The varably of relave commody prces n Argenna. Journal of Money, Cred and Bankng, Volume 15, No. 4, pp hp://dx.do.org/10.307/19916 Caraballo, María Angeles and Carlos Dabús (008), The deermnans of relave prce varably: Furher evdence from Argenna. Cuadernos de Economía, Volume 45, No. 13 (November), pp hp:// Chang, Erc C. and Joseph W. Cheng (000), Furher evdence on he varably of nflaon and relave prce varably. Economcs Leers, Volume 66(1), pp hp://dx.do.org/ /s (99)00186-x Cho, Ch-Young (010), Reconsderng he relaonshp beween nflaon and relave prce varably. Journal of Money Cred and Bankng, Volume 4(5), pp hp://dx.do.org/ /j x Cukerman, Alex (1983), Relave prce varably and nflaon: A survey and furher resuls. Carnege-Rocheser Conference Seres on Publc Polcy, Volume 19, pp hp://dx.do.org/ / (83) Cukerman, Alex and Paul Wachel (198), Relave prce varably and nonunform nflaonary expecaons. Journal of Polcal Economy, Volume 90, No. 1, pp hp:// Fscher, S. (198), Relave prces varably and nflaon n he Uned Saes and Germany. European Economc Revew, Volume 18(1), pp hp://dx.do.org/ / (8) Glejser, H. (1965), Inflaon, producvy, and relave prces A sascal sudy. The Revew of Economcs and Sascs, Volume 47(1), pp hp://dx.do.org/10.307/19415 Grer, Kevn B. and Mark J. Perry (1996), Inflaon, nflaon uncerany, and relave prce dsperson: Evdence from bvarae GARCH-M models. Journal of Moneary Economcs, Volume 38(), pp hp://dx.do.org/ /s (96) Lach, Saul and Danel Tsddon (199), The behavor of prces and nflaon: An emprcal analyss of dsaggregaed prce daa. Journal of Polcal Economy, Volume 100, No., pp hp://

14 58 Paksan Economc and Socal Revew Okun, Arhur M. (1971), The mrage of seady nflaon. Brookng Papers on Economc Acvy, No., pp Parks, Rchard W. (1978), Inflaon and relave prce varably. Journal of Polcal Economy, Volume 86, No. 1, pp hp:// Ukoha, O. Obas (007), Relave prce varably and nflaon: Evdence from he agrculural secor n Ngera. Afrcan Economc Research Consorum Research Paper 171. Vnng, Danel R. and Thomas E. Elwerowsk (1976), The relaonshp beween relave prces and he general prce level. Amercan Economc Revew, Volume 66, No. 4, pp hp://