Relative Agricultural Price Changes in Different Time Horizons. Guedae Cho (North Dakota State University)

Transcription

1 Relaive Agriculural Price Changes in Differen Time Horizons Guedae Cho (Norh Dakoa Sae Universiy) MinKyoung Kim (Norh Dakoa Sae Universiy) Won W. Koo (Norh Dakoa Sae Universiy) Paper prepared for presenaion a he American Agriculural Economics Associaion Annual Meeing, Monreal, Canada, July 27-30, 2003 Address of Corresponding Auhor: Guedae Cho Cener for Agriculural Policy and Trade Sudies Deparmen of Agribusiness and Applied Economics Norh Dakoa Sae Universiy 301F Morrill Hall Fargo, ND Phone #: Fax #: address: gcho@ndsuex.nodak.edu Copyrigh 2003 by Guedae Cho, MinKyoung Kim, and Won W. Koo. All righs reserved. Readers may make verbaim copies for his documen for non-commercial purpose by any means, provided ha his copyrigh noice appears on all such copies.

2 Relaive Agriculural Price Changes in Differen Time Horizons Absrac: Using a monhly daa covering from 1974:1 o 2002:12, his paper explores he linkage beween changes in macroeconomic variables (real exchange rae and inflaion rae) and changes in relaive agriculural prices in differen ime horizons (1, 12, 24, 36, 48, and 60 monhs). Conrolling for facors likely o deermine he long run rend of relaive agriculural prices, he resuls show ha long-erm changes in real exchange rae has had a significan negaive correlaion wih he long-erm changes in relaive agriculural prices. Conversely, changes of he general price have a role in explaining shor-erm changes in relaive agriculural price a bes. Keywords: Relaive agriculural price, exchange raes, inflaion raes

3 Relaive Agriculural Price Changes in Differen Time Horizons 1. Inroducion Changes in relaive commodiy prices in an economy occur coninuously in response o changes in many real facors of demand and supply of differen commodiies. Alhough his proposiion has received subsanial heoreical suppor, many economiss have ried o discern he poenial effec of macroeconomic facors on relaive commodiy prices. In heir raional expecaion model based on he assumpion of imperfec informaion, Lucas (1973) and Barro (1976) show ha unanicipaed inflaion can creae a misconcepion of absolue and relaive price changes, which implies ha he inflaion leads o a dispersion of prices among differen commodiies. A posiive relaionship beween he inflaion rae and relaive price dispersion is also discovered in he menu cos models, which assumes he exisence of a menu cos when changing prices (Shehinski and Weiss, 1977; Ball and Romer, 1989; Ball and Mankiw, 1995). In agriculural economics, many sudies (e.g., Frankel, 1986; Grennes and Labb, 1986; Roberson and Orden, 1990; Saghaian e al., 2002) have examined he poenial effec of moneary shock on changes in relaive agriculural prices. However, hey have concenraed on he issue of he speed of price adjusmen, while sudies examining he issue of he size of adjusmen are sparse. Moreover, because hese sudies have focused only on he effec of moneary shock on he relaive commodiy prices holding long-run

4 money neuraliy hypohesis 1, concenraion is limied on shor-erm changes of relaive agriculural prices. We insis in his paper ha, in an open economy, he U.S. dollar movemens have been imporan role in explaining he long-erm movemens of relaive agriculural price because he changes can cause differen degrees of supply shock on differen indusry secors. In fac, under he assumpion of well-working foreign exchange markes, he possibiliy of U.S. dollar movemens influencing relaive agriculural prices migh no be easily acceped because he macroeconomic shocks only cause a emporal overshooing problem of he nominal exchange rae (e.g., Dornbusch 1976). However, many empirical sudies abou foreign exchange markes sugges ha here is some degree of inefficiency in he markes (e.g., Frankel and Froo, 1987, 1990; Froo and Frankel, 1989; Io, 1990; Frankel and Rose, 1995). As a resul, here have been large and persisen deviaions of he nominal exchange rae from is moneary fundamenals (Dornbusch, 1987; Rogoff, 1996) 2. Therefore, he large and persisen flucuaion of he U.S. dollar, which canno be explained by moneary variables, can cause differen degrees of supply shocks in differen indusry secors. Moreover, due o he persisency, unlike o he moneary shock, i causes relaive long-erm variaion of he relaive commodiy price. 1 An excepion is Saghaian e al. (2002). They find ha he money is no neural o relaive agriculural price even in he long-run. 2 This is called nominal exchange rae misalignmen problem in he relevan lieraure.

5 There are some reasons we believe why U.S. agriculural prices are expeced o be more sensiive o he U.S. dollar movemen, relaive o he prices of oher indusries and non-radable goods. Firs, he agriculural secor is heavily involved in inernaional rade (more han 30 percen of domesic agriculural producs is expored). Second, he demand for agriculural producs is inelasic compared o oher manufacuring producs (Kileson and Poole, 2000) 3. As a resul, supply shock induced by exchange rae movemens can more sensiively affec he agriculural prices han oher manufacuring and non-radable good s prices. The main objecive of he paper is o idenify he imporan macroeconomic facors which explain he changes of relaive agriculural price in differen ime horizons (1, 12, 24, 36, 48, and 60 monhs). To examine he issue, we develop a imeseries model which idenifies unobservable real facors affecing variaion of relaive agriculural prices in he long run. Unlike he shor-erm overshooing problem of he agriculural price induced by moneary shock, we found he long-erm changes of relaive agriculural price in he U.S. have been srongly correlaed wih he U.S. dollar movemens. The paper is organized as follows. Secion 2 includes a brief discussion abou exchange rae movemens under he floaing sysem. In Secion 3, we presen empirical models o examine his issue. The main empirical findings are presened in Secion 4. The paper is summarized in Secion 5. 3 Recenly, Kim and Koo (2002) find ha he U.S dollar movemens affec he performance of U.S agriculure expors differenly han oher indusry secors, which implies possibly differen degrees of domesic supply shocks induced by he U.S dollar movemens.

6 2. Exchange Rae Movemens under he Floaing Sysem Before examining his issue empirically, we discuss he quesion of wheher U.S. dollar movemens can affec he long-erm variaion of relaive agriculural price on heoreical grounds. In fac, he possibiliy migh no be easily suppored by sandard macroeconomic models. To explain his poin, consider he heory of purchasing power pariy (PPP), which is one of he fundamenal assumpions in he flexible price moneary model. Under he assumpion of a fully inegraed world goods marke, he model assumes he following PPP condiion: P (2.1) = θ S or * P P P = * S R =θ, where P and respecively; S and * P are he aggregae price levels in he home and foreign counries, R are he nominal and real exchange raes beween he home and foreign counries (i.e., unis of home currency required o buy one uni of foreign currency); and θ represens facors ha cause he nominal exchange rae o deviae from he PPP, such as ransporaion coss and rade barriers, which are assumed o be consan in he long run. An imporan poin o consider is ha real exchange raes ( R ) are assumed o be a fixed consan (θ ) in boh he shor-run and he long-run. Therefore, from his poin of view, we do no expec ha movemens of real exchange rae can cause any cyclical long-erm variaions in real variables, such as agriculural expor and real agriculure prices. Under he sicky price moneary model (Dornbusch, 1976), shor-run deviaion of nominal exchange raes from he PPP can be explained by sickiness of nominal wages

7 and prices. In oher words, he speed of adjusmen of he financial marke in response o a nominal shock is assumed o be faser han ha of commodiy marke, which could possibly cause emporary overshooing of nominal exchange raes. However, according o he model, he nominal shock should cause only emporal overshooing of nominal exchange raes, which migh cause shor-erm volailiy raher han long-erm cyclical variaions in real exchange raes. However, recen empirical evidence in inernaional macroeconomics and finance appears o conradic he Dornbusch model by revealing he srong possibiliy of a persisen deviaion of nominal exchange raes from PPP. This would indicae ha here is a possibiliy of long-erm flucuaion of real exchange raes. Empirical evidence also demonsraes he possibiliy of some degree of inefficiency in he foreign exchange marke. The mos compac form of explanaion for hese deviaions is he possibiliy of a raional speculaive bubble 4. If he nominal exchange rae moves, i will drif in he same direcion for a long ime unless an imporan economic even changes he direcion of expecaions held by foreign exchange marke paricipans (Frankel and 4 Speculaive bubbles are defined as a phenomenon of nominal exchange rae movemens ha are no based on economic fundamenals, bu raher are based in selfconfirming expecaions (Frankel and Rose, 1995). Alhough many economiss believe ha speculaive bubbles are one of he imporan sources of unexpeced movemen of nominal exchange raes during he pos-breon Woods era, here is no a universally acceped reason for wha sars a bubble or wha causes hem o burs. Poenial explanaions for he sources of speculaive bubbles are he influenial effec of noise raders in foreign exchange markes (De Long e al., 1990, 1991); heerogeneous beliefs of economic agens (Har and Kreps, 1987); and sysemaic forecasing error (Froo and Frankel, 1989). More deailed discussion of his issue is summarized in Frankel and Rose (1995), and Taylor (1995).

8 Rose, 1995; Frankel 1996). Under boh he sicky price model and he raional speculaive bubble model, we can define he real exchange rae as: (2.2) P P = * S R =θ + f, where f relaes o unobservable sochasic facors ha cause flucuaion in he real exchange raes. The disinguishing feaures of he wo models are as follows. Under he sicky price moneary model, f migh be serially correlaed, bu he coefficien of auoregression is far less han one, so ha he deviaion of real exchange raes from an arbirary consan θ would die ou wihin a shor-ime period. By conras, under he assumpion of a raional speculaive bubble, f could possibly have a uni roo or near uni roo. Therefore, under he assumpion of he exisence of a raional speculaive bubble, real exchange rae movemens can have an explanaory power for long-erm variaion of relaive agriculural price via differen degrees of domesic supply shocks. 3. Empirical Model Derivaion In he relevan lieraure (e.g., Vining and Elwerowski, 1976; Parks, 1978; Fisher, 1981; Lach and Tsiddon, 1992; Bomberger and Makinen, 1993; Debelle and Lamon, 1997), economiss have examined he issue of he effec of nominal shocks (inflaion rae) on changes in differen commodiy prices. The empirical quesion in his case is wheher he changes in he general price level are correlaed wih he variabiliy of relaive price changes in an economy. The relaionship beween price change

9 dispersion among differen commodiies (or iner-marke price change dispersion) and general inflaion raes is ypically esimaed wih he following model, (3.1) RPD = α + β ln p + γ ln z + η where RPT is a measure of price change dispersion of differen commodiy groups; ln p is a rae of general inflaion; and ln z are raes of changes of oher relevan variables. Iner-marke price change dispersion is usually measured by a variaion (or sandard deviaion) of changes of relaive prices compared o general inflaion raes such as 5 : N 1 (3.2) RPD = ( π i π ) N 1 where π ln p ln p ) is he rae of change of he ih commodiy group; i ( = i i 1 π ln p ln p ) is an inflaion rae for he period; and N is he number of he ( = 1 commodiy groups. Alhough he empirical model specificaion (3.1) wih he measure (3.2) is appropriae o examine he effec of he general inflaion rae on he relaive price dispersion problem a he macroeconomic level, i is no appropriae o examine he issue of he relaive price change of a specific commodiy group compared o oher commodiy groups. Because we are concenraing on he price variaion problem of a specific commodiy group (agriculural producs) in comparison o oher groups in 2 5 This definiion is used in Parks (1978) and Fisher (1981).

10 differen ime horizon, we should derive he empirical model o fi our economic quesion. To do ha, we firs assume ha here are long-run relaionships beween general price level and he price level of each commodiy group. For insance, consider here are only wo goods in an economy, agriculural and non-agriculural producs. a (3.3) ln p = α 0 + β 0 ln p + η na (3.4) ln p = α 1 + β1 ln p + µ where p is he general price level; a p is he price level of agriculural goods; na p is he price level of non-agriculural goods; a a na na = w p w p and a + w na = 1 p + w ; a w and na w are weighs of he componens of he deflaor for each commodiy group; and η and are observable and unobservable sochasic componens including macroeconomic shocks, and idiosyncraic shocks for each commodiy group, which can affec he real price movemens of each group. In he long-run, he relaionship beween he general price level and he price of a commodiy group i are deermined by relaive supply and demand condiions beween he groups of he commodiies (Kileson and Poole, 2000). For insance, by Engel s law, if he income elasiciy of he agriculural goods is less han ha of he non-agriculural group, he coefficien β 0 is expeced o be less han β 1. Subracing (3.4) from (3.3), we have µ (3.5) a na ln p ln p = ( α α1) + ( β 0 β1) ln p + ( η µ ) 0 If we do no consider he long-run coefficiens β 0 and β 1, which are expeced o be deermined by unobservable real facors (e.g., differen degrees of income elasiciies),

11 we migh easily obain a saisically significan effec of general price level on relaive price movemen of differen commodiy groups. However, his resul suffers from an omied variables problem, and i is difficul o conclude ha he general price movemens are relaed o he relaive price movemens. To eliminae his possibiliy, we rewrie he equaion (3.3) and (3.4) as a (3.6) ln p β 0 ln p = α 0 + η na (3.7) ln p β 1 ln p = α1 + µ. Subracing equaion (3.7) from equaion (3.6), we have (3.8) a na (ln p β ln p ) (ln p β1 ln p ) = ( α 0 α1) + ( η µ ). 0 If we decompose he sochasic erm η and µ as he macroeconomic shocks, such as inflaion and exchange rae shocks, and he unobservable commodiy group specific idiosyncraic shocks ( η = θ + λ0 ln p + π 0 ln r + ω µ = θ1 + λ ln p + π 1 ln r + ε where 0 ; 1 r is real exchange rae, and ω and ε are unobservable commodiy group specific idiosyncraic shocks), we have a na (3.9) (ln p β 0 ln p ) (ln p β1 ln p ) = κ + γ ln p + δ lnr + ζ where κ = α 0 α1 + θ 0 θ1; γ = λ 0 λ1 ; δ = π 0 π 1 ; and ζ = ω ε. By differencing equaion (3.9) wih lag lengh k, we develop our final empirical model as a na (3.10) (ln p β 0 ln p ) (ln p β1 ln p ) = κ + γ k ln p + δ k lnr + ζ k k where k ln z = ln z ln z k for any variable z. Wih he model specificaion (3.10), we examine he quesion of which macroeconomic facors cause more deviaion in he price of a commodiy group from is

12 long-run equilibrium level in comparison o oher commodiy group. If food and agriculural prices are more (less) sensiive o he changes in general price level han prices of oher commodiy groups, he esimaed coefficien γ is expeced o be posiive (negaive). If U.S. dollar appreciaion causes more supply shock in he domesic food and agriculural markes han oher commodiy groups, he expeced sign of δ is negaive. Differen lag lenghs are imporan in examining he main hypohesis of he paper. If we believe ha he inflaion rae causes only shor-run effecs on he changes in relaive price differences, he significance of he esimaed coefficiens should be die ou where k is large enough. This means ha changes in general price level canno explain he changes of relaive prices beween differen commodiy groups in he longrun. The real exchange rae, however, can explain relaively long-erm changes of relaive prices. As we discussed before, he misalignmen problem of he U.S. dollar is prolonged and persisen; once he U.S. dollar appreciaes (depreciaes), i coninues he rend for several years in a row. Therefore, we expec ha he supply shocks generaed by U.S. dollar movemen also coninue for several years, which can beer explain relaive price changes in a longer ime period han inflaion rae. In pracice, we use wo-sep esimaion procedure. In he firs sage, we esimae he coinegraion vecor, which explains long-run relaionship beween he general price level and he price level of each commodiy group. In he second sep, he equaion (3.10) is esimaed by replacing he esimaed long-run coefficiens obained in he firs sep.

13 4. Daa Seasonally adjused monhly consumer price indices for food iems and all iems are used as proxy variables of he agriculural price and general price level. Consumer price indices of all commodiies less food iems, service iems, and all iems less food iems are seleced for comparison. We believe he consumer price index of commodiies less food iems can represen he manufacuring prices, while he consumer price index of service iems can represen he price level of non-radable goods. These daa were colleced by he Bureau of Labor Saisics (BLS) web sie ( Toal rade weighed real exchange raes beween he Unied Saes and major imporing counries are used as a proxy variable for movemens in he U.S. real exchange rae. The daa are obained from he Economic Research Service (ERS) of he U.S Deparmen of Agriculure (USDA) web sie ( Because he rade weighed real exchange index represens he U.S. dollar value compared o currencies of imporing counries, an increase in he index represens an appreciaion of he U.S. dollar. Finally, he sample consiss of 348 observaions exending from 1974:1 o 2002: Empirical Resuls As a firs sep of he analysis, he long-run coinegraion vecors in equaion (3.3) and (3.4) are esimaed by he following procedures. Firs, we examine he saionariy of each variable wih wo differen uni-roo ess: he Said-Dickey (1984) and Philips- Perron (1988) ess. Second, because he es resuls sugges ha all he price indices are difference saionary, we esimae he coinegraion vecor using Park s (1992) Canonical

14 Coinegraion Regression (CCR) mehod, which is more efficien han he leas squares esimaor suggesed by Engle and Granger (1987) Uni-Roo Tess Preliminary graphical invesigaion suggesed ha all he price indices have obvious ime rends so ha, under he alernaive of rend saionariy, he Said- Dickey (SD) (1984) and Phillips-Perron (PP) (1988) ess were applied. Table 1: Uni-Roo Tes Resuls: Sample period 1974:1~2002:12. SD(1) SD(3) SD(5) PP(1) PP(3) PP(,5) All Food All less food Commodiy less Food Service Noes: Criical values for 1, 5, and 10 percen significance levels are 3.99, -3.43, and 3.14 for SD and PP es under he alernaive of rend saionariy. The criical values come from MacKinnon (1991). Because hese ess are sensiive o he choice of order of auoregression, we repor es resuls based on differen orders of auoregression: one, hree, and five. The resuls presened in Table 1 sugges ha all he series are firs difference saionary raher han rend saionary. Thus, a coinegraion approach is used o obain long-run relaionship beween variables 6. 6 We also used he Park s G(p, q) es (1990) under he null hypohesis of rend saionariy. The es resuls also sugges he variables are firs difference saionary raher han rend saionary.

15 5.2. Canonical Coinegraion Regression To obain coinegraion vecors in he equaions (3.3) and (3.4), we applied Park s Canonical Coinegraion Regressions (CCR). Park s nonparameric mehod may have some advanages as compared o Johansen s (1988) Maximum Likelihood (MLS) approach. The CCR mehod does no require a normaliy assumpion and any assumpion abou he lag specificaion. Park and Ogaki (1991) show ha, in Mone Carlo simulaions, he CCR procedure consisenly ouperforms he ML approach in small samples. Asympoically, he CCR and ML approach will give he same resuls, if he number of lags in vecor auoregression (VAR) represenaion is rue for Johansen s approach. We also applied he Park s H(p, q) es for esing coinegraion relaionships. Park s H(p, q) es is compued by he CCR residuals. Under he null of he 2 coinegraion, H(p, q) ess have asympoically χ disribuions wih q-p degrees of freedom, while under he alernaive of no coinegraion, he es saisic diverges o infiniy. Therefore, unlike convenional ess (e.g., Augmened Dickey Fuller es), we can conclude he esimaes are coinegraion vecor when he es saisics fail o rejec he null hypohesis. In our model, each variable is reaed as he firs difference saionary wih drif. Because of he drif, each variable can possess a linear deerminisic rend as well as a sochasic rend. Therefore, we applied H(1,q) es saisics o he null hypohesis of sochasic coinegraion 7. 7 More deailed discussion abou he conceps of deerminisic and sochasic coinegraion is presened in Park and Ogaki (1998).

16 Table 2 presens he esimaed coinegraion vecors and H (1, q) es resuls 8. In he case of he food price, he esimaed coefficien is , which indicae he disproporionae increase of nominal food price compared o general price level during he sample period. In he case of oher prices, he esimae coefficiens are generally more han one (1.0460, , and ), indicaing hese are more proporionaely increased han general price level. As we discussed before, hese resuls migh be due o he differen income elasiciies and produciviy growh rae of each commodiy group. Table 2: CCR Resuls (Sample: 1974:1~2002:12) Food Consan Trend ln p H(1,3) H(1,4) H(1,5) a (0.029) a (0.0001) a (0.008) (0.120) (0.203) (0.247) All less food a (0.014) a ( ) a (0.004) c (0.092) (0.133) (0.226) Commodiy less Food (0.239) a (0.0002) a (0.059) (0.203) (0.280) (0.248) Service a (0.027) ( ) a (0.008) c (0.069) (0.146) c (0.097) Noe: Numbers in parenhesis are he esimaed sandard errors; a, and c denoe significan a he 1, and 10 percen levels. The corresponding coinegraion ess suggesed by Park (1990) canno rejec he null of coinegraion a he five percen level; we conclude ha he esimaes of CCR represen long-run relaionship beween variables in all cases. 8 To implemen CCR and Park s ess, Gauss rouines programmed by Ogaki (1993) is used. In his program, QS kernel and Andrews (1991) auomaic bandwidh selecor is used o obain long-run covariance parameers.

17 5.3. Relaive Price Changes in Differen Time Horizons If macroeconomic variables are imporan o explain changes in relaive prices, he variables should cause more deviaion in one price from is long-run equilibrium level han in oher prices. Wihou considering hese long run relaionships deermined by unidenified real facors, he regression resuls could be biased due o he omied variable problem. To avoid his possibiliy, we consruc price series as deviaions of heir long-run equilibrium levels using he esimaed coinegraing vecors for each price variables, and hen esimae he model (3.10). We presen he resuls showing he changes of relaive agriculural price compared o hree seleced price series (commodiy less food iems, service iems, all less food iems) in six differen ime horizons (1, 12, 24, 36, 48, and 60 monhs). Because preliminary es resuls sugges ha, in he case of k=1, here are auoregressive condiional heeroskedasiciy (ARCH) ype errors, we repor he resuls of GARCH (1, 1) model suggesed by Bollerslev (1986) in his case. In oher cases, however, he serial correlaion is more serious problem so ha we repor he resuls wih a heeroskedasiciy and auocorrelaion consisen (HAC) sandard error suggesed by Newey and Wes (1987). Food vs. Non-Food Commodiy Iems The firs case is he relaive price movemen beween food iems and commodiy less food iems, which is expeced o represen he relaive price movemens of he agriculural and oher manufacuring goods. The esimaion resuls are presened in Table 3. We find a significan linkage of one-monh changes in general price level and one-monh changes in relaive agriculural prices. The esimaed coefficien is negaive

18 ( ) and is significan a he one percen level. However, his linkage is disconneced when he ime horizon is lenghened. The changes in general prices do no have any explanaory power for more han one-monh changes in relaive food and agriculural prices. None of he esimaed coefficiens excep for he one-monh changes are saisically significan a he en percen level. Table 3: Esimaion Resuls: Food vs. Non-Food Commodiy k = 1 k = 12 k = 24 k = 36 k = 48 k = 60 Consan b (2.095) (1.489) (1.374) (1.023) (0.972) (1.244) ln p a (-3.752) (-1.588) (-1.497) (-1.267) (-1.102) (-1.464) lnr (1.260) b (-2.045) a (-3.663) a (-5.897) a (-6.847) a (-5.341) DW-saisics Adj-R Noes: z-raios are in parenhesis in he case of he k=1: In oher cases, Newey-Wes HAC sandard errors are used o calculae he -raios; a, and b denoe significan a he 1, and 5 percen levels. In he case of he real exchange raes, however, he explanaory power increases when he ime horizon is lenghened. The sign of he coefficiens are negaive for he welve-monh changes, and he absolue sizes of he coefficiens and significance levels are increased from a k=12 o a k=48. A k=60, he significance levels and sizes of he coefficiens become smaller han hose a k=48. The adjused R 2 also increases from a k=1 o a k=48. The negaive signs imply ha real appreciaion of he U.S. dollar causes a decrease in food and agriculural price

19 compared o oher manufacuring commodiy prices, which is consisen wih our expecaion. Food vs. Service Iems Table 4 presens he relaive food price movemen compared o ha of service iems, which is expeced o represen he prices of non-radable goods. In he case of he one-monh changes, inflaion rae has a saisically significan explanaory power. The esimaed coefficien is posiive (0.2783) and significan a he en percen level. However, he general price changes do no have an imporan role in explaining he long-erm changes of he relaive food and agriculural price. In any case excep for he one- monh changes, he esimaed coefficiens are no saisically significan. Table 4: Esimaion Resuls: Food vs. Service k = 1 k = 12 k = 24 k = 36 k = 48 k = 60 Consan a (-4.899) b (-2.143) a (-2.664) a (-2.720) a (-3.239) a (-3.672) ln p a (4.092) (0.876) (0.987) (1.105) (1.487) (1.485) lnr (1.032) (-1.541) b (-2.013) b (-2.404) a (-3.169) a (-3.468) DW-saisics Adj-R Noes: z-raios are in parenhesis in he case of he k=1: In oher cases, Newey-Wes HAC sandard errors are used o calculae he -raios; a, and b denoe significan a he 1, and 5 percen levels. In he case of he real exchange rae, i does no have explanaory power in explaining relaively shor-erm, one-monh and welve monhs, changes in relaive

20 agriculural prices. However i has saisically significan explanaory power in he cases of he 24, 36, 48, and 60 monhs changes. The significance of he esimaed coefficien increases when he ime horizons are increased. Food vs. Non-Food Iems Finally, we presen he esimaion resuls of relaive price movemens beween food iems and non-food iems in Table 5. As whole, he economic implicaion of he resuls is similar o he previous cases. In he case of he general price changes, we do no find any significan resuls in any of he ime horizons excep he one-monh changes. Table 5: Esimaion Resuls: Food vs. Non-Food Prices k = 1 k = 12 k = 24 k = 36 k = 48 k=60 Consan (0.204) (-0.241) (-0.749) (-1.178) (-1.456) (-1.466) ln p a (-2.807) (-0.292) (-0.169) (0.192) (0.396) (0.042) lnr (0.655) b (-2.642) a (-3.864) a (-4.492) a (-5.409) a (-5.122) DW-saisics Adj-R Noes: z-raios are in parenhesis in he case of he k=1: In oher cases, Newey-Wes HAC sandard errors are used o calculae he -raios; a, and b denoe significan a he 1, and 5 percen levels. However, in he case of he real exchange rae, he significance of he variables increases when he ime-horizon is lenghened, similar o he previous cases. The esimaed coefficien is , which is no saisically significan a k=1. However, he sign of he coefficiens are changed o negaive from k=12, and he absolue sizes of he

21 coefficiens and significance levels increase from a k=12 o a k=48. The adjused R 2 are also increased from a k=1 o a k=60. The signs of he coefficiens are all negaive and significan. 5. Conclusion Since he influenial paper by Schuh (1974), agriculural economiss have long recognized he poenial effec of exchange rae movemens on he agriculural secor in he Unied Saes. When exchange raes became floaing, mos agriculural economiss believed ha he new marke-based sysem could subsanially miigae he misalignmen problem. Alhough some agriculural economiss (e.g., Gardner 1981; Tweeen, 1989) have recognized he poenial insabiliy problem of he U.S. agriculural secor afer experiencing unexpeced U.S. dollar movemens during he pos-breon Woods era, researches have been concenraed on he linkage beween relaive agriculural prices and inflaion raes, while he poenial linkage beween exchange raes and relaive food and agriculural prices has been largely ignored. This lack of sudy migh be due o influence of he radiional macroeconomic view of he flexible exchange rae sysem. If he foreign exchange marke has been working properly, nominal exchange raes properly align inflaion raes beween counries, which migh preven permanen over-valuaion (or under-valuaion) of he U.S. dollar under a fixed exchange rae sysem. However, in realiy, empirical sudies have suggesed much evidence agains he moneary economic view of he floaing exchange rae sysem. Cyclical misalignmens of he U.S. dollar have been persisen and subsanial during he flexible exchange rae sysem. In fac, conroversial proposals relaing o

22 inernaional moneary reform (e.g., Williamson, 1989; Krugman, 1989; Mundell, 1992; McKinnon, 1995) 9 show how seriously hese problems are considered in his area. The economic implicaion of he presen sudy is simple. The main source of he variaion in relaive agriculural price is he variaion in real exchange rae movemens, especially long-erm variaion. Conversely, variaion of he general price has a role in explaining shor-erm changes in relaive agriculural price a bes 10. Considering he fac ha he misalignmen problem of he U.S. dollar has been cyclical and prolonged, he long-erm linkage beween he variaions in real exchange rae and relaive agriculural prices implies ha he U.S. agriculural secor has faced a prolonged insabiliy problem due o he U.S. dollar movemens. The resuls imply ha he U.S. moneary policy alone migh no be enough o preven he possibiliy of an insabiliy problem in he U.S. agriculural secor caused by U.S. dollar movemens in he fuure. If an imporan source of he misalignmen problem is some degree of inefficiency in he foreign exchange marke, inernaionally coordinaed moneary policy mus be imporan. 9 More deailed discussion concerning differen proposals on inernaional moneary reform is summarized in Frankel (1996). 10 However, we do no compleely rule ou indirec effec of he U.S. moneary policy on movemens of he U.S. dollar, and, hence, on relaive agriculural prices alhough we do no find srong linkage beween inflaion rae and relaive agriculural prices. Eichenbaum and Evans (1995) find ha around 17 percen variaions in he U.S. real exchange raes can be explained by he U.S. moneary policy variaion during he flexible exchange rae sysem. Rogers (1999) also found US moneary policy has been responsible for a minimum 20 percen variaion of real exchange raes beween he dollar/pound during he period of

23 References Andrews, D. W., Heeroskedasiciy and Auocorrelaion Consisen Covariance Marix Esimaion. Economerica, 59 (1991): Ball, L., and N. G. Mankiw, Relaive-Price Changes and Aggregae Supply Shocks. Quarerly Journal of Economics, 110 (1995): Ball, L., and D. Romer, The Equilibrium and Opimal Timing of Price Changes. Review of Economic Sudies, 56 (1989): Barro, R. J., Raional Expecaions and he Role of Moneary Policy. Journal of Moneary Economics, 2 (1976): Bollerslev, T. Generalized Auoregressive Condiional Heeroskedasiciy. Journal of Economerics, 31 (1986): Bomberger, W. A., and G. E. Makinen, Inflaion and Relaive Price Variabiliy: Parks Sudy Reexamined. Journal of Money, Credi, and Banking, 25 (1993): Debelle, G. and O. Lamon, Relaive Price Variabiliy and Inflaion: Evidence from U.S. Ciies, Journal of Poliical Economy, 105 (1997): De Long, J. B., A. Shleifer, L.H. Summers, and R. J. Waldman. Noise Trader Risk in Financial Markes. Journal of Poliical Economy 98 (1990): De Long, J. B., A. Shleifer, L.H. Summers, and R. J. Waldman. Survival of noise raders in financial marke. Journal of Business 64 (1991): Dornbusch, R. Expecaions and Exchange rae Dynamics. Journal of Poliical Economy, 84 (1976): Dornbusch, R., Exchange Rae Economics. Economic Journal, 79 (1987): 1-8. Eichenbaum, M., and Evans, C. Some Empirical Evidence on he Effecs of Moneary Policy Shocks on Exchange Raes. Quarerly Journal of Economics 110 (1995): Engle, R. F., and C. W. J. Granger, Co-inegraion and error correcion: Represenaion, esimaion, and esing, Economerica 55 (1987): Fisher, S., Relaive Shocks, Relaive Price Variabiliy, and Inflaion, Brookings Papers on Economic Aciviy, 2 (1981):

24 Frankel, J. A. Expecaions and Commodiy Price Dynamics: The Overshooing Model. American Journal of Agriculural Economics (1986): Frankel, J. A. Recen Exchange-Rae Experience and Proposals for Reform. American Economic Review, 86 (1996): Frankel, J. A. and K.A. Froo. Using Survey Daa o Tes Sandard Proposiions Regarding Exchange Rae Expecaions. American Economic Review, 77 (1987): Frankel, J. A. and K.A. Froo, Chariss, Fundamenaliss, and Trading in he Foreign Exchange Marke American Economic Review (Papers and Proceeding) 80 (1990): Frankel, J. A. and A. K. Rose, Empirical Research on Nominal Exchange Raes, Handbook of Inernaional Economics, vol. III, edied by G. Grossman and K. Rogoff (1995): Froo, K.A. and J. A. Frankel (1989), Forward Discoun Bias: Is i an Exchange Risk Premium? The Quarerly Journal of Economics 104: Gardner, B. On he Power of Macroeconomic Linkages o Explain Evens in U.S. Agriculure. American Journal of Agriculural Economics 63 (1981): Grennes, T. and J. S. Lapp. Neuraliy of Inflaion in he Agriculural Secor, Journal of Money and Finance, 5 (1986): Har, O. D. and D. M. Kreps. (1987), Price Desabilizing Speculaion. Journal of Poliical Economy 94 (1987): Io, T., Foreign Exchange Rae Expecaions: Micro Survey Daa, American Economic Review, 80 (1990): Johansen, S. Saisical analysis of coinegraing vecors. Journal of Economic Dynamics and Conrol 12 (1988): Kim, M., and W. Koo (2002), How differenly do he Agriculural and Indusrial Secors Respond o Exchange Rae Flucuaion? Agribusiness & Applied Economics Repor No. 482, Norh Dakoa Sae Universiy. Kliesen, K.L. and W. Poole, Agriculure Oucomes and Moneary Policy Acions: Kissin Cousins? Federal Reserve Bank of S. Louis, Quarerly Review, May/June (2000): Krugman, P.R. Exchange-Rae Insabiliy. Cambridge, MA: MIT Press, 1989.

25 Lach, S. and D. Tsiddon, The Behavior of Prices and Inflaion: An Empirical Analysis of Disaggregaed Price Daa, Journal of Poliical Economy, 100 (1992): Lucas, R. E. Jr., Some Inernaional Evidence on Oupu-Inflaion Tradeoffs. American Economic Review, 63 (1973): MacKinnon, J.G. (1991) Criical Values for Coinegraion Tess, Chaper 13 in Longrun Economic Relaionships: Readings in Coinegraion, edied by R.F. Engle and C.W.J. Granger, Oxford Universiy Press. McKinnon, R.I. The Rules of he Game: Inernaional Money and Exchange Raes. Cambridge, MA: MIT Press, Mundell, R. (1992) The Global Adjusmen Sysem. in M. Baldassarri, J. McCallum, and R. Mundell (eds.). Global Disequilibrium in he World Economy, New York: S. Marin s: Newey, W. K., and K. D. Wes, A Simple Posiive Semi-Definie, Heeroskedasiciy and Auocorrelaion Consisen Covariance Marix. Economerica, 55 (1987): Ogaki, M. CCR: A User Guide. Working Paper no Rocheser, N.Y.: Univ. Rocheser, Cener Econ. Res., Park, J. Y. Tesing for Uni Roos and Coinegraion by Variable Addiion, Advances in Economerics, Vol. 8, ed. By G. F. Rhodes and T. B. Fomby. Greenwich, CT: JAI Press (1990), Park, J. Y. Canonical Coinegraing Regressions, Economerica, 60 (1992): Park, J. Y., and M. Ogaki. Inference in Coinegraed Models Using VAR Prewhiening o Esimae Shor run Dynamics. Working Paper no Rocheser, N.Y.: Univ. Rocheser, Cener Econ. Res., Park, J. Y., and M. Ogaki A Coinegraion Approach o Esimaing Preference Parameers, Journal of Economerics, 82 (1998): Parks, R.W., Inflaion and Relaive Price Variabiliy, Journal of Poliical Economy 86 (1978): Phillips, P. C. B., and P. Perron, Tesing for a Uni Roo in Time Series Regression, Biomerika 75 (1988):

26 Roberson, J.C., and D. Orden Moneary Impacs on Prices in he Shor and Long run: Some Evidence from New Zealand, American Journal of Agriculural Economics 72 (1990): Rogers, J.H. Moneary Shocks and Real Exchange raes. Journal of Inernaional Economics 49 (1999): Rogoff, K. The Purchasing Power Pariy Puzzle. Journal of Economic Lieraure, 34 (1996): Saghaian, S.H., M. R. Reed, and M.A. Marchan, Moneary Impacs and Overshooing of Agriculural Prices in an Open Economy, American Journal of Agriculural Economics, 84 (2002): Said, S. E., and D. A. Dickey, Tesing for Uni Roos in Auoregressive-Moving Average Models of Unknown Order, Biomerika 71(1984): Schuh, G. E. The Exchange Rae and U.S. Agriculure. American Journal of Agriculural Economics 56 (1974): Shenshinski, E., and Y. Weiss, Inflaion and Coss of Price Adjusmen. Review of Economic Sudies, 44 (1977): Taylor, M.P. The Economics of Exchange Raes, Journal of Economic Lieraure 33 (1995): Tweeen, L. Farm Policy Analysis. Wesview Press, Vining, D.R., and T.C. Elwerowski, The Relaionship beween Relaive Prices and he General Price Level, American Economic Review, 66 (1976): Williamson, J. The Case for Roughly Sabilizing he Real Value of he Dollar. American Economic Review, 79 (1989):