Document 6568595
Transcription
Document 6568595
一的一 An Economic Analysis of the Dynamics of Taiwan's Rice Sector Tso-kwei Peng* (彭作委〉 Introduction Rice is a commodity with unique economic and political significance for Taiwan's 17 millino people. It is grown by almost 75% of Provinces' 860 ,000 farm families and since World War II , it has generated an important portion of Taiwan's foreign exchange earnings. The analysis of factors that affected domestic consumption and price of rice is necessary for carrying out programs to achieve equitable income to rice producers and for determining 出e long-run economic ou tlook for the rice sector. The specific objectives of this study are (1) to iden tify and measure quantitatively the basic demand structure of the Taiwan rice sector; (2) to estimate the short-run and long-run impact multiplier matrix. The analysis deals only with aggregate demands for rice. The behavior of rice supplies is not analyzed and assumed given i.e. , no attempt is made to explain changes in production in this pape r. In attempting to quantify the important underlying demand structure of the Taiwan rice sector , a very simple and highly aggregate econometric model is formulated in the first part , of the study. The reason is adequate empirical data with which to estimate a more complex and disaggregated model are lacking on hand. *The au thor is a professor , Research Institute of Agricultural Economics , National Chung Hsing University , Taichung, Taiwan , R.O.C. ( 1 ) 一 10。一 The Mod eI The sources of supply and types of utilization of rice in Taiwan are shown , in simplified diagrammed form , in figure 1. The physical variables are identified by a letter symbol in each box. The subscript “ t" of ktter symbols refers to the current time period , while "t-]" refers to the preceding time period. As indicated in figure 1 , the sources of total current sUpply (Qt) consist of current domestic production (Ot ), total import (IM t ) and total carry-in stock from the previous period (S t- 1). Domestic productìon and carry-in stock are the major sources of total supply , imports of rice accounts for only a negligible portion of total. The total available supply is distributed into different utilizatìon outlets-current domestic consumption (qt) , export (EX t ) and carryout stock (St). The total carryout stock conS!sts of commercial stock (SC t ) and Government stock (SG t ), but commercial stock is ignored owing to lack data. There are three m 茍 or types of domestic rice consumption: (1) rice consumed as food (Y ABt); (2) rice used for processing wine (YBA t ) and (3) rice used as seed (YAC t ). The total aggregate rice supply is normally in excess of domestic requirements for food or other domestic used. The excess either enters export channels or is retained as a carryout stock. Of total consumption , rice consumed as food is the major component , only a negligible amount was as seed uses. An Econometric Mod eI As indicatεd in the preceding sector , imports of rice and rice used as seed account for only a smalI part of the tota l. The analvsis of these two sectors , therefore , were excluded from the mode l. Also , no attempt was made to incorporate any analytical explanations of commercial stock. Farm Price 胡 d Government Purchase Price Relation In view of the role of rice as the traditional staple food and as a wa宮e good in Taiwan , the Government establishεd rice ( 2 ) 一 101- Fìg. 1 Taìwan Rice Supply and Utiliza tÌ on Production 。 Z SE、 Export Carry-out Domestlc Consumption EX t St qt Governmentll Commerciall I Seed IIProcessingllF ood üses Uses 1 1 Uses 11 Stock 11 Stock SG t SC t YAC t ( 3 ) YBA t YAB t 一 102 一 policies since World War II in order to stabilize the market price of rice as basis for economywide price stabilization. Taiwan domeStic rice market has not operated under a free competitive market situation. Vnder the Government policies for rice procurement and distribution , average rice price received by farmers are influenced largely by the Government purchase price and stock release. Therefore , in the model the fo Il owing farm price and the Government purchase price relation is postulated: PRF t = f 1 (APG t , 5 t , Ot) 一一一一一一一---------一一一一 (2.1 ) where PRF t = average rice price received by farmer at time t (NT$jKg) APG t = average rice purchase price at time t (NT$jKg) Ot = total prodl叫ion at time t (m. t.) St = Government stock at the end of time t (m. t.) Under normal condition , the average rice pr吋iCe reCei扎丸vεd by farmers 恥 WI山 II change in the same direction 站 as 由 t he change in 出 t hε Gove 凹r‘-nmεnt purchase pr吐ice and total producti叮 on 爪, negati扎丸vely related to changes in Go 叭丸vernment stock , i.e.: a-a R-PA E-A F-G > AU gP R.L 一一一~L g Ot PA3-a R-S F-t >0 +EEW- < AV Food Consumption Relation On the basis 0 1' classical demand theory , the quan tity , demanded of a particular commodity is a function of its price, thεprices of other commodities , and income. But classical theory gives no suggestion as to whether the demand function is linear or nonlinea r. Engel's law suggests that the relation between food and income tends to be curvelinea r. Examination of Taiwan per capita consumption 0 1' rice is associated with a ( 4 ) 一 103 一 rise in per capita income only at lower income le\ els. Beynnd a-certain low income level , per capita rice consumption declines as income increases. Therefore , the per capita income variable was transformed using a cornmon logarithm as following: Y ABt = f2(PRR t' PF t' logIB t , TEM t ) 一一一 --(2.2) where YAB t = per capita consumption for food at time t (kg) PRR t = average retail price of rice (NT$/Kg) PF t = average retail price of flour (I、、叮、 T吋 EMt = 品a丸v宅甘 E盯rage temperature at time t (oC) 10gI~t = logarithm form of per capita income at time t (NT$/Kg) ln the ao叭'e formula , we would normally expect that changes in per capita consumption of rice , YAB t , wiII (1) negative relation with changes in retail price of rice , PRR t' and temperatur亡, TEMt. and (2) positively or negatively related to changes in income depending upon the levels of income and to changes in retail price of flour , i.e.: ~ YAB t <(\笠坐上<仇芒竺玉之A ~PRRt v. ~TEMt v ~ 10gIB t <V ~YABt > ~ PF <0 t Processing Utilization Relation Rice is an important material for processing to be rice wine in Taiwan. Annual demand for rice as raw material for processing is around 45 , 000 ton in terms of brown rice. It s demand related to the retail price of rice and the per capita income , I. e. : YBA t = f 3 (PRR t , IBt) 一一-------_-一一一一一一一一一 (2.3) -( 5 ) -104 一 where YBA t = domestfc use of rice for processing wine at time t (m. t.) PRR t = retail price of rice (NT$fKg) KBt = per capita income (NT$) The other source of rice for processin 思 wine is imports if the domestic price is too high. As a consequence , wc would normally expect the partial derivative of. YBA t wÍth respect to PRR t to be negative and positively or negatively related to changes in incom 亡, l. e.: A , 〉一< a-aB-B A 一 -d < ιVA-7 A-R Y-p B-R a-a nu Government Inventory Rclation The principal objectives of rice policies in Taiwan have been (1) to prod盯 e as much rice as possible from limited agricultur訓 .resourse; (2) to control (procure) as much rice as possible for the Government; (3) to export as much rice as possible to earn foreign excha呵e and (4) to stabilize the market price of rice as a basis for economywide price stablization. Moreover , because rice can be stored for a relatively long period , it is possible that certain amounts of rice accumulated in a gi\'en year will be carried in to the succeeding year. Therefore , a lagged stock variable is included in the following postulated Government inventory relation: SG t = f4 (PRR t , Ot' EX t , SGt_ t)一一-------------一 (2 .4 ) where SG t = Government rice inventory at the end of time t (m. t.) PRR t = retail price of rice (NT$fKg) Ot = total rice production at time t (m. t.) EX t = total export of 丘 ce at time t (m. t.) ( 6 ) 一 105 一 The method of operating the inventory for rice has been an implicit offer by the Government to buy (included compuIsory purchase) the amount of rice produced as much as possible at the harvest period that would export and/or release to market when the price increase during the marketing year. As the result , we would expect the Government inventory is positively related to retail price , production , and !agged inventory , but negatively r eI ated to export volume , i.e.: ~SG+ 一---"一 >0 ~PRRt - -, ~SG+ 一一_" >。 ~ Ot ~ -, ~SG+ ~SG寸 一一一ι->0 一一一~ ~SGt_1 - -, ~EXt <0 Export Relation Large portions of Taiwan export of rice were channeled through the Government agencies. The pu中 oses of export are (1) to earn foreign exchange; (2) to adjust the excess supply of rice. As a consequence , we would expect normally total export to be positively related to 出 e total output (Ot) and total car句, in stocks (SG t _1 ). On the other hand , if per capita domestic rice consumption is increased andfor retail price is risen , then we would èxpect smaller exports of rice. Hence the variables enterring 出 e following postulated export relation are: EX t = f5 (YAB t' Ot' SG t _1 • PRR t ) 一一----------一--一 (2.5 ) where EX t = total export of rice at time t (m. t.) YAB t , Ot' SG t _1, PRR t are the same symboI with previous models According to the above formulation , we would expect the changes of export with respect to the change of explain variables as following: ( 7 ) 一 106 一 ~EX會 一一一ι<0 ~ YAB t ~EX. 一一→一 >0 '~Ot ~ EX. ~ SG t _1 一一一_. ' >0 ~ EX. 一一一ι<0 ' ~PRRt Price Iden 位ty The connecting link between farm price and retail price ~s the marketing margin. For rice , the marketing margin is defined as the difference between the retail price in polished rice and the farm price of paddy rice. Therefore , the processing cost in terms of farm labor wage rate as the margin , the relation is following: PRR t = f6 (PRF t' WR t ) 一一一一一一一----------一一一 (2.6) where WR t = farm level wage rate (NT$/day) Labor market appears to be the most competitive market in Taiwan. For this reason , wage rate differentiation among different labor can be treated as relatively homogeneous in quality. Moreover , rice is processed at rural area. As a result , the processing cost can be in terms of farm level wage rate. The change of retail price normally expect positively related to the changes of farm price and wage rate , i.e.: a-a R-nk p-p R-F > AV ~PRRt ~WRt > 0 Statistical Estimation ln the preceeding section , six relations were formulated for the mode l. The postulated relations do not include all the relevant variables in the system. On the contrary , only main variables for which reliable empirical data were available were considered in each relation. The influences of such omitted variables are treatcd as disturbances and explici tI y recognized by in troducing a random disturbance term , Uit' into each of ( 8 ) 一 107 一 these six relation. For simplif可ling the estimation procedures , all postulated stochastic relation in the formulatεd system are assumed to be linear in parameters. In estimating the parameters of the formulated system , the variables were divided into the following sets of endogenous variables and predetermined variables: Endogenous Variables PRF t = average rice price received by farmers at time t (NT$JKg) YAB t = domestic per capita use of rice for food at time t (Kg) YBA t = domestic use of rice for processing at time t (m. t.) SG t = Government rice inventory at the time t (m. t.) EX t = total J;Î ce exports at time t (m.t.) PRR t = average rice retail price at time t (NT$JKg) Predetermined Variables The set of predetermined variables consisted of the lagged endogenous variables in the system and the following exogenous variables: APG t = average rice Government purchase price at time t (NT$JKg) 。 t = total rice production a t time t (m. t. ) PF t = average flour retail price at time t (NT$JKg) TE:\l t =瓜lerage temperature at time t (oC) logIB t = logarithm form of per capita income (NT$ per capita) WR t = farm level wage rate (NT$Jday) The complete structural system formulated in the preceding section can be summarized and written in the following simple ( 9 ) 一 1 的一 matrix form: AYt + BYt-1 + CX t = U t 一一一一 ---一一一一一一一一一一一一 (3.1 ) where hγ= 附to 叫r 向 YAB t YBA t SG t EX t PRR t 川 = vect叫 of 1幣 d endogenous 叫仙 s = 尸 I PR 盯 F← YAB tι-1 YBA t _1 SG 令 t -1 EX~ t-11 !I PRRt-l J Xt= 附間 of exogenous vari枕 s= í A悶 。t PF t TEM t IBt logIB t WR t l ( 10 ) 一 109 一 A = matrix of parameter associated with endogenous variables 1 。 。 a14 。 。 。 1 。 。 。 a26 。 。 1 。 。 a36 。 。 。 1 a4S a生 6 。 aS2 。 。 1 aS6 a61 。 。 。 。 1 B = matrix of the parameters associated with the the lagged endogenous variables 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 b 44 。 。 。 。 。 。 。 。 。 。 。 。 。 。 ( 11 ) 一 11。一 C == matrix of the parameters associated with the exogenous variables cll c12 。 。 。 。 。 c18 。 。 c23 c24 。 c26 。 c28 。 。 。 。 c35 。 。 c38 。 c42 。 。 。 。 。 c48 。 c52 。 。 。 。 。 c58 。 。 。 。 。 。 c67 c68 and U1t U2t U4t U5t U6t There are two distinctive features of formulated structural sy 咐m (3.1): (1) the matrix of the p 訂ameters associated with the endogenous variables in the system , A , is not a triangular matrix; (2) the matrix of the paramaters associated with the lagged endogenous variables 扭曲 e system , B, is not a null matrix. By difinition , the formulated structural system (3.1) is a simultaneous dynarnic system. Estimation Methods The choice of an estimation method to be used in estimating the parameter of a stochastic structural system depends ( 12 ) -111- critically on the specification of the mode l. The 2SLS method was used to estimate the system because of the simultaneous determination of prices and consumption. Estimated Structures The empirical estimates of the formulated structural system (3.1) were listed as following. The figures in parentheses beldw the estimated parameters are the standard errors of the estimates , d.w. is Durbin-Watson statistic , h is D!l rbin's hstatlstlc. Farm Price and Purchase Price Relation PRF t = 5.858 - 0.000026 St + 0.0000008ιOt+ (0.0000098)b (0.0000012)d 0.8243 APG t (0.06167)a d.w. = 2.067 Food Consumption relation Y ABt = 3070.01 - 0.4生 97 PRR t - 6 1. 5292 PF t (0.60555)d 一 (55 .4 989)d 716.10410gIB t - 18.5131 TEM t (676.1 仿 )d (23.6174)d d.w. = 2.0869 Processing Utilization Relation YBA t = -2078.27 - 320.306 PRR t + 4.9736 IBt (75.3014)a (0.832303)a d.w. = 2.015068 ( 13) 一 112- Government Inventory Relation SG t =-348107.0- 0.080579 EX t + 158 1. 23 PRR t + (98.6961)c (0.371412)e 0.0135899 Ot+ 0.84405 SG t _1 (0.22327)a (0.023277)d h == 1.54 d.w. = 2.15799 Export Relation EX t = 243357.0 - 298.799 YAB t - 1216 .4 3 PRRt + (572.942)b (167.925)C 0.01588 Ot + 0.303379 SG t _1 (0.0190089)d (0.125994)a d.w. = 2.092978 Price Identity Relation PRRt ==-0.0643387+ 1. 2146 PRF t + 0.224125 WR t (O.062043)a (O.0244679)a d.w. == 2.076291 where t-value is significant from zero at 0.5% significance level t-value is significant from zero at 1% of significance level t-value is signigicant from zero at 5% of significance level t-value is significant from zero at 1 0% of significance level e: t-value is significant from zero at 20%of significance level d.w.: no autocorrelation at 1% of significance level h: no autocorrelation at 1% of significance level a: b: c: d: ( 14 ) -113- In the above estimated structures , the signs of all estimated coefficients are in accordance with the theoretical and logical expectation specified in the formulation of the mode l. Analysis and Implications Elasticities Estimated of elasticities of food consumption can be evaluated by means of the coefficients of price , income and retail price of flour , and the means of the data of the observations. The elasticities are listed as following table: Table 1 Price Elasticity , Cross Elasticity and Income Elasticity of Food Consumptoin Price Elasticity空 Income Elasticityb 一 0.0214 -0.037 Note: ~YABt PRRt ~PRRt YAB t Cross ElasticityC 一 0.07 a: En= 盯 b:E1 = ABt 尸 ~logIBt ~YABt PF t c: E RF = 一一一一-﹒三三三三 。 PF t YAB t The demand elasticity is ,一 0.0214 , very inelastic owing to rice is the main staple food for Taiwan's consumer. The income elasticity for rice demand shows a negative sign which indicates 由at rice is an inferior good. Both resu Its provide a ve可 im portant policy implications for Government to protect rice producers and to stabili扭曲e price variation. ( 15 ) 一 114 一 Impact Multiplier In making economic policy decisions , one usually asks questions such as these: (1) if a policy instrument was changed during a given time period , what would be the effects on the related policy target of on other' important variables in the system during the same time period? (2) what would be the impacts of such a change during each of the successive 、time periods? (3) what would be the total impacts over a long period of time? To answer such questions , we need a knowledge of the underlying economic structure. In the framework of the formulated model , the above questions are equivalent to the general problem of finding the solutions of the short-run and long-run impact multiplier matrices of a dynamic system. Short-Run Impact Multipliers To derive the short-run impact multiplier matrix for the dynamic system (3.1) , we consider the followi時 reduced form of the underlying structure (3.1): Yt=D 1Yt-l +D 2 x t +V t 一一 -------四----一--一一一一一 (4.1 ) where 一 tdh BCt -K 'in/ 回 DDV +-L By the definition , the short-run impact multiplier is defined 品 the impact of a unit change of exogeneous variable during a given time period on the endogenous variable during the same time period. That is: ~Yt 一一= D2 = I d 2ij I axthf is the short-run impact multipliers. ( 16 ) 一 115 一 One important policy instrument variable among the exogenous variables of the system (3.1) is the Government purchase price variable (APG t , the Government has adopted the support price policy since 1973). The short引m impact multipliers of the support price are summarized in table 2. Table 2 Short-Run Impact Multipliers: Impacts of the Government Purchase Price on the Endogenous Variables in the System Endogenous Variables Identification Average farm price (NT$/Kg) Per capita consumption (Kg) Processing demand (m. t.) Government inventory (m.t.) Export (m. t.) PRF t 0.8645 YABt -0 .4 7146 YBA t -336.22232 SG t 1,546.06199 EX t 1,418 .1 3625 Retail price (NTJIKg) PR~ Short-run Impact Multiplier of Support Price (NTJ/Kg) 1.05002 These estimates of the short-run Government purchase price impact multipliers imply that an increase of NT$ 1 per Kg in Government purchase price , with all other predetermined variables in the system held constant , willlead to (1) an increase of NT$ 0.86 per Kg in average farm price; (2) a decline of 0 .4 7 Kg per capita in rice food demand; (3) a decline 336 m. 1. in rice for processing; (4) an increase of 1 ,546 m. 1. in Government stock; (5) an increase of 1 ,418 m.t. in total export and (6) an increase of NT$ 1. 05 per Kg in average rice retail price during the same period. ( 17 ) -116 一 An increase of Government purchase price (and also farm price and retail price) leading to higher exports is explained by 由e operation of a再justment stocks. When Govemment stocks mounted to r eI ativ eI y high , the policy usuaIIy was to reduce th凹n through export program to earn foreign exchange. The surprising result is that the percentage of retail price increase is higher than that of Govemment purchase price increase. In other words , the retailers transfer whole cost to consumer and make an unreasonable profit. Long-Run Impact Multipliers Given the dynamic system (生 .1) and given the initial conditions of endogenous variables (yo) and the time-path of exogenous variables (門, for t== 1 ,2 ......,的 can be determined as fo Il owing: yl == D1y o + D2 x l q4 X 唔,A YK= EA 官 qJh qι yqh-- D , vdo + Dq4x + D D i DPo+D仇 +D內xk-l + ------一一-47.ID2X1 一一一----( 4.2) If an exogenous variable is raised by one unit and remains at its new level in successive time periods , i.e.: xl = x2 == ....… ...H.H.--H.H.-...=xk=x*..一一一一一一一 (4.3) Given (4.3) , the relation (4.2) becomes: k-l YK=DTYJ(I+DI+Df++D1)DJ一--(4 .4) Taking the partiai derivative of yk with respect to x*'iι , ( 18 ) 一 117- ;::)Yk -7=(I+DI+Di+Di + . + ;::)x D~-l) D 2 -一一一一 (4.5) _j_y.,. 3 _j_n. k-l But 1 + D 1 + Dî2 + Dí + ........... + Dr .l can be expressed as _j_ (1 - D 1r 1 (1 - Df) ,帥的tion (4 .4) can be w耐心: IT k D x Yk == DîY o + (1 - D 1 r-1.l (1 - n. Dî) 2 For a 叫le dynamic 叮stem, the matrix k will approach a null matrix as k increases. Therefore , the endogenous variable vector will a approach the long-run stationary equ i1ibrium state: Dr lim yk=(I-D1) I D2X* , k →∞ Therefore , the Iong-run impact multiplier matrix of a stable dynamic system is: (1 - D1r1D2 The estimated long-run multiplier , as table 3 , shows that a sustained increases of NT$ 1 per Kg in Govemment purchase price wilI generate in 出e long-run (1) an increase in 9 ,913 m.t. Govemment stocks; (2) an increase in total exports of 1 ,589 m.t. 1n making policy decision , it is useful to know the relative magnitudes between the long-run and short-run impact multiplier resulting from a change in policy instrument variable. Table 4 shows that the long-run effect of a change in Govemment purchase p~ice on Govemment rice stock will be 6 .4 times as large as 出e coηesponding short-run effects. But the Iong-run impact multiplier of the total export is only 1.1 times as large as the short-run effec t. 1n other words , the Government stocks wilI accumulated to a very high amount which wiU depress average farm price. ( 19 ) -118- Table 3 The Long-Run Impact Maltiplier Endogenous Variables Identification Long-run Impact Multiplier of Government Purchase Price (NT$/Kg) Average farm price (NT$ /Kg) Per capita consumption (Kg) Processing demand (m. t.) Government inventory PRF t 0.8645 YAB t -0 .4 7146 YB~ -336.22232 SG t 9 ,913.76771 Export (m.t.) EX t 1,589.49268 Retail price (NT$肢的 PRRt 1.05002 Table 4 Relative Magnitudes between Long-Run and Short-Run Impact Multiplier Endogenous Variable Identification Relative Effect of Govemment Pu rchase Price Government inventory SG t 6 .3 5 Export EX t 1.1 2 ( 20 ) -119- Summary and Conclusions The purpose of this study has been to analyze the basic demand structure of the Taiwan rice sector and to estimate the short-run and long-run impact multiplie r. The sign of variables in whole system are compatible with theory and logic expectation. The formulation of demand for the Taiwan rice sector could be improved by incorporated as a whole and by means of different estimation technique. The result of emperical work shows that rice is an inferior good and demand elasticity is very inelastic which is consistent with the prior research (Chen , 1980). These results provide a very important policy implications for Government to protect rice producers and to stabilize the price variaties. 1mpact of the Government purchase price on retail price indicate that the percentage of retail price increase is higher than that of Government purchase price increase. 1n the long-run the Government stocks accumulate to a very high amount but export increase only slightly. This will increase the fiscal burden of the Govern口lent. Multiplier analysis as a menas to examining policy alternatives in commodity markets are limited to partial equilibrium analysis. One can answer the question concerning the impact of a change in an exogenous variable on the endogenous variables while all other variables are held constant. This type of analysis can be performed to evaluate short-run and long-run effects.. On the other hand , the limitations of multiplier analysis is where the exogenous variables undergo constant or varying rates of change or the underlying model is nonlinear. ( 21 ) -120- References Chen , W. 旺, An Economic Study on Govemment Rice Stock Operation in Taiwan , unpublished Ph.D. dissertation , Universi ty of Illinois , 1980 Kmenta , J. , Elements of Econometrics , New York: John Hopkins Book , 1971 Labys , W.C. , Dynamic Commodity Models: Specification , Estimation , and Simulation , N ew Y ork: Lexington Books , 1973 , Quan討tative Models of Commodity Markets , Cambridge: Ballinger Publishing Co. , 1975 Mo , W.Y. , An Economic Analysis of the Dynamics of the U.S. Wheat Sector , USDA , Technical Bulletin , No. 1395 , 1968 Reutherger , S. ,“ Analysis of a Dynamic Model , with Particular Emphasis on L. R. Projection ," JFE , 1966 (48): 88-106 Theil , H. , Pinciples of Econome 甘ics , New York: John Wiley & Sons , 1971 Vandenborre , R.J. ,“ Dynamic Impact Multiplier: A Case Study of White Dry Edible Bean ," AJAE , 1968 (50): 311-327 ( 22 )