An Input Tax Instrument to Control Nonpoint Pollution in the Ebro Basin

Transcription

1 An Input Tax Instrument to Control Nonpont Polluton n the Ebro Basn Abstract The problem wth nonpont source polluton s the uncertanty about the bophyscal processes and the asymmetrc nformaton between the socal planner and polluters. The mplementaton of polces to control nonpont polluton s a complcated task because the regulator does not have enough nformaton to acheve effcent results. Some control nstruments such as nput-based taxes or ambent-based taxes are suggested n the lterature. The objectve of ths paper s to evaluate the effcency of usng an nput tax nstrument wth a nonunform tax rate to abate nonpont polluton. Emprcally, the model s used to analyze salnty problems n the Ebro Basn (Northeast of Span) where there are heterogeneous farmers. Dfferent types of farmers produce several types of crops wth dvergences n technologes and bophyscal feld characterstcs. The use of an nput tax wth two tax rates can be feasble wth reasonable mplementaton cost, and qute effectve for abatng salnty emssons loads. Keywords: nonpont polluton, nput tax, non-unform tax rates, salnty. JEL classfcaton: D62, C61, Q25, Q53. 1

2 An Input Tax Instrument to Control Nonpont Polluton n the Ebro Basn 1. Introducton Agrcultural polluton s dentfed as one of the most problematc types of polluton n the world. Agrcultural producton generates a runoff of pestcdes, nutrents, sedments, salts and dsease organsms nto esystems. The most common type of agrcultural polluton s nonpont polluton. Nonpont polluton occurs when the pont from whch the polluton s emtted s undentfable, the man characterstc of nonpont polluton s the dffculty n dentfyng who s causng emssons and the place from whch these emssons have been dscharged. Some problems assocated wth nonpont polluton are the uncertanty n bophyscal processes and the asymmetrc nformaton between the regulators and the polluters. These problems mpose mportant dffcultes to the mplementaton of polces to control nonpont polluton. Poneer studes about nonpont polluton are the works of Grffn and Bromley (1982), Shortle and Dunn (1986), and Segerson (1988). The frst two works focus on the use of nputbased nstruments to control nonpont polluton. Segerson (1988) focuses on the use of ambentbased nstruments. However, the problem wth ths approach s that n most cases the regulator does not have enough nformaton about the polluton that each agent s emttng because the scarcty of nformaton about the transport and fate processes wth nonpont polluton. The exstence of ths problem makes the mplementaton of an ambent tax dffcult n general, t s easer to use nput tax polces. Salnty emssons are a type of nonpont polluton. The use of salne water n rrgaton areas or the use of salne sols n agrcultural producton generates a runoff of salne mnerals n the ecosystems. Ths runoff of salne mnerals to water bodes causes them to decrease n qualty. These salts accumulate n the root zone of the plants and negatvely affect agrcultural productvty. The problem of salnty s an mportant ssue n rrgated agrculture worldwde, the rrgated acreage degraded by salnty problems reaches 80 mllon hectares, whch s 30 percent of all rrgated land n the world. In Span the acreage affected by salnty s around one mllon hectares, of whch 312,000 ha are located n the Ebro Valley (Alberto et al. 1986). The salnty of the Ebro Valley presents a serous problem because these 312,000 ha affected by salnty represent almost 50 per cent of the rrgated surface n the basn. The Ebro Valley s located n the North-East of the Iberan Pennsula. Ths terrtory covers around 20 per cent of the surface of Span. The Ebro rver has a total length of 910 km and flows nto the Medterranean sea. The Ebro rver receves water from dfferent rvers, and the entre system covers around 12,000 km. Along the rver basn there are several lakes, aqufers, 2

3 and mountanous areas such as Pyrenees. Because the length of the rver basn, ths area has an enormous dversty. The emssons of salnty substances to the water bodes are harmful to ths area s bodversty. Nonpont polluton from agrcultural producton s a negatve externalty, resultng n negatve mpacts that reduce socal welfare. Ths study analyzes the effcency of an nput tax polcy to abate salnty emssons loads from agrculture, n the Ebro Valley. In the model there are a group of heterogeneous agents who produce dfferent knds of crops, wth dfferent characterstcs of sols, and a varety of rrgaton systems. The sols are categorzed as salne, moderate salne, and no salne. Salne and moderate salne sols generate a runoff of salts and, when the farmers rrgate the land ths runoff goes to the rvers and other plots causng damage to the ecosystems. In the case of the Ebro basn, large quanttes of dssolved salt enter nto the rver. The orgn of ths salne substance s surface runoff from the geologc formatons of the basn. The farmers need water to rrgate and ths rrgaton causes salne to return to the rver. The augmented salnty of the rver affects the ecosystem and can cause major damages to downstream users. The next secton presents the methodology used to analyzed the nput tax regulaton based on one of the poneer papers n ths area, Grffn and Bromley (1982). The thrd secton descrbes the area of study, whle the fourth descrbes the emprcal model used n ths study and the estmaton of the producton and emsson functon. The ffth part presents emprcal results and smulatons. Fnal remarks conclude. 2. Methodology Followng the deas proposed by Grffn and Bromley (1982), f polcy-makers want to mplement a regulaton to control nonpont polluton, they need to take nto account bophyscal nformaton about the producton alternatves and the emssons. These authors proposed the mplementaton of an nput tax to control nonpont polluton n a multple farmer model. In ths work the authors establsh that the regulator has nformaton about the farmers proft functon and can estmate the emsson functon. Under the assumptons of no transacton costs and perfect nformaton about the farmers technology and runoff functon, Grffn and Bromley show that the nput tax s an effcent polcy nstrument to control nonpont polluton. The model can be expressed as the maxmzaton of the prvate benefts of a group of heterogeneous farmers n ndexed by = 1,..., n. In the prvate maxmzaton problem farmers maxmze the benefts obtaned from crop producton less the fxed and varable costs of producton. In the absence of regulaton the farmers problem s to choose the nput vector x to 3

4 maxmze ther prvate net beneft functon. In ths case, farmers are not takng nto account the negatve externalty that ther actvty generates, and ther choce problem s, max Π ( x ) = p f ( x ) p x F (1) y x Ths formulaton represents the maxmzaton of the prvate beneft the -th farmer, where p y s the output prce, f (x ) s the producton functon that s dfferent by each farmer and depends on the choce of nputs of each farmer x, the varable cost s p x x, and F s the fxed costs of producton. The emsson and producton functons dfferent, for each farmer to show the dfferent characterstcs of sol, type of crops, and technology. The model assumes that farmers are rsk-neutral and the prces are exogenous varables determned by the market. Ths model assumes perfect nformaton, that s, the regulator knows the type of agent and hs characterstcs. The frst order condton of the prvate problem descrbes the optmal choce of producton actvtes, Π ( x ) x = p y f ( x ) p x = 0 (2) Ths expresson represents the typcal economc rule, margnal benefts equal margnal costs. In ths case, the margnal costs are only the prvate cost and not the total cost to socety. The farmer s gnorng the external cost of hs actvty and does not consderate the socal costs of hs actons. Envronmental damage costs are the alteraton n the ecosystems caused by the runoff of harmful substances from lands nto the water body. The damage functon D=D(a) s an ncreasng functon of ambent polluton because damages are drven by polluton (D (a)>0). The nonpont polluton emssons are a functon of the pollutng nputs x, and ste characterstcs ε, r (x, ε). The ste characterstc element ε, shows the stochastc nature of the runoff due to the fact that factors as weather or ste characterstcs affect the emssons runoff. In ths work we assume that the emssons functon s a determnstc one n order to smplfy the analyss, so r (x ). The ambent polluton functon depends on the farmers nonpont emssons, a = a(r 1,..., r n,). So the damage functon can be expressed as a) D[ a( r,..., r )] ( 1 n D =. We assume that ths functon s an addtve one, the margnal damage for one farmer does not depend on the polluton emssons functon of the other farmers. So the damage functon s D (a(r )), each farmer generates a damage that depend on hs nput use. The socal planner s problem ncludes the farmers prvate benefts and the polluton externaltes. Socal welfare here (SW) s defned as the dfference between the prvate benefts of 4

5 the farmers and the cost of the socal damages they are causng. The socal planner s problem s thus to, n [ Π ( x ) D[ a ( r ) ] max SW = (3) = 1 In ths case the frst order condton of the maxmzaton problem s: SW ( ) r = Π = 0 x D a x a r x (4) The frst order condton shows that the margnal prvate profts of usng one addtonal unt of nput must be equal to the margnal cost (envronmental damage) of usng ths nput. In the absence of regulaton, farmers wll produce untl ther margnal net benefts are zero (equaton 2). In the socal planner s problem, the nternalzaton of damages means that margnal prvate net benefts must equal margnal damages (equaton 4). The regulator may ntroduce an nput tax n the farmers prvate problem (equaton 1) to force the agents to nternalze the external costs. An nput tax on the pollutng nput wll reduce polluton. Wth the tax, the farmers problem consst of maxmzng prvate benefts less the nput tax, where t s the tax rate. max NB = Π ( x ) t x (5) The frst order condton s equal to: NB x = Π ( x ) t = 0 (6) When the tax s equal the margnal damage, the frst order condton s equvalent to equaton (4) obtaned n the socal planner problem. The regulator could nduce farmers to nternalze the external costs by usng nput tax equal to the margnal damage. Ths tax rate s the Pgouvan rate, and the equatons (4) and (6) show that, to nduce effcent decsons, regulator should set the tax rate equal to the value of the margnal envronmental damage,.e, t D a r = (7) a r x The regulator can calculate the value of the tax wth nformaton about transport and fate functon, and about the ambent functon D(a). To calculate the value of the optmal tax rate, the regulator uses nformaton about the optmal quantty of emssons. The regulator fxes a lmt for ambent polluton, T. The lmt of ambent polluton s that value that permt to mantan good water characterstcs and good water qualty. And the regulator s problem becomes, 5

6 max s. t. n NB = Π ( x ) = 1 (8) D( a) T It s mportant to pont out that the model by Grffn and Bromley does not address two ssues. The frst one s that nonpont polluton s a stochastc process drven by stochastc varables affectng water runoff, such as weather. The second ssue s the lack of nformaton on nonpont polluton emssons loads because of the scarce knowledge about the polluton transportaton and fate processes. As s shown before, n ths model the polluton s assumed to be determnstc. To ntroduce stochastc components n the model, the polluton functon becomes stochastc. One specfcaton s a = a( r 1,...,, ε ), where ε represents envronmental stochastc elements and elements related to the lack of nformaton on the transport and fate process. Ths analyss shows the frst best polcy that can be appled to abate nonpont polluton. The problem of mplementng ths technque to mtgate nonpont polluton s the amount of nformaton needed: the nput tax depends on the partcular producton and polluton functons, transport and fate processes, and spatal dstrbuton of aquatc ecosystems. Therefore, the nput tax rate must be dfferent for each farmer who generates dfferent emssons and damages. Moreover the problem can be poltcally complex because of the dffculty to desgn and mplement dfferent tax rates for each farmer. The frst-best scenaro s the scenaro n whch the allocaton of resources s effcent and t s not possble to make one agent better-off wthout makng the other worse-off; ths s the defnton of Pareto optmalty. To acheve an effcent allocaton of resources t s necessary to have specal market condtons such as: compettve markets, no externaltes and no publc goods, free movement of goods, and perfect nformaton. A market falure occurs when one or more of these characterstcs are not acheve. Emprcally, t s dffcult to fnd a market havng all of these characterstcs. Therefore, an mportant result s the theory of the second-best. The secondbest theorem shows how when there are two or more market falures, to correct one of them could not be effcent because t may make thngs worse. In a second best world, a coordnaton of polces s requred to attan an effcent outcome (Bennear and Stavns, 2007). To acheve effcency n the polluton problem the regulaton outcome must be the pont where margnal costs of the regulaton s equal to the margnal beneft t generates. Nonpont polluton s an externalty problem wth no nformaton n the market, so to acheve a frst-best stuaton s a dffcult, f not mpossble, task. r n 6

7 In the case of heterogeneous farmers, each one produces a dfferent quantty of emssons because n the polluton functon emsson loads depend on the quantty of water used. The model represents dfferent types of crops whch have dfferent requrements of water. Usng a more effcent rrgaton system emts a smaller amount of polluton to the rver. The type of sol determnes the quantty of salne substances that the farmers wll emt. In ths context the optmal tax rate must be dfferent for each farmer. The tax rate nternalzes the socetal cost of emssons by farmers, but, because they are heterogeneous, the rate must be dfferent for each type of farmer. Followng equaton (7) the tax rate s equal to the margnal damage of each farmer; and, wll be dfferent for each farmer snce ther emssons are dfferent, so the optmal tax rate wll also be. Implementaton of dfferent tax rates can be dffcult for the admnstraton because nformaton s needed about the partcular characterstcs of each farmer. In the context of ths work socal planner has nformaton about the crop and rrgaton system of each farmer but there s a lack of nformaton about the knd of sol that each farmer has. Other dffcultes are the transacton costs of mplementaton and the admnstraton costs of the tax. The exstence of transacton costs and the scarcty of nformaton makes t dffcult to mplement an nput tax wth dfferentated tax rates for each type of farmer. Usually, regulators use a unform tax rate for all agents, whle tryng to mnmze the neffcency. Ths s a second best polcy because the most effcent alternatve would be dfferent tax rates reflectng the heterogenety n polluton between agents, but the transacton costs and poltcal hdrances make ths alternatve unfeasble. The approach of ths work s to study the use of a second best polcy to reduce polluton n the Ebro basn. The objectve of the study s to analyze an nput tax over heterogeneous agents, comparng the effcency of unform and non-unform tax rates. In ths study two dfferent tax rates are used because of the dffcult of applyng dfferent tax rates to every agent. 3. The Study Area The Ebro rver s the longest rver wth the bggest hydrographc basn n Span. It covers 20 per cent of the surface of Span. The Ebro basn s located n the north-east of Span and occupes an area of 85,362 km 2, wth a longtude of 910 km. Ths area borders wth the Pyrenees n the north, the Iberan Chan n the south-west, and the Catalonan Coastal n the south-east. Of ths total of approxmately 85,000 km 2, around 450 and 500 km 2 are respectvely n Andorra and France, and the rest s n Span. The Ebro basn covers sx Spansh regons: Cantabra, Castlla-Leon, La Roja, Navarra, Aragon, and Catalona. The part of the Ebro basn studed n ths work s located n Aragon. 7

8 FIGURE 1 The clmate s contnental wth extreme temperatures, very cold wnters and very warm summers. The ranfall s around the natonal mean wth 550 mm a year, dstrbuted between the seasons of sprng and autumn. It s mportant to hghlght the frequent presence of wnds call cerzo, wth a hgh dryng power, whch gves rse to large evapotranspraton, accentuatng the dryness of the area. It s thus necessary to use rrgaton to provde supplementary water to the ranfall. A total of 780,000 ha n the Ebro basn are dedcated to rrgated agrculture. Ths s 28 per cent of the total cultvated area. The agrculture n ths area s characterzed by a large dfference across farmers n terms of both crops and rrgaton systems. The area of nterest for ths study s the Arba basn, n partcular. The Arba s one of the rves that flows nto Ebro rver, and s located n the north-west part of the Aragon regon. The ntensve agrculture n ths area was developed n the mddle part of the twenteth century wth the constructon of an mportant canal to provde water to ths area. The converson n part of ths area to ntensve agrculture has generated the neffcent use of rrgaton water that generates salne return flows to the ecosystems. It s also one of the rvers nsde the Ebro basn wth hgh salnty levels. FIGURE 2 The Arba basn extends over 151,000 ha of the Ebro basn where 53,000 ha are from rrgaton agrculture. The prevalng rrgaton technology s surface rrgaton (53,000 ha), followed by sprnkler (10,000 ha) and drp (200 ha). The man crops planted n the Arba basn are alfalfa (17,000 ha), wheat (12,000 ha) and corn (9,000 ha). Other crops are barley (4,500 ha), sunflower (3,700 ha) and rce (2,300 ha). The salnty n the Ebro basn orgnates from geologcal processes related to the formaton of ths area. The returns from the rrgaton dstrcts n the Arba have hgh salnty loads, whch come from the geologcal materals that form the area sols. The problem has worsened wth rrgaton, whch ncreases water crculaton n sols and generates larger exports of salts that further deterorate the water qualty of the Arba and Ebro rvers. Lke the rest of the European countres, Span must acheve the Water Framework Drectve goals. The Drectve fxes some qualty standards common to all countres n the European Unon. The purpose of the Drectve s to protect and mprove the qualty of all water 8

9 bodes ncludng surface water, water n transton, coastal areas and underground water. Ths objectve should be acheved by the year 2015 n all European countres. Ths s another reason to try to reduce the quantty of salnty substances that agrculture emts to the water bodes n the Arba basn. Due to the hgh level of salnty and, therefore polluton due to runoff the Arba basn s a partcularly good area n whch to test the effcency of mplementng a non-unform nput tax to control salnty emssons. 4. Emprcal Model Ths secton descrbes the bo-economc model used n the analyss. There are dfferent producton and polluton functons for each type of farmer. Producton functons depend on the type of crop produced (corn, wheat or alfalfa), on the rrgaton system (flood or sprnkler), and on the type of sol (salne, moderate salne or non salne). The polluton functons depend on the type of sol and the rrgaton system. Agrcultural producton generates a negatve externalty n the Arba basn through water percolaton n the sol whch brngs salnty loads to rver courses. Ths salnty affects the aquatc ecosystems and water users downstream. Farmers do not take nto account the harmful effect that they are causng. The equaton of prvate proft of farmer ( = 1,...,n) s Π = p s, y y px x k + where y = f x ). In ths model farmers produce a crop y, wth only one productve nput that s ( the rrgaton water use x, whch s the contamnant nput. The crop and nput market prces are respectvely p y, p x, and k and s are the fxed cost of producton and subsdes that dfferent crops receve. 1 In the model t s assume that farmer s a prce-takng. Farmers only consder ther prvate beneft, but not the socal damage (D(a)) from polluton a. Farmers do not consder all producton costs (socal cost) and they are producng a quantty of output hgher than the socal optmum, where producton, water used, and salnty loads would be lower. To produce ther current quantty of output they are usng a quantty of water hgher than the optmal and they are generatng salnty runoff hgh than the optmal. The regulator can mplement an economc nstrument to correct ths damage and to acheve an effcent level of water use and crop producton. In ths case the socal welfare W(x), where x s the vector the nput water, s gven by the expresson: n [ Π( x ) aa( x )] W ( x) = λ (9) = 1 1 Ths prvate proft (π ) s the quas-rent, that s the dfference between the producton actvty revenues and subsdes less the producton costs as nput costs, salares, etc. 9

10 where П s the prvate proft of the farmers less the socal damage λ a a(x ). λ a s the cost of the envronmental damages from salt emsson loads a(x ), and where the emsson load from each farmer a(x ) depends on the pollutng nput rrgaton water x. In ths model the polcy that the regulator s mplementng n order to correct the external effect s an nput tax. Wth the nput tax the restrcted prvate beneft of farmers becomes, n [ Π ( x ) t x ] t Π ( x) = (10) = 1 where t s the tax rate that depend on the type of farmer and x s the water use. The regulator chooses the optmal tax rate, the set of values of tax rates that are equal to the margnal damage that each farmer s generatng. As the margnal damage s dfferent by each farmer the tax rates too. Each farmer s dfferent dependng on hs characterstcs: type of crop, type of sol and rrgaton system. The types of crops consdered n ths work are wheat, corn, and alfalfa. These crops are the most abundant n the study area. The farmers can have three dfferent types of sol: salne, moderate salne, and non-salne. The runoff of salts are mportant n the salne and moderate sols and practcally zero n the case of non-salne sol. The last characterstc of farmers taken nto consderaton s the type of rrgaton system, there are three technologes n exstence but n ths work we only consder flood and sprnkler, snce the thrd type, drp, s practcally not n use n the study area. The use of sprnkler technology generates less salne runoff than flood; t s a more effcent system that requres less water. The producton functons take a quadratc polynomal specfcaton as s common n the lterature due to ts propertes of substtuton between nputs, maxmum level of producton, and no convergence problems n estmaton. Crop producton functons are specfed as quadratc f 2 ( x ) = α 0 + α1 x + α2 x, where x s rrgaton water used by farmer. The parameters of producton functons have been calculated from the estmatons of Uku (2003). In hs work Uku calculate the data to estmate dfferent producton functons, among them for wheat, corn, and alfalfa, wth the EPIC crop growth smulaton package. EPIC (Envronment Polcy Integrated Clmate) smulates the relatonshp between the crop growth and varables as sol, weather, water use, and crop management. The EPIC model have been calbrated to represent the crop producton functons n the study area. The results of the model have been tested wth two systems surveys dstrbuted to farmers n the Ebro basn and checkng the result wth feld experments. Usng the results of Uku have been estmated the producton functons use n ths work. 10

11 Table 1 shows the parameters of the producton functons for corn, alfalfa and wheat under dfferent rrgaton systems (flood and sprnkler). TABLE 1 The polluton functons a(x ) relate salnty emsson loads wth nput rrgaton water. The specfcaton s lnear, and s gven by the expresson: a ( x ) β + β x. Salnty loads depend = 0 1 on the sol salt contents and the water percolaton from crop cultvaton. Accordngly, dfferent polluton functons have been defned for each combnaton of sol type and rrgaton system. The data to estmate the salnty polluton functons have been obtaned by usng the Hoffman (1986) formula, modfed by Quílez (1998). The equaton from Qulez relates the ntal salnty n the sol wth the fnal salnty, when there s a leachng process from percolaton: CE CE0 = cl + pr cl, where CE s the sol fnal salt concentraton, CE ps 0 s the ntal sol salt concentraton, cl s the leachng fracton, pr s percolaton, and ps s sol depth. Clearng up the equaton, the change n salnty concentraton n the sol s gven by 2 : [ CE cl (( pr ps) cl) ] CE 0 CE = CE + (11) Ths equaton leads to the expresson: s = 6 { CE[ CE cl (([(1 ef ) x] ps) + cl) ]} [ ( 1 ef ) x] where salnty emssons s depend on the rrgaton water used x, ths functon also ncludes the effect of the rrgaton system n the coeffcent ef that s the rrgaton system technology effcency. The reason s that the rrgaton technology generates dfferent levels of percolaton, snce the rrgaton effcency s 0.6 for flood and 0.75 for sprnkle rrgaton. Equaton (12) was adapted by Mema (2006) from the one of Qulez, n order to obtan values for salnty runoff n tons for quantty of water n cubc meters. Table 2 shows the parameters on the electrcal conductvty of saturated extract parameters CE, the sol depth ps, and the washng coeffcent cl for the three types of sol salne, moderate salne and no salne. (12) TABLE 2 2 The problem s that the electrcal conductvty s measured n saturated extract (CEe), wth a saturaton percentage equal of 50 percent (5.000 m 3 /ha for a 1 meter deep sol), and n order to calculate salnty emssons ths electrcal conductvty of saturated extract s multpled by percolaton. The correct procedure would be to use nstead the electrcal conductvty of the mass of percolated water, and not the electrcal conductvty of the sol saturated extract. As a result, the estmaton of the salnty emssons s only an approxmaton. 11

12 Ths equaton has been used to generate the data of salnty emssons s, for quantty of rrgaton water use x. The data obtaned wth ths equaton corresponds to one type of sol and rrgaton system. Wth ths equaton we obtan the runoff of salts or salne materals to quantty of rrgaton water use for the farmers. Usng dfferent volumes of rrgaton water n each type of sol and wth each rrgaton system dfferent salnty emssons are obtaned for each type of sol under each rrgaton system. A lnear relatonshp between water rrgaton and salnty runoff have been calculated n order to smplfy the problem. 3 The lnear emsson functon s equaton s: a ( x ) β + β x. Two polluton functons have been estmated for each type of sol, one for = 0 1 flood and other for sprnkle rrgaton. Table 3 shows the values of the coeffcents for the emssons functon equatons. A total of sx dfferent functons have been calculated showng the dfferences n the salnty runoff by each type of sols and rrgaton system. TABLE 3 Dfferent heterogeneous farms are defned by combnng the type of crop, rrgaton technology, and sol characterstcs. For each crop the producton functon s concave and the salnty polluton functon s lnear, and both functons only depend on the rrgaton water nput n a determnstc manner. The negatve externalty generated by the producton actvtes of farmers affect the envronment but n ths case have no drect mpact n the other farmers. The farmer s problem s expressed as, max Π ( x ) = p ( α x ) p x k + s (13) y 0 + α1 x + α 2 2 Farmers are prce-takers, and ther crop producton actvtes generate an ndvdual polluton at the source whch s not observable. The fxed costs and the subsdes are dfferent by each type of crop as same as the crop market prces. Farmers are ratonal and there s not cooperaton between them. Table 4 shows the coeffcents of prces, subsdes, and fx costs, by each type of crop study n the work. x TABLE 4 The dfference between the socal planner s problem and the prvate maxmzaton problem s the ntroducton of the socal cost of the negatve externalty n the maxmzaton functon. The regulator ntroduces an nput tax and the problem becomes, 3 SHAZAM (Whte 2002) s a econometrc software to statstcal technques. One of the prmary objectves of the program s the estmaton and testng of regresson models. 12

13 max t 2 ( x ) = p y ( 0 + α1 x + α 2 x px x k + s t x Π α ) (14) The dfference between equatons (13) and (14) s the last term t x, whch s the nput tax. The tax rate s dfferent for each farmer snce t s taxng the damage that each farmer generates. The regulator can calculate the dfferent tax rates by usng the optmal quantty of emssons. The value of the envronmental damage from salnty polluton s not known, but ths value has been approxmated by calculatng the costs of extractng the salts from the water. CIRCE estmates that the value of clean water that can be use for human consumpton s around /kg. 4 Ths value s the prce of the emsson or the economc value of the damage that farmers are causng to the ecosystems. Mathematcally, the socal planner problem wth nformaton about the value of polluton damage (λ * a ) s, n * [ Π ( x ) a ( a ( x ))] max W ( x) = λ (15) = 1 Wth the frst order condton the regulator can calculate the optmal nput use (x * ), that s, the effcent quantty of nput. The frst order condton of ths maxmzaton s expressed as, W ( x) x = n * [ Π ( x ) a D' ( a ) a x ] = 0 = 1 λ (16) The socal planer has nformaton about the proft functons of each farmer and has an approxmaton of the cost of cleanng salnty to consumpton water, so socal planner also knows the costs of the damage. Wth ths nformaton socal planner can calculate the quantty of emssons socally optmal to socety. The effcent quantty of emssons that each farmer should be emttng and the quantty of output socally desrable to produce can be determned. The socal planner uses ths quantty of emssons to calculate the optmal tax rate for each agent dependng on hs emssons or damage functon. Then the regulator calculates the optmal tax rate by: D' ( a( x * )) * = t x (17) Ths tax s the rate that must be appled to farmers n order to obtan a level of emssons socally optmal. In ths work we smulate a total of eghteen farmers wth dfferent characterstcs (of crops, sols, and rrgaton technology). In table 5 are collected the values of the taxes by each type of farmer s characterstcs. TABLE 5 4 Ths value s estmated by CIRCE (Center of Energetc Research). The cost of desalnzaton s /m 3 and the runoff of salt are 1.2 kg/m 3 then the extracton costs of salts s /kg. 13

14 To mplement an nput tax nstrument wth sx dfferent tax rates s nfeasble for the regulator due to the transacton costs and the nformaton problem about the exact type of sol of each farmer. It s relatvely easy for a government to know the type of rrgaton system that each farmer s usng, but t s dffcult to know the type of sol that each farmer has. Furthermore, farmers can have more than one type of sol. To mplement a unform tax rate for all farmers s neffcent snce ther emssons are dfferent. In ths work we choose two dfferent tax rates wth the objectve to mnmze the neffcency that s generated and not to choose the sx tax rates. The regulator can dfferentate the farmers by ther rrgaton technologes and he can apply dfferent tax rates based on ther respectve technologes. The tax rates chosen for both technologes are the rates determned for moderately salne sol. Then the nput tax are t s = /m 3 and t f = /m 3, where t s s the tax on the farmers that use sprnkler and t f s the tax on the farmers wth flood (see Table 5). The smulatons of the results by each farmer have been calculated usng the program GAMS wth the solver CONOPT. General Algebrac Modelng System (GAMS) have been desgned for modelng optmzaton problems. The smulatons to analyze the effcency of usng an nput tax n the salnty polluton problem have been made wth ths program. For each smulaton there are data of socal welfare, farmer s profts, water use, emssons and tax payment by each farmer. Socal welfare s defned n the model as, SW = n t [ Π x ) + t x ] = 1 ( (18) Socal welfare s the prvate profts when farmers have nternalzed the socal cost that they are generatng (equaton 14), plus the total tax that socety receves. Socal welfare s the summatory of all ndvdual prvate profts wth the nput tax plus the tax that each farmer pay to socal planner. Ths tax s suppose to go to the socety as compensaton for the emssons that agents are generatng. 5. Analyss of results The prevous sectons analyze the theoretcal and emprcal model used n ths work. Ths secton descrbes the results obtaned wth the smulatons. Three dfferent cases are analyzed: frst-best, wthout regulaton, and nput tax. The prmary objectve s to compare the effcency of usng a non-unform nput tax nstrument n the control of runoff of salne materals nto the Arba rver basn. The work also shows secondary objectves relate wth the problem of transacton costs and the change to a more effcent rrgaton technology. As smplfcaton, we assume that farmers are cultvatng n one hectare of land, therefore, all the presented results wll be per hectare. Combnng all of the dfferent characterstcs produces a total of 18 smulatons. 14

15 For each smulated stuaton (combnaton of sol, rrgaton technology, crop, and polcy) we obtan the results of: socal welfare ( ), prvate profts ( ), water use (m 3 ), salnty emssons (T), and yeld (T). In the case of the smulaton of the nput tax there are also results for the tax payment ( ) and tax rate values. The specfcaton of the parameters and functons have been descrbed n the prevous secton. Ths secton began wth a descrpton of the emprcal stuatons for analyss, next we present a dscusson of the results and conclude wth a presentaton of the most sgnfcant outcomes of the study. Analyss of data Table 6 presents the results n the case where there s not regulaton and farmers are maxmzng ther prvate profts wthout ntroducng the polluton externalty. Ths scenaro s the base scenaro because t s the stuaton n whch no regulaton exst n the market, so farmers are producng wthout consderng the polluton costs. The objectve s to see how socal welfare mproves when the socal planner ntroduces a regulaton n the market. TABLE 6 Ths table shows that the alfalfa s the crop wth the hghest level of water consumpton, therefore the worst for water contamnaton of the three crops. Ths s also true for all rrgaton technology and all types of sols. The least productve crop s wheat. Wheat s also the crop that uses the least amount of water, so t s the crop whch produces the lowest level of polluton. Although the most productve crop s alfalfa, t s not the most proftable n all cases, for flood the most proftable s alfalfa and wth sprnkler technology the most proftable s corn. The lowest profts are obtaned wth wheat. In ths table we can observe that sprnkler technology s more effcent n the producton of all crops than flood. Wth sprnkler technology, farmers need to use a lower quantty of water to obtan the same quantty of output. Consderng that rrgaton water s the contamnant nput, the reducton of the water use reduces the quantty of emssons to the ecosystem. In the table we can observe that wth sprnkler technology the emssons level decreases around 50 per cent n 15

16 the case of corn, and more than 50 per cent n the case of alfalfa and wheat. 5 contamnant crop s alfalfa follow by corn at fnally wheat. The most One of the most mportant parameters n the smulatons s that of socal welfare. The dfference n the emssons level generates a greater reducton n the socal welfare between salne sols, and moderate and non-salne sols. The emssons are around 80 per cent greater n the producton n salne sols than n non-salne sols. Socal welfare s lower n salne sols as a consequence of the large amount of contamnatng emssons n these sols. Table 7 shows the smulaton of the frst-best scenaro. In ths case, farmers have ntroduced n ther proft functon the total damage that ther actvty s generatng. As n the scenaro wthout regulaton, ths table shows the varables prvate profts, crop yeld, water use and emssons, and socal welfare. The values for all these varables are the optmal ones. TABLE 7 Comparng ths scenaro wth the last one we can observe that socal welfare ncreases for all crops and rrgaton technologes. Wth flood the socal welfare ncreases around 50 per cent wth salne sol, 20 per cent n the case of moderate salne sol, and a 10 per cent n the case of non-salne sol. Snce the sprnkler s a more effcent rrgaton technology, the ncrease n welfare between the frst-best and the stuaton wthout regulaton s low. Wth sprnkler the ncrease n welfare n salne sol s around 20 per cent, 10 per cent n moderate salne sol, and approxmately 5 per cent n non-salne sol. 6 Crop yeld decrease for all farmers because they are consderng the socal costs of water and they are usng a lower quantty of water, so producng a lower quantty of output. But as n the scenaro wthout regulaton, the most productve crop s alfalfa and the lowest s wheat. Prvate profts are lower for all farmers n the socally optmal stuaton because they are takng nto account the socal cost of the emssons. The reducton n prvate profts are greater for all farmers wth flood rrgaton technology snce polluton costs are hgher. The most proftable crop s corn, for all rrgaton technologes, followed by alfalfa and wheat. We can observe how the reducton n profts between these two scenaros s mportant for farmers wth salne sol. To 5 It s mportant to note that the most effcent rrgaton technology s drp. However, drp technology cannot be mplemented for all types of crops. Furthermore, ths technology s not wdespread n the study area, makng the avalablty of data nsuffcent for ths study to analyze. 6 These percentages are smlar to all crops studes n the work. 16

17 mplement ths regulaton can be a dffcult task as the number of farmers wth salne sol s great, and because t s poltcally unfeasble to cause a large reducton n farmers profts. Comparng these two scenaros (frst-best and wthout regulaton) t can be observe a great reducton n the water use, so n the emssons, for farmers wth salne sols. Ths reducton s lower for farmers wth moderate salne sols, and very small for farmers wth non-salne sols. The rrgaton system permts farmers to produce more effcently the same quantty of output wth less nput. Wth the sprnkler technology the runoff of salne materals to the ecosystem decreases by a sgnfcant proporton. Another mportant reason to use sprnkler technology s water scarcty problems; farmers can be more effcent n the use of water when they use a more effcent technology, n ths case, the sprnkler. Table 8 shows the results of the nput tax smulaton. The nput tax rates column shows the value of the nput rates used n the study, to flood and to sprnkler. The values of the tax rates chosen are the optmal values for both rrgaton systems when farmers have a moderate salne sol. The decson to chose these rates s because t s the ntermedate value between a hgh rate for all these farmers wth salne sol and a low value for farmers wth a nonsalne sol. For each type of rrgaton system and sol, there are dfferent tax rates snce not all farmers are emttng the same quantty of salne materals to the ecosystems. Ths tax rate s the optmal one to force the farmers to emt the optmal quantty of polluton (frst-best). The nput tax rate does not depend on the type of crop, so there are a total of sx dfferent tax rates for each type of sol and rrgaton technology. TABLE 8 To mplement sx dfferent tax rates s a dffcult task. The prncpal problem s an nformaton one, the dffculty s to dentfy the type of sol that each farmer has. However, the socal planner can easly dentfy the type of rrgaton technology. So t s feasble to mplement two dfferent tax rates nstead of one, n order to ncrease the tax effcency. These two tax rates are optmal for sprnkler and flood technologes, n whch farmers have a moderate salne sol. In table 8 we can see that wth the mplementaton of an nput tax nstrument the socal welfare mproves under all sol types wth respect to the stuaton wthout regulaton. In the case of farmers wth moderate salne sol socal welfare s the optmal, because the tax rate s the optmal. For farmers wth salne and non-salne sols, the socal welfare s lower than frst-best snce the tax rates are not the optmal ones. Tax rates chosen are hgher for non-salne sols and 17

18 lower for salne sols. Keepng n mnd that the socal planner does not have nformaton about sol salnty for each farmer. The tax payment s greater for all farmers wth flood technology because they use more water and pollute more. The socal planner mposes a lower tax rate for farmers who use sprnkler technology because they are more water effcent. The farmers wth sprnkler technology almost pay half that of those wth flood. Imposng a lower tax rate for sprnkler users, the socal planner stmulates a change to a more effcent and cleaner technology. The prvate profts decrease wth respect to the no regulaton stuaton because the tax forces the farmers to consder the polluton costs. Wth respect to the frst-best smulaton, the profts of the farmers who have salne sols are hgher. Ths s due to the fact that the farmers are not nternalzng all the damage that ther actvty s causng. In the case of farmers wth moderate salne sol the profts are equal to the optmal. Fnally, for farmers wth non-salne sol, the profts are lower because they are payng a hgher tax that the optmal one. Lke n the frst-best scenaro the most proftable crop s corn followed by alfalfa and wheat. Snce farmers do not take nto account the characterstcs of the sol n ther prvate maxmzaton, the profts, water use, and tax payments are the same for farmers wth the same type of crop and same technology, as table 8 shows. Water use s hgher than optmal for farmers wth salne sols, optmal for farmers wth moderate salne sols, and lower than optmal n the case of non-salne sol farmers. The emssons level s dfferent snce sol characterstcs are dfferent. The emssons n a moderate salne sol are the optmal, hgher than optmal for salne sols, and lower than optmal for non-salne sols. Important results The results of the smulaton are organzed n four parts. The frst result analyzes the effcency of usng an nput tax nstrument to control nonpont polluton. The second result focuses on the utlzaton of a second-best nstrument. Ths result dscuss the effcency of usng a non-unform tax rate nstead of unform one. The thrd analyzes the effect on the emssons of a technologcal change n the rrgaton systems. Usng effcent rrgaton systems, farmers effcency wll mprove, and, thus, the socal welfare. Lastly, the fourth shows the transacton cost problem. Result 1. Ths study shows that a dfferencated nput tax s an effcent nstrument n the control of nonpont polluton. 18

19 As the last secton explans, the prncpal problem wth nonpont polluton s an nformaton one. In the case of nonpont polluton t s dffcult to dentfy the agent who s generatng the damage and the exact orgn of these emssons. Ths problem makes dffcult the mplementaton of polces n the control of nonpont polluton. Dfferent polces can be appled n the control of nonpont polluton. Ambent-based nstruments are charges or subsdes n total polluton n the water bodes or n the ecosystems. These knds of nstruments can be more effcent because they tax the farmers for the emssons that they are generatng, and are taxng agents for the drect damage that they are causng. In ths lne Segerson (1988) proposed a tax-subsdy nstrument that fxes an envronmental standard, penalzes farmers when they emt over the standard, and subsdzes farmers when they emt below the standard. Ths standard s measured n the water bodes. The penalzaton or compensaton depend on the aggregate level of emssons n the water bodes for all groups of farmers, snce the dffculty to know the exact contrbuton of each one. When the socal planner taxes the contamnant nput n order to control the emssons to the ecosystem, t s not penalzng the emssons that the agent s emttng to the water body. Nevertheless, f the socal planner taxes the emssons n water bodes t s drectly taxng the envronmental degradaton. However the effcency of ambent-based nstruments, n context of ths work, are dffcult to appled. The problem wth these polces obtanng nformaton about all the transport and fate processes of salts between the farmers land to the water bodes. The mpact of a contamnant substance n the ecosystem depends on several factors as the geology of the regon, type of substance, clmatc effects, etc. Ths nformaton s rarely avalable n the real world, so the mplementaton of ambent taxes s normally unfeasble. An ambent-based nstrument would be effcent n the salnty control problem. Assumng the possblty to measure wth precson the quantty of salne materals that arrve to the rver, t would be necessary to dentfy the emssons of each farmer. Farmers emt dfferent quanttes of polluton due to ther salnty sol characterstcs. So, the socal planner would need to know the sol characterstcs of all farmers; ths s unfeasble n practce wth current nformaton. In ths context, an nput tax nstrument s a better approach to control nonpont polluton snce the mplementaton of other polces s dffcult. As we can see n table 8, an nput tax nstrument acheves good results, lmtng the quantty of water used and the emssons level. Socal welfare ncreases wth the nput tax wth respect the stuaton wthout regulaton, and the results are near the frst best. 19

20 Another mportant beneft of ths nstrument s that t helps the socal planner control water scarcty. In the Ebro basn one mportant problem s the scarcty of water resources. The effect of the nput tax s a reducton n the quantty of water that farmers use as consequence of the ncrease n the prce. Therefore, a tax on water use acheves the objectve of reducton n the emsson level and permts a more effcent use of rrgaton water therefore preservng the resource. Result 2. One of the most mportant results of ths work s the comparson between the use of an unform nput tax rate and a non-unform one. When there are heterogeneous agents, socal welfare mproves when the socal planner can dfferentate the tax nstead of usng a unque tax. As Shortle et. al (1998) show n ther work, usng non-unform tax rates can be more cost-effcent n polluton control because t s possble to deal dfferent polluters wth dfferent characterstcs and dfferent level of emssons. When farmers are heterogeneous usng a unform tax rate generates neffcences because each farmer pollutes dfferently. A unform tax rate s appled when all of them have the same characterstcs and a unque tax can nduce the farmers to acheve the optmal level of emssons. Wth a unform tax all farmers wll reduce the same amount of emssons. In a heterogeneous world, when taxes are non-unform, farmers wll not reduce the same amount of emssons, they wll reduce proportonally to ther emssons level. The problem of mplementng non-unform tax rates s an nformaton one. It s complcated for the socal planner to have enough nformaton n order to dentfy the types of agents and ther partcular levels of emssons, and more n a nonpont polluton context. Nevertheless, there are stuatons n whch the socal planner can easly obtan nformaton about farmer type. He can dentfy the farmers rrgaton technologes and mplement an nput tax based on ths dfference. 7 As table 8 shows the mplementaton of an optmal tax requres the use of sx dfferent tax rates. To mplement sx dfferent tax rates s practcally unfeasble, however the use of two, determned by each rrgaton technology, s not. Table 9 shows the comparson between three scenaros: an nput tax wth unform tax rate, a non-unform tax rate (wth two rates), and the frst-best. Ths table compares the varables socal welfare, quantty of emssons, and prvate profts between these dfferent alternatves. The 7 In the Spansh case, there are rver basn authortes that have nformaton about the type of rrgaton system of each farmer. So, the socal planner can obtan the necessary nformaton to mplement a non-unform tax rate based on that dfference. 20

21 tax rates chosen to smulate the scenaro n whch the socal planner uses a unform tax rate s the optmal value for sprnkler n a moderate salne sol (0.0159). Results are smlar for any unform rate. TABLE 9 Socal welfare s greater n the frst-best scenaro, snce t s the optmal smulaton. Comparng the tax scenaros, t can be observed that for sprnkler technology the results are the same wth unform tax rate as wth two dfferent tax rates. The explanaton of ths s that the tax rate chosen n the unform case s the optmal one for sprnkler technology, so t s the same rate for the non-unform case. For the smulatons for farmers wth flood, the results are dfferent when the socal planner s mplementng one tax rate and when he s mplementng two tax rates. In all these smulatons we can observe that the socal welfare s lower when the socal planner only mplements one tax rate. When farmers use flood technology they are usng more water and generatng a hgher percolaton, so ncreasng the salts runoff. In ths case t s necessary to rase the tax for those farmers who are pollutng more. Table 9 shows how the results of welfare and emssons mprove wth non-unform tax rates than wth a unque one. For farmers wth flood technology and non-salne sol, welfare decreases, for these farmers the optmal tax rate n flood s (table 8), and t s closer to the unque tax rate (0.0159) than the dfferentated one ( ). In any case, the total welfare gans are hgher wth the non-unform nput tax. The total emsson wth the non-unform tax are closer to the frst-best scenaro than the unform tax. The optmal level of emssons that the socal planner wants to acheve s T (frst-best value). Wth a unform nput tax the total emssons are T, and wth a non-unform nput tax the total emssons are T. Therefore, t s more effcent to use a non-unform nput tax. For a comparson of the prvate profts, farmers wth flood technology wll prefer the mplementaton of a unque rate because ther profts are hgher. Ths s because the tax chosen s not hgh enough, so farmers are not nternalzng the real damage. But ths result depends on the tax rate chose, wth a hgher one the results for profts would be the opposte. Wth two dfferent tax rates the results for socal welfare are hgher than for the unform tax. The total welfare n the frst-best scenaro s equal to 10, Wth a unform tax the welfare s 9, , and wth a non-unform nput tax the total socal welfare s 10,005. So, lke wth the emssons the results are closer to frst-best usng a non-unform tax. 21

22 Result 3. An nput tax can be benefcal to force farmers to change the rrgaton technology to an effcent one. A non-unform nput tax encourages farmers to change to a more effcent technology. A change n the rrgaton technology s benefcal for the farmers n a medum run perod. The results n table 8 show that farmers who use sprnkler technology have hgher benefts than farmers who use flood technology. The socal planner has ncentves to urge farmers to change to a more effcent rrgaton system to avod two problems: polluton problem and water scarcty. An nput tax nstrument can correct these problems because farmers wth flood technology are payng around twce the tax of farmers wth sprnkler technology. By crops, farmers wth corn are payng a tax 44 per cent hgher to use flood technology, farmers wth alfalfa pay 50 per cent more to use flood technology, and farmers wth wheat are payng 47 per cent more to use flood technology. In ths work we assume that the necessary nfrastructures to change to sprnkler technology are adapted. The estmated costs to covert to sprnkler technology s between 4,000 to 6, Tax payments shown n table 8 are per hectare and per year. Each farmer n the study area has around 18.2 hectares. 9 Wth ths data we can estmate that farmers are payng to year around 200 /ha more wth flood technology that wth sprnkler, ths s n corn and n alfalfa. In the case of wheat ths quantty can be around 100 /ha. Ths value s calculated wth the dfference n the nput tax plus the dfference n the water between technologes. Analyzng ths data we can see that for a farmer wth flood and the hectares (18.2), a farmer that produce corn or alfalfa wll pay annually approxmately 3,640 more than a farmer wth sprnkler technology. In the case of wheat around 1,820. These values represent approxmately a 33 per cent of the total profts wth flood technology to corn and alfalfa; and a 42 per cent of total profts wth wheat. Wth the value obtaned for the profts (table 8) and the number of hectares the total profts for a farmer wth corn s, 10, wth flood and 17, wth sprnkler. Wth ths results t s easy to see the ncentves that have farmers to change to a more effcent rrgaton system, snce annually a farmer wth sprnkler s wnnng 6, more than a farmer 8 Data estmated for the Department of Sols and Irrgaton Department, n the Agrfood Research and Technology Center, Zaragoza, Span. Ths value depend of the number of hectares that farmers are changes, n ths case we can speak about scale economes, prce decrease as ncrease the hectares. 9 Ths data s publshed by the Spansh Mnstry of Agrculture, Hechos y Cfras (see bblography). Ths estmaton of the hectares of land for farmer change a lot wth respect the source consulted, but the value chose s the smaller one. Therefore, the results for ths secton do not change. 22

23 wth flood technology. In the case of alfalfa the dfference n profts s 5, , and for wheat the dfference s 4, As show these results farmers have a great ncentve to change ther rrgaton system of flood to sprnkler. Ths s benefcal for the socety n the sense that farmers are usng a scarce resource more effcently, thereby reducng the emssons to the ecosystems. In ths context, s mportant to be n mnd that to change the rrgaton technology the necessary nvestment s very great. Assumng a costs of 4,000 /ha and wth 18.2 hectares farmers wll need to pay around 72,800 to modernzed ther technology. The socal planner can ntroduce any knd of subsdes or credts to make easy the technologcal change. Result 4. A non-unform nput tax polcy must be appled whenever the transacton costs are low enough. Therefore, f the transacton costs of mplementng two tax rates are hgh t wll be more cost-effcent to use a unque tax rate. In the context of ths work, transacton costs are expected to be low due to the abundant nformaton regardng farmers technologes. Ths work does not take nto account the transacton and admnsterng costs of mplementaton of an nput tax polcy. Problems of farmers heterogenety, number of farmers, and technologcal dfferences can make costly the mplementaton of a nonpont polluton polcy. The mplementaton of a non-unform tax rate s assocated wth hgher transacton costs than a unform tax rate. Here t s relatvely easy to dentfy the farmers technology, so transactons costs wll not be much hgher that for unform tax rate. The problem wth non-unform polces s an nformaton one. For a socal planner to know the dfference between each farmer s a dffcult task. The dfference n ths work s that socal planner has free access to the nformaton and, he can mplement a non-unform tax rate wthout expendng more resources. 6. Conclusons Nonpont source polluton presents the dffculty of dentfyng the pollutng agent and the exact locaton of emssons. The mplementaton of polces to control nonpont polluton s a dffcult task snce the necessary nformaton s often scarce. Ths study consders one type of nonpont source polluton, salnty runoff. An emprcal model based on the theores of Grffn and Bromley (1982) s adapted for the case to the Ebro Rver basn n eastern, central Span. Ths area s composed of dfferng sol salnty levels and farmers utlzng dfferent rrgaton technologes. Snce each farmer emts a dfferent level of salne to the rver basn, a model allowng for heterogeneous characterstcs across farmers s 23

24 mplemented wth the goal of ascertanng the effcency of an nput tax nstrument to mtgate polluton n the study area. An nput tax nstrument s easer to appled by the socal planner because does not requred as nformaton as other polcy nstruments as ambent-based ones. Ths work explores the effcency of usng a non-unform, second-best nput tax nstrument nstead of a unform one. The socal planner has enough nformaton about the type of rrgaton technology that each farmer s usng, and can mplement greater penalzatons to farmers usng more polluton generatng technologes. Ths paper shows how a tax nput wth two dfferentated tax rates by rrgaton technologes, s an effcent nstrument to control salnty runoff. Wth ths knd of nstrument farmers wll pay a lower tax whenever ther technology would be more effcent. An non-unform nput tax that dfference the farmers accordng ther rrgaton technology generate an ncentve to change to a cleaner technology. Ths nstrument persuade farmers to change of flood to sprnkler, mprovng the socal welfare. 7. References Alberto F., R. Aragües, and D. Quílez Balance de Sales en la Cuenca del Ebro. M. Marño (eds.). Sstema ntegrado del Ebro. Estudo nterdscplnar. Documento de trabajo. Madrd. Baumol W. and W. Oates The Theory of Envronmental Polcy. Segunda Edcon. Cambrdge Unversty Press. New York. Bennear L. S. and R. N. Stavns Second-best theory and the use of multple polcy nstruments. Envronmental Resource Economcs, 37: Boyd J Water Polluton Taxes: A Good Idea Doomed to Falure?. Resources for the Future. Dscusson Paper Boyer M. and J. J. Laffont Toward a Poltcal Theory of the Emergence of Envronmental Incentve Regulaton. The RAND Journal of Economcs, vol. 30, no 1: Brooke A., D. Kendrck, A. Meeraus, and R. Raman GAMS Tutoral by R. Rosenthal. GAMS Development Corporaton. Washngton. Claassen R., and R. D., Horan Unform and Non-Unform Second-Best Input Taxes. Envronmental and Resource Economcs 19: Fullerton D. and A. Wolverton The two-part nstrument n a second-best world. Journal of Publc Economcs 89, no. 9-10: Gardner R. and R. A. Young An Economc Evaluaton of the Colorado Rver Basn Salnty Control Program. Western Journal of Agrcultural Economcs 10 (1): Grffn R. and D. Bromley Agrcultural Runoff as a Nonpont Externalty. Amercan Journal of Agrcultural Economcs 64: Helfand G. E Standards versus Standards: The Effects of Dfferent Polluton Restrctons. The Amercan Economc Revew 91, no. 3:

25 Hoffman G Gudelnes for recalmaton of salt-affected sol. Appled Agrcultural Research 1 (2): Horan R. D. and J. S. Shortle Envronmental Instruments for Agrculture. Envronmental Polces for Agrcultural Polluton Control. Edted by J. S. Shortle and D. Abler. CABI Publshng. Machn J. and A. Navas Envronmental Impact of Salnty by Dffuse Contamnaton on the Water Qualty of the Arba Rver. Cuadernos de Investgacón Geográfca 28: Martínez Y Análss económco y ambental de la contamnacón por ntratos en el regadío. Ph.D. Dssertaton. Unversty of Zaragoza. Span. Martnez Y. and J. Albac Agrcultural polluton control under Spansh and European envronmental polces. Water Resources Research, vol. 40. Mema M Las polítcas de control de la contamnacón dfusa en el Valle Medo del Ebro. Ph.D. Dssertaton. Unversty of Zaragoza. Span. Palmer K. and M. Walls Optmal polces for sold waste dsposal: taxes, subsdes, and standards. Journal of Publc Economcs 65: Qulez D Salndad en las aguas superfcales de la cuenca del Ebro: Análss del mpacto potencal de regadío de Monegros II. Ph.D. Dssertaton. Unversty of Lleda. Span. Segerson K Uncertanty and ncentves for nonpont polluton control. Journal of Envronmental and Management 15: Smth R. B. W. and T. D. Tomas Multple Agents, and Agrcultural Nonpont-Source Water Polluton Control Polces. Agrcultural and Resource Economcs Revew. Shortle J. S. and J. Dunn The Relatve Effcency of Agrcultural Source Water Polluton Control Polces. Amercan Journal of Agrcultural Economcs 68: Shortle J. S., R. D. Horan, and D. G. Abler Research Issues n Nonpont Polluton Control. Envronmental and Resource Economcs 11 (3-4): Shortle J. and D. G. Abler Envronmental Polces for Agrcultural Polluton Control. CABI Publshng. Unted Kngdom. Spraggon J Exogenous targetng nstruments as a soluton to group moral hazards. Journal of Publc Economcs, 84: Spraggon J Testng ambent polluton nstruments wth heterogeneous agents. Journal of Envronmental Economcs and Management 48: Tomas T., K. Segerson, and J. Branden Issues n the Desgn of Incentve Schemes for Nonpont Source Polluton Control. Uku S Analss economco y medoambental de los sstemas de rego: Una aplcacón al regadío de Bardenas. Ph.D thess, Dep. of Appl. Econ., Unv. of Zaragoza, Zaragoza, Span. Weersnk A., J. Lvernos, J. F. Shogren, and J. S. Shortle Economc Instruments and Envronmental Polcy n Agrculture. Canadan Publc Polcy-Analyse e Poltques, vol. XXIV, no. 3. Whte K. J SHAZAM - For Wndows, Verson 9.0. Xepapadeas A Envronmental polcy under mperfect nformaton: Incentves and moral hazard. Journal of Envronmental Economcs and Management 20: Xepapadeas A "Envronmental Polcy Desgn and Dynamc Nonpont-Source Polluton." Journal of Envronmental Economcs and Management 23, no. 1:

26 8. Tables and Fgures. Fgure 1. Map of Span and Ebro Basn Fgure 2. Map of Arba rver basn 26