The liberalization of Tariff Rate Quotas under oligopolistic competition

Transcription

1 The lberalzaton of Tarff Rate Quotas under olgopolstc competton Margherta Scoppola (Unversty of Macerata) Workng Paper 09/ 11 s a Research Proect on European Unon polces, economc and trade ntegraton processes and WTO negotatons fnanced by the Italan Mnstry of Educaton, Unversty and Research (Scentfc Research Programs of Natonal Relevance, 2007). Informaton about the proect, the partners nvolved and ts outputs can be found at

2 The lberalzaton of Tarff Rate Quotas under olgopolstc competton 1 Margherta Scoppola Dpartmento d stud sullo svluppo economco Unverstà d Macerata (Italy) scoppola@unmc.t Abstract: Several proposals for lberalzng agrcultural Tarff Rate Quotas (TRQs) have been proposed wthn the WTO Doha Development Round. The lterature suggests that the effectveness of lberalzaton depends upon whch of the three elements of a TRQ constrants mports. Most contrbutons assume perfect competton. Ths paper uses an olgopoly two-stage capacty constraned model, n whch the mode of competton s endogenous, to nvestgate the mpact of dfferent lberalzaton optons of TRQs,.e. the reducton n the n-quota and out-of-quota tarffs the expanson of the quota and reforms n the TRQ admnstraton system. The results show that consderaton of the nature of competton between traders may provde unconventonal conclusons about the effectveness of the varous TRQ lberalzaton optons; further, the paper shows how an ncrease n the number of rvals, under certan crcumstances, may be more effectve n ncreasng trade than reducng tarffs. Keywords: tarff rate quotas, olgopoly, trade lberalzaton JEL code: Q18; F13; L13 1 I wsh to thank Govann Anana and Luca Salvatc for helpful comments on an earler draft of the paper. Fnancal support receved by the New Issues n Agrcultural, Food and o-energy Trade (AGFOODTRADE) (Small and Medum-scale Focused Research Proect, Grant Agreement no ) research proects funded by the European ommsson, and by the European Unon polces, economc and trade ntegraton processes and WTO negotatons research proect funded by the Italan Mnstry of Educaton, Unversty and Research (Scentfc Research Programs of Natonal Relevance 2007) s gratefully acknowledged.

3 1. Introducton A growng share of agrcultural trade take places wthn Tarff Rate Quotas (TRQs). In 2004, 1,429 TRQs were notfed to the WTO (WTO, 2006); more than 15% of agrcultural tarff lnes n the EU are currently covered by TRQs, and other countres use TRQs even more ntensvely; n Norway, for example, almost 40% of agrcultural tarff lnes nvolve TRQs. The Uruguay Round Agreement on Agrculture ntroduced several TRQs n order to guarantee mnmum market access where there had prevously been no sgnfcant mports because of hgh protecton. Whether the mplementaton of TRQs has really mproved market access s an open queston and several proposals for lberalzng agrcultural TRQs have been proposed n the WTO Doha Development Round. Four maor ssues dealng wth TRQ lberalzaton are currently under dscusson n the WTO negotatons (WTO, 2008): the reducton of MFN tarffs, as ths also mples a decrease n out-of-quota tarffs; the reducton of the n-quota tarffs for exstng TRQs; the expanson of the quota for TRQs appled to mports of senstve products; and the reform of the method of TRQ admnstraton. TRQs are also used ntensvely under preferental trade agreements, especally by developed countres to lmt preferental access to products from developng countres. One of the core ssues n negotatons between (potental or actual) members of preferental trade agreements s how to lberalze agrcultural trade when TRQs are n place. In both contexts - multlateral and preferental agreements - a key ssue s whch strategy for TRQ lberalzaton s more desrable from the pont of vew of trade and welfare. The lterature on TRQs has offered a number of suggestons about the possble mpact of a reducton n the n-quota or the out-of-quota tarffs, of an expanson of the quota or of an mprovement n the effcency of TRQ admnstraton (e.g. oughner et al, 2000; Skully, 2001a, 2001b). Most contrbutons rely on the assumpton that nternatonal markets for agrcultural products are compettve, even though ths s often not so. There are varous cases n whch nternatonal markets are domnated by few large traders. The few papers on TRQs that have consdered market power n the tradng ndustry (e.g. de Gorter, oughner, 1999; Gervas, Suprenant, 2000; Hraanova, de Gorter, 2005; Rude, Gervas, 2006) all assume the product to be homogenous and the nature of the competton between frms exogenous to the models and, n most cases, ths s assumed to be à la ournot, to avod the ertrand paradox. omparatve statc s performed by holdng the mode of competton constant and, thus, mplctly assumng that t s not affected by trade lberalzaton. However, there are good reasons to expect ether prce or quantty competton on nternatonal agrcultural markets dependng under certan condtons (arter, Mac Laren 1997); and trade polcy s one of the factor determnng such condtons. Furthermore, product homogenety has also been questoned by some authors. 2 Ths paper ams to brdge ths gap, by nvestgatng how the presence of a TRQ nfluences the nature of 2 As for grans, evdence has been found of prce competton for wheat (Thursby, Thursby, 1990) and of quantty competton for malt barley (Dong et al., 2006). Lavoe (2005) has found evdence of product dfferentaton by examnng U.S. and anadan wheat exports 1

4 competton between olgopolstc frms tradng a dfferentated product and analysng the outcome of trade lberalzaton when strategc nteracton between traders s taken nto consderaton. A two-stage capacty constraned model, n whch the mode of competton s endogenous and affected by the trade polcy, s used to model the workng of TRQs and the mpact of trade lberalzaton. Most capacty constraned models bulds on the early work by Kreps and Schenkman (1983) who developed a duopoly two-stage game n whch n the frst perod frms choose capacty, whle n the second compete on prce; they show how, n ths settng, the outcome of the game s ournot. A more recent paper by Magg (1996) has developed a duopoly capacty constraned model n whch the mode of competton s endogenous, to address strategc trade polcy ssues; n hs paper the capacty constrant faced by frms s flexble, as frms can ncrease quanttes n the second perod, albet at a hgher cost. The author has shown that the outcome of ths partcular game ranges from ertrand to ournot dependng upon the value of the costs of adustng capacty n the two stages. The reason behnd the use of a two-stage capacty constraned model s that, under a TRQ, traders have to choose ther capacty n advance wth respect to the perod n whch they actually mport; whatever the method chosen by the government to allocate the rght to mport wthn the quota, frms have to plan n advance the number of lcences they need f they ntend to mport wthn the quota. In most cases, frms know the capacty chosen by rvals, that s, the amount of lcences allocated. Ths may work as a commtment to mport that quantty n the second stage, especally when the cost of mportng out-of-the quota s very hgh. The cost of ncreasng capacty n the frst perod s gven by the cost of obtanng lcences plus the n-quota tarff. In the second perod, frms observe the capacty chosen by rvals and compete on prce; at ths stage, they may stll ncrease capacty, but only by mportng out-of-the quota at the cost of a hgher tarff. Wthn ths settng, f the gap n the cost of ncreasng capacty n the two perods s hgh enough, then the commtment to mport the amount allocated s effectve, and the outcome of the game s ournot. onversely, the narrower the gap between the cost of ncreasng capacty n the two perods, the weaker the commtment to mport the capacty chosen n the frst stage, and the closer the outcome to ertrand. To take nto account the specfc features of the workng of TRQs the settng n ths paper dffers n two man respects from Magg (1996). The duopolstc game has been extended to the case of N frms, n order to reproduce more realstcally the structure of the tradng ndustry for several agrcultural products. Secondly, ths paper also consders the case n whch the frm faces a celng n expandng ts capacty at the frst stage. apacty constraned models assume that frms n the frst stage can ncrease ther capacty wthout lmts, but ths s not always the case under a TRQ. A frm may not obtan all the lcences t needs, because the ndustry as a whole s constraned by the quota celng; n that case, the number of allocated lcences no longer works as a commtment to mport that quantty. TRQ lberalzaton, by affectng the values of the tarffs and/or the cost of obtanng lcences, changes the costs of adustng capactes n the two stages of the game and, thus, the effectveness of 2

5 the capacty constrant whch, n turn, determnes the mode of competton. Hence, the outcome of trade lberalzaton also depends upon whether and how the polcy change affects the nature of competton between frms. The paper frst nvestgates the mpact of dfferent optons of TRQ lberalzaton the reducton of out-of quota and n-quota tarffs, the expanson of the quota and the reform of TRQ admnstraton - on the mode of competton; then, the effects of dfferent lberalzaton strateges on trade, welfare and competton are llustrated by means of a numercal example. The results show that consderaton of the nature of competton between traders may provde unconventonal conclusons about the effectveness of the varous TRQ lberalzaton optons. The paper s organsed as follows; the next secton presents the model, the thrd analyses how dfferent lberalzaton scenaros affect competton and the fourth llustrates the trade and welfare effects together wth the mpact on frms market power. The last secton offers some concludng remarks. 2. The model The model consders N frms mportng a dfferentated product n one country. The assumpton that agrcultural products are dfferentated appears reasonable n the context of many products covered by TRQs. A large part of them are processed goods whch may be qute smlar, though not perceved as dentcal by consumers who treat the brands as close, but mperfect substtutes. Even when consderng trade n raw products, there s no agreement about ther homogenety. Some papers have emphassed the mportance of product dfferentaton n the case of commodtes such as cereals and beef (e.g. arter, Mac Laren, 1997; Lavoe, 2005; Ramos et al. 2007). Although t s less clear that consumers treat raw products mported by dfferent frms as mperfect substtutes, there are however good reasons for makng ths assumpton. One s that consumers often treat products as mperfect substtutes because of ther dfferent geographcal orgn; when export monopoles are n place, lke n the cereal and beef ndustry, products whch consumers perceve as dfferent are n fact mported by dfferent frms (Lavoe, 2005). There are also examples of agrcultural tradng frms whch have been nvestng over tme n dfferentaton strateges by developng a brand; bananas s a good example. The number of traders s exogenous to the model. Frms are assumed to be symmetrc: ths means that they face dentcal cost and demand curves. 3 Each trader mports quantty q and faces the followng nverse demand functon: p = α βq λ q (1) 3 Upward and downward ndustres are not modelled here, the mplct assumpton beng that they behave compettvely. We do not dstngush between exportng and mportng frms; whle such dstncton s crucal to address the ssue of how rents are dstrbuted along the tradng channel (oughner et al., 2000), t s less relevant for the purposes of ths paper,.e. to address the ssue of how the outcome of TRQ lberalzaton may change when takng nto account changes n the competton between frms. 3

6 wth p beng the prce, and β > λ. The drect demand functon s: q = a bp + c p (2) N frms face dentcal margnal tradng costs, m, whch are assumed to be constant. The mportng country employs a TRQ, wth Q, t and T beng the mport quota, the n-quota tarff and the out-of-quota tarff, respectvely. One mportant feature of a TRQ regme s the admnstraton system. The lterature has emphassed varous neffcences created by the most common systems used to allocate the lcences to mport wthn the quota, such as the frst-come-frst-served crtera and lcence-on-demand wthout lcences tradablty (de Gorter, oughner, 1999; Skully, 2001 b; Hraanova et al. 2006). These allocaton methods do not select supplers and operators on the bass of ther costs; rather, hgher-cost supplers and operators have an ncentve to enter the market to capture quota rents and, n ths way, they dsplace lower-cost supplers and operators. Imperfect nformaton and transacton costs can ncrease the costs faced by traders and lmt the access of some of them to the lcence allocaton. Hraanova and de Gorter (2005) have analysed how the exstence of quota rents nduces frms to engage n wasteful rent-seekng actvtes. In ths model, the assumpton of symmetry precludes the possblty of neffcences due to the allocaton of lcences to hgh-cost traders. Yet other sources of neffcences are consdered by assumng that each trader faces constant cost l, whch captures all costs lke nformaton and transacton costs 4 to obtan the rght to mport wthn the quota. The cost of the lcence plus the n-quota tarff s never greater than the out-of-quota tarff,.e. l+ t T. The profts of the trader are: f q Q then Π = q( p t l m) Π = f q > Q then Π = qp ( t+ l+ m) Q ( q Q)( T + m) wth Q Q =. N There are two stages n ths game. As n Kreps and Schenkman (1983), n the frst stage frms choose capacty, whle n the second after observng ther rvals capacty, frms compete on prce. Kreps and Schenkman (1983) have shown how the outcome of ths partcular two-stage game s exactly the ournot outcome. In ther paper, however, quanttes n the second stage are fxed and cannot be changed, whle under a TRQ frms can always ncrease ther capacty by mportng out-ofthe quota. (3) 4 Ths settng s also a useful way to represent cases n whch lcences are transferable ( l captures the cost of purchasng the lcence on the market) or aucton quotas - a method whch s hardly ever used, accountng for less than 1% of agrcultural TRQs n force (WTO, 2006) wth l correspondng to the cost of purchasng the lcences from the government. 4

7 In ths paper, frms choose the amount of lcences they wsh to obtan n the frst stage; by dong so, they commt themselves to mport a certan volume of the product n the second stage of the game. In the second stage frms may stll choose to ncrease ther capacty, but only by mportng outof-quota. In other words, t s assumed that the amount of product to be mported wthn the quota can be chosen only n advance (n the frst stage). Ths reflects reasonably well the way n whch the allocaton of lcences works under most commonly used methods; n fact, lcences are allocated well n advance wth respect to the date of the effectve entry of the product nto the mportng country, and most traders consder that the only way to ncrease capacty afterwards s to mport out-of-the quota. The amount of lcences obtaned n the frst stage works as a commtment of the frm to mport a certan amount of the product n the second stage. As Magg (1996) has shown, n ths settng the prevalng mode of competton and the outcome of the game depend upon the effectveness of ths commtment; as the effectveness of the commtment ncreases, the game moves from a pure ertrand to a ournot outcome. The effectveness of the capacty commtment depends upon the relatve costs of ncreasng capactes n the two stages. There are two possble benchmarks for the soluton of ths game. On one hand, f the cost of ncreasng capacty n the second stage s equal to that ncurred by frms n the frst stage, then the capacty chosen n the frst stage does not work as a commtment devce; n ths case, the capacty constrant s neglgble and frms compete on prce,.e. the outcome of the game s ertrand. On the other hand, f the cost of ncreasng capacty n the second stage s very hgh, then the capacty commtment becomes relevant and, as n Kreps and Schenkman (1983), frms compete on quantty,.e. the outcome of the game s ournot. It s worth notng that the lcences work as a capacty commtment only when the equlbrum quanttes are not greater than the quota ( q Q ). When q > Q,.e. demand and cost condtons are such that there are out-of-quota mports, frms acqure all avalable lcences and mport addtonal quanttes at a hgher cost (.e. the out-of-quota tarff). Under these crcumstances, the amount of purchased lcences Q does not represent the capacty chosen by frms. The margnal cost of adustment n the frst perod s equal to that of the second perod (.e., the out-of-quota tarff). The capacty constrant s neffectve and ths mples that the outcome s always the ertrand one; n other words, frms perceve the polcy as a pure tarff regme wth the tarff beng equal to T. The subgame perfect equlbrum, therefore, mples that: p ( m, T) f q Q, q Q and l t T T* p = > + < p ( m, t, l) f q Q and T T* (4) wth p e p beng the ertrand and ournot equlbrum prces, respectvely, and T * beng a crtcal value of the out-of-quota tarff defned mplctly by p ( m, T) = p ( m, t, l). As aforementoned when q Q, f the value of the out-of-quota tarff s equal to the cost of ncreasng 5

8 capacty n the frst stage ( l+ t = T < T* ) there s no capacty constrant and the outcome of the game s ertrand wth long run cost equal tol+ t+ m. If l+ t < T < T* then the equlbrum can be dentfed as the outcome of a one-shot ertrand game n whch frms compete on prce and have long run costs equal to m+ T. The basc dea s that when the cost of ncreasng capacty n the frst stage s lower than n the second stage ( l+ t < T), nvestment n capacty n the frst stage nduces a hgher prce ( p ( m, T )) than the one that would preval n a pure ertrand one-shot game ( p ( m, t, l )). Fnally, when T T* the capacty commtment s effectve and frms compete à la ournot. The ertrand prce p ( m, T) may be obtaned by maxmsng (4) by holdng the rvals prce fxed, when costs are mt:, p1 max Π = p q ( m+ T) q (5) whch yelds the followng reacton curve: a+ c p + b( T + m) p = 2b Gven the symmetry assumpton, the equlbrum prce s: a+ b( T + m) p = 2 b c( N 1) (6) (7) The ournot equlbrum prce s obtaned by solvng the maxmsaton problem: max Π = qp ( m+ t+ lq ) (8) q from whch the reacton curve s: α ( t+ l+ m) λ q q = 2β (9) The ournot equlbrum prce p ( mtl,, ) s: αβ + ( t+ l+ m)( β + λ( N 1)) p = (10) 2 β + λ( N 1) The crtcal value of the out-of-quota tarff T * above whch the mode of competton shfts to ournot may be obtaned from (7) and (10), and depends upon the values of the demand parameters, of margnal cost, and of lt, and N : 5 Ζ + +Α T* = Β+ 3 λn( β λ) + λ N wth ΖΑ, and ( l t) N( l t) 2 λ ( α m) 2 2 λ N ( l t) 2 2 Βdependng upon demand parameters and costs. (11) 5 Detals are provded n the Appendx. 6

9 When parameters satsfy all constrants, f T < T * ( T > T* ), p1 < p1 ( p1 > p1 ) whch mples q1 > q1 ( q 1 < q 1 ). Whle l and t, as shown n the Appendx, both postvely affect the value of the crtcal tarff the hgher ther values, the weaker the capacty commtment, the hgher T * - the mpact of the number of rvals depends upon demand parameters, costs and N, as wll be llustrated n the numercal example n the next secton. The possble outcomes of the game are summarzed n Table 1. For gven values of demand parameters and m, the outcome depends upon the maxmum amount of lcences avalable to each operator ( Q ) and upon the values of ltn,, and T whch ultmately determne the effectveness of the capacty commtment (.e. whether the out-of-quota tarff s greater than the crtcal tarff or not). One outcome (denoted n the Table wth number 4) cannot be a subgame perfect equlbrum, whle n all other cases the game admts a unque equlbrum. When Q > q, q, the outcome of the game depends upon the value of the out-of quota tarff wth respect to the crtcal tarff: 6 f T T* ournot prevals (outcome 1 n the Table) and, as expected, the prce s determned by the n-quota tarff, whle T s redundant (equaton (10)). ut when T < T* (outcome 2) the equlbrum s ertrand (equaton (7)) and the prce s determned by the out-of-quota tarff; ths means that, although mports are lower than the quota, the n-quota tarff s redundant, whle T determnes the prce level. Ths result dffers from the general conclusons of the lterature on TRQs accordng to whch the out-of-quota tarff s bndng when mports are hgher than the quota (oughner et al. 2000; Skully, 2001a). In ths model the bndng nstrument s determned by the knd of competton between frms; f condtons are such that ertrand prevals, then, whatever the level of mports wth respect to the quota, the bndng nstrument s the out-of-quota tarff. onversely, when Q < q, q, frms operate as f there s no capacty constrant and the outcome s always ertrand (outcomes 5 and 6), wth long run margnal costs equal to T case, as expected, the bndng nstrument st, whle t s redundant. Thngs are less smple when Q s n-between or equal to q and q. If T T* and + m. In ths q < Q < q then nether ournot nor ertrand can be Nash equlbrum (outcome 4 n the Table). Ths s because f frms choose capacty q, mports are lower than the amount of lcences avalable to each operator; but n ths case, frms are n equlbrum by choosng the ournot capacty; however, f they choose q they mport a volume whch s hgher than the quota and n that case the equlbrum s ertrand whch mples the capacty q. If T < T* and q < Q < q (outcome 3 n Table 1) then 6 Hereafter q = q ( p ) and q = q ( p ). 7

10 the outcome s ertrand: frms choose q, mports are hgher than the quota and the bndng nstrument s T. Further, when Q = q the ertrand capacty s a Nash equlbrum when T < T* ; conversely, f T T* the optmal capacty would be to mport q but ths would mply to mport out of the quota and thus to compete à la ertrand; thus, nether ertrand nor ournot could be a Nash equlbrum. Fnally, when Q frms compete à la ertrand (outcome 2). = q, f T T* the outcome s ournot (outcome 1) and f T < T * 3. Lberalzng TRQs: the mpact on the mode of competton The lterature on TRQs has underlned how lberalzaton may have two mportant mpacts: t may mprove market access and t may affect quota rents (Skully, 2001a). The effectveness of lberalzaton n terms of market access depends upon whch nstrument s bndng. The general concluson s that one should reform the bndng nstrument: an expanson, for example, of the quota would not produce any ncrease n trade f the bndng nstrument s the n-quota or the out-of quota tarff, whle ths could well ncrease trade f the quota s bndng. oughner et al. (2000) have stressed the mportance of possble regme swtches followng TRQ lberalzaton. An expanson of the quota s fully effectve f after lberalzaton the bndng nstrument s stll the quota; f the quota expanson s such that the domestc prce determned by the mport quota alone becomes lower than the world prce plus the n-quota tarff, then the new bndng nstrument s the n-quota tarff. Therefore, any ncrease n the quota besdes the pont at whch ths regme swtch occurs s neffectve and rather a combnaton of quota expanson and n-quota reductons could be a more effectve lberalzaton strategy (de Gorter, oughner, 1999). Lberalzaton may also affect the amount of quota rents and, thereby, neffcences due to the exstence of quota rents. An expanson of the quota when ths s bndng may ncrease (reduce) rents f mport demand s suffcently elastc (nelastc), whle a reducton n the out-of-quota tarff always reduce rents f T s the bndng nstrument (Skully, 2001a). Hraanova and de Gorter (2006) have shown that, under certan crcumstances, wth lcence-on demand an expanson of the quota may reduce neffcences as the addtonal lcences go the lower cost frms. The model developed n the prevous secton s here used to analyse how consderaton of the mode of competton between frms may affect the effectveness of the varous TRQ lberalzaton strateges. Hence, we frst nvestgate the mpact of lberalzaton on the mode of competton and then, n the next secton, the trade, competton and welfare effects. Fve possble lberalzaton optons are taken nto account; 7 n addton to the reducton n T and t and the ncrease n Q, two further 7 The ssue of TRQ tarffcaton, that s, the elmnaton of the TRQ and ts replacement wth a tarff equvalent, under a duopoly capacty constraned model s addressed n Scoppola (2008). 8

11 scenaros are consdered. Frst, the government may reform the TRQ regme by mprovng the effcency of the allocaton system and ths results n a reducton of the cost l of obtanng lcences. Second, reforms of TRQs may also result n the entry of new frms (an ncrease n N ). For varous reasons, traders that are as effcent as the ncumbents may not enter the market because of varous barrers n obtanng lcences; the removal of these obstacles may mply an ncrease n the number of rvals on the market. The mpact of the dfferent lberalzaton strateges crucally depends upon the ntal equlbrum and, more specfcally, whether ntally there are out-of-quota mports. When under the TRQ q, q > Q, frms behave as under a tarff and the mode of competton s always ertrand. In general terms, f lberalzaton does not mply a regme swtch, none of the lberalzaton optons affects the mode of competton, although they may have trade and welfare effects: a decrease n t does not affect prces and quanttes, even though t ncreases frms profts and reduces taxpayers welfare. A reducton n T would reduce the prce, ncrease mports, and reduce taxpayers welfare; an expanson of the quota would not ncrease trade but would ncrease frms profts and reduce taxpayers welfare. The mpact of an ncrease n the number of frms does not change the mode of competton; the prce decreases and mports ncrease as a result of the hgher competton on the market. A reducton n l would not affect prce and quanttes, even though t would ncrease frms profts. In other words, mprovements n the admnstraton system may not have any trade effect f ntally there are out-of-quota mports or the capacty constrant s weak; rather, the only effect would be to ncrease traders profts. Thus, when ntally there are out-of-quota mports, the capacty constrant s not effectve and the predctons of ths model are rather smlar to those of prevous lterature. Thngs are dfferent when ntally there are no over quota mports ( q, q < ). Out of the Q fve possble lberalzaton methods, n four cases the lberalzaton of the TRQ may affect frms behavour. The mode of competton depends upon the value of the crtcal tarff T * wth respect to the out-of-quota tarff T. An ncrease n Q does not affect ether T * or T and, thus, does not change the behavour of frms, nor t have any trade or welfare effects. onversely, the reducton n t or l and the ncrease n the number of frms nfluence the value of T * and, thus, may mply a change n the mode of competton. The reducton of T, f T T*, may also change the mode of competton. The mpacts of the dfferent lberalzaton optons when ntally there are no out-of-quota mports are llustrated by means of a numercal example; although the choce over the values for 9

12 parameters clearly nfluences the magntude of the changes, t does not nfluence the drecton of change and, thus, the general fndngs of the smulatons. 8 The mpact of a reducton n t on the crtcal tarff s shown n Fgure 1. The reducton of the n-quota tarff n ths model mples, everythng else held constant, an ncrease n the effectveness of the capacty commtment, and thus a move towards the pure ournot outcome. As the n-quota tarff decreases, the crtcal tarff also declnes and may become lower than the out-of-quota tarff. In Fgure 1, the shft to a pure ournot equlbrum occurs when t < 3.2. The reducton n l has a smlar effect (Fgure 2), as t also strengthens the effectveness of the capacty commtments and, thus, may shft the mode of competton toward ournot. A decrease n T, everythng else held constant, reduces the gap between the costs of adustng capactes n the two perods and, thus, reduces the effectveness of the capacty commtment. Ths s shown n Fgure 3: when T becomes lower than T * there s a shft from ournot to ertrand. Fnally, the mpact of N on the value of the crtcal tarff s reported n Fgure 4. In ths example, wth N < 18 the quantty mported by each frm s lower than Q ; the mode of competton s thus determned by the value of the crtcal tarff whch, n our example, for low values of N ( N < 5 ) s lower than the out-of-quota tarff; thus, frms compete on quantty. As N ncreases the crtcal tarff ncreases and the equlbrum shfts to ertrand. When N 18 the reducton n the amount of lcences avalable to each operator s such that the equlbrum quantty becomes hgher than Q ; whatever the value of the crtcal tarff, ertrand competton prevals. Hence, the model suggests that f N s suffcently low ntally then, everythng else held constant, an ncrease n the number of rvals weakens the capacty commtment and the outcome shfts to ertrand. Postve competton and trade effects, whch would occur n any case wth the ncrease n N, therefore, are hgher than under the assumpton of a fxed mode of competton, as they are amplfed by the shft n the mode of competton. 9 The model shows that four out of fve polcy optons for TRQ lberalzaton affect frms market power; n general terms, ths declnes f the reform mples an ncrease n the number of frms, but market power ncreases f lberalzaton nvolves a reducton n costs, ether by reducng the nquota or the out-of quota tarff or by mprovng the effcency of the allocaton system. Ths mpact on frms market power can be amplfed or mtgated by possble changes n the mode of competton, as wll be seen n more detal n the next secton. 8 The followng values for varables and parameters are assumed throughout the paper (baselne scenaro): α = 400, β = 0.8, λ = 0.1 l = 3, m = 2, t = 5, T = 15, N = 15. The ndvdual quota s set at Q = 140 whch, n the baselne, s hgher than both q and q. 9 It s worth notng that these results hold for a wde range of values for parameters. As underlned n the appendx, a suffcent condton for Fgures 1 to 4 to behave as n ths numercal example, whatever the values of other parameters and varables, s that α m > 1. 10

13 4. Trade, welfare and competton effects of TRQs lberalzaton Trade and welfare effects under the varous optons for TRQs lberalzaton are reported n Table 2 together wth frms market power, measured by the Lerner ndex. In the numercal example, we assume that n the baselne there are no out-of-quota mports ( q, q < ); as aforementoned, Q when ntally frms mport out-of-the quota, the capacty constrant s not effectve and frms behave as under a tarff; the fndngs of the model are, by and large, smlar to those predcted by the prevous lterature. Table 2 does not consder the case of an ncrease n the quota Q because ths would have no effects on trade, welfare and competton. Fnally, lberalzaton n our example never mples a TRQ regme swtch; whatever the change n polcy, the equlbrum mports are always lower than the ndvdual quota. learly, a swtch n the regme would mply that: a) under certan crcumstances, the mode of competton may change, as mentoned n the prevous secton; and b) the mpact of lberalzaton would be that brefly depcted n the prevous secton and smlar to the predctons of other models. Table 2 also ncludes mpact of lberalzaton on rents. In ths framework there are two sources of rents: normal rents arsng from the exercse of market power by frms, but there are also addtonal rents due to the TRQ. The reason why n ths model addtonal rents arse s that the TRQ, by ntroducng a capacty constrant, nduces a less compettve envronment. Wth no capacty constrant lke under a smple tarff regme - the outcome of the game would be the one-shot ertrand wth margnal costs equal to ( m+ t) (.e. p ( m+ t) ). Under the TRQ, everythng else held constant, the prce s hgher than p ( m+ t) ; f the capacty constrant s weak, then the prce s p ( m+ T), whle when the constrant s strong the equlbrum prce s p ( m+ t+ l). Thus, the unt rent due to the exstence of the TRQ s the dfference between the equlbrum prce p ( m+ T) or p ( m+ t+ l) and the prce p ( m+ t). Table 2 reports the changes n the rents due to the TRQ. The reducton n t, when there are no out-of-quota mports, s usually mentoned as an example of effectve lberalzaton, as under perfect competton the prce s determned by the margnal costs. In ths model, the reducton of the n-quota tarff may or may not affect mports dependng on the mode of competton: f the outcome of the game s ournot, then the equlbrum prce p s negatvely affected by a decrease nt (equaton (10)) and mports ncrease; however, f the outcome s ertrand then a reducton oft has no trade effect, as the equlbrum prce s determned by the costs n the second stage of the game (T ). In our example, as Table 2 shows, a small reducton n the n-quota tarff (up to 20%) has no trade effects, because the mode of competton s ertrand. However, a larger reducton makes the capacty commtment more effectve and shfts the outcome of the game to the pure ournot soluton. Under ths mode of competton, frms behavour s drven by the margnal costs of ncreasng capacty n the frst perod and, thus, a decrease n t reduces prces 11

14 and ncreases mports. A small reducton of the n-quota tarff, even n the absence of trade effects, mproves the welfare of frms whle taxpayers are worse-off; frms market power ncreases as a consequence of the reducton n the costs. Wth a larger decrease n t, the prce declnes, mports ncrease and market power contnues to ncrease; nevertheless, consumers and frms gan because of the decrease n the prce and costs, respectvely. Despte the consderable losses n government revenues, total welfare mproves as a consequence of TRQ lberalzaton. It s worth notng that, unlke perfect compettve models, here rents due to the TRQ are postve even when the quota s not bndng; f l+ t < T, then the prce s always hgher than the one that would preval wthout the quanttatve restrant. A reducton n the n-quota tarff, by strengthenng the capacty constrant, even ncreases these rents. In perfect competton models, the reducton of the out-of-quota tarff s neffectve as the prce s determned by the margnal costs, t + l + m, whereas ths may not be the case n ths model. The reducton of the out-of-quota tarff may have trade effects f the mode of competton s ertrand, whle f ntally frms compete à la ournot t s neffectve. In Table 2, whent s hgh and the outcome of the game s ournot, a reducton n T does not affect prce and mports; however, a 30% reducton shfts the mode of competton away from ournot and, thus, further reductons ncrease mports and decrease prces. Unlke the predctons of perfect competton models, ths model suggests that, f the capacty constrant s weak, a reducton n the out-of-quota tarff may ncrease mports, even f there are no out-of-quota mports. There wll be welfare effects anyway, wth a shft of resources from taxpayers to frms that wll beneft from an ncrease n profts. Rents due to the TRQ do not change untl the mode of competton s ournot; however, f there s a swtch toward ertrand (.e. a sharp reducton nt ) then rents due to the TRQ reduce. Thus, unlke the prevous lterature, the mplcatons of ths model are that, f the reducton n T s hgh enough, then lberalzaton may produce postve results both n terms of market access and rents, even f there are no out-of-quota mports. Improvements n the admnstraton of the TRQ by lowerng cost l do not have any mpact on trade f the mode of competton s ertrand. In our example, ths happens f l s reduced by less than 30%; the prce and the quanttes do not change, but market power and profts of frms ncrease; taxpayers and consumers are not affected by lberalzaton and, as a consequence, global welfare mproves. On the contrary, when frms compete à la ournot, that s, when l s so low that the gap between the cost of adustng capactes n the two perods s very wde, then a reducton n l reduces prce and ncreases mports. As can be seen n Table 2, f l s reduced by more than 60%, frms, consumers and taxpayers all beneft from lberalzaton and overall welfare mproves consderably, despte an ncrease n market power from the reducton n costs. Rents due to the TRQ ncrease anyway as a consequence of the strengthenng of the capacty constrant. Thus, the man mplcaton of the model s that lberalzaton by means of an mprovement n the admnstraton of TRQs that leads 12

15 to a reducton n frms costs may be trade effectve only under certan crcumstances,.e. when the out-of-quota tarff s hgh enough - wth respect to the n-quota tarff plus the cost of obtanng lcenses - to make the capacty commtment effectve and frms compete à la ournot. Alternatvely, mprovements n TRQ admnstraton have no trade effects and yeld only an ncrease n rents. Reforms resultng n an ncrease n the number of rvals on the market always have sgnfcant postve effects, whatever the mode of competton. Table 2 shows how an ncrease n the number of frms, everythng else held constant, always reduces prce, ncreases mports, mproves overall welfare and reduces market power. Overall, reforms mprovng the ablty of frms to access lcences, by ncreasng the number of frms and the degree of competton on the market, may be a more effectve way to lberalze trade than reducng tarffs. Ths provdes a further argument for the dea that trade lberalzaton to be effectve should be mplemented along wth an mprovement of competton on nternatonal markets. 5. onclusons The am of ths paper was to nvestgate the mpact of TRQs and of ther lberalzaton when nternatonal agrcultural markets are domnated by olgopolstc traders. A two-stage capacty constraned model has been used n whch the mode of competton s endogenous and the constrant s flexble; the presence of a TRQ, n fact, ntroduces a capacty constrant to traders that, however, s flexble, as traders can always ncrease ther mports out of the quota, albet at a hgher cost. uldng on Magg (1996), the paper has shown how the mode of competton between frms and the fnal equlbrum depend upon the costs of adustng capactes n the two perods, that s, the values of the n-quota and out-of-quota tarffs and the cost of the lcences to mport wthn the quota. Despte the smplfyng assumptons, ths paper has shown how the lberalzaton of TRQs may have rather dverse results when olgopolstc competton s taken nto account. The most sgnfcant fndng of the paper s that consderaton of the nature of competton between traders could undermne a number of usual conclusons about the (n)effectveness of the varous lberalzaton optons. The paper has shown, for example, how under certan crcumstances (e.g. a weak capacty constrant), even f mports are lower than the quota, a reducton n the out-of-quota tarff s effectve, whle the reducton n the n-quota-tarff does not ncrease trade. These unconventonal results are bascally the consequence of takng nto consderaton strategc nteracton between frms. Although the olgopoly framework here used s a two-stage (fnte) game and cannot be consdered as a pure dynamc game, the two-stage structure, nevertheless, represents an ntermedate step toward dynamcs (Vves, 2001, p. 199). It s ths nherent, albet lmted, dynamc whch dves the results; the fndng that a reducton n the out-quota-tarff could be effectve even when there are no out-of-quota mports, s due to the fact that, f the capacty constrant s weak, then a frm s prcng decson s affected by the margnal cost rvals face n the second perod of the game, that s, the cost of mportng out-of-the quota, rather than those of the frst perod (the 13

16 cost of mportng n-quota). Ths result would not be obtaned n statc games, assumng ether ertrand or ournot. A second feature of the model developed here, whch helps to explan the results, s the endogenety of the mode of competton. Trade polcy may affect the nature of the competton among frms by changng the effectveness of the capacty commtment. Ths means that by changng trade polcy nstruments the n-quota and out-of-quota tarffs and the cost of lcences the government nfluences the strategc nteracton between frms. Ths has a number of nterestng mplcatons; frst, trade lberalzaton affects frms market power and, n some cases, t may even ncrease t; second, even f before lberalzaton a certan polcy change mght be effectve, t may become neffectve as a consequence of the lberalzaton tself, f ths leads to a change n the mode of competton. Overall, the fndngs of the paper suggest that, n markets where traders are concentrated and there s good reason to beleve that they exert market power, then lberalsng TRQs by reducng the tarffs may not produce the desred results n terms of mprovement n market access. Ths paper has shown that there are many cases - more than those suggested by the lterature assumng perfect competton - n whch ths s not the case. The man polcy mplcaton, therefore, s that n order to lberalze TRQs effectvely, one should not only consder the level of mports (actual or potental) wth respect to the quota, but also assess the nature of competton n the tradng ndustry, as ths may crucally nfluence the outcome of lberalzaton. The choce of whch nstrument to lberalse should then follow from ths. Fnally, the model has shown how a reform removng barrers to the entry of newcomers may be more effectve n lberalsng trade than a reducton of tarffs. Ths adds a further argument to those already supportng the vew that trade lberalzaton to be effectve needs to be accompaned by an ncrease of competton on nternatonal markets. References oughner, D., de Gorter, H. and I. Sheldon The Economcs of Tarff-Rate Quotas n the Agrcultural Agreement n the WTO, Agrcultural and Resource Economcs Revew 20, arter,., Mac Laren, D., 1997 Prce or Quantty ompetton? Olgopolstc Structures n Internatonal ommodty Markets, Revew of Internatonal Economcs 5, 3, de Gorter, H., oughner, D., 1999 U.S. Dary Polcy and the Agreement on Agrculture n the WTO, anadan Journal of Agrcultural Economcs, 47: Dong, F., Marsh, T., Stegert, K., State Tradng Enterprses n a dfferentated product envronment: the case of global maltng barley markets. Amercan Journal of Agrculture Economcs 88 (1), Gervas J.P., Suprenant, D., (2000) An Economc Investgaton of the Import Lcensng Methods and TRQs n Agrculture, anadan Journal of Agrcultural Economcs, 48: Hraanova, J., De Gorter H., Rent Seekng wth Poltcally ontestable Rghts to Tarffrate Import Quotas. Revew of Internatonal Economcs 13(4),

17 Hraanova, J., De Gorter H., Falck J., The Economcs of Admnsterng Import Quotas wth Lcence-on-demand n agrculture. Amercan Journal of Agrcultural Economcs 88, Kreps, D., Schenkman J., Quantty Precommtment and ertrand ompetton Yeld ournot Outcomes. ell Journal of Economcs 14, Lavoe, N Prce dscrmnaton n the context of vertcal dfferentaton: An applcaton to anadan wheat exports. Amercan Journal of Agrcultural Economcs 87: Magg, G., Strategc Trade Polces wth Endogenous Mode of ompetton. Amercan Economc Revew 86, Ramos, MP., ureau, J.., Salvatc, L Shppng The Good eef Out: EU Trade Lberalzaton to Mercosur Exports, WP Tradeag proect, Rude, J., Gervas, J-P Tarff-Rate Quota Lberalzaton: The ase of World Prce Uncertanty and Supply Management, anadan Journal of Agrcultural Economcs 54, Scoppola, M Tarffcaton of Tarff Rate Quotas under olgopolstc competton. The case of the EU mport regmes for bananas, WP Tradeag proect, Skully, D. 2001a. Economcs of Tarff-Rate Quota Admnstraton. Washngton D: Unted States, Department of Agrcultural, Economcs Economc Research, Servce Techncal ulletn Number 1893, Aprl. Skully, D. 2001b. Lberalzng Tarff-Rate Quotas, n Mary E. urfsher (ed) Agrcultural Polcy Reform n the WTO - The Road Ahead, Economc Research Servce, U.S. Department of Agrculture. Agrcultural Economc Report No Thursby, M., Thursby, J., Strategc trade theory and agrcultural markets: an applcaton to anadan and US wheat exports to Japan. In: arter,., Mcalla, A., Sharples, J. (Eds.), Imperfect ompetton and Poltcal Economy. The New Trade Theory n Agrcultural Trade Research.Westvew Press, oulder, O, pp Vves, X., Olgoply prcng. Old deas and new tools. The MIT Press: ambrdge Mass. WTO Tarff rate quota admnstraton and methods and tarff quota fll, Note by the Secretarat TN/AG/S/22, 27 Aprl 2006; WTO: Geneva. WTO Revsed draft modaltes for agrculture, TN/AG/W/4/Rev.3 10 July 200, WTO: Geneva. 15

18 Table 1. The outcomes of the game Q = Q /N Mode of competton ndng Instrument Mode of competton ndng Instrument Q q, Q > q ournot (1) t ertrand (2) T q < Q q ertrand (3) T q Q < q? (4)? l+t < T* T l+t T < T* Q < q, Q < q ertrand (5) T ertrand (6) T 16

19 Table 2: Trade, welfare and competton effects of TRQs lberalzaton Increase n N aselne 20% 40% 60% 80% 100% Prce Total mports Total profts onsumers' welfare Taxpayers' welfare Total welfare Market power Rents due to the TRQ Reducton n the n-quota tarff aselne 20% 40% 60% 80% 100% Prce Total mports Total profts onsumers' welfare Taxpayers' welfare Total welfare Market power Rents due to the TRQ Reducton n the out-of-quota tarff aselne 10% 20% 30% 40% 50% Prce Total mports Total profts onsumers' welfare Taxpayers' welfare Total welfare Market power Rents due to the TRQ Reducton n l aselne 30% 60% 100% Prce Total mports Total profts onsumers' welfare Taxpayers' welfare Total welfare Market power Rents due to the TRQ Note: n talcs when frms compete à la ournot 17

20 20 ournot ertrand 15 T T* t Fgure 1: The crtcal tarff as a functon of the n-quota tarff T* ournot ertrand 14 T l Fgure 2: The crtcal tarff as a functon of the cost of obtanng lcences T ertrand ournot T * 5 Fgure 3: The out-of-quota tarff and the mode of competton 18

21 18 17 ournot 16 ertrand T* T Q > q Q < q N Fgure 4: The crtcal tarff as a functon of the number of frms 19

22 Appendx a) The ournot equlbrum The maxmzaton problem s the followng: max Π = q ( α βq λ q t l m) q Frst order condtons are: Π = α 2 βq λ q ( t + l + m) = 0 q From whch we obtan the reacton curve α ( t+ l+ m) λ q q = 2β The equlbrum quantty s gven by: q = α ( t+ l) λ from whch: q α ( t+ l) λ q 2β ( α ( t+ l+ m)) = 2 β + λ( N 1) 2β y substtutng the equlbrum quanttes n the demand (1) the ournot prce (10) s obtaned. b) The ertrand equlbrum The maxmzaton problem s: max Π = ( a bp + c p )( p T m) p Frst order condtons are: Π = a 2 bp + c p + b( T + m) = 0 p from whch we obtan the reacton curve: p = a+ c p + b( T + m) 2b Gven the symmetry assumpton: 20

23 a+ c( N 1) p+ b( T + m) p = 2b from whch a+ b( T + m) p = 2 b c( N 1) c) The crtcal tarff The crtcal out-of-quota tarff T * s defned mplctly by p ( m, T) = p ( m, t, l). The closed form expresson for T * s: β ( l+ t) + βλ( 5+ 3 N)( l+ t) + λ ( N 1)( α m+ l( N 3) 3 t+ Nt) T* = 2 β + βλ(3n 5) + λ ( N 3N + 2) whch can be wrtten as: Ζ + +Α T* = Β+ 3 λn( β λ) + λ N wth T * > 0 ( l+ t) ( l t) N( l t) 2 λ ( α m) 2 2 λ N ( l t) Ζ= 2β + 3λ 5βλ, λ(3 β λ(( α m) 4)) 2 2 Α= +, Β= 2β + 2λ 5βλ Gven that β > λ, Ζ> 0. A suffcent condton for Α to be nonnegatve s that α m 1. The varables l and t postvely affect the value of the crtcal tarff f: 2 2 T* Ζ+Α N + λ N = > ( l+ t) Β+ 3 λn( β λ) + λ N Gven that Ζ> 0 and that the denomnator s always postve, a suffcent condton for s Α 0. The mpact of N on T * depends upon demand parameters, costs and N. The functon T*( N ) can be wrtten as: Ω+Ψ N +ΣN T*( N) = W +Μ N + ϑn 2 2 Provdng that λ 2, l+ t and λβ ( λ) are greater than zero, then Σ, Μ and ϑ are postve. Functon T*( N ) behaves as n Fgure 4 f Σ, ΜΨ, and ϑ are all greater than zero. Thus, a suffcent condton for T*( N) to behave as n Fgure 4 s that Ψ > 0, that s, Α > 0. 21