State Trading Enterprises as Non-Tariff Measures: Theory, Evidence and Future Research Directions

Transcription

1 State Trading Enterprises as Non-Tariff Measures: Theory, Evidence and Future Research Directions Steve McCorriston (University of Exeter, UK) Donald MacLaren (university of Melbourne, Australia) Paper presented at the Global Trade Analysis Conference, Geneva, 7 th -9 th June, 01 April, 01

2 State Trading Enterprises as Non-Tariff Measures: Theory, Evidence and Future Research Directions 1. Introduction State trading enterprises (s) are widely used aong the ajor agricultural iporters and exporters including, inter alia, Canada, China, India, Indonesia, Japan and South Korea, aong any others. The concerns with s are that, in iporting arkets, they inhibit arket access while, in the exporter context, they provide unfair advantages aong copeting exporters. If these concerns are justified, then s can be viewed as non-tariff barriers to achieving undistorted trade and, in principle, their effects can be easured in the for of tariff or export subsidy equivalents. The potential for s to distort trade has been recognised in the current OECD MAST initiative (see van Tongeren et al. (009)). However, the treatent of s in this context is inadequate, the characterisation of an being liited to its onopoly status and thus its classification as anti-copetitive. This characterisation is overly-siplistic because it does not fully account for the ways in which s ay distort trade, if at all. The deterinants of this trade distortion turn out to be ore coplex than the onopoly characterisation and, as such, this does not fully capture the heterogeneity of s as they exist across countries and coodity sectors. Therefore, s in the iporting country context are better described by the ore neutral ter non-tariff easure. Nevertheless, assessing their trade-distorting effects pose significant conceptual and easureent challenges which we address in this paper. We highlight the challenges in analysing the issue in the context of s as a nontariff easure in agricultural trade and identify the key deterinants of their tradedistorting effects. These deterinants are not only exclusive rights but also the objective function of the, its possible co-existence with other instruents, its efficiency relative to that of private firs, the characterisation of the doestic arket (covering both distribution and procureent) and the nature of the non- benchark. As such, the treatent of s in the MAST initiative is overly-siplistic and does not recognise sufficiently the iportance of these key features of s as they are currently eployed in agricultural trade. Taking these issues together, suggests that at the heart of the analysis is a coparison of equilibriu outcoes across alternative characterisations of arket 1

3 structure. However, we ephasise that these effects nevertheless can be suarised in tariff equivalents (and export subsidy equivalents for the corresponding exporting case) that provide qualitative and quantitative easures of the effect of s on trade when easured with respect to a given counterfactual arket structure. A discussion of these challenges is followed by the outline of a theoretical fraework that explicitly allows these easures to be derived. It is iportant to note that a sall country with perfect copetition in all arket segents is only one possible coparator benchark As well as outlining a fraework to deal with iporting s in various guises, a related objective of the paper is to present soe evidence on the potential trade distorting effect of a specific. This involves a case-study of the wheat sector in Japan. The ain point to establish is that it is possible to address the trade-distorting effects of s via a echanis that is transparent (in ters of the underlying concerns about the alternative arket structures that ay replace the ), which can accoodate partial refors to s, and where other features of the specific coodity sector ay be iportant, for exaple, doestic price support. The final aspect of the paper suarises the ain features of aking the trade-distorting features of s ore transparent and, in doing so, highlights the inadequacies of the current classification of the in MAST. Future directions for research on s are also highlighted. In the latter context, we highlight how the presence of an affects the nature and extent of risk in coodity arkets, an issue that is of growing interest and concern given recent events on world coodity arkets. The paper is organised as follows. In Section, we detail the status of s in the context of the WTO. In section 3, we outline the ain challenges in addressing the issue which is followed by a presentation of a ore foral theoretical fraework that highlights these challenges. Although s can arise on both the export and iport sides of arkets, we focus here on iporting s 1. We eploy this fraework to derive forally tariffequivalent easures of their trade-distorting ipact. In Section 4, we apply this fraework to a case study of the Japanese wheat sector. In the final section, we suarise and conclude and outline future directions in dealing with issues. 1 We have analysed exporting s elsewhere see McCorriston and MacLaren (007).

4 . State Trading Enterprises in the Context of the WTO In the General Agreeent on Tariffs and Trade (GATT 1947) it was recognised that state trading enterprises had the potential to distort international trade because of governent involveent in their activities. Thus the intention in Article XVII of the GATT was to ake s behave in accordance with noral coercial considerations through defining rules that constrained how they could behave legitiately (WTO, 01b). It is iportant to note that 'coercial considerations' are not synonyous with 'profit axiisation'. This distinction is a fundaental one because, as is discussed below, s have a range of objectives, only one of which is profit axiisation. Without this distinction between coercial considerations and profit axiisation, s could never behave as private firs would and hence the intention of the Article would be unachievable. Nevertheless, what constitutes coercial considerations reains iprecise. With Article XVII drafted to ensure that s behaved in the sae way as private firs with respect to their effects on international trade, it is not surprising that "the WTO does not seek to prohibit or even discourage the establishent or aintenance of state trading enterprises, but erely to ensure that they are not operated in a anner inconsistent with WTO principles and rules." (WTO, 01a). Despite this reassurance, the behaviour of s engaged in international trade in agricultural products continues to cause soe disquiet aongst governents that do not use s because they suspect that, in practice, the activities of s do have the potential to restrict iports or to expand exports. However, these alleged distortions ake sense only if the counterfactual has been defined, a point that is developed below in sections 3 and 4. While in GATT 1947, and especially in Article XVII, the rules that s were to follow had been clearly specified, exactly what kind of entity qualified as an had not been defined. This deficiency, aongst others, was rectified in GATT 1994 in the In the report of the Appellate Body that dealt with the coplaint by the United States that the Canadian Wheat Board was acting in a anner inconsistent with Article XVII, it was ade clear that 'coercial considerations' did not ean 'profit axiisation' as the United States had argued. For details of the legal arguents, see WTO (004). It is interesting to note that the third parties to this dispute were Australia, Chile, China, Chinese Taipei, the European Union, Japan and Mexico, the iportant users of s at that tie being Australia, China and Japan. 3

5 Understanding on the Interpretation of Article XVII where a working definition is provided as follows: Governental and non-governental enterprises, including arketing boards, which have been granted exclusive or special rights or privileges, including statutory or constitutional powers, in the exercise of which they influence through their purchases or sales the level or direction of iports or exports. There are two iportant features of this definition that distinguish an fro a private fir. First, it is not ownership per se that atters, but the nature of the exclusive rights or special privileges that are bestowed on an entity. Second, it is the effect of these exclusive rights or special privileges on trade flows that is of significance. Consequently, the organisations that can be characterised by this definition range fro state-owned governent entities that have onopoly rights with respect to doestic sales and, by extension, onopsony rights with respect to procureent, through to a nuber of privately-owned iporting firs that have been given iport licenses but that otherwise copete with a private sector on sales in the doestic arket. 3 In order to ensure transparency about the existence, objectives and functions of s, Mebers are required to notify the Council for Trade in Goods and to provide regular updates. 4 The inforation is collated by the Working Party on State Trading Enterprises (WTO, 011) and it ensures that s are not being used in ways that are inconsistent with the WTO. Transparency is also enhanced through the ability of Mebers to ake counternotifications, i.e., to notify the Council for Trade in Goods of the existence of other Mebers' s that have not been notified by the latter. In suary, the GATT-legal status of s is now better understood than it was prior to the establishent of the WTO in However, it is only recently that they have been subject to econoic analysis that perits their trade-distorting effects to be easured and which provides a fraework within which to easure the tariff equivalent of this particular for of non-tariff easure. 3 Fro notifications ade by governents to the WTO about, inter alia, the existence and objectives of the s, a nuber of distinct types of can be identified. These include statutory arketing boards, regulatory arketing boards and canalizing agencies (WTO, 01b). In practice, they operate to achieve given policy objectives and often do so in conjunction with other policy instruents. For a discussion of the types of, the objectives and the eans used to pursue the, see OECD (001). 4 Details of the Mebers that have notified are to be found in (WTO, 011, Table 1). In 1995, 59 countries out of 97 notified the WTO of their s. 4

6 3. Deterinants of the Effects of s on Market Access Analysing the effect of state trading enterprises on trade poses several challenges. In this section, we outline the nature of these challenges and how they are likely to ipact on identifying the trade distorting effect of s. As noted above, the focus here is on the iport case; the analytical issues nevertheless spill over to the export case. We first suarise the essential challenges in dealing with the trade effects by posing a series of questions that constitute the necessary ingredients for the theoretical fraework. This is followed by a ore detailed exposition of a theoretical odel that encopasses these features 5. The Main Issues (i) What is the benchark arket structure? With the presence of the, the arket structure has been deterined by the governent. In the siplest case, an ay be the sole actor on the arket such that it has both onopsony and onopoly status. The case is not restricted to this specification as private firs could be licensed by the governent to operate in (segents of) the arket. But it poses the obvious question: if there were no, what would the underlying arket structure look like? Perceptions of this structure ay differ, ranging fro a copetitive outcoe (i.e., if the were disbanded, there would be a large nuber of private firs that would enter the arket and there would be no arket power distortions either in procureent or in sales) through to concerns that the non- outcoe would be doinated by private firs that could exert arket power (in the siplest case, the onopoly/onopsony would be replaced by a private onopoly/onopsony). This uncertainty about the arket structure in the counterfactual has two iportant iplications for the odelling of the trade effects of s. First, in addressing the antionde arket structure, we want to allow for these different perceptions to exist so that the analyst does not ipose an unsuitable arket structure and pre-deterine the outcoe. In doing so, this suggests a fraework such that the ipact of s can be easured and where the analysis can also reflect debate about these potentially different perceptions. Second, the odelling fraework has to be sufficiently general and transparent to allow for 5 This discussion and the theoretical fraework that follows draw on a series of papers: see, in particular, McCorriston and MacLaren (005); addressing iporting s in a norative context, see McCorriston and MacLaren (011a); the doestic and trade distorting aspects of s are addressed in McCorriston and MacLaren (011b). 5

7 these differences and where the tariff-equivalent effect can clearly reflect a range of copetitive outcoes ranging fro the private onopoly/onopsony case through to the copetitive case. (ii) How do the objectives of the differ fro those of the private fir? A siple characterisation of the issue is that because arket structure is anipulated by the governent, the arket structure issue is all about the nuber of firs. Since in the ost straightforward case, the arket is characterised by the existence of the only, the issue reduces to the trade ipact of the existence of a onopoly/onopsony agent doinating the arket. Indeed, this could be the case; but it is overly-siplistic. s, and the anipulation of arket structure that the presence of the involves, are essentially instruents of governent policy and, as such, the pay-off function of the ay reflect the underlying ais of governent policy. So, while it is reasonable to assue that the pay-off function for the private fir is the axiisation of profits, the pay-off function for the could reflect welfare axiisation, or re-distribution in favour of producers or consuers, or the axiisation of profits. Alternatively, it ay reflect soe weighted cobination of the arguents in a welfare function. In this context, the issue parallels the industrial organisation/public econoics literature where the objective of a public fir is different fro the characterisation of a private fir since the pay-off function differs between these two cases 6. The iportant point then is that the issue is not just about the nuber of firs but also about the differences in the nature of the pay-off function between these private fir/ cases. This difference has a crucial ipact on the outcoe. If the (single) private fir is a arket interediary between consuers and producers, a profit axiising fir will exploit both onopoly and onospony power on both sides of the arket. This would then be reflected in two arket distortions because both doestic procureent and sales will be too low. However, a welfare axiising will take these two distortions into account, resulting in procureent and sales both being higher, procureent prices greater and consuer prices lower copared to the profit axiising case. In this setting, the welfare axiising replicates the copetitive outcoe and, therefore, this can hardly be classified as uncopetitive. Note that the nuber of interediaries in these two cases 6 The welfare axiising objectives of the public fir is evident in theoretical approaches as surveyed by de Fraja and Delbono (1990) though these odels deal alost exclusively with the closed econoy case. 6

8 equals one but the outcoe is very different because of the characterisation of the pay-off function differs across both cases. However, the private fir case ay lead to ore iports copared with the case. even though there is only a single arket interediary in both exaples, since there is ore doestic procureent with the welfare axiising and a lower level of iports because the exploits the ters of trade distortion. Most governent intervention in agricultural arkets does not, of course, reflect the ai of welfare axiisation but rather re-distribution in favour of farers or consuers. Thus, the pay-off function for the ay reflect the governent's weighted welfare objective or welfare bias, the bias reflecting the overall nature of governent policies. The nature of this bias will therefore partially offset the (but not necessarily full) extent of the procureent or consuption distortions that ay otherwise exist with a profit-axiising single interediary. In turn, the bias will affect the levels of procureent and consuption and, by extension, the ipact on trade and arket access. (iii) The Characterisation of Exclusive Rights It is often assued that state trading iplies state ownership. This is not the crucial characteristic associated with state trading in the context of the WTO as has been discussed above in Section. The ain characteristic relates to exclusive rights. Thus, while a state trading enterprise ay indeed be state owned, a private fir (or a given nuber of the) ay be defined as a state trading enterprise if it has been allocated exclusive rights by the governent. It is how these exclusive rights subsequently ipact on trade flows and arket access that is at the heart of the concerns of the WTO. Of course, when these exclusive rights interact with the re-distributive characterisation of the 's pay-off function, and how the trade outcoe copares with the outcoe in the private fir benchark, only then are we able to address the broader concerns associated with s. Exclusive rights ay cover all or segents of a given arket. Suppose the iporting country can be specified where there are two sources of procureent for the raw product: doestic production and iports. These two sources of the coodity are then distributed to consuers. We assue that the private firs/ act as interediaries between the procureent arkets and consuers. At the consuer stage, the product ay or ay not be differentiated. The nature of exclusive rights deterines which segents of the arket the can operate in. There are three obvious possibilities: (i) the has exclusive 7

9 rights over both sources of procureent and, in turn, over distribution to consuers; (ii) the has exclusive rights over iports but only private firs procure fro the doestic agricultural sector such that the and the private firs copete in the distribution arket; and (iii) the has sole rights over doestic procureent but private firs can procure iports and, like the previous case, both private firs and the copete in the distribution arket. The latter case would be the ost obvious case where we have, say, a doestic arketing board but note, that even though the is not directly involved in iports, it is still a concern in a WTO context since the exclusive rights that apply in the doestic context can still potentially affect arket access. In the forer case, the has joint exclusive rights whereas in the latter two cases, the has partial exclusive rights; while in the forer, it is only the that characterises the arket, in the latter two cases, the co-exists with private firs. Note also where there are exclusive rights and where the pay-off function for private firs and the can differ, we can characterise the arket as a ixed oligopoly. There are possibly ore coplex characterisations of exclusive rights. For exaple, in the allocation of tariff-rate quotas, the can co-exist with private firs (China and the role of COFCO would be an exaple of this case) 7. Alternatively, the state trader ay co-exist with private firs in procureent in the doestic arket but have exclusive rights over iports (India and the role of the Food Corporation of India would fit with this characterisation). (iv) How do you easure the tariff equivalent effect? The discussion in points (i)-(iii) has highlighted soe of the central concerns in odelling an -related arket and how these outcoes ay copare with the private sector antionde. As should now be clear, the issue is not only about nubers per se but it also relates to the potential differences in the pay-off functions of the and private firs and to the designation of exclusive rights that deterine in which segents of the arket the can operate. It is the cobination of these three characteristics of s that lead to the potential ipact on trade. The one reaining challenge is to capture the potential trade distorting effects of s in a single tariff equivalent easure. 7 See McCorriston and MacLaren (010a, 010b) for an assessent of COFCO on international coodity arkets. 8

10 There are two parts to this. The first is to be clear that when we are addressing the issue (with the various coponents of it that we have detailed above), we are essentially coparing alternative arket structures. Assue, for exaple, we had an with a producer-biased pay-off function that had joint exclusive rights. To answer the question about the trade-distorting effect, we are essentially asking what is the level of arket access that would arise in this particular characterisation of the arket copared with one which was characterised by the existence of private firs only? Given the discussion above, this private fir only case ay be ore or less copetitive. More directly, we are coparing how arket access differs across alternative and discrete characterisations of arket structure. Siilarly, with an that is consuer-biased and has exclusive rights to iport only, the specific question is what is the level of arket access that arises in this characterisation of arket structure copared with the private sector benchark? It is the coparison across these discrete characterisations of arkets and how the level of arket access varies across each case which is at the heart of the conceptual issue in assessing the trade distorting effect of s. Given this discrete difference in arket structures, how does one capture the trade distorting ipact in a single easure? The concept we eploy is to define an iplicit tariff that will bring about a correspondence in the level of iports across these alternative characterisations of arket structure. For exaple, suppose we had a producer-biased that had joint exclusive rights; we can solve out for the level of iports that would arise in this specification of the arket. We then ask the question: what is the level of the iplied tariff that would have to be iposed on private firs in the anti-onde case to give the sae level of iports as in the characterisation? The level of this iplicit tariff then becoes the tariff-equivalence easure. There are several advantages to this easure. First, it can bring about a concordance for any given characterisation of an (reflecting both exclusive rights and the pay-off function) with the private fir benchark. Hence, it can capture a large part of the heterogeneity of s that exist both within (i.e. across coodity sectors) and across any iporting countries. Second, we are not prescriptive about the nature of the underlying benchark: if one analyst/trade negotiator has a different perception of what the arket would look like in the absence of the fro another, we can capture these different perceptions in our tariff-equivalent easure. Third, we can allow for other policy 9

11 instruents that ay exist or alternative features of the iporting country in question. For exaple, the iporting country ay be sall or large: the fraework used can capture this and will be reflected in the iplicit tariff easure. Siilarly, if the country eployed doestic price support, again the fraework is flexible enough to account for this feature. Finally, but arguably ost iportantly, the iplicit tariff equivalent can be easured. This is principally done by eploying partial equilibriu odels that can be calibrated with country-specific data, or at least data which are broadly reflective of a given agricultural arket. While, of course, there ay be arguents about the precision of the tariff equivalent easures in these cases, the arguent and discussion can then focus on the factors that are ost likely to ipact on the trade-distorting outcoe that arises fro the presence of the however specified. With this easureent of the trade-distorting effect coes several benefits. The ost obvious is transparency. Associated with this transparency coes the ability to copare across different characterisations of s either by coodity sector or by country, or both. We can also address the question of refors to s. Across any countries, the role of s has been frequently changed, either the change in the pay-off function and/or the change in exclusive rights that apply to the. So, when refor has been undertaken, it does not necessarily involve the coplete de-regulation of s such that the arket looks like the private fir outcoe but rather where the s exclusive rights change and it co-exists with private firs. By easuring the changes in the iplicit tariffequivalent easure, we can then approxiate the trade liberalising effect of these changes to arket structure which refor involves. In the exaple below, we highlight these issues with reference to the role of the in the wheat sector in Japan. Putting These Challenges into a Foral Fraework These ain challenges are analysed in a foral fraework that allows easureent of the trade-distorting effect of s. We start with identifying the pay-off functions for the. (i) The Pay-Off Function The welfare function, W, for the is given by: W = α1 PS+ α CS+ α 3π which, noralising on α 3, can be re-written as: 10

12 W = α p PS+ αccs+ π (1) where PS is producer surplus, CS is consuer surplus, π is the 's profits and the αs are the policy weights. We leave for the present the specific characterisation of the source of profits, but note that if α α = 0, we have an that acts like a private fir and p = c solely axiises profits. Alternatively, for α α = 1, we have a welfare axiising p = c. Adjusting the weights in the welfare function will therefore reflect the overall bias of governent policy; with α > 1, α < 1, reflecting re-distribution towards producers (as in p c developed countries) and α < 1, α > 1 reflecting re-distribution towards consuers (as in any developing countries). p c (b) Identifying the Ipact of s The principal ai is to copare arket access with the with what would have been the case in the private fir benchark. This benchark range fro onopoly/onopsony or, to varying degrees, oligopoly/oligopsony, or it can converge on a copetitive outcoe. Specifically, we pose the question, what would the iplicit tariff that would have to be iposed on private firs to induce the to reproduce the outcoe with the? To see this, consider the following profit functions for a representative doestic fir where subscript h ( ) relates to profits fro the sale of doestically-procured (iported) agricultural coodities, p ( p ) is the retail price of the doestically-procured h (iported) coodity, q ( q ) is the quantity of doestically-procured (iported) h coodity and p A and p w are far-level and world prices respectively. Then, π = ( p p ) q h h A h e π = ( p p t ) q w () The variable e t is the iplicit policy easure we will use to identify the trade distorting effects of the which we will solve for explicitly in the following section. This variable is defined as the iplicit tariff equivalent that would have to be iposed on private fir iports to replicate the level of iports that would arise in the state trading case ( Q ). Explicitly, aggregating over n private firs (such that effect solves: e Q t = Q = nq ), the tariff equivalent ( ) Q (3) 11

13 To illustrate the intuition, take a siple exaple. Suppose the private fir set-up were characterised by a sall nuber of firs that could exert oligospony power in the procureent of agricultural supplies and arket power in the sale of the in the doestic arket. By restricting the level of doestic procureent, farers are potentially worse off, the ability of the private firs to do this being greater because they can price discriinate between the doestic and iport arkets in the procureent of these supplies. Against this, consider the effect of an and assue the objective function of this is to axiise producer surplus and profits fro the sale of doestically-procured and iported coodities in the doestic arket. Assue also the has joint exclusive rights, i.e., it has sole rights over doestic procureent, iports and the sale of the coodities in the doestic arket. In this case, doestic procureent by the ay increase beyond that procured by the private firs;. if iports decrease as a consequence,, there will be a positive tariff equivalent effect. Clearly, the agnitude of the trade distorting effect will depend on the objective function of the and the nature of the private fir benchark the replaces. In general, however, the sign of t e is abiguous. In addition, these effects will also depend on the nature of exclusive rights that apply to the, i.e., the extent to which it has arket power in the procureent arkets and in the sale of coodities to doestic consuers. These are considered next. (c) Exclusive Rights In the private sector benchark, a representative fir chooses how uch to procure doestically and how uch to iport before final sale to consuers. Total profit π (in equation ()) is ade up of the two coponents π h and π. Products sourced fro different arkets ay or ay not be differentiated. Assuing arket segentation, the representative fir can therefore act as a discriinating oligopsonist in the procureent arket and as an oligopolist in the output arket. As the nuber of copeting firs ( n ) increases, the ability to exert arket power in either of these arkets diinishes. Assuing Cournot behaviour, the representative fir (i) chooses quantities to axiise joint profits as given by: π = π + π = ( q (4) i hi i e p h p A ) qhi + ( p pw t ) These exclusive rights of the can take several fors though here we deal with the two ost obvious. First, we assue that the has sole rights in the procureent of both i 1

14 doestic and iported coodities and in the sale of these coodities. The extent to which it acts as a textbook onopsonist/onopolist will depend on the nature of its objective function as given by (1) with the profit coponent of the welfare function detailed in equation (1) for this being given by: 8 h h A h w π = π + π = ( p p ) Q + ( p p ) Q (5) An alternative characterisation of an is where it has sole rights to iport and it is excluded fro doestic procureent. The private sector procures and sells doestic output and copetes with the in the output arket. Thus the has exclusive rights over iports but, depending on the size of n, the output arket ay be oligopolistic. The size of n will also deterine the extent of oligopsony power exerted against doestic producers. But, in contrast to the previous case, neither the private firs nor the can price discriinate in the procureent arket. In this case, the profit function for a representative private fir, i, is given by: π = ( p p ) q (6) hi h and, for the, the profit coponent of the welfare function in (1) is given by: A w hi π = ( p p ) Q (7) Clearly, there are cases of s which ay differ fro the above two cases. For exaple, the ay have sole rights in the procureent of iports but it has to copete with the private sector in the doestic procureent arket. As an alternative, the private sector ay also be responsible for procureent of iports either exclusively or in copetition with the. While these variations can be dealt with readily in the proposed fraework, we focus on the two cases above as these are the ost transparent in ters of the analytics and are also readily applicable to the any exaples of iporting s currently being used. Deriving the Tariff Equivalent Effect The objective is to derive the tariff equivalent effect ( e t ) in a fraework that allows us to consider the issues discussed above. To proceed, we derive the first-order conditions for the private fir case inclusive of the iplicit policy instruents and derive equilibriu quantities. Then, for a given characterisation of the and, accounting for the general welfare function given in equation (1), we derive its corresponding first-order conditions and the equilibriu quantities. We then set the equilibriu quantities in the n private fir 8 The standard onopoly/onopsony case would arise only with α =α = 0. p c 13

15 benchark and the to equal each other, and derive an explicit expression for the tariff equivalent. Given that our ai is to derive explicit easures of the trade distortion, we assue a specific functional for. 9 Assue utility is given by: U = + u Q h, Q ) (8) ( where is the outside good and u( Q, Q ) is quadratic and is given by: h u( Q, Q ) = a Q + a Q 0.5( b Q + b Q + γ Q Q ) h 1 h 1 h h The inverse deand functions that are derived fro this utility function are given by: p p a b Q γq h = 1 1 h (9) a b Q γq = h (10) where: subscripts h and refer to the hoe produced and iported good respectively; b b γ > 0 iplies that the goods are not perfect substitutes; Qh = nqh and Q = nq 1 represent sales of the doestically produced and iported good respectively; q h and q are the quantities of goods procured in hoe and iported, respectively, and sold in hoe by the typical fir; and n is the nuber of copeting firs. To capture the potential for arket power to be exerted in the procureent arket, we assue upward-sloping, inverse supply functions. For the doestically-procured coodity, this function is given by: p = f + (11) A kq h and for the iported good the function is: p = F+ (1) w KQ If K = 0, we have the sall country case and there is no potential for ters of trade effects in the purchase of iports. However, there ay still be an effect on iports because the quantity procured doestically by the will differ fro that of the private firs and, hence, will affect the quantity iported. 9 Pursuing the approach with general functional fors does not develop the analysis to a stage that can be readily ipleented. This is because we evaluate the doestic and trade effects on specific levels of iports, doestic procureent and doestic sales. In order to derive these for the cases that we investigate, explicit values for the volues procured and sold need to be derived. Therefore, to avoid adding unnecessary algebra to the paper, we proceed directly to the case with a specific functional for. 14

16 (i) Private Fir Benchark Given the profit function in (4), the first-order conditions for profit axiisation for a representative i fir are given by: ( b + k)( n+ 1) γ ( n+ 1) γ ( n+ 1) q + K)( n+ 1) q a1 f = a F t 1 h ) e ( b Aggregating over n firs, equilibriu quantities are given by: (13) e φ ( a1 f ) γ ( n+ 1)( a F t ) Q h = nqh = n φ3 (14) e φ 1 ( a F t ) γ ( n+ 1)( a1 f ) Q = nq = φ3 (15) 3 1 ) where φ1 = ( b1 + k)( n+ 1), φ = ( b + K)( n+ 1) and φ = φφ γ ( n+1. (ii) with Joint Exclusive Rights The axiises the weighted welfare function given in (1) with the profit coponent as given in (4). The first-order conditions for the are: b1 ( α c ) + k( α p ) γ ( α c ) γ ( α c ) Q h b + K ( α c ) Q a = a 1 f F Note that the relative weights on producer and consuer welfare are captured in these firstorder conditions. It should also be noted that, if the were solely interested in axiising profits and acted like a private fir, then the left-hand side atrix of (16) would be identical with the left-hand side atrix in (13) for n = 1. In this case, the iplicit policy easure would be zero and there would be no difference between the and the private onopoly/onopsony. Fro equation (16), the corresponding equilibriu quantities procured and sold by the are given by: Q Q h λ ( a1 f ) γ ( α c )( a F) = (17) λ 3 3 λ1 ( a F) γ ( α c )( a1 f ) = (18) λ where λ1 = b1 ( α ) + k( α ), λ = b ( α ) + K and λ = λλ γ α. c p c 3 1 ( c ) (16) 15

17 (iii) with Exclusive Rights to Iport Only In this case, the has sole rights to iport, it is excluded fro doestic procureent but has to copete for sales with the private sector that can only procure doestically. The first-order conditions in this case relate to the profit function for a representative fir given by (6) and the 's welfare function by (1) but with the profit coponent given by (7). This gives: b ( αc ) + K γ n(1 αc ) Q a F = (19) γ ( b1 + k)( n+ 1) q a h 1 f Aggregating over the n firs that can procure only doestically, the equilibriu quantities are given by: ' [ b ( α c ) + K]( a1 f ) γ ( a F) Qh = nqh = n (0) λ4 Q φ ( a F) nγ (1 α )( a f ) = (1) ' 1 c 1 λ4 where λ4 = [ b ( αc ) + K] φ1 n(1 α c ) γ and where φ 1 is defined as above and with the prie distinguishing equilibriu quantities fro the earlier two cases. Given the equilibriu quantities derived above, we now can proceed to derive the explicit easures that identify the trade distorting effects of s. We do this for the two characterisations of the s. (iv)ipact of an with Joint Exclusive Rights To derive the tariff equivalent effect, using the definition in (3), set (18) equal to (15) and solve out for gives: t e JE e t, where the subscript (JE) refers to the joint exclusive rights case. This ( a F)( nλ3φ1 φ3λ1) ( a1 f ) γ[ nλ3( n+ 1) φ3( α c )] = () nλφ 3 fro which it can be concluded that the tariff equivalence is a function of n, α and α. 1 p c (v) Ipact of an with Exclusive Rights to Iport Only The sae procedure as above can be used to derive the corresponding easures when the has exclusive rights to iport only and it copetes in the output arket with the private sector that procures its output fro the doestic agricultural sector. The 16

18 corresponding tariff equivalent easure is given below, where the subscript (MO) refers to the iport only case: t e MO ( a F)( λ4nφ1 φφ 1 3) ( a1 f )( λ4nφ1 φφ 1 3) = (3) nλ φ fro which it can be concluded that the tariff equivalence is no longer a function of 4 1 α p. (vi) s and Other Governent Policies Of course, s are not the only instruent of governent policy and they are seldo used in isolation. Rather, the anipulation of arket structure often goes hand-in-hand with other governent distortions aied at re-distributing incoe. For exaple, a governent ay use price support to guarantee a iniu price to producers. In such cases, the iportant issue to address is the arginal effect of the, i.e., the nature of the transfers caused by the when other distortions are accounted for. The above fraework can be readily adapted to deal with such cases. Consider, for exaple, the trade distorting effect that would arise in the case of the with joint exclusive rights. With a guaranteed far level price, we can aend the inverse supply function to be given by k = 0 and f with the latter being set at the guaranteed price level. Re-writing () we have: t e JE ' ' ' ' F)( nλ3φ1 φ3λ1 ) ( a1 f ) γ[ nλ3( n+ 1) φ3( α c )] ' ' nλ3φ3 ( a = ( ) ' ' ' where φ 1 = b1( n+ 1), φ 3 = φφ 1 γ ( n+1), λ1 = b1( α c ) and λ3 = λλ 1 γ ( αc ) with all other variables as given above. The tariff-equivalent effect is now a function of f as well as n and the policy weights. ' ' ' Intuitively, the ipact of the is changed in the presence of other instruents. Consider, for exaple, the role of the with no price support. Relative to an uncopetitive private sector, the corrects the distortion that would arise fro too uch buying power in the procureent arket, with a producer-biased increasing procureent. As such, since for a producer-surplus axiising it is the average rather than the arginal outlay curve that atters, it increases doestic procureent, giving rise to a positive producer subsidy equivalent. But with a guaranteed price, the inverse supply and arginal outlay functions are flat (at least in a certain range) so that this distortion does 17

19 not need correcting to the sae extent by the. The can only effectively exert its control on the arket via its procureent of iports and the sale of doestic and iported coodities to consuers. Since the effect of the is now on exercising ters of trade effects and onopoly power in the doestic arket, it ay procure less than the n > 1 private firs. Given that an is essentially associated with anipulating arket structure to achieve a given objective, the role of price support has already (partially) fulfilled this role and the arginal effect of the is reduced. 4. Case Study: Ipact of s in the Wheat Sector in Japan The theoretical odel gives iportant insights into the key factors that likely deterine the ipact of s in iporting countries 10. However, following the tradition of the trade policy literature, the above odel can be calibrated using price and quantity data and assued values for the key elasticities. The calibration follows fro the work of Dixit (1988). The iportant point to note is that the fraework offers a potentially useful tool with which policy akers can iprove the transparency of the trade effects of s that can arise in different environents. Irrespective of the specific results, we nevertheless show that the direction of the effects is consistent with the discussion provided above. The odel is calibrated using data relating to the wheat arket in Japan. This is an iportant case study for several reasons. First, Japan is a key player in the on-going Doha Round negotiations and is coonly seen as a country where there is liited arket access for iported coodities, that the governent s policy is strongly targeted at the interests of producers and away fro consuers and where it has used state trading enterprises to anage the procureent of doestic and iported agricultural coodities and their subsequent doestic sale. Specifically, the Japan Food Agency in the past has been the doinant feature of the rice and wheat arkets in Japan. Second, in recent years, the Japanese governent has changed the exclusive rights that apply to the. In detail, the Japan Food Agency that prior to 00 had exclusive rights to procure doestic and iported wheat for sale on the doestic arket has now exclusive rights to iport only and the private sector now procures doestically-grown wheat and copetes with iported wheat in the output arket. However, it should also be noted that the Japanese governent also guaranteed prices to farers with the Japan Food Agency procuring at governent set 10 For ore detailed propositions that arise fro this fraework, see McCorriston and MacLaren (011a). 18

20 prices. Coinciding with the changes to the exclusive rights that applied to the, the Japanese governent also reoved this price support with additional copensation now being given to farers via the Incoe Stabilisation Fund. As such, the refor of the in Japan can be characterised as a ove fro the with joint exclusive rights coexisting with price support to an with rights to iport only with no price support. 11 We calibrate the paraeters of the theoretical odel based on price and quantity data for the Japanese wheat arket in 000. These data are shown in Table 1 along with assued values for the key elasticities. However, as the previous discussion above shows, the ipact of the is highly dependent on the bias in the welfare function which, in this case, likely reflects the overall bias in Japanese agricultural policy. To this end, we use the estiates of Lee and Kennedy (006) who evaluate the relative weights on producer and consuer welfare that appear consistent with the wheat policies pursued by the Japanese governent. Therefore, α andα are assued to equal 1.5 and 0.75, respectively. We p c also set n=10 and, assue in the case where the has iport rights only, the arket coprises the sae nuber of participants in the output arket, i.e., n=9 plus the. The calibrated paraeters are also shown in Table 1. Based on the calibrated paraeters, we can therefore evaluate the effects of the using (), ( ) and (3) above and highlight how the doestic and trade effects of the Japan Food Agency have changed following aendents to the nature of exclusive rights that apply. In considering how these effects relate to the change in exclusive rights, we assue that the bias in Japanese agricultural policy has reained unchanged. The results are reported in Table. We report three cases: first, where the Japan Food Agency with joint exclusive rights co-existed with doestic price support; second, assuing the joint exclusive rights were aintained but price support was reoved; and third, where we have the Japan Food Departent with iport rights only and doestic price support has been reoved. There are two iportant insights that arise fro these results. First, the Japan Food Agency did indeed cause a trade distortion, although the extent of it was significantly reduced following the change in exclusive rights that apply. When the had joint exclusive 11 Since these payents are lup-su in nature, they do not create any additional distortions and so are not odelled directly here. 19

21 rights, the tariff equivalent was around $517 per tonne this aounts to around a 86 per cent of the world price. Following the change in the exclusive rights, this effect would have fallen to a tariff equivalent of $34 per tonne or 57 per cent ad valore. Second, in the joint exclusive rights case, with the governent guaranteeing prices, the inverse supply curve is flat and there is no oligopsony proble for the to correct. Nevertheless, on the consuer side the arket is less copetitive so overall sales go down. As a result, the procures less fro doestic producers, less is sold to doestic consuers, iports are lower and so the trade distorting effect is positive. Note, however, that in the case where we have the with joint exclusive rights without doestic price support, the negative effects on arket access also increase, the tariff equivalent is now $688 per tonne (115% ad valore). Table 1: Calibration Data and Paraeters Japanese Wheat Market, 000 Paraeter Value Calibrated Value Paraeter Deand elasticity 0.5 b Elasticity of substitution 5 b Doestic retail price $903/tonne γ Retail price of iported $103/tonne a good Sales of doestically tonnes a 5115 produced coodity Sales of iported good tonnes k Export supply elasticity 5 f Doestic supply elasticity 0.5 K.764E-05 Doestic producer price $800/tonne F Iport price $596/tonne Data on doestic production, sales and iports coes fro FAO. Iport prices are calculated as unit values fro iport value and volue data fro the FAO. Doestic producer and retail prices are sourced fro USDA. No available studies produce deand and supply elasticities though the values chosen are broadly coensurate with elasticity data in other developed countries. Table 6 deals with sensitivity analyses with respect to these chosen paraeters. 0

22 Table : Doestic and Trade Effects of the Japan Food Agency in the Wheat Market (US$ per tonne) Joint Exclusive Rights with doestic price support Joint Exclusive Rights with no doestic price support Tariff Equivalent 517 (86%) 688 (115%) Iport Only 34 (57%) Table 3 reports the results fro a sensitivity analysis of the initial calibration. The first row reports the results fro the initial calibration and the assuptions associated with it. Throughout the reainder of the table, we report how the tariff equivalent effects vary when the underlying assuptions are changed. In the case where joint exclusive (JE) rights applied, the ipact on trade, the copetitiveness of the underlying benchark and the assuption about the doestic deand elasticity atter ost. When the can iport only (MO), the tariff equivalent easure is particularly sensitive to the underlying copetitiveness of the arket. Finally, it should be noted that the exercise reported here is not intended as a definitive evaluation of the effects of the that applies in Japan. Rather, and drawing fro the theoretical odel outlined above, the iportant point is that a theoretically-consistent fraework can be used to evaluate the effects of s that arise in different environents. For exaple, by changing the relative weights in the objective function, we can capture the likely bias of policy in developing countries, and by capturing explicitly the exclusive rights that apply, the odel can be applied using observable data to increase the transparency of the likely effects of s and, in turn, infor policyakers of the trade effects of these s. 1

23 Table 3: Doestic and Trade Effects of the Japan Food Agency-Sensitivity Analysis (US$ per tonne) Note: Tariff Equivalent JE nps MO Benchark Case a More Copetitive Doestic Market b Weaker Ters of Trade Effects c More Elastic Doestic Deand d JE JE ps and nps refer to the cases where the has joint exclusive rights but with and without doestic price support respectively. MO refers to the case where the has exclusive rights to iport only. a In the benchark case, we use the calibrated data presented in Table 1. We also assue n=10 and α = 1. 5 and α = We vary the underlying benchark with the sae calibrated paraeters p c or keep the sae nuber of private firs and re-calibrate the odel to allow for different values for the elasticity data. b Nuber, n, of firs in benchark set equal to 50. c Elasticity of iport supply set equal to 0. d The elasticity of deand is increased fro 0.5 to Suary and Future Research Issues The principal ai of this paper was to outline the challenges in addressing state trading enterprises as non-tariff easures. s indeed have the potential to restrict arket access and therefore act as a non-tariff barrier to trade. However, this does not necessarily arise only because an has single desk status. To put it bluntly, to assue that the issue is solely to do with the nuber of participants in the arket, as in the treatent of s in the MAST prograe, is overly siplistic. There are any factors that contribute to the trade-distorting outcoe of s. The ai of this paper has been to outline the several challenges in analysing the trade effects of s and to present a fraework that can incorporate these various challenges. Iportantly for policy analysis, the fraework lends itself to a single and transparent tariff-equivalent easure that can be applied to arkets in which state enterprises are to be found. Even if there are different perceptions regarding what the non- arket ay look like, the fraework outlined here can nevertheless accoodate these perceptions as well as accoodating other features of the arket that also ipact in assessing the trade-distorting aspect of s, e.g., the presence of doestic price support easures, and the 'sall' versus 'large' country assuption.