2.1 The case study: Peru's price range system

During the 1990s, the Peruvian economy underwent a massive program of structural adjustment and economic liberalization. As part of this program, and in conjunction with Uruguay Round commitments, Peru lowered its tariffs and other trade barriers on agricultural products (Trivelli, Reference Trivelli2003: 5). This trade liberalization exposed Peruvian farmers to increased competition from world markets.

Worried about the negative impact on its farmers, Peru established a PRS in June 2001.Footnote ² The PRS covered 45 agricultural products, which were separated into four general categories: milk, yellow maize, rice, and sugar.Footnote ³ The PRS operated as follows:

• • Within each category, there was one tariff line designated as the ‘marker’ product and all others are designated as ‘associated’ products.

• • A floor and ceiling price were determined for each marker product on the basis of monthly average f.o.b. (freight on board) prices for the past 60 months on the international reference market for that marker product. These prices were to be updated semi-annually and converted to a c.i.f. (cost, insurance, and freight) floor and ceiling prices.Footnote ⁴

• • In addition, a reference price was to be calculated every two weeks, reflecting the average international market price for the product.Footnote ⁵

• • The reference price was then compared to the c.i.f. floor and ceiling prices to determine whether a ‘variable additional duty’ should be levied or a ‘tariff rebate’ should be issued.

Whenever the international reference price (denoted by p^R) fell below the c.i.f. floor price p, a ‘variable additional duty’ was levied against the imported good. Letting b denote the import costs associated with marker products, the additional variable duty AD was calculated according to the following formula:

$$AD = (1+b)\; ({\underline p} - p^R)$$

Assuming that the additional cost associated with the variable additional duty was passed on to the consumer, this policy kept Peruvian prices for the good above world market prices. It therefore protected Peruvian farmers from negative price shocks on the global market.

In contrast, whenever the reference price rose above the c.i.f. ceiling price $\bar p$, a tariff rebate was issued to the importer. The tariff rebate was calculated as follows:

$$TR = \left( {1\; + \; b} \right)\; \left( {\,p^R - \bar p} \right)$$

Assuming that the tariff rebate was passed on to the consumer, this policy kept Peruvian prices for the good below world market prices. It therefore served to protect Peruvian consumers from undesired price increases on the global market.

The stated objectives of the Peruvian PRS were three-fold:Footnote ⁶

• First, the PRS was designed to counter the adverse effects to domestic agricultural production on account of ‘distortions … due, in particular, to the agricultural policies implemented by the main food producing and exporting countries’ and as ‘reflected in the uncertainty and instability of domestic prices’.

• Second, the PRS was ‘a stabilization and protection mechanism that makes it possible to neutralize the fluctuations of international prices and limit the negative effects of the fall in those prices’.

• Third, the PRS constituted ‘an appropriate means of improving the levels of competitiveness of domestic producers, by giving the market clear signals with regard to trends in prices, thereby allowing economic agents to operate efficiently and productively’.

In this sense, Peru's PRS serves as a quintessential example of a small developing country's government's employment of an instrument to safeguard against undesired price fluctuations associated with increased agricultural trade liberalization. Peru's Ministry of Economy and Finance noted that the PRS was intended to protect both domestic producers and consumers, through the establishment of a floor and ceiling prices for calculating a variable additional duty or tariff rebate, as necessary, to fulfill its various stabilization and competitiveness objectives.Footnote ⁷

2.2 Potential WTO legal controversy arising due to an agricultural PRS

A PRS, however, encounters potential problems when examined in the context of the WTO's Agreement on Agriculture. The drafters sought to limit the types of border measures designed to restrict the volume or distort the price of imports of agricultural products.Footnote ⁸ To that end, they decided it would be best to require that border measures and other non-tariff barriers designed to restrict agricultural imports be converted into ordinary customs duties.

Article 4.2 of the Agreement on Agriculture states: ‘Members shall not maintain, resort to, or revert to any measures of the kind which have been required to be converted into ordinary customs duties, except as otherwise provided for in Article 5 and Annex 5.’ Article 4.2 includes a footnote 1 that provides an illustrative list of the types of measures that might violate this provision. These include ‘quantitative import restrictions, variable import levies, minimum import prices, discretionary import licensing, non-tariff measures maintained through state-trading enterprises, voluntary export restraints, and similar border measures other than ordinary customs duties’.

To date, the bulk of WTO cases interpreting Article 4.2 have concerned quantitative import restrictions applied against agricultural imports.Footnote ⁹ The PRS, however, does not act as a quantitative restriction. Instead, the two types of measures that it most closely approximates are (1) variable import levy, and (2) minimum import price.Footnote ¹⁰

Through 2013, only one WTO dispute, Chile–Price Band System, has opined on these issues. The Appellate Body (AB) found that Chile's PBS, which shared similarities with Peru's PRS, was inconsistent with Article 4.2 of the Agreement on Agriculture.Footnote ¹¹

2.3 Background on the Guatemala‒Peru FTA and subsequent WTO dispute

Anticipating potential legal problems with its PRS, Peru sought guarantees from its PTA partners that they would not challenge the PRS through WTO dispute settlement. One example is the Guatemala‒Peru FTA, signed on 6 December 2011. Article 9 of Annex 2.3 explicitly recognizes Peru's right to maintain its PRS.Footnote ¹² The FTA also clarified that in the event of an inconsistency between the FTA and WTO agreements, the FTA ‘shall prevail to the extent of the inconsistency, unless otherwise provided in this Treaty’.Footnote ¹³

More than a year later, neither party had ratified the treaty. On 12 April 2013, Guatemala proceeded to file a WTO challenge, alleging that Peru's PRS violated several WTO commitments:Footnote ¹⁴ These included:

• • Article 4.2 and footnote 1 of the Agreement on Agriculture, to the extent that the additional duty constitutes and encompasses elements of a variable import levy, a minimum import price, or a measure similar to the two;

• • Article II:1(b) of the GATT 1994, since the measure constitutes a duty or charge other than an ordinary customs duty that was not included in Peru's schedule of concessions;

• • Article X:1 and X:3(a) of the GATT 1994, regarding failures in the publication and administration of the trade regulations related to the PRS.

Should the Panel find that the duties arising from the PRS are ordinary customs duties, Guatemala argued, in the alternative, that Peru's actions would then amount to a violation of Articles 1, 2, 3, 5, 6, and 7 of the Customs Valuation Agreement.

Why did Guatemala bother to marshal resources to challenge Peru's PRS when it had previously agreed to recognize its legality in the bilateral FTA? The overall economic stakes were not large.Footnote ¹⁵ Based on our analysis of available data, bilateral trade was significant in only two sugar products subject to the PRS.Footnote ¹⁶ The WTO litigation appears then to reflect the lobbying power of Guatemala's sugar producers.Footnote ¹⁷

By the time the Panel was composed in September 2013, neither party had yet ratified the FTA, although Guatemala's Congress gave approval two months earlier. Guatemala initiated ratification procedures in December 2013, with Peru receiving a communication on 5 March 2014.Footnote ¹⁸ At that point, had Peru's President simply initiated ratification procedures and sent notice to Guatemala, the FTA would have entered into force. Peru's FTA ratification process does not require Congressional approval, only Presidential action.Footnote ¹⁹ Peru's President refused, viewing the filing of the WTO dispute as having ‘created uncertainty with regard to the existence of the balance negotiated’ in the FTA.Footnote ²⁰ Thus, the bilateral FTA, as of this writing, has not yet entered into force.

On 27 November 2014, the Panel issued a report finding Peru to have violated Article 4.2 of the Agreement on Agriculture because the duties arising from the PRS constitute a variable import levy that was not permissible under the Agreement.Footnote ²¹ In addition, the Panel also ruled that the duties arising from the PRS violated GATT Article II:1(b) because they are not ordinary customs duties and were not declared as other duties or charges.Footnote ²² The Panel exercised judicial economy with respect to the GATT Article X:1 and X:3(a) claims.Footnote ²³ Finally, the Panel did not find it necessary to rule on Peru's argument that the FTA modified the rights between the two parties with respect to the PRS, resulting in Guatemala's waiver of its WTO rights concerning the PRS; the Panel emphasized that the FTA had not yet entered into force and therefore its provisions were not legally binding.Footnote ²⁴ Both Peru and Guatemala filed a notice of appeal. The Appellate Body ruling, issued on 20 July 2015, largely upheld the Panel's rulings except with respect to the Panel's ruling concerning the similarity of the PRS to a ‘minimum import price’ within the meaning of the Agreement on Agriculture.Footnote ²⁵

3.1 A stylized model of a small importing nation

Suppose Peru institutes a PRS under which it adjusts the additional duty applied to an imported commodity (call it sugar) so as to ensure that the domestic (tariff-ridden) price always lies within the interval $[{\underline p}, {\bar p}] $. Let the world price p ^w of sugar fluctuate between three different states.Footnote ²⁶ In the high state (i = H) p ^w = p _H, in the medium state (i = M) p ^w = p _M, and in the low state (i = L) p ^w = p _L, where the following relationships hold (i) p _H ⩾ p _M ⩾ p _L, (ii) $p_H \gt \bar p$, (iii) $p_M \in [{\underline p}, {\bar p}]$, and (iv) p _L < p. Thus, by design, the high price p _H exceeds the upper bound ${\bar p}$ of the PRS, the low price p _L lies below the lower bound p while the medium price p _M lies strictly inside the interval $[{\underline p}, {\bar p}]$.

Let θ_i denote the probability of state i so that we have:

(1)$$p^w {\rm \;} = \left\{ {\matrix{ {\,p_L \; {\rm with\; prob}\; {\rm \theta} _L} \cr {\,p_M \; {\rm with\; prob}\; {\rm \theta} _M} \cr {\,p_H \; {\rm with\; prob}\; {\rm \theta} _H = 1 - {\rm \theta} _L - {\rm \theta} _M} \cr}} \right.$$

Let AD _i denote the additional duty applied by Peru in state i. Then, the variable duty imposed by Peru under its PRS is given by:

(2)$$AD{\rm \;} = \left\{ {\matrix{ {AD_L = \; {\underline p} - p_{\rm L} \; if\; p^w = \; p_{\rm L}} \cr {0\; if\; p^w = \; p_{\rm M}} \cr {AD_H = \; {\bar p} - p_{\rm H} \; if\; p^w = \; p_{\rm H}} \cr}} \right.$$

Observe that AD _H < 0 < AD _L; i.e. the variable levy is positive when the world price is low, whereas it is negative when it is high. Finally, when the world price falls within the price band $[{\underline p}, {\bar p}]$ established by the PRS, the additional duty imposed by Peru equals zero.Footnote ²⁷

Treating Peru as a small open economy (i.e. as a country that takes world prices as given), the domestic (tariff-included) price in Peru under various states of the world is simply the sum of the world price and the duty applied:

(3)$$p = \left\{ {\matrix{ {p_L + AD_L = {\underline p} \; if \; p^w = \; p_L} \cr \quad {p_{\rm M} \; if\; p^w = \; p_{\rm M}} \cr {p_H + \; AD_H = {\bar p}\; if\; p^w = \; p_H} \cr}} \right.$$

Thus, by design, the variable duty imposed ensures that the domestic price always stays within the PRS band [$\bar p$, p].

Let domestic (inverse) demand function in Peru for sugar be given by:

(4)$$d\left( p \right) = a - \; p$$

and the domestic supply function by:

(5)$$s\left( p \right) = \; p$$

Then, Peru's import demand curve for sugar is the difference between domestic demand and supply:

(6)$$m\left( p \right) = d\left( p \right) - s\left( p \right) = a - 2p$$

In the absence of trade, we would have m(p) = 0 which yields p* = a/2 as the autarkic equilibrium price in Peru. Given the nature of the dispute, we assume m(p _i) > 0; i.e. Peru imports the good under all states of the world, an outcome that is guaranteed by the inequality p ^H < p*, which we assume holds.

Given a domestic price p, the surplus consumers enjoy from the good equals:

(7)$$CS = \mathop \int \limits_p^a \left( {a - p} \right)dp = \displaystyle{1 \over 2}\left( {a - p} \right)^2 $$

so that expected consumer welfare can be written as:

(8)$$ECS = \mathop \sum \limits_i \theta _i CS\left( {\,p_i} \right)$$

Similarly, the surplus of domestic producers equals:

(9)$$PS = \mathop \int \limits_0^p pdp = \displaystyle{1 \over 2}p^2 $$

which gives expected producer surplus as:

(10)$$EPS = \mathop \sum \limits_i \theta _i PS\left( {\,p_i} \right)$$

The tariff revenue collected by Peru under each state of the world equals:

(11)$$TR = \left\{ {\matrix{{t_L m( {\underline p} ) = ( {\underline p} - p_L )m( {\underline p} )\; if\; p^w = \; p_L} \cr {0\; if\; p^w = \; p_{\rm M}} \cr {\; t_H m\left( {\bar p} \right) = - \left( {\,p_H - \bar p} \right)m\left( {\bar p} \right)\; \; if\; p^w = \; p_H}\cr}} \right.$$

which yields expected tariff revenue as:

(12)$$ETR = \theta_L ( {\underline p} - p_L )m( {\underline p} ) - \theta_H (\,p_H - {\bar p})m\left( {\bar p} \right)$$

We can now write down Peru's expected total welfare under the PRS as a weighted sum of consumer surplus, producer surplus, and tariff revenue:

(13)$$EW = \mathop \sum \limits_i \theta _i {[CS\left( {\,p_i} \right) + \lambda PS\left( {p_i} \right)]} + \lambda [ {\theta_L ( {\underline p} - p_L} )m( {\underline p} ) - \theta_H (p_H - {\bar p} )m( {\bar p} )]$$

where the parameter λ ⩾ 1 measures how much extra weight the government puts on producer interests and tariff revenue relative to consumer surplus. When λ = 1, the government weights all components of welfare equally, and, as λ increases, the relative weight placed on consumer welfare declines while that placed on other components increases.

Using the formulae in (6) through (13), we can directly calculate:

(14)$$\displaystyle{{\partial EW} \over {\partial {\underline p}}} \vert_{\lambda = 1} = - 2\theta_L \left( {\underline p} - p_L \right) \lt 0$$

i.e. if the government weights all components of welfare equally (λ = 1), then it is optimal to reduce the lower bound of the PRS to its minimum possible value. This implies that it is optimal for Peru not to impose any minimum price on the commodity. Similarly, we have:

(15)$$\displaystyle{{\partial EW} \over {\partial \bar p}} \vert _{\lambda = 1} = 2\theta _H \left( {\,p_H - \bar p} \right) \gt 0$$

i.e. it is optimal for Peru to raise the upper bound of the PRS band. It is worth emphasizing that equations (14) and (15) imply that in the standard trade model it is optimal for a small importing nation to allow maximum variability in its local price if its objective is to maximize expected national welfare.

Thus, in the canonical trade model of a small open economy, the mere presence of price fluctuations does not generate a welfare-based rationale for the use of a PRS like the one employed by Peru so long as the objective of the government is to maximize expected aggregate welfare. The intuition for this result is that the gains from trade enjoyed by an open economy depend upon whether and how much the world price of a commodity differs from its autarkic price. In our simple model above, the autarkic price equals p* = a/2. Imagine the world price equals Peru's autarkic price, i.e. p ^w = p* = a/2. Would it gain from trade? The answer is a clear no since domestic production and consumption under free trade would exactly equal that under autarky. Now, consistent with the dispute, suppose p ^w < p*. Then, under free trade, Peru would import this commodity and enjoy gains from trade. While domestic producers would lose, the increase in consumer surplus would dominate their loss. Indeed, the lower the world price relative to Peru's autarkic price, the larger would be its gains from trade. A PRS band supported by fluctuating duties reduces the extent to which the domestic price in Peru differs from its autarkic price and therefore reduces its overall gains from trade.

The careful reader may observe that the above discussion seems to ignore the fact that the duties applied raise revenue. While revenue is part of domestic welfare, for the case of a small open economy, the burden of an import duty is borne entirely by local consumers ‒ see the formula for the domestic price in equation (3). As a result, from an aggregate welfare perspective, the variable duty system instituted to support the PRS simply redistributes income within Peru. When p ^w = p _L, the duty t _L raises the domestic price in Peru to p and redistributes income from consumers to producers and the government; whereas when p ^w = p _H, the rebate t _H benefits local consumers at the expense of the other two groups by lowering the domestic price to $\bar p$.

Figure 1 in the appendix provides a graphical illustration of why the expected welfare of a small importing nation declines when it puts in place a PRS to lower domestic price variability. When p ^w = p _L, the duty t _L raises the domestic price to the minimum acceptable price p and consumer surplus declines by the amount U + V + W + X. Domestic producer surplus increases by U and government collects revenue equal to area W. The net welfare loss suffered by Peru when p ^w = p _L then equals U + W − (U + V + W + X) = −V − X, which is simply the deadweight loss of the duty t _L. Now consider the case where p ^w = p _H so that the rebate t _H lowers the local price to $\bar p$. Then, when p ^w = p _H consumers gain the area A + B + C + D + E due to the rebate that lowers the domestic price to p; producers lose A + B; while the government loses B + C + D + E + F in order to cover the subsidy payments needed to lower the local price to p. Adding these gains and losses together, we find that the net welfare effect of the PRS on Peru is once again negative and it equals − (B + F). Thus, regardless of whether the world price is below or above the PRS interval $[ {\underline p}, {\bar p} ]$, the net welfare of a small importing nation declines due to the PRS.

Figure 1. Effects of the PRS system on a small open economy

The above analysis has focused on the case where the government cares equally about all components of welfare (i.e. λ = 1). In the real world, it is possible that the government weighs producer interests and tariff revenue more than consumer surplus. Suppose the importing government weighs producer interests and revenue more than consumer surplus (i.e. λ > 1). Then, the first-order conditions $\displaystyle{{\partial EW} \over {\partial {\underline p}}} = 0$ and $\displaystyle{{\partial EW} \over {\partial \bar p}} = 0$ can be solved for the endpoints of the optimal PRS band. We have:

$${\underline p}^{\ast} = \displaystyle{{2\lambda p_L - \left( {\lambda - 1} \right)a} \over {3\lambda - 1}}$$

where p* is increasing in λ and:

$${\underline p}^{\ast} - p_L = \displaystyle{{\left( {\lambda - 1} \right)\left( {a - p_L} \right)} \over {3\lambda - 1}}$$

from which it follows that:

$${\underline p}^{\ast} \gt p_L \Leftrightarrow \lambda \gt 1$$

so that it is indeed optimal for the government to impose a PRS that raises the minimum price p _Lto p* if it weighs producer surplus and revenue more than consumer surplus. However, at the same time, we have:

$${\underline p}^{\ast} - p_H = \displaystyle{{\left( {\lambda - 1} \right)\left( {a - p_H} \right)} \over {3\lambda - 1}}$$

i.e. if λ > 1, it is optimal for the government to also raise the maximum price p _Hto p*. In other words, if p* > p _L, it must also be the case that p* > p _H. Thus, the mere presence of political economy pressures that cause the government to weight producer interests (and revenue) relatively more than consumer welfare cannot yield a PRS that raises the domestic price when the world price is low while also lowering it when the world price is high. Allowing for λ > 1 only causes the government to raise the domestic price in all states of nature so that it cannot account for a PRS that offers a rebate to local consumers when the world price is high.

3.2 Some other relevant economic considerations

Given the results yielded by the standard model of a small importing nation, it is worth asking whether certain modifications in the assumptions of this model can make a PRS welfare-improving. The model makes two important assumptions that may be suspect in the context of agricultural trade. First, it assumes that all agents are risk neutral. In the context of agricultural commodities, consumers as well as producers might be risk averse. Second, the standard model assumes that production is completely flexible and can be adjusted up and down seamlessly depending upon the observed world price. We next investigate whether altering these assumptions can provide a rationale for why a small importing nation might want to lower fluctuations in the domestic price via import tariffs.

Consider first the role of risk aversion. In their analysis of Chile‒Price Band System, Bagwell and Sykes (Reference Bagwell and Sykes2004) note that since a PRS insulates the domestic economy from price fluctuations it would be attractive to local producers (as well as the government) if producers care not only about the average or expected price but also about its volatility.Footnote ²⁸ The standard model of a small open economy presented above ignores the effects of price volatility since all agents (producers, consumers, as well as the government) are assumed to be risk neutral and their welfare can be evaluated in terms of their expected payoffs.

Indeed, greater price variability is actually beneficial in the standard model because it increases a country's expected gains from trade by causing the world price to diverge more from its autarkic price. If risk-averse producers are willing to sacrifice high price episodes in order to avoid low price ones, a PRS can be optimal from a domestic viewpoint. However, since a PRS suppresses the overall expected gains from trade, for a PRS to increase welfare the benefits of reduced price variability would need to be sufficiently large ‒ i.e., a small amount of risk aversion on the part of local producers will generally not be enough to overcome the lower expected gains from trade implied by reduced price variability under a PRS.Footnote ²⁹

A well-established strand of the rather extensive literature on price stabilization measures argues that to identify the full impact of such measures one needs to examine their effect on not just the country instituting the system but also the rest of the world. The model presented above cannot capture such external effects since, by definition, a small country has no impact on the rest of the world. Bagwell and Sykes (Reference Bagwell and Sykes2004) have argued that for the case of a large importing country, not only could risk-averse domestic producers be better off under a PBS relative to a fixed conventional tariff but the trading partner may also prefer a PBS if it leads to a lower average tariff. But, as Bagwell and Sykes (Reference Bagwell and Sykes2004) note, while reducing price volatility in the domestic market a PBS may simultaneously amplify fluctuations in the net price received by foreign exporters. Thus, for such a system to be optimal from a joint welfare perspective, domestic producers would need to be more risk averse than foreign exporters so that shifting risk abroad would be efficient. Bigman (Reference Bigman1987) considers a two-country framework in which the free trade relative price is endogenous and shows that the use of variable levies by one country causes price instability to be exported from its market to that of its trading partner, on the basis of which he argues that variable levies can ‘destabilize thy neighbor’. In other words, reduced variability in one country's market is achieved at the expense of increased variability in its trading partner's market. If this spillover leads the trading partner to also institute a price stabilization measure of its own, then both countries can be worse off relative to a scenario where neither uses such measures. Indeed, price variability can even increase when price stabilization measures are used by both countries relative to when they are completely absent, thereby making such measures counter-productive.Footnote ³⁰

As Turnovsky (Reference Turnovsky1974) notes, the classical argument in favor of price variability in competitive markets formalized by Oi (Reference Oi1961), Massell (Reference Massell1969), and Waugh (Reference Waugh1966) rests on an important assumption: i.e. though producers (and consumers) face uncertainty, they make supply and demand decisions based upon actual prices whereas in the real world many critical production decisions need to be made before the resolution of price uncertainty. This is particularly true in the context of agricultural commodities where farmers make a variety of decisions well before actual prices for their crops are revealed. Thus, while the results of the simple model presented above clearly illustrate the benefits of price variability, they do not adequately capture the impact of uncertainty on production decisions. Eaton and Grossman (Reference Eaton and Grossman1985) show that free trade is not optimal for a small open economy facing uncertain terms of trade provided some factors of production are immobile ex-post and insurance markets are incomplete. In such an environment, tariff protection can serve as an incomplete substitute for the lack of insurance. In a similar vein, a PRS like the one instituted by Peru can help insure that the welfare of consumers and producers does not fall below a certain minimum threshold. But as Eaton and Grossman (Reference Eaton and Grossman1985) note, trade policy intervention does not constitute a first-best solution to the problem of income redistribution; for example, an income tax is preferable because it can redistribute income without distorting prices faced by domestic agents.

To summarize, although the canonical model of a small open economy suggests that a PRS is always welfare sub-optimal, this may not be the case once certain underlying assumptions of this model are altered. In particular, the economic justification for a PRS in a small open economy turns on whether producers and/or consumers are risk averse, factors of production are immobile, and insurance markets are incomplete. In the presence of such factors, a PRS may prove to be welfare-enhancing, albeit it is not necessarily the first-best option.

Variable import levy

In Chile–Price Band System, the AB explained the nature of a ‘variable import levy’ which is disallowed under WTO rules. Variability alone is not determinative.Footnote ³¹ The AB declared that ‘the presence of a formula causing automatic and continuous variability of duties is a necessary, but by no means a sufficient, condition’.Footnote ³² Instead, the AB emphasized the need for ‘additional features’ including ‘a lack of transparency and a lack of predictability’ that ‘contribute to distorting the prices of imports’.Footnote ³³

A first question raised by this line of jurisprudence is the frequency with which a duty must change in order for it to be considered variable. The AB's past language refers to a formula which changes ‘continuously’. The Panel found that Peru's PRS, which employed a mathematical formula that resulted in duty changes every two weeks, met this condition.Footnote ³⁴ But suppose that the reference price employed in the PRS formula were tweaked monthly, quarterly, or annually? At what point, if any, would the change be infrequent enough such that the PRS formula would no longer meet the ‘automatic and continuous variability’ requirement?

A second question is whether the introduction of some degree of discretion in the application of the formula would matter. What if the PRS formula did not result in a set duty, but instead merely set a range from which authorities could choose? Suppose the PRS formula, instead of arriving at an absolute additional duty, churned out a range, mandating that the additional duty lie between the interval [AD _L, AD _H], where AD _L represents the lower bound for the additional duty and AD_H represents the upper bound. The upper and lower bound for the range are to be calculated through the use of different coefficients (c _L, c _H), to be added to the formula discussed in Section II.1, where c _L < c _H. In other words, AD _L = (1 + b + c _L)(P _floor − P _r) and AD _H = (1 + b + c _H)(P _floor − P _r). The implementing authority, whether an agency within the executive branch or the legislature itself, is free to select any duty rate within that range as the additional duty. The same notion of an upper and lower bound would exist for the tariff rebate. After selecting the duty/rebate rate, the authority would then publish the rate in a gazette for a fixed period of time to provide advance notice before it is enacted. Would such a scheme provide the necessary discretion for the PRS to not be considered an automatic formula? And, if so, would this matter in considering whether a PRS is a ‘variable import levy’?

These hypotheticals suggest that the concept of a variable import levy, as elaborated upon in Chile–Price Band System, requires further clarification.Footnote ³⁵ By explaining the concept further, the AB can give guidance as to how, if ever, a WTO member could design a PRS scheme that is not a ‘variable import levy’ deemed illegal under Article 4.2 of the Agreement on Agriculture.

Our view is that a shift toward emphasizing the design, architecture, and structure of the PRS as determinative may be the right direction in which to guide the jurisprudence.Footnote ³⁶ This continues to allow for consideration of whether the formula results in automatic and continuous changes in which the results are predictable and transparent, but also opens the possibility for consideration of additional factors beyond what is laid out in Chile–Price Band System. It also falls in line with the AB's approach toward examining duty schemes under other circumstances, most notably GATT Article III:2.Footnote ³⁷

4.2 Minimum import price

In order to stand, a PRS must also not be similar to a minimum import price. In Chile–Price Band System, the AB clarified that ‘[t]he term ‘minimum import price’ refers generally to the lowest price at which imports of a certain product may enter a Member's domestic market’.Footnote ³⁸ It further noted that the Panel had clarified that ‘[these] schemes generally operate in relation to the actual transaction value of the imports’.Footnote ³⁹ This notion was further reinforced during the dispute's compliance proceedings.Footnote ⁴⁰

Because a PRS operates in relation to a reference price and not the actual transaction price, it is possible, at least in theory, for imports under a PRS to enter a market below the floor price set by a PRS. For example, consider a scenario where the actual transaction price (P _t) is lower than the floor and the reference price, i.e. P _t < P _r < P _floor. This may occur if prices continue to decline in the intervening period following when the reference price was last readjusted. So long as the difference between the transaction price and floor price is larger than the sum of the original duty plus the additional duty, then the actual transaction price will fall below the floor price. Whether this proposition holds true depends, in part, on the difference between the transaction and reference price. The greater this divergence, and the lower the rate of the original duty, the more likely it is that this will hold.

Thus, the proper assessment is not whether a PRS imposes a minimum import price but whether it is similar to such a policy. In assessing similarity, the AB's jurisprudence suggests that one considers two factors: (1) the structure and design of the PRS, and (2) its effects.Footnote ⁴¹

Unlike the Panel in Chile–Price Band System, the Panel in Peru–Agricultural Products found that the measure imposed by the PRS was not similar to a minimum import price.Footnote ⁴² Guatemala has appealed this ruling.Footnote ⁴³

While the Chilean and Peruvian schemes employ different names (PBS vs. PRS), a closer examination reveals that the structure and design of the two systems are largely similar. Peru also occasionally refers to its scheme as a bando de precios (PBS). However, three points of contrast are worth noting in the two schemes' design: First, the Peruvian reference price is updated every two weeks, as opposed to weekly for the Chilean reference price. Second, the Peruvian reference price is based on the price in a pre-designated market, whereas the Chilean reference price is the lowest price found of several markets in relationship to one another. Third, the Peruvian scheme requires reference prices be converted to c.i.f. terms, thereby incorporating freight and insurance costs, whereas the Chilean reference price is left on lower f.o.b. terms. All three differences increase the likelihood that the actual transaction price will diverge from the reference price by a greater degree when prices fall.

The Panel, however, did not emphasize these differences in its ruling. Instead, it focused primarily on the second factor – its actual effects. Whereas Chile conceded that it was ‘unlikely’ for imports to enter the Chilean market at a price below the floor price,Footnote ⁴⁴ Peru submitted evidence that in approximately 57% of the fortnightly periods since the PRS came into force, various transactions entered Peru at a price lower than the reference price and floor price.Footnote ⁴⁵ These amounted to more than one-third of trade transactions recorded over these periods.Footnote ⁴⁶ The Panel found this evidence to be decisive in illustrating that the PRS did not impose a de facto threshold and was therefore not similar to a minimum import price.Footnote ⁴⁷

The Panel's points to three questions for the AB to clarify:

• First, did the Panel erred in treating this analysis as primarily an effects-based test?Footnote ⁴⁸ As noted above, the Panel ruling failed to elaborate upon on the design elements of the Peruvian PRS that make it more likely for actual import transactions to occur below the implicit threshold stipulated in the PRS. The AB may wish to clarify the level of analysis that a Panel must undertake if it is to find that the structure and design supports the respondent's argument that a PRS is not similar to a minimum import price.

• Second, when it comes to considering how the PRS affects the price of actual transactions, how ought the effect of the PRS be measured? The evidence may be expressed in terms of the percentage of total import volume (in either value or quantity), the percentage of total transactions, and/or the percentage of days/weeks/other periods in which imports priced below the threshold enter into the market. Are all of these acceptable metrics? Or should greater emphasis be placed on one over the other? Note that Peru submitted evidence only with respect to the latter two metrics, but not with respect to total imports (except for sugar).Footnote ⁴⁹

Why does the exact metric matter? At first glance, it may seem impressive that below-threshold transactions occurred in 57% of the fortnightly periods in which the PRS was in effect. But whether this is truly the case depends, in our view, on the percentage of total import volume represented by such transactions rather than the actual number of transactions themselves or the percentage of periods in which they occurred. In the one instance (sugar) where Peru submitted statistical evidence on the total volume of trade, only 3% of imports entered below the alleged threshold price.Footnote ⁵⁰ We believe that the AB ought to require a respondent to demonstrate effect through data highlighting total volume of trade (rather than the other two metrics) and set forth a brighter line for the percentage of total trade exceeding the alleged threshold price.

Third, should the respondent, when seeking to prove the actual effect of a PRS, be asked to present evidence on a product-specific level? For a PRS to not amount to a measure similar to a minimum import price, the respondent ought to demonstrate that imports below the alleged de facto threshold entered the market with some regularity for all of the tariff lines. In this case, Peru appears to have focused its evidence on the four marker products rather than on the associate products.Footnote ⁵¹ The AB would do well to scrutinize the data submitted for the other products to ensure that transactions below the alleged threshold price exist for all 45 affected tariff lines.

To summarize, the Peru–Agricultural Products dispute raises a number of pressing questions over the evidentiary burden necessary for a WTO member to demonstrate that a PRS does not give rise to a measure similar to a variable import levy or minimum import price. In clarifying these questions, the AB will make clearer how a WTO member ought to design such a scheme and what types of statistical evidence it ought to collect, if it hopes to employ a PRS in line with Article 4.2 of the Agreement on Agriculture.