Parallel Trade and its Ambiguous Effects on Global Welfare

1. Introduction

The regulation of parallel trade is a fiercely debated issue in the global trading system. At the heart of this debate is the ambiguous nature of the welfare effects of parallel trade, an aspect which has been noted by Ganslandt and Maskus (2004), Maskus and Chen (2004), and Valletti and Szymanski (2006). A compromise reached under World Trade Organization (WTO) law permits countries to freely decide whether to allow or ban parallel trade.¹ Advocates of strong patent rights for pharmaceuticals support a global policy banning parallel trade. Danzon (1998) argued that if parallel trade of pharmaceuticals were permitted, profits in the pharmaceutical industry would decrease, thereby reducing incentives to invest in research and development (R&D) for new drugs. Grossman and Lai (2008) opposed this argument. Nevertheless, as pointed out by Maskus (2001), policymakers in many developing countries support an open regime of parallel trade and place a greater emphasis on the affordability of pharmaceuticals than on promoting R&D abroad.

Maskus and Ganslandt (2002) put forward that parallel trade freedom may increase prices in low-income countries and result in smaller markets not being served. We will show that this assertion is correct for specific combinations of parallel trade costs and heterogeneity of countries in terms of market size. However, we will also show that in this case parallel trade freedom still exerts a positive effect on global welfare even though small markets remain unserved. This conclusion contrasts with previous work on the welfare effects of parallel trade by Jelovac and Bordoy (2005) and Malueg and Schwartz (1994).²

There are three main strands of literature on the determinants of parallel trade.³ First, Maskus (2000) and Maskus and Chen (2004) suggested that multinational companies build markets through exclusive territorial dealership rights in order to vertically control the operations of their licensees. They may, however, find it difficult to enforce private contractual provisions prohibiting sales outside the authorized distribution chain so that parallel trade occurs. Our study of parallel trade is closely related to this strand of literature. Second, Brekke et al. (2009) and Ganslandt and Maskus (2004) suggested that in certain industries such as the pharmaceutical industry national governments intervene in private markets by regulating prices. As this results in international price differences, there is a potential for arbitrage between markets as put forward by Grossman and Lai (2008) and Richardson (2002). A third determinant of parallel trade, as suggested by Maskus (2000), is the incentive of parallel importing firms to free ride on investments in marketing as well as on the before- and after-sales services of official licensees and authorized distributors.

2. Simple Model of Parallel Trade

We consider a model with two countries, denoted by subscript j = A,B, and two firms. In country A there is a monopolistic manufacturer m. In country B there is a single authorized independent firm r that is responsible for the distribution and retail of the manufacturer's product. We assume that efficient international distribution of the product requires the manufacturer to build a market in B through exclusive territorial dealership rights. Suppose that the exclusive distributor in B has already established costly distribution channels. The demand in A is q_A = γa − bp_A, and the demand in B is q_B = a−bp_B, with γ > 1. Let γ measure the heterogeneity of the two countries in terms of markets size. Let Π denote the profit of the manufacturer and π the profit of the distributor, respectively. For simplicity, we assume that the marginal costs of production are equal to zero in both countries. The exclusive distributor sells to consumers in B at first, but may also engage in parallel trade from B to A. Arbitrage by individual consumers between B and A is legally prohibited. The marginal costs of engaging in parallel trade are denoted by t. They include, following Berka (2009), Li and Maskus (2006), and Maskus and Chen (2004), parallel-trade specific distribution and advertising costs as a result of re-packaging and re-labeling as well as import duties on parallel trade. We assume that the parallel import product is a perfect substitute for the product sold by the original producer in A. We analyze a situation in which the manufacturer is awarded the right to prevent parallel trade as a benchmark.

Double Marginalization without Parallel Trade

Suppose that the manufacturer can become involved in the retail of the product in A. The distributor in B has a monopoly on the local retailing business. We assume that retailing in B does not involve any costs, except for the costs incurred by the distributor in buying the product from the manufacturer. In the first stage, the manufacturer sets a wholesale price for the distributor, who then sets a retail price p_B in the second stage. The manufacturer is awarded the right to prevent parallel trade of the product from B, i.e. he is awarded an exclusive importation right. The manufacturer maximizes . The distributor maximizes . We obtain the following equilibrium quantities supplied () and retail prices () for the two countries, wholesale price in B () and profits:

(1)

Double Marginalization with Parallel Trade

Suppose that the manufacturer cannot prohibit or at least contractually limit parallel trade. In the first stage, the manufacturer chooses the wholesale price , , at which he sells the product to the distributor in B. In the second stage, the distributor chooses the retail price p_B, p_B ∈ [0, ∞), in B. In the third stage, the manufacturer m and the exclusive distributor r simultaneously choose the prices at which they sell the product in A in a Bertrand model of duopoly, e.g. , , and , , respectively.

Proposition 1. Parallel trade does not occur in any sub-game perfect Nash equilibrium in a double marginalization game with Bertrand price competition in the last stage.

The proof is straightforward as the manufacturer's marginal cost of production is equal to zero and . In the second stage, the distributor anticipates that he will be driven out of the market in A in the third stage. He maximizes the profit generated in B. Working backwards to the first stage and adopting the Kuhn–Tucker method, the maximization problem of the manufacturer is given by:

(2)

and

We obtain two solutions that satisfy all first-order conditions. First,

(3)

Note that as t > 0. More specifically, . The higher the parallel trade costs (and thus the less profitable the parallel trade) the higher is and the lower . However, is only satisfied for specific values of the parameter t. From follows . For to be satisfied it is sufficient that . Henceforth, we will refer to t_h = (a/2b)(γ − 1) as the threshold for high trade costs. To summarize, (, , , , ) only satisfies all first-order conditions if t ≤ t_h. If, however, t > t_h, it is not a solution for the maximization problem of the manufacturer as given by (2) as in this case.

We obtain the second solution:

(4)

Note that () is equal to the monopoly price (profit-maximizing wholesale price) in a double marginalization game in which parallel trade is prohibited. Intuitively, if the two countries are virtually homogeneous (γ → 1) and the parallel trade costs are so high that t > t_h, the distributor will not be willing to engage in parallel trade.

3. Effects of Parallel Trade Freedom on Profits

There are three types of trade costs, denoted by subscript k = h, i, l. Subscript h denotes high trade costs, t > t_h, subscript i denotes intermediate trade costs, t_l ≤ t ≤ t_h, and subscript l denotes low trade costs, t < t_l. For high trade costs, we obtain the following equilibrium prices and quantities in both countries:

(5)

For intermediate trade costs, we obtain:

(6)

As we will see below, for low trade costs, country B will not be served. We obtain the following equilibrium price and quantity in country A:

(7)

In both situations, with and without parallel trade, equilibrium prices and quantities in countries A and B are identical if t > t_h. The threshold for low trade costs is denoted by t_l. The distributor will only be willing to sell the product in B as long as (). From this follows: . Intuitively, if trade costs are very low, t < t_l, potential competition from parallel trade is so fierce that the manufacturer has to charge such a high wholesale price in B in order to deter parallel trade that the distribution of the good in B becomes unprofitable.

To summarize, we consider three scenarios. First, parallel trade costs are so high, t > t_h, that parallel trade is not a worthwhile activity for the distributor and thus a non-credible threat. Parallel trade freedom then does not have any impact on profits, consumer surplus as well as national and global welfare. Second, the manufacturer strategically sets prices in order to deter parallel trade if t_l ≤ t ≤ t_h. However, the wholesale price will be sufficiently low so that the distribution of the product in B is still a worthwhile activity. Third, the manufacturer will charge such a high wholesale price in B in order to deter parallel trade that the market in B ends up not being served if t < t_l.

4. Effect of Parallel Trade Freedom on Global Welfare

From the equilibrium prices, quantities, and profits given in section 3, it is straightforward to derive consumer surplus and welfare in both countries if parallel trade is prohibited, and if it is permitted, respectively. We obtain the following levels of global welfare—being the sum of national welfare in A and B—if parallel trade is prohibited (W**), and if it is permitted ():

(8)

We derive the welfare effects of parallel trade freedom by subtracting global welfare if the manufacturer has the right to prevent parallel trade from global welfare under parallel trade freedom. Recall, however, that even if parallel trade were permitted, the (non-credible) threat of parallel trade would not have any impact on global welfare if t > t_h.

Welfare Effects of Parallel Trade Freedom for Intermediate Trade Costs

Proposition 3. Parallel trade freedom increases global welfare if trade costs are intermediate and (i) γ ≥ 5/2 or (ii) .

Proof. (i) Let the net effect of parallel trade freedom on global welfare be denoted by ΔW_i:

(9)

Note that ΔW_i is a quadratic function of t. Note also that ΔW_i = 0 at t_h = (a/2b)(γ − 1). Hence, in order to show that ΔW_i ≥ 0, it is sufficient to show that ΔW_i is a monotonically decreasing function of t for t_l ≤ t ≤ t_h. From follows . As is the unique maximum, ΔW_i decreases in t for any . Taking into account that ΔW_i = 0 at t_h = (a/2b)(γ − 1), it follows that ΔW_i > 0 for . We analyze for which values of γ. It is straightforward to see that if γ ≥ 5/2. Furthermore, ΔW_i monotonically decreases in t for t_l ≤ t ≤ t_h. Hence, taking into account that ΔW_i = 0 at t_h, ΔW_i ≥ 0 if γ ≥ 5/2.

(ii) If γ < 5/2, we cannot apply the same logic as in the previous case as t_l = (a/2b)(γ − 5/2) would be negative. Since t is non-negative, we set t_l = 0 in this case. For γ < 5/2, ΔW_i has its unique maximum at , which is positive as γ < 5/2. Hence, the question arises as to whether ΔW_i is positive or negative at t_l. If we can show that ΔW_i is positive at t = 0, this would imply that ΔW_i is also positive for t_l ≤ t ≤ t_h, taking into account that ΔW_i = 0 at t_h. By setting t = 0 in (9) we obtain ΔW_i = (a²/72b)(4γ² − 11γ + 7). Note that ΔW_i ≥ 0 if γ ≥ 7/4. Consequently, if γ ≥ 7/4, ΔW_i is positive between zero and t_h. Thus, ΔW_i ≥ 0 if 7/4 ≤ γ < 5/2. □

Intuitively, parallel trade freedom harms both the manufacturer as well as the distributor. Parallel trade freedom is also detrimental to consumers in country B because it leads to a higher local retail price and a lower quantity sold. Hence, consumers in country A are the only beneficiaries from parallel trade freedom. As long as γ is sufficiently large, 7/4 ≤ γ, the positive effect of parallel trade freedom on consumers in A ceteris paribus more than outweighs the sum of its negative effects on the manufacturer, the distributor, and consumers in B.

In contrast, Numerical Example 1 illustrates that the following proposition holds.

Proposition 4. Parallel trade freedom can exhibit negative welfare properties if trade costs are intermediate and γ is sufficiently low [1 < γ < 7/4].

Recall that ΔW_i = 0 at t_h = (a/2b)(γ − 1). By looking at (9), it becomes apparent that ΔW_i has another null at t = (a/2b)(7/2 − 2γ) > 0 as γ < 7/4.

Numerical Example 1. We set a = 100, b = 1/2 and γ = 13/8.

Figure 1 shows that ΔW_i = 0 at t = 25 and at t = t_h = 62.5. Furthermore, ΔW_i has its unique maximum at . Also, t_l = 0. We can see from Figure 1 that ΔW_i < 0 ∀t ∈ (0, 25), which suggests that Proposition 4 holds. If countries A and B are virtually homogeneous, 1 < γ < 7/4, consumers in A will benefit less from parallel trade freedom. In this case, the effect of parallel trade freedom on global welfare will be negative if trade costs are intermediate.

Welfare Effects of Parallel Trade Freedom for Low Trade Costs

Proposition 5. Parallel trade freedom increases global welfare if trade costs are low and γ is sufficiently high, γ > 5/2, even though the smaller market remains unserved.

Proof. For t_l > t, the effect of parallel trade freedom on global welfare is:

(10)

As we can see, ΔW_l is a quadratic function of t. Recall that γ must be greater than 5/2 as t > 0. For smaller values of the parameter γ, we would automatically end up in one of the other scenarios mentioned above. By differentiating (10) we obtain . is the unique maximum as . Furthermore, . By setting t = 0 in (10), we obtain ΔW_l = (a²/288b)(20γ² − 16γ − 67). We can see that ΔW_l > 0 at t = 0 if γ > 5/2. By setting t = t_l = (a/2b)(γ − 5/2) in (10), it follows that ΔW_l = (a²/8b)(γ − 2). ΔW_l > 0 as γ > 5/2. Consequently, taking into account that ΔW_l is a quadratic function of t, , ΔW_l > 0 at t = 0, and ΔW_l > 0 at t_l, it is straightforward to see that ΔW_l > 0 if t_l > t and . □

Contrary to intuition, parallel trade freedom still has a positive effect on global welfare in this case even though the market in B remains unserved. If γ is sufficiently high, the positive effect of parallel trade freedom on consumers in A ceteris paribus more than outweighs its negative effects associated with lower profits for the manufacturer and the closure of the market in B.

5. Conclusion

Our two-country model of parallel trade based on the work of Maskus and Chen (2004) suggests that the threat of parallel trade can limit the monopolistic manufacturer's scope for price discrimination and thus increase retail-market competition. While this finding is in line with the prior literature on parallel trade, such as Ganslandt and Maskus (2004) and Malueg and Schwartz (1994), we find that the manufacturer's strategic pricing to avert parallel trade depends on the level of heterogeneity of countries in terms of market size and on the level of parallel trade costs. If parallel trade costs are high, potential competition from parallel trade does not arise, and the manufacturer will always charge the monopoly price in the domestic market. If, however, parallel trade costs are low, potential competition from parallel trade arises, and the manufacturer advantageously sets prices in the domestic and foreign market in order to avert parallel trade. In an empirical, policy-oriented article, Maskus and Ganslandt (2002) have put forward that parallel trade may increase prices in low-income countries and lead to the closure of smaller markets. Our analysis confirms this assertion under the condition of low trade costs and sufficient heterogeneity of market size in both countries. Because of the threat of severe competition from parallel trade, the manufacturer sets a prohibitively high wholesale price in the foreign market resulting in unprofitable product distribution abroad. Consequently, it would be desirable for policymakers in the foreign country to discourage parallel trade and to encourage price discrimination in order to open the otherwise unserved market. Nevertheless, we also show that even in this case global welfare is positively affected through parallel trade freedom. It is in this respect that we believe our analysis is different from existing works such as Maskus and Chen (2004), Valletti (2006), and Valletti and Szymanski (2006).

As to the general welfare properties of parallel trade freedom, it increases global welfare if countries are sufficiently heterogeneous in terms of market size. Yet, for intermediate trade costs and virtually homogeneous countries, parallel trade freedom can have negative welfare properties. In this case, the negative effect on the manufacturer, the distributor, as well as on consumers in the foreign country more than outweighs the positive effect of parallel trade freedom on domestic consumers. Considering these ambiguous welfare effects, we suggest that neither parallel trade freedom nor a ban, but instead a rule of reason is justified from an economic perspective.