The Case for Free Trade

The traditional argument in favor of free trade goes back to the origins of economic science and to the ideas of Adam Smith and David Ricardo. International trade can be considered just another transformation activity available to the national economy: exports are transformed into imports according to their international relative prices. From the viewpoint of the consumers, international trade expands the set of goods and services that they can afford, either by reducing their price, or by making new commodities available, or by a combination of both. From the viewpoint of efficiency in production, free trade allows and encourages specialization in activities and technologies in which the country possesses a comparative advantage. The product of these activities can then be traded for those commodities for which domestic production is relatively inefficient. Thus, free trade allows a better allocation of resources with the corollary that welfare is enhanced. It is worth noting that a country will achieve a higher welfare with free trade than with protection, even if its trade partners engage in protectionist policies. This classical argument in favor of free trade is based on the same foundations as the case for market allocation of resources and against government intervention in the price setting process—import tariffs and protectionist practices are just a particular case of price distortions and restrictions to trade. Like the argument for free markets, it remains at the core of mainstream economic theory and has undergone many refinements.⁴

Many economists, however, believe that the classical argument for free trade does not capture some of the main gains that a country can obtain from free trade policies. Conventional arguments for free trade are static in nature while there is a widespread belief that international trade and economic growth are closely related. The new theories of economic growth seek to identify those policies and mixeds of the economic environment which are key to achieve, not only a more efficient use of existing resources, but a higher rate of output growth in the long run. In the traditional theory of growth, the removal of inefficient policies, such as barriers to trade, could prompt a one-time upward shift in the level of output but could not affect its long-term growth rate. In contrast, recent growth theory shows how free trade can significantly contribute to increase permanently the growth rate of an economy.⁵

Recent analytical and empirical developments have identified a variety of sources of economic growth. The main emphasis has been in the role of human capital accumulation, learning by doing, investment in re-search and development, and expansion of the span of available products. These factors have the potential to enhance the use of available resources, such as labor and nonhuman capital, reversing the tendency toward diminishing marginal returns. Moreover, some of these newly analyzed factors seem to exhibit positive externalities which produce increasing returns to scale at the macroeconomic level.⁶

Open trade policies allow individual countries to profit from growth enhancing factors. Imports of inter-mediate products may embody advanced technology and are the result of human capital accumulation and research and development investments realized elsewhere. Closer economic links increase the transmission of new technologies. Expanded competition forces domestic firms to accelerate the pace of absorption of new technologies and devote adequate resources to develop new or improved products. Access to world markets permits a wider diversification of output as well as a larger span of available inputs.

Protectionist policies also implicitly hamper exports. An increase in import tariffs can result in an exchange rate appreciation. The preferential treatment they afford to import-substituting industries reallocates resources toward production for the protected domestic market and away from export-oriented industries. Also, an important part of the bias against exports takes the form of general increases in costs and low productivity of factors.

Arguments in favor of protectionist policies generally assume an unrealistically enlightened government that will be able to choose the right industries to protect and that will manage trade and domestic policies in an optimal manner over time. In practice, governments have not been very successful in “picking winners”—determining which industries to support—because this depends on details of production technology and market structure which governments typically know little about.⁷ Trade policies often get heavily influenced by special-interest politics rather than consideration of national costs and benefits. Even when initially based on sound economic theory, protectionist policies create powerful special-interest groups that tend to perpetuate and extend privileged situations. Sophisticated programs of trade intervention and selective tariffs, even when based on technically solvent arguments, will probably be captured by interest groups and transformed into instruments of income redistribution in their favor. Beneficiaries and would-be beneficiaries of protectionist trade policies are likely to devote a potentially large amount of resources to “rent seeking”—activities aimed to maintain and extend existing protection, influence the allotment of import quotas, etc.—further magnifying distortions and inefficiencies in the domestic allocation of resources. In this context, an easy-to-monitor commitment to free trade may be the optimal viable policy.

Revenue Generation: A Nonprotectionist Case for Tariffs

The rationale for tariffs is sometimes unrelated to protectionist trade policies. Unlike nontariff barriers—quotas, competitive subsidization, voluntary export restraints, etc.—import or export tariffs yield tax revenue to the government. In some countries, taxes on international trade constitute the main and most stable source of government revenue. The need to raise revenue may outweigh, in some cases, the negative effect that tariffs have owing to their protectionist side effects.

Trade taxes are not optimal instruments to raise revenue. It can be shown that a combination of domestic taxes levied neutrally with respect to domestic and foreign products and that yield the same revenue can cause a lower efficiency loss.⁸ When the revenue objective is taken as given, the necessary taxes should be designed to minimize losses in efficiency and growth potential. The efficiency losses generated by taxes on international trade stem from the “wedge” they introduce between international and domestic prices. The inward-oriented bias they cause can produce large inefficiencies and seriously hamper growth. Domestic taxation such as consumption or income taxes can meet the revenue target with lower rates, broader bases, and without a protectionist bias. Taxes on domestic consumption can also be collected on imports at the border. This will fulfill the revenue function of a tariff and be equally easy to administer without protecting domestic producers. Similarly, taxes on luxury items are better designed as domestic excises—collected at the border when applied to imports—than as tariffs.

The argument against tariffs as revenue-raising instruments, however, rests on the availability of trade-neutral domestic taxation. In some cases, either the necessary domestic taxes do not exist and need an implementation period or they cannot be raised beyond existing levels. An argument in favor of trade taxes is their lower administrative cost. The World Development Report⁹ estimates that the administrative costs of levying trade taxes are between 1 percent and 3 percent of revenue collected. The corresponding costs for VAT and income taxes are estimated at 5 percent and 10 percent, respectively. Therefore, countries which are constrained by the weakness of their tax administration, the immature stage of market institutions, and the lack of qualified personnel and accounting sophistication of taxpayers resort to tariffs as a relatively straightforward method to raise revenue. Thus, budgetary pressures combined with exceptionally limited tax handles can dictate the use of trade taxes for revenue purposes.

In the medium- and long-term, a strategy aiming at sustainable growth should encompass the use of domestic tax instruments and implementation of administration improvements necessary to phase out any reliance on taxes on international transactions. While administrative costs of collecting trade taxes are low, the economic social cost of increasing trade taxes is generally higher than that of raising domestic taxes.¹⁰ Correspondingly, when tariffs are used with a revenue objective, the rate should be low enough to avoid any significant protectionist bias.

Balance of Payments Objectives

Tariffs or across-the-board import surcharges are often applied for balance of payments reasons by countries facing an imbalance in the external sector. An external imbalance indicates that the real exchange rate (the ratio between the domestic and foreign prices of tradable commodities) is not sustainable. It can be caused by differentials in inflation rates that are not reflected in the exchange rate, deterioration in the terms of trade, etc. By imposing a tariff, governments may try to avoid the necessary domestic adjustment.

Under a floating exchange rate regime, the tariff will prompt a revaluation offsetting the trade effect of the tariff increase. Due to that, tariffs are used to correct external imbalances under fixed exchange rate regimes or when the government tries to avoid a devaluation. The aim is to imitate the effect that a devaluation would have on the trade balance. While a devaluation would have beneficial effects on exports, however, a tariff impairs exports through increases in costs. Since resources will move away from the export sector, the measure can, in fact, worsen the trade balance in the long run.

When a country faces a balance of payments problem, the optimal response depends on the cause of the imbalance. Generally, a policy of domestic adjustment, including fiscal measures, that addresses the root of the imbalance, is necessary.

Antidumping and Countervailing Duties

Under GATT rules, antidumping and countervailing duties can be imposed by country members to protect their domestic producers from injury owing to “dumping” of goods by foreign suppliers or to offset trade-distorting subsidies set by trade partners. The legitimacy of these measures is based on the defense of competition and fair trade. Nevertheless, the imposition of duties is not always the optimal response by the importing country. Moreover, the general consensus among trade economists is that, in most cases, countries have used antidumping and countervailing provisions to implement protectionist policies.¹¹

From the perspective of the importing country, dumping may be harmful if it involves predatory pricing or intermittent dumping. Predatory pricing occurs when prices are set below cost by foreign exporters with the objective of driving domestic firms out of business. Once domestic competition is eliminated, prices are set at monopoly levels. In practice, this type of operation is very uncommon because the preconditions for its success are rarely met. Foreign suppliers must have a stable monopolistic position on the international market and domestic suppliers must face substantial re-entry barriers. Intermittent dumping is the disposal of occasional surpluses by exporting them at exceptionally low prices. In either case, before the introduction of antidumping duties, the potential harm to competing domestic producers needs to be weighed by the government of the importing country against the eventual benefits to consumers and importers.

In many cases, legitimate market behavior is labeled as dumping by importing countries and antidumping measures are undertaken with protectionist objectives. For example, price differentials between domestic and export markets do not necessarily imply dumping behavior. Price differentiation is a profit maximizing strategy routinely implemented by firms that operate in several markets. Many actions that would not be considered unfair practices domestically are, however, prosecuted with alacrity when pursued by foreign exporters.

Countervailing duties are introduced with the objective of offsetting export subsidies implemented in the exporting country. Subsidies to “strategic” industries have been justified by proponents of the new trade theory as means to gain competitive advantage in these industries.¹² Under GATT rules, subsidies are allowed on exports of primary products only. In practice, however, the subsidy component of the price of imported commodities is difficult to determine.

The incidence of antidumping and countervailing actions—including duties, voluntary export restraints, etc.—increased substantially in the late 1970s. Finger and Nogues¹³ report that between 1980 and 1985, the European Community and seven other countries initiated 1,155 antidumping cases and 425 antisubsidy cases. Table V.1 shows a preponderance of countervailing actions in the United States while antidumping cases are evenly spread among all countries. Bhagwati¹⁴ argues that antidumping and countervailing provisions have been “captured” by protectionist special-interest groups and domestic producers.¹⁵ The design of these provisions facilitates their use for the purpose of putting obstacles to successful foreign suppliers. Often, most of the burden of proof lies with the defendant and different forms of restrictions might be triggered at the moment of filing the complaint, independently of whether it is eventually substantiated. Kelly and McGuirk¹⁶ present evidence of the widespread use of countervailing and antidumping duties in the European Community to protect cartelized domestic markets from foreign competition—for example, in 52 percent of the cases under anticartel investigation in the chemical industry, a parallel anti-dumping action had been initiated. Once antidumping or countervailing duties are established, they tend to become permanent.

Table V.1.

Countervailing Duties and Antidumping Actions Initiated

Source: Bhagwati (1988).

¹Countervailing duties and antidumping actions correspond to U.S. Trade Act, Sections 701 and 731, respectively.

Table V.1.

Countervailing Duties and Antidumping Actions Initiated

			United States 1	Australia	Canada	EEC
Countervailing duties
	1980		8	0	3	0
	1981		10	0	0	1
	1982		123	2	1	3
	1983		21	7	3	2
	1984		51	6	2	1
	1985		39	3	3	0
		Total	252	18	12	7
Antidumping actions
	1980		22	62	25	25
	1981		14	50	19	47
	1982		61	78	72	55
	1983		47	87	36	36
	1984		71	56	31	49
	1985		65	60	36	42
		Total	280	393	219	254

Source: Bhagwati (1988).

¹Countervailing duties and antidumping actions correspond to U.S. Trade Act, Sections 701 and 731, respectively.

View Table

Table V.1.

Countervailing Duties and Antidumping Actions Initiated

			United States 1	Australia	Canada	EEC
Countervailing duties
	1980		8	0	3	0
	1981		10	0	0	1
	1982		123	2	1	3
	1983		21	7	3	2
	1984		51	6	2	1
	1985		39	3	3	0
		Total	252	18	12	7
Antidumping actions
	1980		22	62	25	25
	1981		14	50	19	47
	1982		61	78	72	55
	1983		47	87	36	36
	1984		71	56	31	49
	1985		65	60	36	42
		Total	280	393	219	254

Source: Bhagwati (1988).

¹Countervailing duties and antidumping actions correspond to U.S. Trade Act, Sections 701 and 731, respectively.

The “Infant Industry” Argument

Many departures from free trade, including the import substitution policies followed by many developing countries in the 1960s and 1970s, have been justified by using different variants of this theory. The argument essentially posits that some industries have initially high costs but may, in the long run, acquire a comparative advantage after a temporary period of development. Several conditions have to be assumed for the argument to be valid:

(1) The present value of future social and private returns generated by the infant industry must be higher than that of other industries or alternative uses of resources. The cost of the infant industry should include the loss in efficiency and damage to other sectors owing to distortions caused by protection.

(2) In the absence of protection, private investment in the industry would not occur. That is, expected private returns do not justify investment in the industry. Therefore, a substantial part of the returns or decrease in costs to be generated by the industry must consist of externalities. If future benefits could be privately appropriated by the initial investors, private firms would themselves be willing to incur the initial costs and protection would be redundant. If private investors are willing to undertake the initial investment but they lack the necessary capital, other measures would be more effective than protection (e.g., liberalization of capital markets and foreign investment).

(3) Temporary protection must make initial investment in the industry cost effective for private investors. It might be the case that, even with protection, private investment does not take place. In practice, “import substitution” policies have necessitated many other government interventions besides protection—subsidies, regulation, official prices, etc. When externalities are the reason for protection, sufficient externalities must accrue to the industry itself, to become competitive eventually. If most externalities are spillovers to other industries, the infant will never be able to stand alone.

(4) The protection has to be, in fact, temporary, and must be decreased over time in an optimal manner. In particular, international competition has to be kept constantly at a sufficiently high level to encourage the cost-saving investments that will eventually render protection unnecessary. Policymakers must be able to monitor the evolution of production and market characteristics in the industry. The timetable of protection disarmament has to be well known, credibly announced, and timely implemented. If firms expect the period of protection to be extended, they will not invest in reducing costs and the industry will not be able to survive international competition. Since this is likely to force the government to extend the period of prothe expectation will become self-fulfilling.¹⁷ It is also assumed that vested interests to maintain protection will not arise or will be effectively resisted.

Thus, the validity of the infant industry argument rests on the validity of very restrictive conditions. Even when those conditions are met, free trade combined with targeted industrial policies is usually a superior response. When the cause of the externality can be clearly identified (and it is difficult to make a consistent case for protection when it cannot), a targeted policy is generally superior to across-the-board protection of the industry. For example, if intensive research and development need to be carried out in the early stages, subsidies to this activity would be more cost-effective than wide-ranging protection.

Much of the thrust of the infant industry argument for protection dissipates when foreign direct investment is considered. In cases where the decrease in costs is brought about by initial investment in research and development or in general know-how, the industry can operate at the lowest cost from the outset by attracting foreign investment. Foreign investment and access to international capital markets—which call for nonprotectionist, outward-oriented trade policies—are also the first-best solutions when the problem stems from imperfect domestic capital markets.

Trade economists have consistently pointed out that, in practice, protection has persisted for long periods without evidence of any decrease in costs in the protected industries. On the contrary, there is abundant evidence of resilient inefficiency and high costs in protected industries that have been trimmed only by trade liberalization. Also, there is no evidence that higher rates of protection have been accorded to those industries with greater externalities.

Tariffs and the Terms of Trade

In principle, imposing a tariff may be beneficial for a country that can affect the international price of its imports or exports. If exporters do not act collusively but the country as a whole possesses some degree of monopolistic power on export markets, the imposition of an export tariff can imitate the effect of collusion among exporters. Similarly, an import tariff can implement an advantageous pricing strategy when the country has a monopsony on import markets. The optimal tariff would be obtained by setting the benefits—that is, government revenues and private profits—against the costs of the tariff—that is, domestic distortions and misallocation of resources. Although the welfare of the country could be raised in this manner, the welfare of other countries would be decreased in a larger amount. Thus, the gains to one country are clearly achieved at the expense of others.

Small trading economies, such as those of most developing countries, can hardly affect international prices by trading in manufactured or primary products. Those countries, however, that possess the monopoly of a natural resource, such as oil, can improve their terms of trade by restricting or taxing exports. That is also the case of large industrial countries whose volume of international trade accounts for a large share of the international markets. Nevertheless, tariffs motivated by the terms of trade argument are rarely seen in practice. The reason seems to be the fear to trigger a chain of mutual retaliations that would compromise the world trading system.¹⁸ It can be shown that when the possibility of retaliation is taken into account, the use of tariffs to improve the terms of trade may no longer be optimal.¹⁹ Multilateral trade institutions and the inter-World War experience of trade retaliations seem to have sufficed to enforce restraint in the use of this sort of “beggar-my-neighbor” policy

Strategic Trade and the New International Economics

In recent years, some trade economists have proposed new approaches to the analysis of the gains and costs of free trade.²⁰ In the center of the new theory is the consideration of economies of scale and externalities in some industries. Economies of scale and externalities cause decreasing average costs and thus, larger producers have an advantage over smaller ones. According to this theory, those suppliers that gain initial control of a greater portion of the market will be able to drive competitors out of it and to secure for themselves an oligopolistic position. The “new trade theory” posits that if an industry exhibits these characteristics, a country may be able to gain an oligopolistic or monopolistic power on the international market for the product under consideration. Once a country has achieved that position, it will obtain monopoly profits at the expense of its trade partners. Moreover, if its trade partners pursue free trade policies, the main producer can effectively preclude competition due to its lower costs. By implementing specific industrial policies, such as subsidies and protection, a country can allow selected domestic industries to contest monopolistic positions in international markets and replace existing monopolists.

Free traders have argued that the existence of imperfectly competitive domestic markets greatly strengthens the case for free trade. The larger the market, the less scope there is for monopolistic market power, and free trade greatly increases the size of the market. In fact, free or freer trade has proved to be the best anti-monopoly weapon. Moreover, there is no evidence that international markets are generally oligopolistic. Even in cases where supply is concentrated in a few producers, the threat of new entrants can preclude monopolistic behavior. This threat is particularly present in international markets under free trade.

Krugman²¹ argues that to design strategic trade policies, governments must have extensive knowledge of many details that are possibly beyond their reach. In addition, the general equilibrium effects of strategic trade policies are uncertain. Since promotion of a particular sector implies drawing away resources from other sectors, the pursuance of strategic trade policies in key industries is likely to generate significant inter-sectoral and intertemporal distortions. Empirical analyses based on general equilibrium models indicate substantial efficiency losses which generally exceed the gains from strategic trade policies.²² Furthermore, adherence to a policy of industry subsidization and protectionism creates powerful incentives for rent-seeking behavior and interest group pressures. Under a policy of discretionary protection, in practice, governments are unlikely to resist lobbying and special-interest politics. Many economists that would, in principle, justify protectionism in some selected cases, consider that an adequate implementation of selective protection is politically infeasible. Consequently, the establishment of a blanket policy of free trade may be the best feasible policy.²³

Optimal Tariff Structure Under a Revenue Objective

In most cases, the imposition of tariffs is not an optimal policy.³¹ That is, trade-neutral domestic taxes and subsidies in the case of externalities can raise the same revenue as a given tariff system with lower distortions and efficiency losses.

In practice, tariffs are often introduced by governments seeking to protect domestic producers from foreign competition. Protection of this sort distorts the relation between domestic and international prices prompting an inefficient allocation of resources. As higher prices spread across domestic markets, they impair the exporting capacity of the country through higher costs, lower factor productivity, and currency overvaluation. In the absence of foreign competition, domestic producers often lack the incentive to introduce growth-enhancing new technologies and cost-saving measures. Protection encourages instead unproductive rent-seeking activities such as lobbying and special-interest politics. These activities are undertaken by potential beneficiaries to maintain and extend existing levels of protection. From the revenue point of view, trade-neutral domestic taxation provides a broader base and the possibility of lower tax rates while avoiding inward-oriented biases. According to these considerations, the optimal tariff would be an across-the-board zero rate tariff.

Nevertheless, owing to revenue considerations, a low tariff has been frequently advised. When designing the optimal tax mix to meet given revenue objectives, the distortions introduced by a tariff should be weighed against savings in collection costs and reduction of distortions elsewhere in the economy. Collection costs of international trade taxes are generally lower than those of domestic taxes. The World Development Report³² estimates that the average collection costs of trade taxes are between 1 percent and 3 percent of revenue. In contrast, the same source estimates that the cost of domestic taxation can be as high as 10 percent of revenue collected in the case of income taxes and 5 percent of revenue in the case of the VAT.

The case for a low tariff rests on its small economic cost. The economic cost of levying a tax is a more comprehensive concept than the administration cost. The economic cost is the value of the output forgone due to losses in economic efficiency. The economic cost of a tax rises more than proportionally with the tax rate.³³ This is particularly true in the case of taxes on international trade.³⁴ These taxes add protectionrelated efficiency losses to the losses that could be expected from any indirect tax with a relatively narrow base. Efficiency losses related to protection are caused by the inward-oriented bias that tariffs introduce. The discrimination against imported products distorts domestic prices, increases domestic costs, and impairs exporting activities. At low rates, the lower administrative cost of a tariff might prevail over the economic cost of protection. The case for a low-rate tariff is reinforced if existing domestic taxes are costly to administer and highly distortionary, and if fiscal imbalances dictate an urgent increase in government revenue.

Finally, when a country has a complicated tax structure that is difficult to dismantle in the short term, it may become infeasible to increase domestic taxation further. In such cases, the revenue objective and the level of domestic taxation may have to be taken as given.

Once it has been decided to implement a tariff with low rates, there are several reasons to minimize the dispersion of rates.

• Multiple rates substantially increase the cost of administering the tariff. Since low collection costs are an important reason for using tariffs, multiple rates undermine that rationale by increasing collection costs.

• A uniform rate implies uniform effective protection of all domestic industries.³⁵ Provided that exporters are reimbursed the tariff paid on their imported inputs, a uniform rate structure minimizes discrimination between sectors. When inputs are taxed at a lower rate than outputs, industries producing final goods are granted a higher effective protection than other economic activities. In general, a high tariff rate on a product benefits domestic producers of that good and harms those activities that use the product as input. Rate dispersion will produce different levels of effective protection across industries according to the rates that apply to their output and inputs. Exporters occupy a special position in the economy. While the price of their output is generally set by international markets, the price of their inputs—that is, their costs—are increased by protection. Consequently, protectionist policies tend to impair export industries. To minimize this antiexport bias and allow exporters to compete in international markets, it is generally necessary to reimburse the import duties paid on inputs used to produce exports.

• Rate dispersion encourages special-interest groups to devote resources to increase the amount of protection granted to them. Every industry has a strong incentive to press for an increase of the particular rate that applies to their output. In contrast, a clear commitment to uniformity diminishes the incentive to engage in lobbying activities. First, since any increase in the tariff rate gained by a group has to be extended to all imports, the tariff rate increase that any special-interest group is likely to obtain is small. Second, all import-substituting sectors benefit from an increase, irrespective of who devoted resources to lobby for the increase. Therefore, there is an incentive to wait and let other groups bear the cost of lobbying. That is, free riding can be expected to diminish protectionist pressures.³⁶

• The distortions prompted by a tariff grow more than proportionally with the rate. In contrast, revenue grows, at most, proportionally. In fact, revenue should be expected to increase less than proportionally to the increase in rates. This is due to the decrease in demand prompted by higher prices and, perhaps, due to tax evasion and smuggling encouraged by higher tariff rates.

Therefore, to obtain a given level of revenue with the lowest distortionary effect, the tariff should be applied across the board to all imports with a minimum dispersion of rates.

There are also arguments in favor of rate dispersion. The main nonprotectionist argument for differentiated tariff rates is based on the theory of optimal taxation. Optimal taxation is also known as Ramsey taxation after the economist F.P. Ramsey, who in 1927³⁷ derived the characteristics that an optimal tax system should have to minimize tax-induced distortions.³⁸ According to the theory of optimal taxation, those commodities for which demand has a lower price elasticity should be taxed at higher rates. The reason behind this prescription is that if the demand for a commodity is inelastic, it will change relatively little after the imposition of a tax. That is, consumers will absorb the tax burden without substantially changing their economic behavior. Hence, to minimize distortions, taxation should concentrate on those commodities.

Another result of optimal commodity taxation is that final goods should be taxed at higher rates than inputs to production. The reason is that tax-induced changes in the relative price of inputs can lead to the choice of inefficient technologies. Some inefficient technologies may become more profitable to the extent of overtaking efficient technologies. Since the tax will be eventually borne by the final consumer, it is better to tax final consumption directly, thereby avoiding production distortions.

Extending that argument, the theory of optimal taxation recommends higher tariff rates on final consumption goods than on intermediate products. Among final consumption goods, those with a more inelastic demand should have higher tariff rates.

The theory of optimal taxation has found very little application in practice. The amount of information required to design the optimal tariff structure is beyond the possibilities of even the most sophisticated government agency. It requires precise knowledge of demand elasticities for all goods and services. Moreover, in a dynamic economy, the characteristics on which an optimal tax structure should be based—price and cross elasticities of demand, preferences, available products, etc.—are constantly changing. The administrative cost of implementing such a system would probably exceed the revenue collected.³⁹

Tariff rates differ widely across commodities in many countries. Nevertheless, the reasons for such rate spread owe little to the theory of optimal taxation. Special-interest groups and lobbying by import-substituting industries often press governments for higher tariff rates on their products and lower tariff rates on their inputs. Most governments implement tariffs to protect domestic industries and, from a protectionist point of view, the optimal tariff structure does not need to exhibit a uniform rate.

Optimal Tariff Structure Under a Protection Objective

In many cases, policymaking needs to operate within the parameters of a given level of protection. The optimal tariff structure under that constraint will depend on the reasons for protection. For instance, if the objective is to reallocate resources toward a particular sector—such as agriculture—and away from other domestic industries, a uniform tariff may be unable to accomplish that goal. Nevertheless, the case for a tariff structure with a single rate or with very few rates within a narrow band, might still be argued on administrative, political, or efficiency grounds.

A common argument in favor of protection is based on the infant industry theory.⁴⁰ According to this argument, protection is needed to allow some domestic industries to develop. This theory posits that certain industries are initially uneconomic but may become competitive in the long run. These industries, once they are established, generate externalities and experience a decrease in costs that will, eventually, allow them to compete in the world markets. Recently, strategic trade theories have defended protection for selected industries on the grounds that these industries exhibit increasing returns to scale. Therefore, established producers can effectively prevent competition because their costs are lower than those of foreign potential entrants. The markets for their products will necessarily be dominated by a few firms from one country. Protection can allow domestic suppliers to challenge incumbent oligopolists and, eventually, take their place.

If protection is motivated either by the infant industry argument or by strategic trade considerations, the optimal tariff structure would not normally be uniform. Protection should be accorded only to those industries that need it to generate externalities or to contest international markets. When protection must be extended to several industries, the rate of protection accorded to each of them should be a function of its particular protection needs. Overprotection of an industry may eliminate the incentives to invest in cost-saving activities, defeating the very purpose of the tariff.

In the case of a country that possesses monopsonistic power on certain international markets, protection can improve its terms of trade. Since the country can lower the price of some of its imports by restricting demand, a tariff on those imports may improve welfare in the country at the expense of other countries. The optimal tariff structure would restrict protection to those products whose international price the country can affect. The tariff rate on each commodity should be commensurate to the market power of the country.⁴¹

Nevertheless, arguments for protection based on terms of trade are seldom employed in practice. Fear of retaliation seems to be sufficient to dissuade governments from implementing this type of policy.⁴² Observed behavior suggests that when the possibility of retaliation by trading partners is taken into account, a policy of zero tariffs could be optimal.

To conclude, even when tariffs are implemented for protection, uniformity or quasi-uniformity of tariffs may still be defended on a variety of grounds. First, the advantages of a diversified rate structure may not compensate the cost of administering such a system. Second, political economy arguments point to the need for across-the-board rules to curtail unproductive rent-seeking activities. Finally, the practical difficulties involved in identifying the relevant industries or the appropriate rates of protection may preclude the implementation of optimally differentiated rates.

Exemption of Imported Inputs

Exporters can be provided with duty waivers and exemptions from nontariff barriers. To facilitate administration, the coverage of these exemptions may be extended in some cases to inputs employed in the production for domestic markets. Variants of this method have been implemented in Taiwan (rebate based on accounts), Mexico and Morocco (temporary admission), and Indonesia, Thailand, and Korea (duty exemptions). Under these methods, exporters are registered and must submit an export report or plan including a list of necessary inputs to obtain a license. Usually, if the proportion of exports to total production is higher than a given threshold, an exemption of 100 percent of imported inputs is granted.

Efficient implementation of these schemes presents several obstacles. First, it is administratively difficult to extend exemptions to imported inputs employed indirectly. In the absence of a similar exemption for domestic producers of intermediate goods used by exporters, domestic inputs may not be able to compete with direct imports. Second, it is equally difficult to ascertain whether a producer is above or below the threshold on exports that warrants a full exemption of tariffs on inputs. Also, when exporters obtain exemptions only to the extent that inputs are effectively used in the production of exports, it may be necessary to monitor the technical requirements of different production activities. If the use of exempt inputs cannot be reasonably monitored, tax revenue will be eroded and domestic inputs will be displaced by duty-free imports. A possible solution to this problem, implemented in Korea and Taiwan, is to determine and publish regularly a matrix of technical coefficients for the main export commodities that is, then, applied in all cases. Similar methods have been implemented in Bangladesh, India, and Pakistan with assistance from the World Bank. Morocco applies a system of exemptions based on declarations by exporters that are verified by customs officials within six months. It should be noted that when duty-free imports can be deviated into the domestic market where they are sold at (higher) domestic prices, the opportunity cost of inputs to exporters is given by the domestic price. Therefore, even small spillovers of duty-free imports into the domestic market can neutralize the attempt to develop a competitive export sector.

Duty Drawbacks

Under this method, duties and other indirect taxes actually paid on inputs are rebated to exporters. Duty drawbacks have been used in order to meet the needs of small or occasional exporters. The system has also been used when imported inputs represent a small proportion of total inputs in the exporting sector. Duty drawbacks, unlike most direct export subsidies, are allowed by GATT rules. An important industry in which duty drawbacks have been used to isolate exporters in both developing and industrial countries is the auto industry (Brazil, Mexico, the United States, and others). Different versions of this method have also been implemented in Korea, India, Indonesia, Taiwan, and Thailand.⁴⁷ Owing to administrative considerations, a scheme of drawbacks should include a standard rebate for each regular export product and the possibility of filing a claim, supported by evidence, when a higher rebate is sought.

Bonded Factories and Duty-Free Export-Processing Zones

The system of bonded factories involves bringing in the inputs and shipping the exports under customs seal. Under this system, the physical location of the factory is not geographically constrained. Mexico and Mauritius, among other countries, have employed the in-bond manufacturing system.

As an alternative to in-bond manufacturing, many developing countries have established duty-free export-processing zones. The reasons for the implementation of this system have been diverse. Regional development objectives, pressures from domestic special-interest groups, the intent to bypass necessary liberalization of the trade regime, the assessment that the administrative demands of bonded factories were beyond existing capabilities, or the desire to emulate the success of East-Asian export processing zones have been behind the wide popularity of this system. Thomas and Nash report that at least 30 developing countries have introduced different forms of duty-free areas to promote export activities.⁴⁸ In practice, the results have been mixed and many of these initiatives have performed very poorly as a result of unfavorable location, high starting costs, lack of necessary institutions and infrastructure, and shortcomings in their administration. The approach has been most successful where export-processing zones were part of a general environment favorable to exports and foreign investment.

The theoretical analysis of duty drawbacks, tariff exemption of export inputs, and other methods of isolating exporters from inward biases created by protectionist policies is still scarce. It is generally accepted that a first-best approach would require the removal of protectionist policies, eliminating thereby the need to insulate exporters from the damaging effects of protectionism. When some level of protection is taken as given, duty drawbacks may improve the competitiveness of export industries at the expense of introducing new distortions. That is, duty drawbacks are a second-best policy. The main distortion introduced by duty drawbacks is a bias in favor of imported inputs. Domestically produced inputs will generally have a domestic price above their international price owing to the protection accorded to import-substituting firms. When an exporter purchases those domestic inputs, he cannot obtain a refund of the amount by which their domestic price exceeds the international price. In contrast, if the same exporter imports his inputs—which might be identical to those domestically produced—he will receive a duty drawback. Therefore, when inputs are imported—rather than bought from domestic producers—the price effectively paid by exporters is the international price instead of the higher domestic price.

A particular instance of this distortion occurs when different export industries use different amounts of nonimportables, such as labor. Those export industries that are more intensive in the use of nonimportables will be disadvantaged with respect to those that are more import intensive. This bias will operate independently of the comparative advantage of the country, which might very well lie in the intensive use of nonimportables as inputs.

Panagariya analyzes the effects of input tariffs on welfare with and without duty drawbacks.⁴⁹ He finds that when protection is extended to inputs, the implementation of duty drawbacks is welfare-improving up to a point. The reason is that export drawbacks allow a partial correction of the anti-export distortions introduced by protection. This effect generates a structure of exports more in accordance with the underlying comparative advantage of the country. Nevertheless, some exporters may switch from domestic to imported inputs to escape higher domestic prices, as previously discussed. In some cases, this will occur even when the adopted technology is less efficient (at world prices). When tariff rates are high, the latter inefficiency offsets the beneficial effect of the drawbacks. J. Behrman and S. Levy also analyze some conditions under which export drawbacks may be welfare-reducing.⁵⁰ If the country exports several goods and they are produced with different combinations of domestic and imported inputs, those exports that are import-intensive will be favored by tariff rebates. In this sense, export drawbacks are similar to a system of arbitrary export subsidies since they are not based on considerations of comparative advantage. Instead, the drawback system favors those exports with a larger proportion of imported inputs.

Implementation of Export Taxes

The Rationale for Export Taxation

The Residence Principle

This principle—also denoted as the nationality or domicile principle—asserts that natural persons, or individuals, are taxable in the country or tax jurisdiction in which they establish their residence or domicile, regardless of the sources of income. For a legal person, or nonindividual legal entity, establishing residence is less clear-cut, although it has increasingly been tied to the location where its business activities are registered and/or its management and control are effectively exercised. In some countries—notably the United States, Canada, and Australia—the concept of residence is more narrowly and legally defined in terms of nationality. In these countries, a natural person is taxable on the basis of his citizenship rather than physical residence, while a legal person is taxable in the place of its registration or incorporation rather than where it carries out its activities. From the standpoint of taxation principles, taxing factor income based on residence more closely approximates the equity or ability-to-pay principle. In relation to capital income, a wholly residence-based tax, because it focuses on the residence of the saver, can be viewed as a tax on the ownership of capital and thus on savings.⁶²

The Source Principle

The source principle asserts the prior, and even sole, claim of the source country, or country in which the income arises to natural or legal persons, to tax such income without reference to other criteria or physical presence or legal residence. Clearly, taxing income on a source basis is ad rem in nature and approximates the benefit principle. It also amounts to a tax on the location of capital, or investment, because it disregards the status of the investor in favor of the siting of the investment.⁶³

Application of the Residence and Source Principles in Practice

Taxation of income by adherence to pure principle is tantamount to taxing net national product (residence principle) or net domestic product (source principle) because the former taxes residents’ worldwide income and the latter all income produced by domestic factors irrespective of their owners’ residence.

In practice, as noted above, countries have tended not to stay with the pure application of one or other principle, but to apply a mix of residence- and source-based direct taxation, the former for nationals residing in the country and the latter for income earned within the country by nonresidents and/or nonnatural persons. The precise nature of mixes has depended on each taxing jurisdiction’s perception of the relative importance of a number of factors, notably the attractiveness of foreign investment, the revenue implications, the domestic administrative capabilities, and the degree of cooperation with competing jurisdictions that it can expect.

Regrettably, even such mixes have not tended to be uniform, thus resulting in potential double taxation of the same income. For example, developing countries, as a broad group, have generally forgone taxing foreign income of residents primarily on grounds of administrative expediency or because the level of income involved was relatively modest, rather than because of a commitment to the source principle. On the other hand, in taxing the domestic source income of nonresidents, they have had to balance their ever-present revenue needs with the resulting disincentive effects on the foreign investor who computes the user costs of capital and services on an after-tax basis.

For developed countries, the concerns are somewhat different because they generally face a relatively more balanced two-way movement of production factors and resource flows among themselves, and a oneway movement toward developing countries. As a result, they remain much more concerned with issues relating to the incidence of taxes on production factors. They each view such incidence essentially from the perspective of a home country trying to approximate the application of the residence principle to the worldwide income before tax of its residents, while continuing to tax domestic source income of nonresidents at an equivalent rate. This explains their interest in transforming the tax principles of residence and source through the provision of tax credit rather than a tax deduction from income for tax levied in a competing tax jurisdiction, within the framework of tax treaties.

Defining the Problem

Allocating the global income of multinational enterprises, MNEs, among fiscal jurisdictions for tax purposes is a major source of conflict between tax authorities and the enterprises concerned. Much of this allocation is carried out by setting transfer prices.

In the context of international trade, a transfer price is the price for the internal sale of a good or service in intrafirm trade, that is, in trade between branches or affiliates of a single business enterprise located in different countries. Transfer prices are administrative prices set by the management of transnational or multinational enterprises. If management wants its branches and subsidiaries in different countries to operate as independent profit centers, then it may set these prices as close as possible to the prices that would prevail in market transactions between unrelated purchasers and sellers. Tax considerations, however, may influence management to reject such a profit center or “shadow” pricing strategy. The setting of transfer prices will affect the global tax burdens of multinational enterprises because both import and company taxation of enterprises are based on input and output prices, and effective tax rates vary greatly among countries. Their discretion over price setting gives MNEs the opportunity to design sets of transfer prices for intrafirm trade which reduce their global tax liability or achieve other strategic aims, such as overcoming foreign exchange restrictions. Consequently, jurisdictions with moderate or high business income tax rates are vulnerable to tax base erosion on the revenues they receive from taxpayers with transfer pricing opportunities.

The problem for national tax authorities is to establish and enforce rules for the setting of such prices. Considerations of self-interest should lead governments to regulate transfer pricing in ways that minimize conflicts with other jurisdictions and do not discourage future investment, while, at the same time, safeguarding their revenue bases.

International Guidelines and the Arm’s-Length Standard

One way to avoid international conflict over double taxation issues and to foster a favorable business climate in the host country is for the tax authorities to adopt regulatory criteria that command the support of international bodies. Fortunately, a consensus exists among the Organization for Economic Cooperation and Development (OECD),⁷⁰ the United Nations,⁷¹ and the European Community⁷² that a single guideline should be employed both by enterprises for setting transfer prices and by taxing jurisdictions in their assessment. This guideline is the arm’s-length criterion. As will be discussed later, however, the arm’s-length criterion is not without its critics. An arm’s-length price is the market price that would have been negotiated by unrelated parties engaged in the same or similar transactions under the same or similar conditions. To prevent the use of transfer price setting for tax avoidance, most countries have established their right to adjust reported transfer prices to conform to an arm’s-length standard. Implementing the arm’s-length principle has proved to be a difficult task for both the multinationals themselves and the tax authorities who regulate them.

OECD Guidelines for Determining Arm’s-Length Price

Following closely on U.S. practice, the OECD endorses four methods of determining arm’s-length prices. These are (1) the comparable uncontrolled price method (CUP); (2) the resale price method; (3) the cost plus price method; and (4) any other method which is found to be acceptable, known collectively as fourth methods.⁷³

The comparable uncontrolled price (CUP) method assumes that a known comparable uncontrolled price exists for the sale or purchase of the same or similar goods to or from independent third parties. Market conditions associated with a proposed CUP exchange should be similar to those which characterize the actual transfer price exchange. Even if an MNE is aware of the appropriate CUP for a given intrafirm trade transaction, tax authorities in a host country may not be able to identify it owing to the informational asymmetry between MNEs and the authorities. Where no close CUP exists, authorities will be tempted to impose observable spot prices for transactions which are superficially similar. This approach can lead to serious distortion in the allocation of profit within MNEs, across taxing jurisdictions. For instance, MNE affiliates occasionally operate at a loss, particularly in developing companies, for strategic reasons such as maintaining market share or retaining security of access to raw materials during a period of weak market demand.

The resale price method takes, as its point of departure, the price at which a product passes beyond the limits of the multinational enterprise through a sale to an independent purchaser. A markup is next subtracted from this price to reflect the costs and profit margin of the foreign-based unit of the enterprise which purchased the good from the producing unit and resold it to the independent purchaser. The residual is deemed to be the arm’s-length price of the original sale. This method is most applicable to cases where the reseller does not add substantially to the value of the product—but is essentially a distributor. For the method to work effectively, comparable markups of independent distributors need to be found.

The cost plus price method is based on the supplier’s cost of the resources used in production of a particular commodity to which an appropriate profit markup is added. The cost plus method is beset by many problems. Like the resale price method, it ignores demand or business cycle conditions and fails to reflect competitive conditions; in addition, it overemphasizes historical costs. This approach is most readily applied in pricing a subsidiary’s output in cases where only one or a few products are produced, where the subsidiary performs the role of subcontractor, or where a specialized product has been produced, tailored to the needs of a specific customer—that is, where there are no important manufacturing intangibles. Finally, employing this method in a satisfactory manner requires the determination of an appropriate profit markup and the identification of appropriate costs of production.

The fourth methods of acceptable alternatives include a wide mixture of techniques with different rationales, including profit-splitting according to allocation formulas based on sales, and expected or normal returns on capital invested by an affiliate.⁷⁴

MNEs may occasionally favor the application of fourth methods in their own transfer pricing as a means of self-assessing the profit level that a host country’s revenue authority might anticipate from them. If a steady accounting profit is expected by the authorities, irrespective of changes in business conditions, an MNE may reason that it is prudent to generate such an outcome to avoid investigations or the hostility of the host government.⁷⁵

The first three methods may be effective when dealing with tangible commodities. They are, however, often unworkable in cases where payments are made to the parent corporation for the use of intangible services such as intellectual property, management fees, and other services, which are unique to the parent. Furthermore, tangible property often incorporates important intangibles: for example, not all cars are alike.

How Serious a Problem Is Transfer Price Manipulation?

How serious a problem is transfer pricing for developing countries? Statistical evidence on the growth of the intrafirm component of international trade suggests that the proportion of trade likely to involve transfer prices has increased sharply over the last two decades. Hence, the opportunity for transfer price manipulation has increased, and with it, the danger of revenue erosion from high-tax countries. For example, in 1989, 86 percent of U.S. parent-company imports were from foreign affiliates, while 89 percent of U.S. parent-company exports were destined for foreign affiliates.⁷⁶ A related important secular trend in international trade has been the growth in services as a component of total trade. With the increased importance in total trade flows of such intangible items as royalties and license fees, the scope for tax avoidance through price manipulation has increased correspondingly. It is very difficult to determine comparable uncontrolled prices for these items and even to determine where such services are performed. Therefore, manipulative price-setting of these services may be one of the most preferred avenues for global tax avoidance.⁷⁷

Nevertheless, substantive arguments can be made which suggest that MNEs do not abuse their transfer price-setting power to avoid taxes in developing countries. Multinationals may refrain from manipulative transfer-pricing practices in developing countries because (1) marginal corporate tax rates are typically lower in developing than in developed countries, and (2) in cases where the parent enterprise is resident in a country with a foreign tax credit system, corporate taxes paid in a host country can be credited against the tax liability imposed by the home country on remitted profits. At the empirical level, a recent study of transfer-pricing practices in the petroleum industry involving the United States and petroleum-exporting countries concluded that, between 1974 and 1984, MNEs in this industry did not substantially manipulate transfer prices.⁷⁸

To reach a balanced assessment, however, this sanguine picture needs to be contrasted with important qualifications and counterarguments. First, the deductive arguments ignore the role of tax havens which seek to attract the paper profits of MNEs by guaranteeing a regime of little or no taxation. Relatively stateless MNEs may locate their head offices in such havens to facilitate global tax avoidance, though real control and the source of investment lie elsewhere. In addition, MNEs with headquarters in industrialized countries may require branches and affiliates in capital-importing countries to purchase inputs through other subsidiaries located in tax havens. Second, the assumption that marginal corporate tax rates in developing countries are lower than those found in industrial countries is not valid for a number of Asian and African countries.⁷⁹ Finally, in setting transfer prices, MNEs may be influenced by considerations other than global tax minimization. They may inflate their input prices as a means of circumventing exchange controls and other restrictions on profit repatriation. They may also wish to disguise the extent of economic rents actually earned in the host economy. High rents may lead to demands for higher wages for affiliate workers and political demands that the host country revamp its tax regime to capture more economic rent.

Finally, it is important to realize that, while abusive transfer-pricing practices may affect a very small proportion of tax revenues in large industrial countries, they can affect a very significant proportion of the tax revenue of a small, poor country when there is significant foreign investment in that country.

Experience of Developed Countries with Transfer Price Regulation

While policy guidelines in industrialized countries usually favor use of the CUP method and discourage recourse to fourth methods, the empirical evidence on transfer-pricing practices suggest that fourth methods are employed more often than any other approaches. Three ex post studies of methods used in resolving U.S. transfer-pricing cases found that fourth methods were used between one third and almost one half of the cases, while the CUP method was the second most likely method to be selected.⁸⁰

Two promising approaches to curtailing transfer-pricing abuses are international cooperation through the exchange of tax information between revenue authorities in different countries, and advance-pricing agreements (APAs). An advance-pricing agreement provides a means of avoiding uncertainty about the acceptability of MNE practices. This approach, adopted by the United States in 1991, guarantees the enterprises concerned advance approval of their transfer-pricing methodologies, if adequate documentation is provided. The recent report of the European Community on company taxation, known as the Ruding Report, has endorsed the practice of advance-pricing agreements between the tax authorities and MNEs. The Ruding Report suggests that countries involved in intrafirm trade might conclude advance transfer-pricing agreements between each other. This practice has already been initiated by Australia and the United States.⁸¹ Simultaneous audits of affiliates of a single MNE in two or more countries have also been undertaken.

Experience in Developing Countries

Tax administrations in most developing countries are poorly equipped to cope with transfer-pricing issues. In particular, they tend to lack the technical expertise needed to challenge prices set by MNEs using the methodologies recommended by international bodies for the establishment of arm’s-length prices. In many cases, developing countries appeal to international guidelines when making transfer-price adjustments halfheartedly. Tax statutes in many developing countries often equate arm’s-length dealings with un-controlled market prices and omit rules for establishing transfer prices when comparable uncontrolled market prices are unavailable. In such situations, the administrative response is often to make arbitrary price adjustments or adjustments which are the outcome of implicit bargaining, while seeking to cloak actual practice under the veil of internationally approved guidelines.

Some developing countries have enacted ad hoc and simple rules which do not provide comprehensive solutions, but nevertheless address particular facets of the problem. These approaches include alternative minimum corporate taxes based on turnover, use of royalty fees based on the volume of natural resource extraction, and the disallowance of deductions, such as royalty payments and interest payments on equity infusions by foreign-based parents.

Examples of these strategies pursued by some Latin American countries are disallowing the deduction of royalties paid to related foreign entities for technological contributions, restrictions on the size of deductions for expenses incurred abroad, and establishing presumptive income for certain activities such as international transport, where the use of separate accounting is particularly difficult.⁸² These rules have had mixed success. Those that disallow the deduction of royalties or limit deduction of expenses have proved ineffectual, because MNEs will find other methods to repatriate profits from host countries without incurring significant tax liabilities.

Policy Lessons for Developing Countries

As a first step in articulating regulatory policy, developing countries should enact legally enforceable measures against those transfer-pricing practices which are intended to erode their tax bases. In designing these measures, authorities should consider their own national legal traditions. The enactment of transfer price statute provisions will signal that the host government intends to combat manipulative pricing practices and may thereby dissuade potential tax evaders.

If the revenue authorities in a developing country are unable to obtain comparable uncontrolled prices, which of the other two methods sanctioned by OECD guidelines—the resale price method and the cost plus price method—should they use? In principle, the option that should be preferred depends on where the intangibles are located. But practical considerations are likely to be more important in many instances. For example, a developing country will generally have access only to the records of the subsidiary. If a manufacturing subsidiary in the developing country sells to its overseas parent for distribution, the developing country will be unable to apply the resale price method. Nevertheless, developing countries have had some success in adjusting MNEs’ transfer prices under the re-sale price method, when the output is a homogeneous product quoted on an international exchange such as the London Metal Exchange.⁸³ These approaches should be used with caution, if the observed commodity exchange price is for a refined commodity many stages removed from the intrafirm trade transaction. In those cases, adjustment should be made for the downstream value added associated with refining and transportation. Failure to make the necessary adjustments for intermediate value added may produce serious tax distortions and discourage production.⁸⁴

Country experience suggests that royalty and management fees have been an attractive avenue for remitting profits to the capital-exporting country. Many capital-importing countries employ formula restrictions both on the deductibility of such payments and on the actual repatriation of funds under these categories. A less drastic alternative is to impose withholding taxes at moderate rates on nonarm’s-length royalties and other fees paid abroad. Quantitative restrictions or outright disallowance might still be warranted in cases of such payments to affiliates in known tax havens. While the OECD model tax treaty advocates that royalty payments abroad not be subject to a withholding tax, it does so only for the case of arm’s-length transactions.

Revenue authorities should be diligent in identifying sham transactions and questioning transactions involving the purchase of services from related companies in known tax havens. They should also explore opportunities for international cooperation in the form of tax information exchanges or even simultaneous audits with their counterparts in countries where the parents of their most important subsidiaries reside. These developments can be facilitated by negotiating provisions for exchange of information and other types of cooperation in tax treaties.

Countries may wish to make provision in their legislation for advance rulings. This option should be attractive to a great many MNEs because it would avoid capricious outcomes, wasteful litigation and the un-certainty associated with prolonged disputes. When advance ruling and negotiable agreement are difficult to conclude, host countries may wish to explore the use of international arbitration services such as those provided by the United Nations.

In cases where import tariffs are high or where the enterprise becomes liable to sales taxes on its imports, an MNE may attempt to minimize the price of imported goods by underinvoicing. Customs and income tax departments need to cooperate closely in their tax surveillance activities to ensure that subsidiaries do not adopt a double invoicing system with low prices for customs and high prices reported to the income tax authorities.

Critics of the arm’s-length criterion emphasize its lack of workability, particularly for developing countries, the long delays in resolving disputes arising from its application, and the high costs incurred by tax authorities who attempt to apply it rigorously. They offer as an alternative the universal apportionment method which would assign the global profits of MNEs among tax jurisdictions based on apportionment formulas.

These formulas might contain as components the wage bill or value added of affiliates, the book value of assets, and sales. While such an apportionment approach has been successfully implemented at the state level in federal countries, it has not been applied in the international setting except by the countries of the Andean Pact. Using a universal apportionment formula to allocate the global profits of enterprises operating in both high- and low-wage countries will not produce a desirable tax allocation from the perspective of the latter group of countries, if the wage bill is an important component of the formula. Moreover, regionally based initiatives to replace the arm’s-length standard with formula apportionment would be likely to lead to different formulas in different jurisdictions.

Despite the shortcomings of current international guidelines on transfer pricing, they are unlikely to undergo radical alteration in the future. Therefore, the most effective strategy for developing countries that wish to safeguard their tax revenue bases against abuses of international transfer prices is to devote adequate resources to transfer-pricing regulation within the framework of international guidelines.

Defining Branches, Subsidiaries, and Permanent Establishments

A foreign-based enterprise which carries on business operations in a host country, but has not incorporated in the host country, has established a branch. Under a branch, the foreign enterprise usually faces unlimited liability for the debts and other legal obligations of its branch. In contrast, if the foreign-based enterprise incorporated its business in the host country, then a subsidiary exists. Under the subsidiary, liability is limited to the assets of the subsidiary in the host country. An important related concept is that of a permanent establishment.

A permanent establishment, as defined in article 5 of the OECD Model Tax Treaty, is a fixed place of business through which the business of an enterprise is wholly or partly carried on.⁸⁵ This definition excludes certain activities of a foreign resident in a host state such as conducting business in that state through a broker or general agent or maintaining inventories of goods in the host state for delivery to an agent or purchaser. Once a foreign person acquires a fixed place of business or employs resident individuals, the enterprise is deemed to have set up a permanent establishment. The income producing activities of a business are generally taxable once that business sets up a permanent establishment in a tax jurisdiction. A corollary of this guideline is that foreign persons should not be subject to branch taxation if they have not engaged in activity giving rise to a permanent establishment.

In general, international conventions and bilateral treaties take the view that the profits to be attributed to a permanent establishment are those it would have made had it not been dealing with its own head office, but with an entirely separate enterprise under arm’s-length conditions. If separate accounts do not exist, then revenue authorities will apply the arm’s-length principle in ascribing a profit to the establishment.

Advantages and Disadvantages of Operating as a Branch or a Subsidiary

Rationale for a Branch Profits Tax

Many jurisdictions impose a branch profits tax on foreign-based branches to achieve greater neutrality and equity in treatment between branches and subsidiaries and to forestall revenue erosion. Without such a tax, repatriation of branch profits does not give rise to tax liability, unlike the repatriation of dividends by a subsidiary.

A simple branch profits tax at the same or lower rate than the withholding rate on foreign dividends can achieve approximate neutrality between foreign branches and subsidiaries operating in the host country. A lower rate is justified if a branch profits tax is imposed on profits in the year or quarter in which they are earned, whereas the dividend withholding tax is only applied to remittances to shareholders. Countries that impose this form of tax often choose a design which exempts from taxation any income that is rein-vested in the branch. This approach has been adopted under the U.S. 1986 Tax Act.

Some experts advocate that the design should not emphasize internal formal harmonization of the tax system at the expense of administrative simplicity.⁸⁹ For example, the Canadian branch profits tax, which seeks to achieve rigorous parity between the taxation of subsidiaries and branches, involves the use of concepts which have no counterpart in standard accounting. Developing countries, looking for pragmatic solutions with greatest simplicity, should be able to achieve approximate parity with a simple surcharge on branch profits. For example, in the case of a 40 percent income or enterprise profits tax and a 20 percent withholding tax on distributed dividends, the total tax liability on a subsidiary’s income will be 52 percent, assuming full distribution of income. This example suggests that a 12 percent surcharge on branch profits is appropriate. A somewhat lower surcharge rate is appropriate to recognize the deferral advantage enjoyed by subsidiaries, because such a surcharge would be applied immediately, whereas the withholding tax would apply only upon repatriation; it would also recognize that subsidiaries do not generally repatriate all of their profits.

Instituting a branch profits tax on foreign-based permanent establishments may violate the nondiscrimination provisions of an existing tax treaty. In such cases, current treaties will have to be renegotiated. Introducing such a tax also increases the need to restrict deductible interest disbursements to foreign parties.

Preventing Thin or Hidden Capitalization

Thin or hidden capitalization of a subsidiary arises when a foreign investor responds to tax incentives by substituting foreign debt capital for equity capital financing, particularly in cases where the debt financing exhibits some of the characteristics of equity and the debt is owed to a nonarm’s-length lender. While a high debt/equity ratio per se should not be taken as proof of hidden equity capitalization, a high ratio may indicate efforts to achieve a tax advantage through excessive debt financing. This type of financing enables a foreign-owned subsidiary to reduce its taxable profits by deducting interest payments to nonresident creditors to which only nonresident interest withholding taxes would apply. By contrast, if the profits of a subsidiary are transferred by dividends, such payments will not be deductible in calculating the subsidiary’s taxable profits. In addition, to avoid corporate income taxation on what would otherwise be profits, the thin capitalization strategy may enable the parent company to avoid or defer taxation in the home country.

Multinational enterprises can often use thin capitalization techniques in combination with treaty shopping to reduce their global tax burden, because of different treaty provisions for withholding taxes on dividends and interest. Treaty shopping occurs when a combination of treaty and national tax laws involving more than two countries create a favorable tax regime which can be exploited by residents of nontreaty countries.⁹⁰ For example, interest payments might be routed through one or more holding companies in a series of countries linked by suitable treaties, so that the return on investment can be shifted from the original source country to a country where it will bear little or no tax.

Thin or hidden capitalization tax rules are designed to restrict such behavior. Three general approaches have been adopted by countries to restrict thin capitalization. These are (1) the denial of a deduction for interest expense on any related party loan; (2) general anti-abuse arm’s-length approach; and (3) acceptable capital-to-debt ratio approach.

The denial approach discriminates in an arbitrary fashion because it penalizes debt between related parties relative to other forms of debt. It also encourages avoidance by disguising related party debt as unrelated debt. Policing such abuse is difficult, particularly for countries with limited administrative resources.

The anti-abuse approach seeks to look behind the form of a transaction, in light of all relevant circumstances, to see whether the real nature of the contribution is debt or equity. Each debt transaction must be examined separately to determine whether the terms of the debt instrument are those which would have been entered into by unrelated parties.

Under the fixed-ratio or capital structure approach, a fraction of interest paid to foreign creditors is disallowed as a deductible expense when the debt/equity ratio exceeds a specified level. The disallowed portion of interest is determined by prorating with the excess debt/equity ratio. In measuring the equity portion of a capital structure, most jurisdictions will find it administratively difficult to insist on the annual fair market valuation of corporate equity. Therefore, a feasible alternative is to define equity as the sum of retained earnings at the beginning of the year, contributed surplus and paid-up capital, though this approach will not be appropriate for jurisdictions with high rates of inflation. The fixed-ratio approach may be applied universally or only to transactions between related parties. Oldman, Rosenbloom, and Youngman advocate that thin capitalization restrictions apply not only in cases of potential outright tax evasion but also to bona fide shifts in capital structure which would transform what would otherwise be dividend payments abroad into deductible foreign disbursements.⁹¹ Universal limitation rules should be more effective than rules which permit exemptions and give unnecessary discretion to tax administrators in host countries.

Financial theory and practice suggest that the optimal appropriate ratio of debt to equity varies from industry to industry and from company to company. Countries which employ this approach sometimes provide a higher deductible debt/equity ratio for financial institutions, though this refinement introduces additional administrative complexity.

The ratio approach would appear to be a more appropriate tax instrument for developing countries in combating thin capitalization than the anti-abuse or denial approaches, because of its limited use of tax assessing resources and the lack of administrative discretion.⁹²

Recently, a fourth alternative has been suggested for economies in transition experiencing high rates of inflation. Under this approach, the deduction of interest is limited to a specific percent of taxable income gross of interest expenses, rather than being prorated on the debt/equity ratio. To avoid penalizing financial institutions, a variant of this rule would allow all interest incurred to be deducted up to the amount of interest income. The justification for this approach is to prevent both foreign- and domestic-based enterprises from stripping earnings. This approach may be preferable to the ratio approach when accounting equity cannot readily be modified to reflect inflation-induced changes in fair market value.