Journal Issue
Share
Article

Trade Liberalization in Developing Countries

Author(s):
International Monetary Fund. Research Dept.
Published Date:
January 1991
Share
  • ShareShare
Show Summary Details

One of the main objections to proposals of trade reform is that there is an inherent conflict—in the short run, at least–between lowering import restrictions and achieving balance of payments objectives. In particular, because a reduction in the level of protection may be expected to bring about a worsening of the trade balance, it has been argued that liberalization may not be a viable policy option for countries that face foreign exchange or borrowing constraints. That balance of payments considerations are viewed as playing a central role in the decision of whether to liberalize trade flows is brought out clearly in a recent comprehensive study of trade reform in the developing countries (Papageorgiou and others (1986)), which concludes:

[There is] overwhelming evidence attesting to the inference that the fate of a liberalization policy is determined, first and foremost, by developments in the balance of payments position, A significant deficit, involving a substantial loss of foreign exchange reserves, is most likely to abort a liberalization attempt. Without such a loss, on the other hand, even in the presence of other economic hardships in one form or another, liberalization is likely to be sustained. The authors of ten country studies—Argentina, Brazil, Chile, Indonesia, Korea, Pakistan, Peru, the Philippines, Portugal, and Spain—explicitly reach this basic conclusion (pp. 14-15).

The notion that tariff reductions are likely to bring about a worsening of the external position has its origins in the simple idea that lower tariffs lead to a switch in consumption patterns from domestic toward foreign-produced goods, and hence to an increase in imports. With exports being determined by demand in the foreign country, and thus unaffected by the tariff, a tariff reduction would necessarily reduce the trade surplus (where the change in the latter is identically equal to the increase in imports).1

More recently, researchers have argued that, in order to understand the connection between the external current account balance and the level of protection, one needs a model of saving and investment. Since the current account is identically equal to the difference between national saving and investment, a reduction in trade restrictions can only be expected to affect the current account if it brings about a differential response of saving and investment flows. The switch in focus from exports minus imports (net exports) to saving minus investment highlights the potential importance of taking into account the intertemporal effects of policies—including trade liberalization—on the current account.

In contrast to much of the earlier literature, the conclusions that emerge from the saving-investment approach do not suggest that tariff reductions will necessarily lead to a worsening of the external position. Most of the analysis within this approach has focused on the effects of liberalization on saving, with investment effects usually ignored,2 In papers by Razin and Svensson (1983) and Calvo (1987, 1988, 1989), for example, it was emphasized that the effects on the current account of trade reforms depend on how the public perceives their implementation through time. On the one hand, liberalizations that are perceived as permanent are unlikely to result in sharp movements in the current account in one direction or another.3 In addition, a phased reduction in tariffs, in which the rate in future periods is expected to be below today’s rate, might even lead to an improvement in the current account. On the other hand, if lack of credibility is an important characteristic of actual liberalizations, then this would lend support to the view that lowering tariffs will result in a worsening of the external position.4 If liberalization policies are expected to be reversed in the future, consumers will increase their demand for imports (reduce saving) while tariffs are temporarily low, thereby contributing to a deterioration in the trade balance.

Extensions of this basic analysis—again focusing on the saving side of the current account—have also supported the view that trade liberalizations have ambiguous effects on the external balance. Edwards (1988, 1989) and Edwards and Ostry (1990) showed that the incorporation of nontradable goods could cause the comovement between liberalization policies and the current account to become ambiguous.5Murphy (1986) examined the issue of how the accompanying fiscal policies affect the current account response to trade liberalization. He showed that, if the government uses tariff revenues to finance its own consumption, then the current account effects of trade liberalization—which are generally ambiguous—depend on the commodity composition of government spending. 6Rodrik (1987) considered how tariff reductions would affect the current account in economies with imperfectly flexible labor markets. Once again, the conclusion emerged that liberalizations had ambiguous effects on the current account, which depended in this case mainly on the relationship between changes in the tariff and the level of employment.7

Thus, it is apparent that a large body of recent theoretical research argues against the view that trade liberalizations will necessarily cause the external balance to worsen. The agnostic tentative conclusion—that liberalizations may be expected to have ambiguous effects on the current account, which depend on a range of factors including the expected time path of tariff reductions, the importance of nontraded goods, the role of accompanying fiscal policies, and the extent of labor market rigidities—is also consistent with some available empirical evidence relating to the developing countries. For example, in a multicountry study of trade reform in developing countries undertaken by the World Bank (Thomas (1989) and Thomas and others (1990)), it was found that net exports as a percentage of gross domestic product (GDP) actually rose in the period following reform in comparison to the pre-reform period. Further, relative to a control group of developing countries, net exports of the trade reformers also increased. These results, which are of course subject to many caveats (see Thomas (1989, pp. 17-18) and Khan (1990)), suggest that there is no presumption, based on empirical evidence, that reducing trade restrictions systematically results in a worsening of the external position. 8

While previous theoretical research tends to support the view that trade liberalization need not worsen the current account, this paper argues that much of this earlier analysis cannot be applied in a straightforward way to the developing countries, because there are reasons to believe that conditions in these countries depart in important ways from those that are postulated in much of the existing literature. Two such reasons are central to the analysis that follows. First, in contrast to the circumstances of most developed economies, trade liberalization in developing countries begins from an initial position of relatively high levels of protection. 9 Thus, the assumption of low (effectively zero) initial trade distortions frequently adopted in the theoretical literature is not appropriate for an analysis of the effects of liberalization in developing countries.

Second, imports to developing countries tend to be dominated by intermediate products, whereas previous theoretical literature on the relationship between tariffs and the trade balance has concentrated on trade restrictions on final products.10 Evidence presented in Thomas (1989) suggests that only about one fifth of developing country imports consist of final goods, with the remaining four fifths accounted for by imports of intermediate and capital goods. 11 Intermediate inputs alone appear to account for about half the imports of the typical developing country. Moreover, there are indications that liberalizing trade in intermediates has received more prominence in actual trade reform packages pursued by the developing countries than have tariff reductions involving final products, which makes an analysis of this type of measure all the more relevant. 12

Finally, one consequence of examining the relationship between tariff reductions on intermediates and the external balance is that the response of investment will play an important role. While much of the previous literature has focused exclusively on saving effects, it will be argued below that the response of investment to a trade liberalization involving intermediate inputs may be no less important in determining the overall impact on the current account. For this reason, and in contrast to much of the previous literature, the analysis that follows provides a detailed discussion of how such tariff changes affect investment. 13

The main theoretical conclusion of the paper is that the effect on the external balance of trade liberalization involving intermediate inputs depends on both the initial level of protection and on the economic structure (particularly, relative factor intensities across various sectors). From a policy perspective, this conclusion implies that, if one accepts the argument that trade reforms will necessarily lead to a deterioration in the external position, one must also be making a judgment about the economic structure of the economy undergoing trade reform. Given the likely diversity of economic structures and initial levels of protection that exist among the developing countries, it seems unlikely that the pessimistic policy conclusion—that trade reform is not a viable option for economies operating under foreign exchange or borrowing constraints— would apply with equal force to all developing countries at all times.

The remainder of the paper is organized as follows. In Section I, an intertemporal, optimizing, general equilibrium model of a small open economy with initial trade restrictions is developed to investigate the effects of reducing tariffs on intermediate inputs on a variety of macro-economic variables including the real exchange rate, investment, saving, the current account balance, and economic welfare. The main analytical results are presented in Section II—first, in the case of permanent or fully credible liberalization policies, and then, in the case of temporary or imperfectly credible policies. Throughout the section, a number of special cases are presented in order to illustrate some of the important channels through which tariff reductions on intermediates affect the macroeco-nomic variables mentioned previously. Finally, Section III contains the conclusions.

I. The Analytical Framework

The model developed below is the simplest one capable of addressing the main issues that were set out in the previous section. The model is completely real (that is, monetary considerations are not considered), and consists of optimizing agents (producers and consumers) who maximize an intertemporal objective function (profits or utility), subject to technological or budget constraints. 14 In order to analyze saving and investment decisions in as simple a manner as possible, an intertemporal structure is chosen in which there are only two periods, the present (period 1) and the future (period 2). 15 There is no uncertainty, and agents are assumed to have rational expectations (perfect foresight) with respect to all future-period variables.

In addition to two tradable final goods (importables and exportables) and an imported intermediate input, the model incorporates a nontrad-ables sector. The country is assumed to be small in world markets and, therefore, it takes the world prices of all tradable goods as given.16 The inclusion of a nontradables sector permits an analysis of the effects of trade liberalization on the real exchange rate—defined as the relative price of the exportable in terms of the nontradable good—an issue that has been given considerable prominence in the policy literature. 17 Further details of the model are provided in the remainder of this section.

Supply

The supply side of the model consists of competitive firms whose objective is to maximize the present value of current and future profits from production. Four types of goods are considered. Exportables (X), importables (M), and nontradables (N) are produced domestically; while the supply of an intermediate input (m), which is required in the production of final goods X, M, and N, is met entirely by imports from abroad. Profit-maximizing firms produce output of final goods using primary factors (labor, capital, and land) 18 and the intermediate input. It is assumed in what follows that land is sector specific, while labor and capital are mobile across sectors. The reason for making the assumption that capital is intersectorally mobile is that part of the effect of liberalization on the current account will arise through producers’ decisions to allocate capital differently across sectors in response to relative price shifts. In order to study such shifts, it is necessary to assume some degree of intersectoral capital mobility.

The presence of additional factors—other than labor and capital—is important, because, as is well known from the theory of international trade, the assumption that the number of tradable goods (three—an importable and exportable final good and an imported intermediate input) is less than the number of (internationally) nontradable factors (five—labor, capital, and three types of sector-specific land) ensures that the price of nontradable goods will be endogenously determined in a small open economy by the interaction of demand and supply for such goods. 19 This assumption is therefore appropriate if one wishes to analyze the real exchange rate effects of trade liberalization.

With such large numbers of goods and factors, it proves fairly cumbersome to solve the model by directly working through the first-order conditions for profit maximization. A convenient alternative, however, is to use the so-called dual approach (see, for example, Dixit and Norman (1980)). Under this approach, the output supply and intermediate input demand functions are simply the partial derivatives of the economy’s revenue or value-added functions, which are defined as the maximum value of output (net of the cost of the intermediate input), given prices and endowments of factors. Thus, if Ri is the revenue function in period i, and pi and vi are vectors of prices and factor endowments in period i, respectively, with Pji being the j th element of pi, then the supply Sji, in period i of the good whose price is Pji is given by

Defining p i as the vector [1, pi qi, ri], where 1 is the price of the exportable (the numeraire), pi is the price of the importable, qi is the price of the nontradable, and ri is the price of the intermediate input, then

In other words, the demand for intermediates is equal to minus the partial derivative of the revenue function with respect to the price of the intermediate , ti, which is the fourth element of the vector pi20

A further point in connection with the supply side of the model relates to investment. First, it is necessary to assume that one of the three goods is either a pure investment good or, as is more conventional, a composite good that may be used either for investment purposes (that is, to augment the future capital stock) or for current consumption. In order to simplify the analysis, it will be assumed that the composite good corresponds to the numeraire good (X)21

Second, since firms are assumed to maximize the present value of profits from investment, in equilibrium the discounted value of the increase in value added in period 2 from a small investment in period 1, δ∂R2/∂l, is equated to the price of a unit of the investment good, which is unity. Here, δ is the discount factor equal to one divided by one plus the world rate of interest in terms of good X, and I is the level of investment. The condition may be stated formally as

where the initial capital stock, denoted k1, is assumed not to depreciate between the two periods. Equation (3) may be used to define an optimal investment level, I (.), as a function of all variables that affect value added in period 2, namely the relative prices of importables, P2, of nontradables, q2, and of imported intermediates, r2, the vector of factor endowments of land and labor, z2, and the discount factor, ∂:

Standard properties of the revenue function imply that an increase in the discount factor (a fail in the rate of interest) raises the optimal investment level; that is, ∂I/∂δ = I> 0. 23 In what follows, attention focuses not so much on the effects of interest rate changes or factor supply changes (changes in the z2 vector), but rather on how changes in the domestic price of intermediates brought about by changes in commercial policies affect investment.

Since intermediates are an input, along with capital, into the production process, it should be clear that the effect on investment of a change in the tariff on intermediates depends on the technological relationship (whether complementary or substitutable) between these two inputs. If the two inputs are net complements as is conventionally assumed (see, for example, Svensson (1984, pp. 652 and 659)), then a tariff reduction in period 2 (a fall in r2)will tend to raise investment. This will be the maintained assumption for the remainder of this paper, so that ∂I/∂r2 = Ir2 <0.

It should be noted that the relative price of nontradable goods is also an endogenous variable that will respond to commercial policy changes. The effect of changes in q2 on investment will in general be determined by relative factor intensities across the various sectors. To take an example, suppose nontradables are more capital intensive than tradables. Then a rise in q2(a real appreciation) will increase investment. The reason is that the rise in q2 shifts resources from tradables to nontradables, where the latter are—relative to the rest of the economy—intensive users of capital. This tends to raise the demand for period 2 capital, which stimulates investment. The opposite would hold if tradables were relatively capital intensive. 24

Demand

As far as demand is concerned, consumers are assumed to maximize utility, subject to the constraint that the present value of their expenditures does not exceed the present value of their resources. It is assumed that the utility function is weakly time separable, with each period’s subutility function being homothetic. The motivation for this assumption is that it permits a rigorous definition of within-period price indices, which measure the cost of the consumption basket in each period. Given world interest rates, movements in the within-period price indices determine the relative cost of current, in terms of future, consumption, or the consumption rate of interest. The latter, in turn, is a key determinant of saving behavior and, hence, of the current account.

The relevant information for optimal consumption choice may be summarized in the present value (or lifetime) expenditure function:

which gives the minimum lifetime expenditure necessary to achieve utility level, W, for a given set of prices. As can be seen, E(.) is separable between first and second period prices, reflecting the underlying separability of preferences. The functions, πi(1, pi, qi), i = 1, 2, correspond to exact price indices for each period’s consumption basket.

By analogy with the supply side, if one defines Pi to be the vector of prices, [1, pi, qi], with pji as its jth element, then the demand, Dij, in period i for the good whose price is pji is given by

For example, the demand for nontradables in period 1 is given by similarly,•E•q2 equals the consumption of good N in period 2.

In addition to the demand for a single good, one may be interested in the demand for total consumption in a given period. If C1 denotes real consumption spending in period 1, then, by analogy with equation (6)

since π1 is the price of the consumption basket, C1. It follows, therefore, that the value of nominal spending (that is, measured in units of the numeraire good) in period 1 is given by the product of the price index in period 1, π1, and C1; that is, π1Eπ1. Finally, it follows from standard properties of the expenditure function that all goods must be intertemporal substitutes. This means that an increase in π2 must raise real spending in period 1, so that Eπ1π2 > 0. Equally, if one considers a particular good, it must be the case that Eq1q2> 0.

Government

Since the main focus of this paper is on commercial policies, activist fiscal policies are not considered. 25 Accordingly, the sole function of the government in this model is to levy tariffs and to redistribute the resulting revenues to consumers in a lump-sum fashion. This allows one to focus on the important substitution effects (both intratemporal and intertemporal) from trade liberalizations rather than on the combined effects of budgetary and commercial policies.

Accordingly, the budget constraint for the government states that the present value of transfers to the public, G, equals the present value of the revenues from levying tariffs. If ti is the tariff on intermediates in period i, and Ti is the tariff on final good imports, then

where, as mentioned previously, Rjiis the derivative of the revenue function, Ri, with respect to its jth argument (so that, for example, Ri2 = ∂ Ri(1,pi, qi, ri;zi, ki)/∂ pi equals the supply of importable s in period i, with i = 1,2). 26 From the properties of the revenue and expenditure functions discussed previously, the first two terms represent the present value of revenues from tariff collections on final goods, while the last two terms represent tariff collections from imported intermediates. Finally, in this expression, the world prices of importables and intermediate inputs have both been normalized to unity by suitable choice of units.

Equilibrium

The first condition that must hold in equilibrium is the economy-wide budget constraint, which states that the present value of utility-maximizing consumption, E, plus the profit-maximizing investment level, I, be equal to the present value of income from production, R1 + ∂ R2, and rebated tariff revenues, T1(Ep1 - R12) + ∂ T2(Ep2 - R22) - t1R14 - ∂ t1R24, namely:

Clearly, equation (9) allows for trade imbalances in each period but requires that the discounted sum of these imbalances be equal to zero. 27 While equation (9) represents the requirement of external balance (or intertemporal solvency), internal balance is achieved when the market for home goods clears in each of the two periods:

In equations (10) and (11), the left-hand side represents the demand for nontradable goods in a given period, while the right-hand side represents the corresponding supply. 28

Equations (9) through (11) summarize the model’s equilibrium. The three endogenous variables are the level of welfare, W, and the two prices for nontradables, q1 and q2. In general, one may expect that commercial policy changes will influence all three of these variables. Once the effect of tariff changes on real exchange rates (reciprocals of q1 and q2) and welfare is known, the current account (ca)effects of such policies may also be derived, by differentiating the following expression: 29

Thus, the current account in period 1 is equal to saving minus investment. Saving, in turn, is equal to the difference between income (both from production and rebated tariff revenues), R1 — t1R14 + T1(EP1 - R12), and consumption, π1E ππ1. 30 As can be seen, ca depends on all the endogenous variables of the model (through the revenue and expenditure functions) and directly on the various tariff rates. By combining equation (12) with the economy-wide budget constraint (equation (9)), it may be verified that any trade deficit in period 1 must be offset by a trade surplus of equal present value in the second period. 31

II. Tariff Reductions on Imported Intermediate Inputs

This section presents the main comparative static results for the effects of tariff reductions on imported intermediate inputs. The case of fully credible (permanent) liberalizations is considered first, followed by the noncredible case. In the course of the discussion, a number of specific examples are examined, which help to focus on some of the principal channels through which this type of liberalization affects the main macroeconomic variables of interest.

Before proceeding with the details, it is useful to introduce the discussion by contrasting the analysis with the more usual one involving tariff reductions on final goods. In much of the previous literature, a tariff reduction on final goods affects the current account mainly as a result of an intertemporal substitution effect on saving. Reducing the tariff on imports directly lowers the cost of the consumption basket by lowering the domestic relative price of importables faced by consumers. If the trade liberalization is regarded as temporary (or noncredible), imports, and, hence, the consumption basket, are cheap today relative to their expected price in the future. This tends to raise demand for imports (and, other things equal, increase total consumption), hence worsening the trade balance. In contrast, a liberalization that is regarded as permanent will not have much effect on saving (and, hence, the trade balance), because the future price of imports is not expected to differ much from the current price. For this reason, consumers do not perceive an advantage in “dissaving” or borrowing in order to finance purchases of imports today, and the trade balance does not deteriorate.

In the case of trade liberalization involving intermediate inputs, tariff reductions do not directly affect the price of any final good. Thus, the impact on the current account, if there is one, must be through a different channel. There are, in fact, three main channels through which a reduction in the tariff on intermediates affects the current account: (1) real income or wealth effects on saving; (2) intertemporal substitution effects on saving; and (3) investment effects. The wealth effect arises because a tariff reduction allows for a more efficient combination of primary factors and intermediate inputs to be used in the production of an existing output mix; it also allows for a more efficient output mix. Although the first mechanism is fairly clear-cut—liberalization allows producers to use more “intermediate-intensive” production techniques than were previously available—the second is perhaps more subtle. It arises because, with different sectors using intermediates in different intensities (owing to differences in their production technologies), the tariff on intermediates will in general lead to a mix of outputs that does not maximize value added for the country. Important determinants of this output-mix effect are, first, relative factor intensities across sectors, and second, movements in relative prices of final goods between sectors (essentially, movements in the real exchange rate). The latter determines the direction in which resources move when trade is liberalized, while the former determines the extent to which this resource movement increases national income.

How will these movements in real income affect saving? The answer depends on the distribution of gains over time. If they are concentrated in the present, consumption smoothing dictates that saving increases. Conversely, if they are concentrated in the future, saving will decline.

The intertemporal substitution effect on saving arises because liberalization causes the equilibrium real exchange rate (the rate that simultaneously clears the nontradables market—internal balance—and ensures that the discounted sum of trade imbalances is zero—external balance) to change. This, in turn, affects the cost of present consumption in terms of future consumption—the consumption rate of interest—-and, hence, saving.

Two main factors influence the behavior of the real exchange rate— relative factor intensities and welfare effects. If nontradables use intermediates intensively, reducing the tariff on imported inputs causes the supply of nontradables to rise, and their relative price to fall—a real depreciation. The opposite holds if tradables are intensive users of intermediates. In addition, however, liberalization tends to raise the economy’s welfare or real income level. This causes demand for nontradables to increase, which puts upward pressure on their price, thus favoring a real appreciation.

Finally, it may be noted that intertemporal substitution effects are also important in determining real exchange rate behavior. If the price of nontradables falls today, agents will reduce their demand for nontradables in the future and consume more (relatively cheaper) home goods today, as long as there is some degree of intertemporal substitution in consumption. The resulting incipient excess supply of future nontradables favors a reduction in their relative price. Thus, the intertemporal substitution effect generates a positive comovement between the current and future price of nontradables. Of course, to determine the overall effect on saving, it is necessary to know whether the price of the consumption basket rises by more or less in the present than in the future. This depends both on the relative degree of appreciation or depreciation of the real exchange rate in each period, and on the relative magnitudes of the expenditure shares on home goods in the two periods.

As to investment effects, the previous section suggested how changes in the future tariff influence investment behavior today. What was not discussed, however, was the fact that, because of the intertemporal substitution effects mentioned above, liberalizations today could influence the real exchange rate in the future and, hence, investment behavior. For this reason, the factors influencing the real exchange rate—factor intensities and welfare effects—will also affect the response of investment to tariff reductions.

With these preliminaries in hand, it is now possible to proceed with the main derivations. With the initial position being characterized by positive tariffs on intermediates in both periods (assumed to be the only distortion), it is necessary to distinguish the effects of permanent liberalizations from those of temporary (or noncredible) policies. Accordingly, the case of a permanent liberalization will be considered first, followed by a consideration of the temporary case. It will be seen that the implications for the current account of noncredible policies involving intermediate inputs are quite different from those considered in the previous literature dealing with liberalization of trade in final products.

Permanent Liberalizations

Consider an initial equilibrium with a positive tariff on intermediates in both periods (t1, t2 > 0), The government announces a permanent tariff reduction of amount dt in both periods. In this case, the wealth equivalent of the welfare change is given by

Thus, the welfare change is proportional to the period 1 distortion, t1, and the change in the volume of imported inputs in that period, [R442(R462)2/R662+(R432R362R462/R662)dq2/dt]. The intuition of these expressions is as follows.

Consider first the change in the volume of imports of intermediates in period 1. There are two effects. First, there is a direct increase in import levels associated with the reduction in their relative price (represented by the own price effect, R144 > 0 in equation (13)); second, there is an indirect effect operating through the response of the real exchange rate (RER) in period 1. This indirect effect, which is ambiguous in sign, depends on both the RER response, dq1/dt and on the sign of the Rybczynski derivative, R143. 32 The latter is negative (positive) if, relative to tradables (nontradables), nontradables (tradables) use intermediates intensively. 33 The intuition is that if R143< 0, the tariff on intermediates results in too few nontradables being produced and consumed. If lowering the tariff on intermediates results in a depreciation of the RER (dq1/dt > 0), production will shift away from nontradables and an initial distortion will be magnified. In this case, the indirect effect operating through the RER will mitigate against the direct increase in welfare from lowering tariffs. Finally, the adjustment in the volume of imports of intermediate inputs in period 2 reflects these same channels and, in addition, it also reflects changes in the level of investment carried out between periods 1 and 2. 34

To solve for the reduced form of the welfare change, one needs to substitute for the RER effects, which, upon differentiating the equilibrium conditions, may be given by

where

and ▵ is the determinant of the matrix multiplying the RER vector from equations (10) and (11), which is positive.35 It may be noted also that substitution of equations (14) and (15) into (13) yields the result that equiproportionate reductions in t1 and ∂ t2 necessarily increase welfare (see Appendix for an outline of proof).

The intuition of equations (14) and (15) is straightforward. The first two terms are substitution effects, while the last two are income (or distortion-magnification) effects. Consider first equation (14). The first term gives the effect on q1 of the reduction in t1, while the second term represents the effect of the reduction in t2 on q1. Accordingly, the reduction in t1 increases the supply of nontradables if the latter use imported inputs intensively (R243 < 0), and conversely. Therefore, if R236 < 0, this within-period substitution effect generates a real depreciation in period 1, and conversely.

Analogously, the reduction in t2 raises the supply of period 2 nontradables if these are intensive users of intermediates (R243 < 0). In addition, however, the fall in t2raises the period 2 marginal product of capital (under the assumption that intermediates and capital are complementary inputs). If nontradables use capital intensively (R236 > 0), then the fall in t2 has a secondary impact on the supply of period 2 nontradables, which operates in the same direction as the direct effect, and conversely. If, taking into account the indirect effect operating through investment, nontradables are relatively intensive users of intermediates, then R243 - R236R246/R266 < 0, and conversely.

Consider the case where R243 - R236R246/R266 < 0, Then the reduction in t2 increases the supply of period 2 nontradables, which tends to depress q2. Agents will substitute consumption of (relatively cheaper) period 2 nontradables for period 1 nontradables. This further reduces excess demand for period 1 nontradables and, hence, also favors a depreciation of the real exchange rate in period 1. This intertemporal substitution effect explains the second term in equation (14).

The last two terms in equation (14) are income effects. Accordingly, reducing the tariff in periods 1 and 2 reduces a distortion, which raises real income and, therefore, the demand for nontradables. Other things equal, this channel favors an appreciation of the RER.

To sum up, if nontradables are intensive users of intermediates in both periods (taking into account in period 2 the effect of the liberalization on investment) and substitution effects outweigh income effects, then a permanent liberalization causes a real depreciation in period 1. If, however, tradables use intermediates intensively, the real exchange rate necessarily appreciates. A completely analogous interpretation carries over to equation (15).

Having discussed the welfare and RER effects of a permanent liberalization, it is now possible to derive the current account response, which is given by

In equation (16), the first four terms represent the effect on saving. They consist of a real income effect (first term); intertemporal substitution effects due to real exchange rate movements that affect the consumption rate of interest (second and third terms); and a welfare or wealth effect (fourth term). The last two terms in equation (16) represent the effects of the liberalization on investment. Clearly, the response of saving depends on a combination of real income and relative price (that is, real exchange rate) effects, while investment moves according to a direct effect (the fifth term in equation (16)) and a relative price effect (sixth term). In general, the overall impact on the current account cannot be signed and depends on the relative magnitudes of the various effects identified above. However, several special cases of equation (16) help to illustrate the interaction among the various forces at work and the circumstances under which the current account is likely to improve or deteriorate as a result of reducing the tariff rate on imported intermediate inputs. These are investigated below.

Consider first the case in which the government can credibly commit to free trade in the long run—that is, in period 2. In this case, liberalization is interpreted as a reduction in the tariff rate, t1 toward zero, leaving constant t2 at its long-run expected value of zero. The analogue of equation (16) in this case is

where 1 – C1w is the marginal propensity to consume out of wealth in period 2, which is assumed positive. Equation (17) shows that a reduction in t1(with t2 = 0) affects the current account through three main channels. The first two summarize the effects on saving, while the last gives the effect on investment. Accordingly, the first term in equation (17) is the consumption-smoothing effect—the reduction in t1 raises real income by reducing a distortion. Because agents allocate their wealth optimally across periods, part of the real income gain is saved. Thus, the consumption-smoothing effect favors an improvement in the current account. The second term is the intertemporal substitution effect, which depends on how movements in the RER in both periods affect the consumption rate of interest (CRI).36 Finally, the third term is the investment effect. Its sign depends on which sector uses capital more intensively and on the behavior of the RER in period 2.

To gain further insight into equation (17), consider the two-sector analogue of the model under consideration here, in which there are no nontradable goods. In this case, it is straightforward to show that the expression in equation (17) reduces to

In this case, therefore, one obtains the “perverse” result that a trade liberalization necessarily improves the current account if the initial trade distortion is positive. 37 The reason, of course, is that without nontradables, all the prices in the model are exogenous for the small country, so that there are no effects on consumption or investment rates of interest. Although consumption rises as a result of the real income gain from reducing the distortion, its proportional rise will necessarily be less than the proportional rise in the value of period 1 output because of consumption smoothing. 38 Put somewhat differently, while imports of intermediates rise as a result of the liberalization, net exports of final products will necessarily rise by a greater amount, leading to an improvement in the resource balance.

Alternatively, consider the case with nontradables in which initial trade distortions are small. Equation (17) then becomes

where

by the concavity of the expenditure function. In this case, the effect on saving, — k1R143, takes on a particularly simple form. A reduction in t1reduces (raises) saving if nontradables (tradables) use imported intermediates intensively. 39 The intuition is that if R143 < 0, trade liberalization leads to a relative expansion of the nontradables sector and, hence, to a real depreciation in period 1, which lowers the CRI.40 With real income unaffected because of the assumption that initial distortions are small, saving must decline.

However, the behavior of the current account is not given by the response of saving alone. In particular, the reduction in t1 also affects investment behavior through the response of the period 2 RER. Suppose, for example, that R143< 0. Then, as long as t1 is sufficiently small, the reduction in t1 will generate a real depreciation in period 2. This, in turn, lowers the marginal productivity of investment (and, hence, the optimal level of investment) if nontradables are capital intensive, but increases investment otherwise. Alternatively, if 1R143> 0, the reduction in t1 raises q2. In this case, a trade liberalization causes investment to decline if tradables are capital intensive, and to rise otherwise. In general, the current account response depends on the behavior of both investment and saving, which in turn depend on factor intensity assumptions across the various sectors.

Finally, in the general case with nontradables and significant initial distortions, both substitution and income effects will play a role in determining the response of the current account. As is clear from equation (17), consumption smoothing favors an improvement in the current account as income gains are spread across the two periods. Real exchange rate effects (and, hence, the intertemporal substitution effect on saving) now depend both on factor-intensity assumptions and on the relative magnitudes of income and substitution effects. This is also true of the investment response, since now the behavior of the marginal productivity of capital in period 2 does not depend solely on factor-intensity assumptions (as it did in the case without initial trade distortions).

The effects of a reduction in t1 for the case in which the government credibly precommits to free trade in the long run (so that t2 is expected to be zero) are summarized in Table 1a and 1b. If the initial level of tariffs is fairly low (that is, t1 is close to zero), then saving rises if tradables use intermediates intensively, but falls otherwise. Further, in the case in which saving rises, the real exchange rate will be appreciating, and

Table 1aEffect of a Tariff Reduction on Imported Intermediate Inputs: Short-Run Distortions Only(Low initial tariffs)
Sector Using Intermediates Intensively
VariableTradablesNontradables
Welfare++
Real exchange rateappreciatesdepreciates
Saving+
Investment+a -b+b -b
Note: Tariffs are assumed to be sufficiently low so that the distortionary (income or wealth) effects associated with liberalization do not dominate the substitution effects.

If nontradables are capital intensive.

If tradables are capital intensive.

Note: Tariffs are assumed to be sufficiently low so that the distortionary (income or wealth) effects associated with liberalization do not dominate the substitution effects.

If nontradables are capital intensive.

If tradables are capital intensive.

Table 1b.Effect of a Tariff Reduction on Imported Intermediate Inputs: Short-Run Distortions Only(High initial tariffs)
Sector Using Intermediates Intensively
VariableTradablesNontradables
Welfare++
Real exchange rateappreciatesappreciates
Saving++
Investment+ab+ab
Note: Substitution effects are assumed to be dominated by income effects.

If nontradables are capital intensive.

If tradables are capital intensive.

Note: Substitution effects are assumed to be dominated by income effects.

If nontradables are capital intensive.

If tradables are capital intensive.

depreciating for the case in which saving declines. As to investment, the latter rises if nontradables are intensive in intermediates but not in capital or if tradables arc intensive in intermediates but not in capital; in all other cases, investment declines. As can be seen, detailed knowledge of the economic structure is required to predict saving, investment, and therefore current account responses to trade liberalization in this case.

When tariff levels are initially relatively high (Table 1b), the net effect on all variables of interest will depend on the relative magnitudes of both income and substitution effects. The results in Table 1b assume that the income effects dominate, so that irrespective of factor-intensity assumptions, the real exchange rate appreciates (in both periods) and saving rises. With the real exchange rate appreciating in the second period, investment will increase if nontradables are capital intensive, but fall otherwise.

The assumption that the government could credibly precommit to a situation of free trade in period 2 was made mainly in order to isolate some important channels through which liberalization affects the current account, in the absence of the additional complications that arise in the presence of second-period distortions (as reflected in equation (16)). However, in many developing countries, a more reasonable assumption is that some positive level of import protection will remain over the indefinite future. This case is discussed below.

As far as saving is concerned, the main difference between equations (16) and (17) is that, even in the absence of RER effects, a permanent liberalization has an ambiguous effect on saving when there are period 2 distortions, whereas saving necessarily rises when trade distortions are confined to the first period. The intuition is simply that, with t1 and t2 both positive in the initial equilibrium, the liberalization affects the level of real income in both periods. While consumption-smoothing considerations favor an increase in saving when real income in period 1 rises, the expected rise in future income favors an increase in borrowing (dissaving). The net effect of these two tendencies, and hence the overall impact on saving, is in general ambiguous.

Further, in addition to the impact on saving, the adjustment in the current account also reflects the response of investment. Under the assumption that intermediates and capital are complementary inputs, the reduction in the tariff unambiguously results in an increase in investment (and a worsening of the current account balance), even in the case in which there are no nontraded goods and, hence, no relative price effects. This contrasts with the analysis in equation (17), in which there was no effect on investment.

Third, if initial distortions are small, so that income effects may be ignored, then the effect on saving (S) is given by

where

Thus, in contrast to the previous case (equation (17)), a reduction in the tariff need not reduce saving, even if nontradables are intensive users of intermediate inputs and there are no income effects. The reason is that if nontradables use intermediates intensively, a liberalization depreciates the RER in both periods. While the depreciation in period 1 lowers saving via intertemporal substitution, the depreciation in period 2 tends to increase saving. In general, the overall effect is ambiguous.

Fourth, investment behavior does not depend exclusively on the behavior of the period 2 RER as it did in the case with short-run distortions only. Specifically, as indicated earlier, the reduction in t1 directly contributes to a rise in investment in the case in which capital and intermediates are complementary inputs. If tariff levels are initially quite high and income effects dominate, liberalization causes an appreciation of the RER in period 2, which further increases investment if nontradables are capital intensive. If, however, tariffs are not too high initially (so that income effects are small), then the effect of the RER will serve to reduce investment, either if nontradables are intensive in intermediates and capital, or tradables are intensive in both intermediates and capital. In general, the overall response of investment to a liberalization will depend on both these direct and indirect effects.

Table 2a and 2b summarize these results. By way of comparison with Table 1a and 1b, several points can be made. First, if there are no nontradable goods, a reduction in t1 necessarily increases saving if there are no distortions in the second period; in contrast, a permanent liberalization was shown to have an ambiguous effect on saving if trade is initially distorted in both periods. Second, if tariffs are relatively low and there are nontradable goods, a reduction in t1 (with t2 = 0) was shown to increase saving if nontradables are intensive in intermediates, but reduce it otherwise. This need no longer be the case if t1 and t2 are reduced simultaneously, as in a permanent liberalization. Finally, the effect on investment of a permanent liberalization consists of a direct effect (which raises investment under the assumption that intermediates and capital are complementary inputs) and an indirect effect operating through the response of the period 2 RER. This contrasts with the analysis in equation (17), in which the investment effect only depended on the response of the RER in period 2.

Temporary Liberalizations

An important issue that has received considerable theoretical attention and that may have relevance for understanding the effects of actual episodes of trade reform concerns the credibility of policies. In what follows, the effects on welfare, the RER, and the current account of temporary liberalizations are considered. These results may have relevance in situations in which the government cannot credibly commit to a permanent reduction in tariffs. In this case, the public will come to expect that future tariff levels will not be reduced alongside current tariffs (as was the case above). As will be shown, liberalizations need no longer be welfare improving in this case. Moreover, this fact has important implications for the response of other variables, including the current account. 41 Although qualitatively similar results have been obtained in earlier studies (for example, Calvo (1987, 1988, 1989)), these results have emphasized mainly the consumption channel as a means through which temporary policies may reduce welfare. Thus, temporary liberalizations might be immiserizing (that is, welfare reducing) if they led to significant overconsumption during the liberalization years. By contrast, when liberalization takes the form of reductions in tariffs on intermediates, the welfare effects hinge more on how such policies affect production decisions, either as regards investment in physical capital, or in relation to the channeling of resources among the various sectors of the economy.

Table 2a.Effect of a Permanent Tariff Reduction on Imported Intermediate Inputs: Short- and Long-Run Distortions(Low initial tariffs)
Sector Using Intermediates Intensively
VariableTradablesNontradables
Welfare++
Real exchange rateappreciatesdepreciates
Saving??
Investment+a+b
Note: Tariffs are assumed to be sufficiently low so that the distortionary (income or wealth) effects associated with liberalization do not dominate the substitution effects.

If nontradables are capital intensive.

If tradables are capital intensive.

Note: Tariffs are assumed to be sufficiently low so that the distortionary (income or wealth) effects associated with liberalization do not dominate the substitution effects.

If nontradables are capital intensive.

If tradables are capital intensive.

Table 2b.Effect of a Permanent Tariff Reduction on Imported Intermediate Inputs: Short- and Long-Run Distortions(High initial tariffs)
Sector Using Intermediates Intensively
VariableTradablesNontradables
Welfare++
Real exchange rateappreciatesappreciates
Saving??
Investment+a+a
Note: Substitution effects are assumed to be dominated by income effects.

If nontradables are capital intensive.

Note: Substitution effects are assumed to be dominated by income effects.

If nontradables are capital intensive.

To see this, consider the effect of reducing the tariff in period 1 alone (that is, leaving t2 unchanged), which is given by

where

As can be seen, if the period 2 tariff is initially zero, reducing the tariff

in period 1 is necessarily welfare improving. This result was obtained earlier. However, it is equally clear that with an initial trade distortion in period 2, it may no longer be optimal to liberalize in the first period alone. Put differently, the optimal first-period tariff subject to a fixed distortion in the second period (which may simply be the public’s expectation that there will be some positive level of protection that will persist indefinitely) is in general different from zero (here, it is simply equal to the ratio ∂ t2K5/K4).

To see how this might come about, consider an example in which tradable goods are intensive in both capital and intermediates relative to nontradables. 42 In this case, it is easily verified that a reduction in t1 necessarily causes the RER to appreciate in period 2. Note, however, that with t2positive, protection in period 2 has resulted in overproduction of nontradables relative to tradable goods. Essentially, with imports of intermediates restricted below their free trade level, the sector that uses these inputs intensively (tradables) is too small relative to the rest of the economy. The real appreciation in period 2 leads to a further shift of resources toward the nontradable sector. The magnification of an initial production distortion reduces welfare.

Furthermore, under the assumption that tradables are relatively capital intensive, the real appreciation in period 2 causes investment to decline when tariffs are lowered in period 1. But, with t2> 0, the capital stock in period 2 is already below its optimal level under the assumption that capital and intermediates are complementary inputs. Thus, in addition to the magnification of a production distortion, the reduction in t1 tends to magnify an initial investment distortion.

The possibility of an immiserizing liberalization has important implications for the response of the current account. However, the implications are somewhat at variance with those found in the previous literature. At issue is the interpretation one wishes to give to the performance of one macroeconomic indicator, the current account, during the process of liberalization. As previously mentioned, lack of credibility in trade reforms has been said to contribute to a significant worsening of the current account at the same time as it might bring about a reduction in the level of social welfare. Thus, in previous models, the “adverse” movement in the current account has clear welfare implications. In the present case, however, the possibility of an immiserizing liberalization may actually make a significant deterioration of the current account less likely during a noncredible liberalization. To see this, recall that the current account effect is given by

The main qualitative difference between this expression and the one given in equation (17) (where t20) is that now dW/dt1 need no longer be negative. Moreover, in the immiserization case, the welfare effect contributes to an increase in saving, since the reduction in welfare depresses current consumption. Although the presence of a second-period distortion need not imply that the current account will improve, there is at least one case in which the fact that t2> 0 unambiguously contributes to such a perverse outcome. This is the case previously discussed in which tradables are intensive users of both capital and intermediates, and a temporary liberalization results in an appreciation of the RER in both periods and a decline in investment (which, as seen previously, is unambiguously welfare reducing).

While the responses of saving and the current account are in general ambiguous, it can be said unambiguously in this case that the role of the second-period distortion, or put somewhat differently, the consequence of the government’s inability to credibly commit to free trade in the second period, is to make an improvement in the current account more, rather than less, likely (that is, the coefficient on t2 in the reduced-form version of equation (19) is necessarily negative). At the same time, the second-period distortion also makes the immiserization outcome more likely (that is, the larger is t2, the smaller will be the welfare gains, or the larger will be the welfare losses, from reducing t1). This argument may suggest that, in contrast to previous results, the behavior of the current account may not be an appropriate signal to look at when judging the success of a liberalization episode. In particular, depending on the economic structure, the fact that the current account does not deteriorate much, or even improves, may reflect a reduction in efficiency and welfare (stemming from less-than-optimal investment levels and greater-than-optimal production levels of nontradable goods in period 2), rather than the usual static gains in economic efficiency that arise when producers face prices that reflect more closely marginal costs in world markets.

III. Conclusions

This paper has examined an issue of current policy concern in developing countries: namely, how will the current account respond to a reduction in the tariff on imports? Previous analytical research has been unable to offer definite predictions, which seems consistent with the empirical observation that there is no systematic tendency among developing countries for trade liberalization to lead to a deterioration in the external position.

The motivation for adding to the theoretical literature in this area was to see what new channels from liberalization to the current account were present when tariffs were reduced on intermediate inputs rather than on final products. It was shown that the effects on saving depended on detailed information about the economic structure, in particular the relative factor intensities across the various sectors. Further, wealth effects emanating from the reduction in initial distortions were also shown to have an important effect on the response of saving. In addition, the model incorporated investment behavior, and it was shown that investment might rise or fall when tariffs on intermediates were reduced, depending both on the initial level of trade restrictions and on the economic structure of the country. The behavior of investment was also shown to be of some importance in evaluating both the current account and welfare implications of noncredible liberalization policies. In particular, previous literature has argued that noncredible reforms might lead to a situation of overconsumption and, hence, to a lower level of economic welfare. Thus, in such models the “adverse” movement in the current account has clear welfare implications. In contrast, the results obtained here suggest that an equally plausible outcome of noncredible policies would be a decline in investment from an initial situation of underinvestment. In this case, with investment falling, the correlation between movements in the current account and changes in welfare would be opposite to the case in which the liberalization policy led to overconsumption.

To illustrate some of the considerations affecting the relationship between tariff reductions on intermediates and the external current account balance, consider the case of distortions assumed to exist only in the short run. To focus on the role of economic structure, suppose further that the initial tariff levels are sufficiently small so that wealth effects may be ignored. Finally, assume that the tradables section uses intermediate inputs intensively (relative to nontradables). In this case, reducing the tariff on imported inputs leads to a relative expansion of the tradables sector and, hence, to a decline in the relative price of the goods produced by that sector—that is, a real appreciation. This real appreciation raises the cost of current consumption in terms of future consumption, thereby stimulating national saving. In addition, the possibility of substituting consumption across periods implies that the higher relative price of nontradables today will create an incipient excess demand for nontradables in the future. Market clearing will therefore require an appreciation of the real exchange rate in the future. The consequent fall in the future-period relative price of tradables will lower the economy’s demand for future-period capital and, hence, the optimal investment level, if (and only if) tradables are, relative to the rest of the economy, intensive users of capital. In summary, therefore, if tradables use both capital and intermediates intensively relative to the rest of the economy, liberalization leads to an increase in the level of saving and a decline in the level of investment and, hence, unambiguously to an improvement in the external current balance. Equally, however, it is clear that under alternative assumptions (for example, with nontradables intensive in both intermediates and capital), liberalization would cause the external position to deteriorate. What ultimately happens to the current account is thus an empirical issue, and cannot be determined ex ante on theoretical grounds.

The main policy implication of the paper is that, given the likely differences in economic structure and in initial levels of protection that exist among developing countries, trade liberalizations cannot be expected to systematically affect the current account of these countries in one direction or another. As a consequence, the frequently made argument that a trade liberalization in a given country cannot be contemplated because of the adverse consequences for the external position of that country loses much of its force unless it is based on detailed information about that country’s specific economic structure.

Appendix Outline of Proof that a Permanent Liberalization Raises Welfare

To show that equipro portion ate reductions in t1 and ∂t2necessarily increase welfare, define the present value revenue function:

where the maximization is taken with respect to the choice of I. Note that R1is convex in prices because R1and R2are convex in prices. Note further that the partial derivative of R1with respect to a price is the same as the partial derivative of R1, i = 1,2, with respect to that price (by the envelope theorem). Since R1 is convex and Eis concave in prices, the Hessian matrix – RijEij is positive-semi definite (pdf). Call this Hessian matrix A.

Consider the fourth-order principal minor of Ain q1, q2, t1, t2. Call this four-by-four matrix B, Partition B into four two-by-two submatrices, B11, B12, B21, B22. In particular, B22 is a two-by-two matrix whose first-row vector is given by [R1440], and whose second–row vector is [0 R244 – (R246)2/R266].

Consider the inverse of B; call it C, which will also be pdf. Use the formula for the inverse of partitioned matrices (see, for example, Dhrymes (1978, pp. 458-89)) to compute the lower right element of C, C22. Note that the two-by-two matrix, C22, is pdf. Therefore, the diagonal elements and determinant of C22 will be nonnegative. These two facts can be used to sign the expression in equation (13) in the case of equipro portion ate reductions in t1 and ∂ t2. I am grateful to Avinash Dixit for suggesting this proof.

    BrockPhilip L.“Stochastic Tariffs Capital Formation and Economic Liberalization in Developing Countries” (unpublished; Durham, North Carolina: Duke UniversityOctober1987).

    • Search Google Scholar
    • Export Citation

    CalvoGuillermo A.“On the Costs of Temporary Policy,” Journal of Development EconomicsVol. 27 (October1987) pp. 24561

    CalvoGuillermo A.“Costly Trade Liberalizations: Durable Goods and Capital Mobility,” Staff Papers International Monetary FundVol. 35 (September 1988) pp. 46173.

    • Search Google Scholar
    • Export Citation

    CalvoGuillermo A.“Incredible Reforms,” in Debt Stabilization and Developmented. byGuillermo A.Calvo and others (Oxford;New York: Blackwell,1989).

    • Search Google Scholar
    • Export Citation

    CordenW. Max“Protection and Liberalization: A Review of Analytical Issues,”IMF OccasionalPaper 54 (Washington: International Monetary FundAugust1987).

    • Search Google Scholar
    • Export Citation

    CordenW. Max“Trade Policy and Macro economic Balance in the World Economy,”IMF Working Paper88/101 (Washington: International Monetary FundNovember1988).

    • Search Google Scholar
    • Export Citation

    DhrymesPhoebus J.Introductory Econometrics (New York;Heidelberg: Springer-Vedag1978).

    DixitAvinash K. and VictorNormanTheory of International Trade (Welwyn, England: Nisbet1980).

    DjajicSlobodan“Temporary Import Quota and the Current Account,”Journal of International EconomicsVol. 22 (May1987) pp. 34962.

    DornbushRudiger“Tariffs and Nontraded Goods,”Journal of International EconomicsVol. 4 (May1974) pp. 17786.– Open Economy Macroeconomics(New York: Basic Books1980).

    • Search Google Scholar
    • Export Citation

    EdwardsSebastianThe Order of Liberalization of the External Sector in Developing Countries, Essays in International FinanceNo. 156 (Princeton, New Jersey: Princeton University Press1984)

    • Search Google Scholar
    • Export Citation

    EdwardsSebastian“Temporary Import Tariffs, the Real Exchange Rate and the Current Account,”IMF Working Paper88/80 (Washington: International Monetary FundAugust1988).

    • Search Google Scholar
    • Export Citation

    EdwardsSebastian“Tariffs, Capital Controls, and Equilibrium Real Exchange Rates,”Canadian Journal of EconomicsVol. 22 (February1989) pp. 7992.

    • Crossref
    • Search Google Scholar
    • Export Citation

    EdwardsSebastian andJonathanD. Ostry“Terms of Trade Disturbances, Real Exchange Rates, and Welfare: The Role of Capital Controls and Labor Market Distortions,”IMF Working Paper89/21 (Washington: International Monetary FundMarch1989); forthcoming in Oxford Economic Papers1991.

    • Search Google Scholar
    • Export Citation

    EdwardsSebastian andJonathanD. Ostry“Anticipated Protectionist Policies, Real Exchange Rates, and the Current Account: The Case of Rigid Wages,”Journal of International Money and FinanceVol. 9 (June1990) pp. 20619.

    • Crossref
    • Search Google Scholar
    • Export Citation

    EdwardsSebastian andSweder vanWijnbergen“The Welfare Effects of Trade and Capital Market Liberalization,”International Economic ReviewVol. 27 (February1986) pp. 14148.

    • Crossref
    • Search Google Scholar
    • Export Citation

    EngelCharles andKennethKletzer“Tariffs, Saving and the Current Account,”NBER Working Paper1869 (Cambridge, Massachusetts: National Bureau of Economic ResearchMarch1986).

    • Search Google Scholar
    • Export Citation

    EngelCharles andKennethKletzer“Trade Policy Under Endogenous Credibility,”NBER Working Paper2449 (Cambridge, Massachusetts: National Bureau of Economic ResearchNovember1987).

    • Search Google Scholar
    • Export Citation

    EngelCharles andKennethKletzer“Tariffs and Saving in a Model with New Generations,”Journal ofInternational EconomicsVol. 28 (February1990) pp. 7191.

    • Search Google Scholar
    • Export Citation

    FrenkelJacob A. andAssafRazinFiscal Policies and the World Economy(Cambridge: MIT Press1987).

    FrootKenneth A.“Credibility, Real Interest Rates, and the Optimal Speed of Trade Liberalization,”Journal of International EconomicsVol. 25 (August1988) pp. 7193.

    • Crossref
    • Search Google Scholar
    • Export Citation

    KhanMohsin S.“The Macroeconomic Effects of Fund-Supported Adjustment Programs,”Staff Papers International Monetary FundVol. 37 (June1990) pp. 195231.

    • Search Google Scholar
    • Export Citation

    KhanMohsin S. andJonathan D.Ostry“Response of the Equilibrium Real Exchange Rate to Real Disturbances in Developing Countries,”IMF Working Paper91/3 (Washington: International Monetary FundJanuary1991).

    • Search Google Scholar
    • Export Citation

    KruegerAnne O.“Trade Policies in Developing Countries,” inHandbook of International EconomicsVol. 1(1984) pp. 51969.

    LopezRamon andDaniRodrik“Trade Restrictions with Imported Intermediate Inputs: When Does the Trade Balance Improve?”Journal of Development EconomicsVol. 34 (November1990) pp. 32938

    • Crossref
    • Search Google Scholar
    • Export Citation

    MarionNancy P.“Nontraded Goods, Oil Price Increases, and the Current Account,”Journal of International EconomicsVol. 16 (February1984) pp2944.

    • Crossref
    • Search Google Scholar
    • Export Citation

    MurphyRobert G“Tariffs, Non-Traded Goods and Fiscal Policy,” The International Trade JournalVol. 1 (Winter1986) pp. 125—43.

    NearyPeter“Determinants of the Equilibrium Real Exchange Rate,” American Economic ReviewVol. 78 (March1988) pp. 21015.

    OstryJonathan D.“The Balance of Trade, Terms of Trade, and Real Exchange Rate: An Intertemporal Optimizing Framework,”Staff Papers.International Monetary FundVol. 35 (December1988) pp. 54173.

    • Search Google Scholar
    • Export Citation

    OstryJonathan D. (1990a) “Tariffs and the Current Account: The Role of Initial Distortions,”Canadian Journal of EconomicsVol. 23 (May) pp. 34856.

    • Crossref
    • Search Google Scholar
    • Export Citation

    OstryJonathan D. (1990b) “Trade Liberalization in Developing Countries: Initial Trade Distortions and Imported Intermediate Inputs,”IMF Working Paper90/100 (Washington: International Monetary FundOctober).

    • Search Google Scholar
    • Export Citation

    OstryJonathan D. (1991a) “Tariffs, Real Exchange Rates and the Trade Balance in a Two-Country World,”European Economic ReviewVol. 35 (July1991) pp. 112742.

    • Crossref
    • Search Google Scholar
    • Export Citation

    OstryJonathan D.(1991b) “Trade Restrictions with Imported Intermediate Inputs: A Comment” Journal of Development Economics (forthcoming 1991).

    • Search Google Scholar
    • Export Citation

    OstryJonathan D. and Andrew K.Rose“Tariffs and the Macroeconomy: Some Empirical Evidence,”IMF Working Paper89/70 (Washington: International Monetary FundSeptember1989);forthcoming in Journal of International Money and Finance1992.

    • Search Google Scholar
    • Export Citation

    PapageorgiouDemetris and others “The Phasing of a Trade Liberalization Policy,”CPD Discussion Paper1986-42 (Washington: World BankJanuary1986).

    • Search Google Scholar
    • Export Citation

    RazinAssaf andLars E.O.Svensson“Trade Taxes and the Current Account,”Economics LettersVol. 13 (1983) pp. 5557.

    RodrikDani“Trade and Capital Account Liberalization in a Keynesian Economy,”Journal of International EconomicsVol. 23 (August1987) pp. 11329.

    • Crossref
    • Search Google Scholar
    • Export Citation

    RodrikDani“Promises, Promises: Credible Policy Reform via Signalling,”EconomicJournalVol. 99 (September1989) pp. 75672.

    SahayRatna“Trade Policy and Excess Capacity in Developing Countries,”Staff Papers International Monetary FundVol 37 (September1990) pp. 486508.

    • Search Google Scholar
    • Export Citation

    SenPartha andStephen J.Turnovsky“Tariffs, Capital Accumulation, and the Current Account,”International Economic ReviewVol. 30 (November1989) pp. 81131.

    • Crossref
    • Search Google Scholar
    • Export Citation

    SvenssonLars E. O.“Oil Prices, Welfare, and the Trade Balance,” Quarterly Journal of EconomicsVol. 99 (November1984) pp. 64972.

    SvenssonLars E. O. andAssafRazin“The Terms of Trade and the Current Account: The Harberger-Laursen-Metzler Effect,”Journal of Political EconomyVol. 91 (February1983) pp. 97125.

    • Crossref
    • Search Google Scholar
    • Export Citation

    ThomasVinod“Developing Country Experience in Trade Reform,”PPR Working Paper295 (Washington: World BankOctober1989).

    ThomasVinod and others “Lessons in Trade Policy Reform” Policy and Research Series 10 (Washington: World BankMarch1990).

    ThomasVinodU.S. Central Intelligence Agency Directorate of Intelligence Handbook of Economic Statistics(Washington: Government Printing Office1989).

    • Search Google Scholar
    • Export Citation

    van WijnbergenSweder“Tariffs, Employment and the Current Account: Real Wage Resistance and the Macroeconomics of Protectionism,”International Economic ReviewVol. 28 (October1987) pp. 691706.

    • Crossref
    • Search Google Scholar
    • Export Citation
*Jonathan D. Ostry, an Economist in the Research Department, holds a doctorate from the University of Chicago, as well as degrees from the London School of Economics and Political Science, Oxford University, and Queen’s University. He would like to thank Avinash K. Dixit, Dean A. DeRosa, Elhanan Helpman, Mohsin S. Khan, Peter J. Montiel, Assaf Razin, Ratna Sahay, and Carlos A. Vegh, as well as participants at the 1991 meetings of the Canadian Economic Association, for useful comments and suggestions.
1Of course, it has long been realized that there might be offsetting effects stemming from endogenous movements in the equilibrium real exchange rate. Thus, a significant real depreciation associated with tariff reductions might stimulate exports, which would tend to mitigate the worsening in the trade balance caused by the increase in imports. However, the assumption that many developing countries operate with fixed nominal exchange rates, and that wages and prices may not be downwardly flexible in developing countries, led many observers to believe that—in the absence of significant inflation abroad—the real depreciations necessary to prevent a worsening of the external position might not be achievable. See Corden (1987, pp. 21-22) and Dornbush (1980, pp. 65-66) for a description of the effects of a tariff in a static Keynesian model under fixed exchange rates.
2Two exceptions are Brock (1987) and Sen and Turnovsky (1989), Also see Corden (1988) for a less formal discussion of some of these issues.
3Two studies by Engel and Kletzer (1986, 1990) also conclude that a permanent liberalization has an ambiguous effect on the current account that depends, among other things, on the form of the saving function. In their models, a reduction in tariffs could well lead to an increase in savings and thus result in an improvement in the current account. Sen and Turnovsky (1989) also find an ambiguous effect on saving.
4Credibility issues associated with trade liberalization are also examined in Engel and Kletzer (1987), Froot (1988), and Rodrik (1989).
5See also Dornbush (1974) for an early analysis of the connection between tariffs and nontraded goods in the context of a static model.
6In general, if lump-sum taxation is not available to the government, reducing tariff rates on imports will require increases in other distortionary taxes for a constant path of government spending. On the current account effects of changes in distortionary taxes, see Frenkel and Razin (1987, chap. 8).
7See also the related analysis in Edwards and Ostry (1989). A study by van Wijnbergen (1987), which examines the relationship between tariff changes and the current account in economies with wage indexation, is of related interest. In that paper, an ambiguous relationship between tariff reductions and the current account is also established. For a similar conclusion relating to the connection between import quotas and the current account, see Djajic (1987).
8See also Ostry and Rose (1989). The experience with trade liberalization of the Southern Cone countries during the 1970s also does not suggest that the liberalizations per se were the main contributors to the deteriorating external position of these countries. See, for instance, Edwards (1984, particularly pp. 1-2, and 1988, p. 1),
9As Krueger (1984) writes in her survey of trade policies in developing countries: “What distinguishes protectionist policies in developing countries [relative to those in developed countries] is the height ofprotection” (p. 527, emphasis added). Corden (1987) makes essentially the same observation: “Protection in developing countries is generally much higher than in developed countries, covering a much broader range of imports, and is often extremely high by any measure” (p. 3).
10The issue of intermediate products has received extensive treatment in the literature dealing with the effects of oil shocks (see, for example, Svensson (1984) and Marion (1984)), and many results concerning the effects of terms of trade shocks involving intermediate products have their counterparts in the analysis of tariff changes when tariffs are initially positive. A recent paper by Lopez and Rodrik (1990) deals with the intermediate products issue in the context of a model that is similar to the one developed below (see also my comment (Ostry 1991b)). The main differences between their study and the present one are given in footnote 13 below. Another paper that examines the issue of commercial policies with imported intermediate inputs is Sahay (1990).
11“There is also evidence (see, for example, U.S. Central Intelligence Agency (1989)) that intermediate inputs make up a similarly large fraction of total imports of the Eastern European and Soviet economies. Although an analysis of trade liberalization issues in the specific context of these economies is beyond the scope of this paper, the importance of imported intermediate inputs in these countries and the high initial level of distortions indicate that the present analysis might also be relevant to such economies.
12The bias of liberalization measures in favor of intermediates (Thomas (1989, p. 14 and pp. 24-25)) may reflect the view that increased imports of intermediate products enhance productive efficiency and economic growth.
13It may be noted in this context that the related paper of Lopez and Rodrik (1990) explicitly ignores the effects on investment of tariff changes on imported intermediate inputs. In addition to allowing for the endogenous investment response, this paper also considers the effects of tariff changes in the general case in which trade flows may be distorted in both the short and long run. This seems to be the more relevant case in developing countries, where some positive level of import protection has typically been envisaged to remain over the indefinite future (see, for example, the conclusions of the World Bank study referred to previously (Thomas (1989, p. 6)). Of course, when tariffs on intermediates are expected to remain positive in all future periods, the effects of a liberalization will depend on the public’s expectations about the future course of the tariff, and hence, on the credibility of the government’s policies. The paper by Lopez and Rodrik explicitly assumes a situation of free trade in the long run, and hence cannot address the issue of credibility, an issue which is discussed in detail in Section II of the present paper.
14Because monetary considerations are ignored, the model cannot address the issue of how trade liberalization affects the balance of payments or the change in international reserves. The focus instead will be on the impact of tariff reductions on real variables such as the current account, welfare, and the real exchange rate.
15Since perfect capital mobility and perfect foresight will be assumed, the second period may, without loss of generality, be regarded as comprising a larger (possibly infinite) number of future periods.
16For an analysis of the effects of tariff changes (on final goods) in a large-(two-)country version of the present model, see Ostry (1991a). It should also be noted that the present paper does not consider the issue of changes in the relative price between the two final tradable goods (terms of trade shocks), which have received considerable attention elsewhere in the literature (for example, Svensson and Razin (1983) and Ostry (1988)). Therefore, with the international terms of trade constant, it is legitimate to think of the two final tradable goods as forming a single composite tradable commodity. Although the possibility of tariffs on final good imports is allowed for in the specification of the model, the effects of changes in this tariff rate—which have also been analyzed extensively in the literature—are not considered in what follows.
17See, for instance, Dornbush (1974) and the references therein. For a discussion of the appropriate measure of the real exchange rate in models with more than one tradable good, see Neary (1988) and Khan and Ostry (1991).
18There is nothing special about land here. The third factor could equally be some other natural resource that is supplied inelastically.
19The endogenous determination of the price of home goods would also be assured if there were only two types of land, since in this case the number of tradable goods would still be less than the number of nontradable factors. Three types of land were assumed mainly for symmetry, and to capture the idea that each sector used one specific factor.
20The reason for the minus sign preceding the partial derivative of the revenue function is that the intermediate input is like a “negative” output. The demand for the input is therefore equal to minus the partial derivative of the revenue function with respect to ti
21The analysis is not fundamentally different if one of the other goods is chosen to be the composite good. See equation (4) and footnote 22.
22 If the composite good corresponded to good N, then the cost of a unit of capital (and hence the optimal investment choice) would also be a function of q1, the relative price of nontradables in period 1. A rise in q1 would tend to lower investment, and conversely. An analysis of this case, which does not alter the main qualitative results of the paper, is available from the author on request.
23This can be verified by noting that, at the optimum, the marginal product of capital must be a decreasing function of the level of the capital stock.
24This is simply an application of the well-known Rybczynski theorem of international trade. This theorem is also useful for predicting the response of investment to changes in the tariff on final products (see, for example, Dixit and Norman (1980, pp. 49-59)).
25See Murphy (1986) for a discussion of the effects of commercial policies when tariff revenues are used to finance government expenditures.
26Note that the capital stock in period 2, k2 = kl + I, because there is no depreciation.
27Since the trade balance is the main endogenous variable of interest, it is necessary to have at least two periods so that within-period imbalances are possible. For an analysis of trade liberalization in a model where the current account is exogenously fixed by a quota on foreign borrowing, see Edwards and van Wijnbergen (1986) and Edwards (1989).
28Recall that the derivatives appearing on the right-hand side of equations (10) and (11) are the supplies of nontradables, since the third element of the price vector, pi, is the relative price of nontradables, qi.
29It should be noted that the current account and trade account balances are equal in period 1 by the assumption that the economy does not inherit any external debt commitment from periods before the first period.
30Using the property that the revenue function is linear-homogeneous in prices, it may be verified that the definition of saving minus investment is equivalent to the more usual definition of exports minus imports.
31It may be noted that the current account in period 2 is equal to saving since there is no investment in the second period.
32On the interpretation of Rybczynski derivatives when factors outnumber goods, see Dixit and Norman (1980, p. 57).
33Because the relative price between importables and exportables is consta
34Specifically, for a given level of the real exchange rate in period 2, the change in the tariff affects imports according to
where
The second term in the expression for d R2 4/6t clearly reflects the adjustment of investment to the tariff change. In addition, the liberalization alters the level of the equilibrium real exchange rate in period 2. This change in the real exchane rate affects the volume of imports according to
where
Again, the second term in the expression for d R2 4/∂2 measures the responsiveness of investment to the adjustment in the real exchange rate. Clearly, then, the only difference between the response of imports in period 1 and period 2 relates to changes in the level of investment.
35In what follows, A will always be taken to be the determinant of the matrix multiplying the RER vector in equations (10) and (11). Under standard assumptions, its sign is always positive (Dixit and Norman (1980, pp. 131-32)). Note also that in signing the aijs, it has been assumed, in addition to standard curvature properties, that there are no factor intensity reversals, in the sense that the sign of the effect of a change in the tariff on the supply of nontradables is the same in periods 1 and 2. This is somewhat stronger than the assumption that sign (R143)= sign(R243), since in period 2, the tariff affects the supply of nontradables also through an investment effect. Therefore, the assumption of no factor intensity reversals is interpreted as sign(R143) = sign(R243 - R236R246/R266).
36The expressions for dq1/dt1 and dq2/dt1 are provided in Ostry (1990b).
37This is, in fact, a very different result from the case of trade liberalization involving final goods, in which the current account response is in general ambiguous when T1 >0 (see Ostry, 1990a).
38Clearly, what is important here is that the change in t1 does not affect the value of output in the second period because of the assumption that t2 is initially zero. With income rising only in the first period, part of the gain will be saved to finance future consumption.
39This result is established in Lopez and Rodrik (1990) under the special case in which all goods are substitutes,
40Of course, there is also a real depreciation in period 2, which tends to raise the CRI. What the expression for saving tells us is that the effect of the depreciation in period 2 cannot dominate the effect of the depreciation in period 1.
41In general, because the welfare effects of liberalizations are ambiguous in this case, the response of other variables that depend on the welfare effects (the RER, saving, and investment) are also ambiguous, even under specific factor-intensity assumptions.
42The result that temporary liberalizations may be immiserizing does not depend on this specific choice of factor-intensity assumptions.

Other Resources Citing This Publication