- Email: [email protected]
Domestic taste differences, transportation costs and international trade
Journal of International
DOMESTIC
Economics 18 (1985) 65-82. North-Holland
TASTE DIFFERENCES, TRANSPORTATION AND INTERNATIONAL TRADE
COSTS
James R. MELVIN* University
of Western
Ontario,
London,
Ontario N6A 5C2, Canada
Received August 1983, revised version received June 1984 The implications of taste ditIerences among domestic consumers is investigated, and transplantation cost assumptions are found to be crucial to the analysis. In a frictionless world crosshauling occurs, trade volume is indeterminant and tariffs need not have welfare signilicance. With international transportation costs the traditional equivalence of tariffs and tax-subsidy schemes breaks down, and consumers are affected differentially by terms of trade changes. When domestic transportation costs dominate, cross-hauling can occur and tariffs will have regional consequences. Tariffs can change the pattern of international trade, and may internalize trade and result in the wasteful use of resources.
1. Introduction The theoretical international trade literature, in order to focus attention on the exchange of commodities between countries, has abstracted from the purely domestic transactions that are implicitly assumed to take place. Individual countries are usually treated as if they were single economic units with unique excess demand and supply vectors. This simplification has resulted in a neglect of the relationship between transactions carried out among domestics and those carried out between domestics and foreigners. The purpose of this paper is to show that the failure to explicitly examine the decisions of consumers as to whether they will trade with other domestics or with foreigners may result in misleading conclusions. When differences in demand among domestic consumers are introduced, one finds that a number of questions immediately arise, the first being what assumptions are to be made about frictions such as transportation costs. Three cases are considered: equal transportation lcosts among all economic units, higher international transportation costs, and higher domestic transportation costs. It is shown that several of the traditional trade results concerning tariffs, quotas *The first draft of this paper was written while I was on sabbatical leave at the University of New South Wales, Australia, and I am indebted to a Canada Council grant which made that leave possible. Since then versions have been presented at the University of Western Ontario, Princeton University, the University of Rochester and the University of Toronto, and I wish to acknowledge the comments of participants of those seminars. The helpful comments of two referees also substantially improved the paper. 0022-1996/85/%3.30 0 1985, Eisevier Science Publishers B.V. (North-Holland)
66
J.R. Melvin,
Domestic
taste
dl@rences
and taxes must be revised when transportation costs are explicitly introduced. Demand conditions have always played a rather minor role in international trade discussions. One approach has been to assume that the economy behaves as a single individual, or more formally that a community utility function exists. The conditions for the existence of such a community indifference system are now well known.’ Another approach has been to make no specific assumption about domestic demand and to concentrate attention on questions of when Pareto superior situations can be found, as in the gains from trade arguments. Here it is shown that with an appropriate redistribution of income all consumers can be made at least as well off as they were in autarky.2 There have been several papers which have taken explicit account of differences in demand conditions within a country. The first of these were articles by Kenen (1957, 1959) and Johnson (1959). These were mainly concerned with the derivation of the offer curve. Rao (1971) has examined the Stolper-Samuelson Theorem in terms of a model in which there are, in the domestic economy, two groups of consumers with different tastes. More recently, Bhagwati and Brecher (1980) and Brecher and Bhagwati (1981) have analyzed the consequences of foreign-owned factors of production on several of the standard trade propositions. Their model is explicitly in terms of two distinct groups of domestic residents, but because they are concerned with foreign ownership, it is differences in factor endowments which are emphasized. In the present paper attention is focused on taste differences, and we abstract from endowment differences. An additional difference between this paper and those of Bhagwati and Brecher is the issues addressed, for while they were concerned with foreign ownership, transfers, economic growth and tariff policy, attention here is focused on the comparison of tariffs, quotas and taxes, and the interaction between tariffs and transportation costs. On the production side our model is the standard one: there are two goods, X and I: produced with labour (L) and capital (K) under conditions of constant returns to scale. The two factors are in fixed supply for the economy. Thus, we have:
x=~,w,,~x),
(1)
‘For a full discussion see Chipman (1965). ‘One area of the neoclassical analysis which has paid more attention to demand conditions is growth theory, where differences between the saving propensities of capitalists and labourers are found to result in important changes in some of the conclusions.
J.R. Melvin,
Domestic
taste d$erences
67
R=K,+K,,
(3)
E=L,+L,.
(4)
These give the standard production possibility curve TNT’ of fig. 1. On the demand side we have two groups of consumers with tastes identical and homothetic within the group but differing between groups, and with utility functions Ui = Ui (Xi 3 ~),
i= 1,2.
(5)
Each group has an endowment of capital and labour, L, and Ki such that CLi=L and CKi= K-. In what follows, reference to ‘consumer i’ means the group of consumers whose tastes are represented by consumer i. It is assumed throughout that the country is small and takes world prices as given. Y
Fig. 1
One aspect of this model of which advantage is seldom taken in the trade literature is the link between the demand and the supply side formed by the condition that each individual is constrained in his role as consumer by the quantity of the two commodities he receives as factor payments for the
68
J.R. Melvin,
Domestic
taste difirences
provision of capital and labour to the two industries. In a competitive situation he receives his marginal product in quantities of X and Y depending on to which industries he sells his services. The value of the bundles of commodities he receives will be a function only of the commodity price ratio, although the specific endowment point will, in most cases, be indeterminate. If consumers own either labour or capital, but never both, then corresponding to any aggregate production point there will be a unique commodity endowment point.3 Then with known commodity prices unique consumption points can be found for each individual, and a unique aggregate consumption point can be determined. Then, since with given prices the production point is determined, a unique excess demand and supply vector can be derived, and by repeating this procedure for different prices one can construct a unique offer curve. Note that this procedure will provide a unique offer curve even if there are a large number of consumers, each with different utility functions.4 If consumers have endowments of both capital and labour, the story is essentially the same except that in general the commodity endowments (the payments for their factor services) will not be unique, but will depend on to which industry they sell their factor services. The value of this commodity endowment is unique, however (for given commodity prices), and thus the consumption point and offer curve are again uniquely determined. And note that all consumers can have different factor endowments. This paper is concerned with the effect of taste differences-among domestic consumers, and thus to neutralize endowment effects we assume that while individuals have different preferences, they all have identical endowments of capital and labour. It is also assumed that the two groups of consumers are the same size. This simplifies the analysis but does not qualitatively affect the results. When consumers are endowed with both capital and labour, although the value of their commodity endowments are uniquely determined by prices, there are infinitely many endowment bundles possible. One of these commodity endowments is of particular interest. Suppose each consumer group (or for that matter each individual) efficiently produces X and Y using its own endowment of factors. In fig. 1, if TA’T’ is the production possibility curve for the economy, then since the consumer groups are the same size WAW’ will be the commodity endowment locus for both consumers. WAW is a scaled down version of TA’T’ and exactly half the size. For any price line P tangent to TA’T at A’ the same P is tangent to WAW’ at A, where A ‘Consider the specific example where X=K”4L”/4 and Y =K2’3L’/3. It is easily shown that for any aggregate output point X,Y, on the production possibility curve, labourers will receive $X0 and fYO while capitalists will receive 4X, and QYO. 41n general, an offer curve so derived need not be well behaved [see Johnson (1959)], and the trade equilibrium need not be unique.
J.R. Melvin,
Domestic
taste differences
69
is on the line OA’. Tastes for the two consumers are represented by the two indifference curves tangent to P at C2 and Ci. The locus WAW’ is similar to the production possibility curve for nationals derived by Bhagwati and Brecher (1980) and used by Brecher and Bhagwati (1981). Our endowment locus is somewhat easier to construct than theirs, for with our assumption that both consumers own factors in the same proportions as they exist in the economy, both consumers can efficiently produce both commodities at all prices for which both commodities are produced in the economy. With endowments differing as in the BhagwatiBrecher model, the construction of the individual production sets is significantly more complex. In this model the assumptions made about transportation costs are of crucial importance. Three cases are examined. We first consider the neutral case where transportation costs are identical for any pair of potential traders, whether foreign or domestic. The simplest example is the completely frictionless world where all transportation costs are assumed to be zero. The second case assumes that there are transportation costs between foreigners and domestics but none among domestics. This most closely corresponds to the traditional model, but even here significantly different results are derived. The third case assumes that transportation costs between the two domestic consumer groups are higher than they are between domestics and foreigners. This case may have particular relevance for countries such as Canada where population centres are often closer to U.S. markets than they are to each other. 2. Trade in a frictionless world This section considers the case where there can be no presumption as to whether transportation costs among domestics or between domestics and foreigners are higher. The simplest model, and the one considered initially, is where all transportation costs are zero. This assumption is relaxed at the end of the section. Implicit in the usual discussion of trading situations of the kind shown in fig. 1 is the assumption that the second individual will trade the vector AC2=AE with the first individual and that the first individual, as well as engaging in this amount of domestic exchange, will reach equilibrium by trading with the rest of the world to move from point E to point Ci. In this case the observed foreign trade vector for the economy as a whole is A’C = EC1 . However, equilibrium could also be achieved if both individual trade only with the rest of the world. Thus, the second individual could export X and import Y so as to move from point A to C2, while individual 1 exports Y and imports X, trading the vector AC1. Here the economy exports and imports both commodities, and the total volume of trade is no longer the
70
J.R. Meloin,
Domestic
taste
difirences
difference between the excess demands and supplies of the two individuals, but is equal to the sum. The observed trade flow would be the vector C;C;, which we call gross trade. A’C, the trade vector from the traditional model, we refer to as net trade. The real-world phenomenon of the same commodity being both imported and exported, referred to as crosshauling, is sometimes thought to be at variance with the theoretical model in which we have an export good and an import good. We see, however, that this apparent conflict is easily resolved if attention is focused on the total volume of domestic trade rather than just on the net foreign trade vector. Thus we have: Proposition I. The observation of a country both importing and exporting the same commodity is perfectly consistent with the neoclassical, two-sector trade model.
The cross-hauling shown in fig. 1 depends on the fact that domestic consumers have excess demands for different commodities at the trading equilibrium. If tastes were such that both Ci and C, were on the same side of A, then both consumers would have excess demands for the same commodity and there would be no cross-hauling and no distinction between gross and net trade. Note, however, that if there is domestic exchange in autarky (and there will be, given the assumptions of this model) for prices close enough to the autarky price ratio the cross-hauling possibility of fig. 1 must exist. As suggested above, point A in fig. 1 is only one of an infinite number of possible output endowments associated with production at A’ for the economy. Output endowments are the payments to the two consumers as owners of factors of production in units of the two commodities produced, and these endowments depend on to which industry the consumers allocate the services of their labour and capital. Different allocations among industries cannot affect consumers’ income or welfare, and thus all possible endowment points must lie on the price line P through A. The actual production endowment locus depends on the production functions and can be derived, but is of no particular significance. It is some line segment of price line P in fig. 1, and it represents all possible commodity endowments for the two consumers associated with production at point A. The utility of the two individuals is independent of where on P the endowment point actually lies, but gross trade flows depend crucially on where this point is. As one special case, suppose the commodity endowment were Cz for consumer 2, implying an endowment of E for consumer 1.5 Here, gross and net trade ‘Recall
that
&A
= AE.
J.R. Melvin,
DomesGc
cmre d@rences
71
are both equal to EC,, and there is no cross-hauling. Only consumer 1 trades, and thus must trade entirely with the rest of the world. Alternatively, suppose Cz were the endowment point for consumer 1 with E the endowment for consumer 2. Now gross trade could be as large as CIE+ CIC1. Thus we have: Proposition 2. The volume of gross trade is indeterminate. Whether there is cross-haulimg and the amount will depend on how consumers allocate their factor services between industries.
The argument leading to proposition 2 obviously depends on the assumption of a barter system. However, even the introduction of a neutral money with which factors are paid and commodities purchased does not affect the indeterminacy of the gross trade vector. Indeed, money introduces further possibilities in this frictionless world, for now all goods consumed by domestics could be purchased in the foreign country. In this frictionless model the volume of gross trade is indeterminate, except that it must be at least as large as A’C, the amount of net trade. Net trade, however, is unique for any commodity price vector, and it is this net trade which provides the traditional gains from trade. The volume of gross trade has no welfare significance. Nevertheless it is the gross trade vector which is measured in the real world, and it is important to observe that much of the trade we measure may have little or no welfare consequence. In what would be traditionally regarded as autarky, one could observe a significant amount of gross trade in this model. As an illustration of the importance of recognizing the possible existence of gross trade, consider the effects of tariffs. In this frictionless model with crosshauling, the two goods are both exported and imported, and thus a tariff could be imposed on either or both commodities. Suppose that, in terms of lig. 1, both individuals carry out all of their trade in foreign commodity markets, so that the total trade vector is C,C,. First, assume that a tariff is levied on commodity I! The immediate effect of any small tariff will be an increase in domestic exchange. The importer of Y will initially find that domestic production is less expensive than imports and will purchase as much as possible in the domestic market. In fig. 1 consumer 2’s demand for Y can easily be satisfied domestically. Here, a tariff reduces gross trade but does not change net trade, and welfare remains unchanged for both individuals. This analysis suggests that the commodity on which the tariff is imposed may be quite important. Now consider a tariff on X. Again this will result in a switch from foreign trade to domestic exchange, but in this case consumer l’s demand for X cannot be satisfied through exchange with consumer 2. The domestic price of X will thus increase and for the economy as a whole the effects will be the
12
J.R.
Melvin,
Domestic
taste
d@rences
same as in the traditional analysis. The effects on the two individuals will differ, however, and are considered in detail in section 3. Now consider the imposition of a quota. In fig. 1 the initial effect of a quota imposed on either or both commodities would be to internalize trade, for if one individual finds that he cannot satisfy his excess demand in foreign markets, he will turn to the domestic market. In a frictionless model, and presuming that the quota is large enough to permit the satisfaction of net trade flows, the quota will therefore have no welfare implications. In this model there are many possible equilibria, some with small and some with large amounts of trade, among which the individuals in the economy are indifferent, and the imposition of quotas simply limits the range of possibilities. However, if quotas are small enough to prevent the net trade flow from being achieved, the price of the commodity against which the quota has been imposed will increase. Thus, if an unrestrictive quota is imposed and this quota is gradually reduced, the effect could be first to switch trade from international to domestic markets until domestic exchange has been maximized, and thereafter for quotas to have the same effects as tariffs. From the above we have: Proposition 3. Tariffs and quotas may substantially reduce the volume of trade without having any welfare consequences.
This section has considered a model in which it was assumed that frictions between any pair of individuals, whether domestic or foreign, were identical. The special case examined assumed that all frictions were zero. We can imagine consumers and producers to be located at points on the frictionless plane familiar from regional and urban economics. In such a model countries are created by subdividing this plane into two or more regions, and international trade is measured by paying special attention to transactions which occur across the boundaries of these areas. Significantly more realism can be introduced by assuming that the transportation costs between any two economic units is proportional to the distance between them. Welfare, for individuals and for countries, will be maximized by minimizing the total cost of transportation. It is interesting to note that our principal conclusions are largely unaffected by such a reinterpretation of the model. First, we would expect to observe crosshauling as long as both goods are produced in both countries. If X is produced in Hamilton and Detroit, consumers in Buffalo and Windsor will both import X rather than paying the higher domestic transportation costs. The indeterminacy referred to in proposition 2 will be reduced but not eliminated, for there will still be consumers equi-distant from two or more producers. The effect of a tariff on Y in fig. 1 will again be to internalize trade, and although the fob price of foreign goods will be unaffected, those
J.R. Meloin,
Domestic
taste difirences
73
consumers formerly importing Y will suffer a loss equal to the difference between the domestic and foreign transportation costs. Consumers in Buffalo will now pay the cost of transportation from Detroit rather than from Hamilton, and thus the prices faced by some consumers will increase. A tariff on X will first internalize trade and then will have the traditional effect of raising the price to domestic consumers and producers. Large quotas on X and Y will internalize trade, while more restrictive quotas will raise prices as in the case for tariffs. In the frictionless world a tariff on Y or an unrestrictive quota were shown to have no welfare effects. With transportation explicitly introduced this is no longer the case, but the welfare costs are different from those of the traditional analysis. First, the cost is just the difference in transportation rates and will vary from individual to individual. Second, only those consumers of Y who were formerly importers will be affected. Whether these welfare costs are high or low will depend on the transportation costs involved. 3. International
transportation
costs
It is now assumed that international transportation costs are uniformly higher than are domestic transportation costs. This would be the case for countries isolated from major trading partners. For simplicity, it is assumed that the transportation cost differentials internalize trade so that gross trade and net trade are equal, but that they do not create a differential between domestic prices and the world terms of trade. This assumption is equivalent to what is implicitly assumed in the traditional analysis and thus facilitates a comparison of the results. Introducing small transportation costs between domestics and foreigners would complicate the analysis but would not qualitatively affect the results. Finally, note that all the results derived in this section also hold in the frictionless model of section 2. They are presented here so that they can be more easily compared with the traditional conclusions. In the initial equilibrium of fig. 2, Ci and C2 are the consumption points for consumers 1 and 2, resulting in net trade of A’C=AC, - C,A. Now suppose the terms of trade change to P’. While this would be considered an improvement in the terms of trade for the ecomomy, it is clear that while consumer 1 is much better off, consumer 2 has suffered a reduction in utility.6 Furthermore, the fact that trade will make some individuals worse off is not associated with the Stolper-Samuelson result, for we have completely neutralized differences in factor rewards. Indeed, this result can be 6This is similar to a result derived by Brecher and Bhagwati (1981) where an improvement in the terms of trade for the economy could make nationals worse OK Bhagwati and Brecher refer to this as the ‘differential-trade-pattern’ phenomenon.
J.R. Melvin,
Domestic
taste
dlfirences
Fig. 2
thought of as the demand counterpart More formally we can state:
of the Stolper-Samuelson
Theorem.
Proposition 4. Quite apart from the q&t through factor returns, if tastes dijfir trade will harm any individual with an excess demand for the net export good.
Of course trade is not always harmful to one individual. We see from fig. 2 that if prices change by enough, so that both consumption points are on the same side of the production point, then further price changes will change welfare for the two consumers in the same direction.’ It is even possible that the consumer who originally loses could be better off than in autarky; such a situation is shown in fig. 3. This illustrates another difference from the Stolper-Samuelson result where changes in factor rewards are a monotonic function of commodity price changes (as long as both goods are produced). In the case described here, small movements from autarky will always make one consumer worse off but large changes may make all individuals better Off.
It is, of course, well known that gains from trade can be assured only with an appropriate redistribution of income; Samuelson (1939) made this quite ‘Note that proposition 4 will still apply, for now neither consumer has excess demand for the export good.
J.R.
Meluin, Domestic
taste dl@rences
15
Fig. 3
clear in his original proof of the theorem. The impression is often left, however, that redistribution is required because of the Stolper-Samuelson results. Note also that, unlike the Stolper-Samuelson Theorem, the argument of Proposition 4 generalizes easily to any number of individuals. These remarks simply reinforce the well-known but often neglected proposition that no proposal for free trade is complete unless it includes a specific plan for income redistribution. We now consider the imposition of a tariff. Unlike the situation in section 2, the economy imports only X, and a tariff on X has the traditional effect of raising the relative price of X to both producers and consumers. In fig. 4, free trade prices are P, with consumption points Cz and Cr. With a tariff on X, production will move to B on the tariff-ridden price line P’. Consumer 2 can now sell X at price P’ and thus his welfare increases to C;. Trade between the domestic consumers moves consumer 1 to D, and then trade with the rest of the world at price line P results in equilibrium for consumer 1 at Ci.* It is clear that the tariff has had a differential welfare effect on the two individuals, and we thus have a corollary to proposition 3. Note also that had both individuals imported X in free trade, a tariff could first reduce and then increase the welfare of consumer 2. ‘Here, and throughout, in order to avoid problems associated with changing the endowment point, we are assuming that all tariff or tax revenue is distributed, in lump-sum fashion, to the individual Irom whom it was collected. In Iig. 4 the ‘extra’ utility for consumer 1 associated with being above the domestic price line P’ through B is due to the tat$T revenue. Note that even with all the tariff revenue consumer 1 will always be worse oII with a tariK
J.R.
Melvin, Domestic
taste
diferences
Y
Fig. 4
Mundell (1960) argued that a tariff structure can be duplicated by a system of purely domestic taxes. In the present model this is not, in general, true. Consider the free trade position in fig. 4 with consumption points Cz and Ci. Assume a production subsidy to X and a consumption tax on X at the same ad valorem rate as the tariff in the previous example. The production subsidy to X will move production to i3 just as in the tariff case. The consumption tax will also increase the price of X for both individuals but, unlike the tariff, will not permit individual 2 to sell his excess supply of X at the tax-ridden price. Both individuals will still be able to trade with each other or with foreigners at the terms of trade P, and the equilibrium consumption points will be cl; and C;. Several differences between a tariff and an equal-rate tax-subsidy can be seen from fig. 4. First, it can be shown that a tariff will result in less net trade than an equal-rate tax-subsidy system. To illustrate, put point F on price line P such that C”B=BF. Then FC; is net trade with the tax-subsidy system, and DC; is net trade with a tariff. Now C; and C’; lie on the same ray from the origm because of the homogeneity of tastes. Furthermore, the line through DF is parallel to a ray from the origin going through Cl; and Cl, since C;B= BD and c’;B = BF. Thus, since the lines DF and C; C’i diverge, we have that FC’; > DC;, and trade volume is greater with the tax. And note that this conclusion does not depend on how the tariff proceeds are distributed among consumers. Suppose, for example, that all of the tariff revenue is reallocated from consumer 1 to consumer 2 in lump-sum fashion. Consumer 1 has a higher propensity to consume X than does consumer 2, and thus such a shift would reduce aggregate demand for X. With
J.R. Melvin,
Domestic
taste dilfkrences
17
unchanged production this would reduce imports (and reduce the tariff revenue to be redistributed), further increasing the difference between FC;’ and DC;. Thus, for a tariff, imports are maximized when consumer 1 receives all the tariff revenue. The welfare effects are also different, for with a tax-subsidy program both consumers lose, while with a tariff, consumer 2 is better off and consumer 1 worse off. One can, however, redistribute income to generate the same welfare effects for these two policies. This is most easily seen by beginning with the tariff case and transferring the bundle C;C; from consumer 2 to consumer 1. Consumer 1 is then at F and could trade to reach cl;. The equilibrium consumption points would be cl; and C;, the same as for the tax.g The redistribution question is therefore important in this model. Indeed, it is the redistribution of the tariffs and taxes which determine whether and to what extent the two types of policies will differ. There seems to be no reason, however, to suppose that redistributions to make the two policies equivalent will always be undertaken. Thus, in general, we have: Proposition 5. The effects of a tariff and an equal-rate tax-subsidy will not be identical when domestic consumers have qualitatively different excess demands in free trade. lo Net trade will be less with a tariff, and the welfare consequences for domestics will differ.
One final characteristic of the frictionless model of section 2 can now be noted. While a tariff will reduce trade to the essential net-trade component, a tax-subsidy system need not. In fig. 4, for example, in the frictionless model both consumers could still trade with foreigners after the imposition of the taxes. Indeed the tax-subsidy system could actually increase gross trade. It has been shown that if one wishes to restrict trade, taxes may be preferable to tariffs for the former provides a more even treatment of all consumers. The somewhat pdradoxical result that a domestic policy tool may be more appropriate in the international sector stems from the differences in the way in which taxes and tariffs affect the two different consumers in the domestic economy. With a tariff one of the consumers is, in effect, forced to trade at two different price ratios, while with a tax this does not occur. To put the matter slightly differently, with a tax the government collects the same per unit tax regardless of from whom the commodity is purchased, and thus, including the assumed redistribution of income, both individuals 9Note that this transfer has generated international trade. Indeed, any change which increases income for consumer 1 will increase the imports of commodity X. ‘OIf in free trade both individuals have excess demand for the same commodity, then the standard result that tariffs and tax-subsidies are identical will be true, at least for small tariffs. At some level, however, a tariff will eliminate trade for one individual, and higher tariffs will have the effect shown in fig. 4. Thus, as long as tastes differ, there is some tariff for which the standard result does not hold.
78
J.R. Melvin,
Domestic
taste diferenccs
actually trade at the ratio C2C, in fig. 4. With a tariff the government collects revenue on goods purchased abroad but not on domestic exchange, which allows one of the domestic consumers to gain by trading at the tariff distorted domestic prices. In effect this consumer collects the ‘tax’ rather than the government. Thus, the choice between taxes and tariffs is a choice of where we impose the distortions - between domestic consumers or between domestics and foreigners. In his survey of the theory of customs unions, Lipsey (1960) argued that there could be no general presumption that customs unions would be welfare improving, for while a customs union removes some price inequalities it creates others. The same kind of argument applies here. The switch from a tariff to a tax removes some price distortions but creates others. Some individuals will enjoy increased utility and others will have their welfare reduced. In general there can be no presumption as to which is preferred. 4. Domestic transportation costs We now suppose that domestic transportation costs are uniformly higher than transportation costs between countries.” Specifically, we assume that our two groups of domestic consumers are distant from one another but are each close to foreign markets. In Canada, for example, transportation costs between Vancouver and Toronto are much higher than they are between Vancouver and Seattle or between Toronto and New York. To simplify, we make the assumption that foreign transportation costs are zero while domestic transportation costs are finite. We also assume that both domestic regions face the same foreign prices. These two assumptions are not crucial for the results and can easily be relaxed. Some results are immediate. The indeterminacy of section 2 is eliminated, for now all trade is with foreigners, and gross trade is maximized. Crosshauling will occur if the two consumers have excess demands for different goods. Furthermore, because there is no indeterminacy, tariffs and quotas will have the standard effects within a region, but as shown below, will have quite different effects for the economy as a whole. Terms of trade changes will now increase welfare in one region and reduce it in the other, giving rise to regional disparities. Note that such disparities are entirely due to taste differences, for all consumers are still earning identical factor incomes as long as commodity prices are the same for both regions. The initial equilibrium is shown in fig. 5 with free trade prices P, and the trade vector for the economy C2C1.AC1 > AC2 and the net trade vector for the economy is AC. We now introduce a tariff, and note first that because both “See Melvin (1985) for a discussion of the erects of transportation costs and tariffs in a regional model where all consumers have identical tastes but where factor endowments direr among regions.
J.R.
Melvin,
Domestic
taste
difirences
19
Fig. 5
goods are imported, a tariff can be imposed on either or both commodities. We first consider a uniform tariff on both commodities at a rate less than the transportation cost for both goods between the two consuming groups. This amounts to a tariff on X for consumer 1 and a tariff on Y for consumer 2, and the production and consumption changes will be the usual ones. In fig. 5 production moves to B, and B, and consumption to D, and D, for consumers 1 and 2, respectively. I2 The price lines facing the two consumers (both in their role as producers and consumers) are P, and P,. Because both still trade with the rest of the world at terms of trade P, both can consume above producer price lines through B, and B2. Welfare for consumers has been reduced in both regions. Furthermore, production for the economy is inefficient, and factor prices in the two regions are now different. Note also that if a tariff had been applied only to one of the two commodities only one of the regions would have been affected. Tariffs can also change the pattern of trade in this model. As an illustration, note that with free trade the economy is a net importer of X. If a tariff is imposed on X (but not on Y) so that consumption moves to D, for region 1 and remains at C2 in region 2, there will be net imports of commodity Y If the economy was a net exporter of the commodity which used its abundant factor intensively in free trade, it will not be doing so after the tariff is imposed. I3 The same result can also be derived with tariffs on both commodities. Thus we have: ‘*It is again assumed that the tariff revenue is returned to the consumer from whom it was collected as a non-distorting, lump-sum subsidy; thus avoiding endowment changes. 13This could be considered as another possible explanation of the so-called Leontief Paradox.
80
J.R. Melvin,
Domestic
taste dij&wces
Proposition 6. If regions difer in tastes, tariffs can create regional disparities a,rld can change the pattern of net trade.
Now suppose that the tariffs are increased so that they are just equal to the transportation costs. To simplify the diagram assume this equality occurred at the tariff-ridden prices P, and P, of fig. 5. For any small reduction in transportation costs consumers find it profitable to internalize trade rather than to trade with foreigners. When this occurs the gains from foreign trade are lost, and if both consumers are able to satisfy their excess demands domestically, both will consume at points such as F, and I;, on price lines P, and P,. If domestic trade just balances, there is no longer any tariff revenue. The effect of the tariff has been to internalize trade and substitute domestic tariff revenue for domestic transport - transportation which uses resources but does not provide any utility to consumers. We have thus shown: Proposition 7. Tariffs which internalize trade in the presence of domestic transportation costs result in welfare reductions by substituing waste@1 transportation for tar$f revenues.
In general one would not expect trade between regions to balance, and in fig. 5 one individual or the other would be expected to trade with the rest of the world even after trade has been internalized. Nor is it necessary that the tariff internalize trade in both commodities. Thus, some gains from foreign trade would be expected to remain. Of course with a high enough tariff all foreign trade could be eliminated. Certain countries, of which Canada is an obvious example, have pursued a policy of high tariffs as a method of ensuring domestic markets for domestic production. In the Canadian case tariffs have been imposed on manufacturers to protect Eastern industry or, in other words, to force Western consumers to buy Eastern output. It is not clear, however, that the true costs of such protection have been appreciated. Any tariff which is successful in internalizing trade imposes on consumers not only the traditional cost associated with the distortion but also the costs identified in proposition 7. For countries such as Canada it may well be that the major cost of a tariff is the waste of resources associated with unnecessary transportation. Paradoxically, the major cost of the tariff is due to the fact that it results in domestic trade. It can also be seen that the effects of tariffs and economy-wide tax-subsidy systems are very different in this regional model. A federal tax-subsidy system must apply to all consumers and producers, and clearly there is no combination of taxes and subsidies which would produce the tariff situation of fig. 5. The production tax on Y required to generate production at B,
J.R. Melvin,
Domestic
taste
d@%rences
81
would be exactly offset by the production tax on X required to generate Bz, and similarly for the consumption side. If there were a tariff on only one commodity, any tax-subsidy system would affect producers and consumers in both regions, while the tariff does not. Of course taxes and subsidies applied at the regional level could be equivalent in their effects to a tariff or, perhaps more importantly, regional governments could use taxes and subsidies to counteract the undesirable consequences of federal tariffs. Even in this case, however, taxes and subsidies could exactly offset a tariff only if the entire tariff revenue is returned to the region in which they were generated, and there is no reason to expect this to be the case in real world situations.i4 Finally, it is interesting to think of this regional model in terms of the customs union question. Suppose the two regions are two countries, called A and B, and that the transportation costs are now called ‘tariffs’. Note that these ‘tariffs’ are pervasive and apply to all goods which move between A and B. Initially tariffs between both A and B and the rest of the world are small (or zero). A and B now decide to form a customs union. Rather than remove the ‘tariffs’ between them, however, they choose to erect common tariff barriers with the rest of the world that are so high that trade is generated between A and B over the internal ‘tariffs’. Of course such a strange customs union would never be seriously considered. This would seem to be a reasonably accurate description of the tariff policies of countries such as Canada, however.
5. Conclusions
The purpose of this paper has been to investigate the implications of taste differences among consumers within a country for several of the traditional trade propositions. To permit such an investigation, a model was developed where tastes differed among groups but where all individuals were assumed to have the same factor endowments. In a frictionless world it was shown that the volume of trade was indeterminate and that cross-hauling was to be expected. Tariffs and quotes may sometimes have no welfare consequences in such a world. When international transportation costs were introduced it was shown that terms-of-trade changes would have differential welfare effects on consumers, even though endowments were identical. In contrast to the traditional model, tariffs and a tax-subsidy system were found not to be equivalent, for both the volume of trade and the welfare effects will differ. “‘While tari!% on both goods has been assumed, note that the same analysis applies if there is a tarir on only one good, except that now one region bears the entire burden of both welfare losses. The loss will be even larger if we relax the assumption thai the tariB proceeds are returned to the region where they were collected.
82
J.R. Melvin, Domestic taste differences
With domestic transportation costs the dominant friction, cross-hauling will again occur, and tariffs were found to affect regions differently. Tariffs can also internalize trade and result in the wasteful use of resources to provide transportation. It was also shown that tariffs can change the pattern of trade. References Bhagwati, J.N. and R.A. Brecher, 1980, National welfare in’an open economy in the presence of foreign-owned factors of production, Journal of International Economics 10, 103-115. Brecher, R.A. and J.N. Bhagwati, 1981, Foreign ownership and the theory of trade and welfare, Journal of Political Economy 89,497-511. Chipman, J.S., 1965, A survey of the theory of international trade: Part 2, The neo-chtssicaf theory, Econometrica 33,685-760. Johnson, H.G., 1959, International trade, income distribution and the offer curve, Manchester School 27,241-260. Kenen, P.B., 1957, On the geometry of welfare economics, Quarterly Journal of Economics 71, 426-447. Kenen, P.B., 1959, Distribution, demand, and equilibrium in international trade: A diagrammatic analysis, Kyklos, 629-638. Lipsey, R.G., 1960, The theory of customs unions: A general survey, Economic Journal 70, 496 513. Melvin, J.R., 1985, The regional economic consequences of taritfs and domestic transportation costs, Canadian Journal of Economics 18. Mundell, R.A., 1960. The pure theory of international trade, American Economic Review 50, 67110. Rao, V.S., 1971, Tariffs, and welfare of factor owners: A normative extension of the StolperSamuelson Theorem, Journal of International Economics 1,401-415. Samuelson, P.A., 1939, The gains from International trade, Canadian Journal of Economics and Political Science V, 195-205.
DOMESTIC
Economics 18 (1985) 65-82. North-Holland
TASTE DIFFERENCES, TRANSPORTATION AND INTERNATIONAL TRADE
COSTS
James R. MELVIN* University
of Western
Ontario,
London,
Ontario N6A 5C2, Canada
Received August 1983, revised version received June 1984 The implications of taste ditIerences among domestic consumers is investigated, and transplantation cost assumptions are found to be crucial to the analysis. In a frictionless world crosshauling occurs, trade volume is indeterminant and tariffs need not have welfare signilicance. With international transportation costs the traditional equivalence of tariffs and tax-subsidy schemes breaks down, and consumers are affected differentially by terms of trade changes. When domestic transportation costs dominate, cross-hauling can occur and tariffs will have regional consequences. Tariffs can change the pattern of international trade, and may internalize trade and result in the wasteful use of resources.
1. Introduction The theoretical international trade literature, in order to focus attention on the exchange of commodities between countries, has abstracted from the purely domestic transactions that are implicitly assumed to take place. Individual countries are usually treated as if they were single economic units with unique excess demand and supply vectors. This simplification has resulted in a neglect of the relationship between transactions carried out among domestics and those carried out between domestics and foreigners. The purpose of this paper is to show that the failure to explicitly examine the decisions of consumers as to whether they will trade with other domestics or with foreigners may result in misleading conclusions. When differences in demand among domestic consumers are introduced, one finds that a number of questions immediately arise, the first being what assumptions are to be made about frictions such as transportation costs. Three cases are considered: equal transportation lcosts among all economic units, higher international transportation costs, and higher domestic transportation costs. It is shown that several of the traditional trade results concerning tariffs, quotas *The first draft of this paper was written while I was on sabbatical leave at the University of New South Wales, Australia, and I am indebted to a Canada Council grant which made that leave possible. Since then versions have been presented at the University of Western Ontario, Princeton University, the University of Rochester and the University of Toronto, and I wish to acknowledge the comments of participants of those seminars. The helpful comments of two referees also substantially improved the paper. 0022-1996/85/%3.30 0 1985, Eisevier Science Publishers B.V. (North-Holland)
66
J.R. Melvin,
Domestic
taste
dl@rences
and taxes must be revised when transportation costs are explicitly introduced. Demand conditions have always played a rather minor role in international trade discussions. One approach has been to assume that the economy behaves as a single individual, or more formally that a community utility function exists. The conditions for the existence of such a community indifference system are now well known.’ Another approach has been to make no specific assumption about domestic demand and to concentrate attention on questions of when Pareto superior situations can be found, as in the gains from trade arguments. Here it is shown that with an appropriate redistribution of income all consumers can be made at least as well off as they were in autarky.2 There have been several papers which have taken explicit account of differences in demand conditions within a country. The first of these were articles by Kenen (1957, 1959) and Johnson (1959). These were mainly concerned with the derivation of the offer curve. Rao (1971) has examined the Stolper-Samuelson Theorem in terms of a model in which there are, in the domestic economy, two groups of consumers with different tastes. More recently, Bhagwati and Brecher (1980) and Brecher and Bhagwati (1981) have analyzed the consequences of foreign-owned factors of production on several of the standard trade propositions. Their model is explicitly in terms of two distinct groups of domestic residents, but because they are concerned with foreign ownership, it is differences in factor endowments which are emphasized. In the present paper attention is focused on taste differences, and we abstract from endowment differences. An additional difference between this paper and those of Bhagwati and Brecher is the issues addressed, for while they were concerned with foreign ownership, transfers, economic growth and tariff policy, attention here is focused on the comparison of tariffs, quotas and taxes, and the interaction between tariffs and transportation costs. On the production side our model is the standard one: there are two goods, X and I: produced with labour (L) and capital (K) under conditions of constant returns to scale. The two factors are in fixed supply for the economy. Thus, we have:
x=~,w,,~x),
(1)
‘For a full discussion see Chipman (1965). ‘One area of the neoclassical analysis which has paid more attention to demand conditions is growth theory, where differences between the saving propensities of capitalists and labourers are found to result in important changes in some of the conclusions.
J.R. Melvin,
Domestic
taste d$erences
67
R=K,+K,,
(3)
E=L,+L,.
(4)
These give the standard production possibility curve TNT’ of fig. 1. On the demand side we have two groups of consumers with tastes identical and homothetic within the group but differing between groups, and with utility functions Ui = Ui (Xi 3 ~),
i= 1,2.
(5)
Each group has an endowment of capital and labour, L, and Ki such that CLi=L and CKi= K-. In what follows, reference to ‘consumer i’ means the group of consumers whose tastes are represented by consumer i. It is assumed throughout that the country is small and takes world prices as given. Y
Fig. 1
One aspect of this model of which advantage is seldom taken in the trade literature is the link between the demand and the supply side formed by the condition that each individual is constrained in his role as consumer by the quantity of the two commodities he receives as factor payments for the
68
J.R. Melvin,
Domestic
taste difirences
provision of capital and labour to the two industries. In a competitive situation he receives his marginal product in quantities of X and Y depending on to which industries he sells his services. The value of the bundles of commodities he receives will be a function only of the commodity price ratio, although the specific endowment point will, in most cases, be indeterminate. If consumers own either labour or capital, but never both, then corresponding to any aggregate production point there will be a unique commodity endowment point.3 Then with known commodity prices unique consumption points can be found for each individual, and a unique aggregate consumption point can be determined. Then, since with given prices the production point is determined, a unique excess demand and supply vector can be derived, and by repeating this procedure for different prices one can construct a unique offer curve. Note that this procedure will provide a unique offer curve even if there are a large number of consumers, each with different utility functions.4 If consumers have endowments of both capital and labour, the story is essentially the same except that in general the commodity endowments (the payments for their factor services) will not be unique, but will depend on to which industry they sell their factor services. The value of this commodity endowment is unique, however (for given commodity prices), and thus the consumption point and offer curve are again uniquely determined. And note that all consumers can have different factor endowments. This paper is concerned with the effect of taste differences-among domestic consumers, and thus to neutralize endowment effects we assume that while individuals have different preferences, they all have identical endowments of capital and labour. It is also assumed that the two groups of consumers are the same size. This simplifies the analysis but does not qualitatively affect the results. When consumers are endowed with both capital and labour, although the value of their commodity endowments are uniquely determined by prices, there are infinitely many endowment bundles possible. One of these commodity endowments is of particular interest. Suppose each consumer group (or for that matter each individual) efficiently produces X and Y using its own endowment of factors. In fig. 1, if TA’T’ is the production possibility curve for the economy, then since the consumer groups are the same size WAW’ will be the commodity endowment locus for both consumers. WAW is a scaled down version of TA’T’ and exactly half the size. For any price line P tangent to TA’T at A’ the same P is tangent to WAW’ at A, where A ‘Consider the specific example where X=K”4L”/4 and Y =K2’3L’/3. It is easily shown that for any aggregate output point X,Y, on the production possibility curve, labourers will receive $X0 and fYO while capitalists will receive 4X, and QYO. 41n general, an offer curve so derived need not be well behaved [see Johnson (1959)], and the trade equilibrium need not be unique.
J.R. Melvin,
Domestic
taste differences
69
is on the line OA’. Tastes for the two consumers are represented by the two indifference curves tangent to P at C2 and Ci. The locus WAW’ is similar to the production possibility curve for nationals derived by Bhagwati and Brecher (1980) and used by Brecher and Bhagwati (1981). Our endowment locus is somewhat easier to construct than theirs, for with our assumption that both consumers own factors in the same proportions as they exist in the economy, both consumers can efficiently produce both commodities at all prices for which both commodities are produced in the economy. With endowments differing as in the BhagwatiBrecher model, the construction of the individual production sets is significantly more complex. In this model the assumptions made about transportation costs are of crucial importance. Three cases are examined. We first consider the neutral case where transportation costs are identical for any pair of potential traders, whether foreign or domestic. The simplest example is the completely frictionless world where all transportation costs are assumed to be zero. The second case assumes that there are transportation costs between foreigners and domestics but none among domestics. This most closely corresponds to the traditional model, but even here significantly different results are derived. The third case assumes that transportation costs between the two domestic consumer groups are higher than they are between domestics and foreigners. This case may have particular relevance for countries such as Canada where population centres are often closer to U.S. markets than they are to each other. 2. Trade in a frictionless world This section considers the case where there can be no presumption as to whether transportation costs among domestics or between domestics and foreigners are higher. The simplest model, and the one considered initially, is where all transportation costs are zero. This assumption is relaxed at the end of the section. Implicit in the usual discussion of trading situations of the kind shown in fig. 1 is the assumption that the second individual will trade the vector AC2=AE with the first individual and that the first individual, as well as engaging in this amount of domestic exchange, will reach equilibrium by trading with the rest of the world to move from point E to point Ci. In this case the observed foreign trade vector for the economy as a whole is A’C = EC1 . However, equilibrium could also be achieved if both individual trade only with the rest of the world. Thus, the second individual could export X and import Y so as to move from point A to C2, while individual 1 exports Y and imports X, trading the vector AC1. Here the economy exports and imports both commodities, and the total volume of trade is no longer the
70
J.R. Meloin,
Domestic
taste
difirences
difference between the excess demands and supplies of the two individuals, but is equal to the sum. The observed trade flow would be the vector C;C;, which we call gross trade. A’C, the trade vector from the traditional model, we refer to as net trade. The real-world phenomenon of the same commodity being both imported and exported, referred to as crosshauling, is sometimes thought to be at variance with the theoretical model in which we have an export good and an import good. We see, however, that this apparent conflict is easily resolved if attention is focused on the total volume of domestic trade rather than just on the net foreign trade vector. Thus we have: Proposition I. The observation of a country both importing and exporting the same commodity is perfectly consistent with the neoclassical, two-sector trade model.
The cross-hauling shown in fig. 1 depends on the fact that domestic consumers have excess demands for different commodities at the trading equilibrium. If tastes were such that both Ci and C, were on the same side of A, then both consumers would have excess demands for the same commodity and there would be no cross-hauling and no distinction between gross and net trade. Note, however, that if there is domestic exchange in autarky (and there will be, given the assumptions of this model) for prices close enough to the autarky price ratio the cross-hauling possibility of fig. 1 must exist. As suggested above, point A in fig. 1 is only one of an infinite number of possible output endowments associated with production at A’ for the economy. Output endowments are the payments to the two consumers as owners of factors of production in units of the two commodities produced, and these endowments depend on to which industry the consumers allocate the services of their labour and capital. Different allocations among industries cannot affect consumers’ income or welfare, and thus all possible endowment points must lie on the price line P through A. The actual production endowment locus depends on the production functions and can be derived, but is of no particular significance. It is some line segment of price line P in fig. 1, and it represents all possible commodity endowments for the two consumers associated with production at point A. The utility of the two individuals is independent of where on P the endowment point actually lies, but gross trade flows depend crucially on where this point is. As one special case, suppose the commodity endowment were Cz for consumer 2, implying an endowment of E for consumer 1.5 Here, gross and net trade ‘Recall
that
&A
= AE.
J.R. Melvin,
DomesGc
cmre d@rences
71
are both equal to EC,, and there is no cross-hauling. Only consumer 1 trades, and thus must trade entirely with the rest of the world. Alternatively, suppose Cz were the endowment point for consumer 1 with E the endowment for consumer 2. Now gross trade could be as large as CIE+ CIC1. Thus we have: Proposition 2. The volume of gross trade is indeterminate. Whether there is cross-haulimg and the amount will depend on how consumers allocate their factor services between industries.
The argument leading to proposition 2 obviously depends on the assumption of a barter system. However, even the introduction of a neutral money with which factors are paid and commodities purchased does not affect the indeterminacy of the gross trade vector. Indeed, money introduces further possibilities in this frictionless world, for now all goods consumed by domestics could be purchased in the foreign country. In this frictionless model the volume of gross trade is indeterminate, except that it must be at least as large as A’C, the amount of net trade. Net trade, however, is unique for any commodity price vector, and it is this net trade which provides the traditional gains from trade. The volume of gross trade has no welfare significance. Nevertheless it is the gross trade vector which is measured in the real world, and it is important to observe that much of the trade we measure may have little or no welfare consequence. In what would be traditionally regarded as autarky, one could observe a significant amount of gross trade in this model. As an illustration of the importance of recognizing the possible existence of gross trade, consider the effects of tariffs. In this frictionless model with crosshauling, the two goods are both exported and imported, and thus a tariff could be imposed on either or both commodities. Suppose that, in terms of lig. 1, both individuals carry out all of their trade in foreign commodity markets, so that the total trade vector is C,C,. First, assume that a tariff is levied on commodity I! The immediate effect of any small tariff will be an increase in domestic exchange. The importer of Y will initially find that domestic production is less expensive than imports and will purchase as much as possible in the domestic market. In fig. 1 consumer 2’s demand for Y can easily be satisfied domestically. Here, a tariff reduces gross trade but does not change net trade, and welfare remains unchanged for both individuals. This analysis suggests that the commodity on which the tariff is imposed may be quite important. Now consider a tariff on X. Again this will result in a switch from foreign trade to domestic exchange, but in this case consumer l’s demand for X cannot be satisfied through exchange with consumer 2. The domestic price of X will thus increase and for the economy as a whole the effects will be the
12
J.R.
Melvin,
Domestic
taste
d@rences
same as in the traditional analysis. The effects on the two individuals will differ, however, and are considered in detail in section 3. Now consider the imposition of a quota. In fig. 1 the initial effect of a quota imposed on either or both commodities would be to internalize trade, for if one individual finds that he cannot satisfy his excess demand in foreign markets, he will turn to the domestic market. In a frictionless model, and presuming that the quota is large enough to permit the satisfaction of net trade flows, the quota will therefore have no welfare implications. In this model there are many possible equilibria, some with small and some with large amounts of trade, among which the individuals in the economy are indifferent, and the imposition of quotas simply limits the range of possibilities. However, if quotas are small enough to prevent the net trade flow from being achieved, the price of the commodity against which the quota has been imposed will increase. Thus, if an unrestrictive quota is imposed and this quota is gradually reduced, the effect could be first to switch trade from international to domestic markets until domestic exchange has been maximized, and thereafter for quotas to have the same effects as tariffs. From the above we have: Proposition 3. Tariffs and quotas may substantially reduce the volume of trade without having any welfare consequences.
This section has considered a model in which it was assumed that frictions between any pair of individuals, whether domestic or foreign, were identical. The special case examined assumed that all frictions were zero. We can imagine consumers and producers to be located at points on the frictionless plane familiar from regional and urban economics. In such a model countries are created by subdividing this plane into two or more regions, and international trade is measured by paying special attention to transactions which occur across the boundaries of these areas. Significantly more realism can be introduced by assuming that the transportation costs between any two economic units is proportional to the distance between them. Welfare, for individuals and for countries, will be maximized by minimizing the total cost of transportation. It is interesting to note that our principal conclusions are largely unaffected by such a reinterpretation of the model. First, we would expect to observe crosshauling as long as both goods are produced in both countries. If X is produced in Hamilton and Detroit, consumers in Buffalo and Windsor will both import X rather than paying the higher domestic transportation costs. The indeterminacy referred to in proposition 2 will be reduced but not eliminated, for there will still be consumers equi-distant from two or more producers. The effect of a tariff on Y in fig. 1 will again be to internalize trade, and although the fob price of foreign goods will be unaffected, those
J.R. Meloin,
Domestic
taste difirences
73
consumers formerly importing Y will suffer a loss equal to the difference between the domestic and foreign transportation costs. Consumers in Buffalo will now pay the cost of transportation from Detroit rather than from Hamilton, and thus the prices faced by some consumers will increase. A tariff on X will first internalize trade and then will have the traditional effect of raising the price to domestic consumers and producers. Large quotas on X and Y will internalize trade, while more restrictive quotas will raise prices as in the case for tariffs. In the frictionless world a tariff on Y or an unrestrictive quota were shown to have no welfare effects. With transportation explicitly introduced this is no longer the case, but the welfare costs are different from those of the traditional analysis. First, the cost is just the difference in transportation rates and will vary from individual to individual. Second, only those consumers of Y who were formerly importers will be affected. Whether these welfare costs are high or low will depend on the transportation costs involved. 3. International
transportation
costs
It is now assumed that international transportation costs are uniformly higher than are domestic transportation costs. This would be the case for countries isolated from major trading partners. For simplicity, it is assumed that the transportation cost differentials internalize trade so that gross trade and net trade are equal, but that they do not create a differential between domestic prices and the world terms of trade. This assumption is equivalent to what is implicitly assumed in the traditional analysis and thus facilitates a comparison of the results. Introducing small transportation costs between domestics and foreigners would complicate the analysis but would not qualitatively affect the results. Finally, note that all the results derived in this section also hold in the frictionless model of section 2. They are presented here so that they can be more easily compared with the traditional conclusions. In the initial equilibrium of fig. 2, Ci and C2 are the consumption points for consumers 1 and 2, resulting in net trade of A’C=AC, - C,A. Now suppose the terms of trade change to P’. While this would be considered an improvement in the terms of trade for the ecomomy, it is clear that while consumer 1 is much better off, consumer 2 has suffered a reduction in utility.6 Furthermore, the fact that trade will make some individuals worse off is not associated with the Stolper-Samuelson result, for we have completely neutralized differences in factor rewards. Indeed, this result can be 6This is similar to a result derived by Brecher and Bhagwati (1981) where an improvement in the terms of trade for the economy could make nationals worse OK Bhagwati and Brecher refer to this as the ‘differential-trade-pattern’ phenomenon.
J.R. Melvin,
Domestic
taste
dlfirences
Fig. 2
thought of as the demand counterpart More formally we can state:
of the Stolper-Samuelson
Theorem.
Proposition 4. Quite apart from the q&t through factor returns, if tastes dijfir trade will harm any individual with an excess demand for the net export good.
Of course trade is not always harmful to one individual. We see from fig. 2 that if prices change by enough, so that both consumption points are on the same side of the production point, then further price changes will change welfare for the two consumers in the same direction.’ It is even possible that the consumer who originally loses could be better off than in autarky; such a situation is shown in fig. 3. This illustrates another difference from the Stolper-Samuelson result where changes in factor rewards are a monotonic function of commodity price changes (as long as both goods are produced). In the case described here, small movements from autarky will always make one consumer worse off but large changes may make all individuals better Off.
It is, of course, well known that gains from trade can be assured only with an appropriate redistribution of income; Samuelson (1939) made this quite ‘Note that proposition 4 will still apply, for now neither consumer has excess demand for the export good.
J.R.
Meluin, Domestic
taste dl@rences
15
Fig. 3
clear in his original proof of the theorem. The impression is often left, however, that redistribution is required because of the Stolper-Samuelson results. Note also that, unlike the Stolper-Samuelson Theorem, the argument of Proposition 4 generalizes easily to any number of individuals. These remarks simply reinforce the well-known but often neglected proposition that no proposal for free trade is complete unless it includes a specific plan for income redistribution. We now consider the imposition of a tariff. Unlike the situation in section 2, the economy imports only X, and a tariff on X has the traditional effect of raising the relative price of X to both producers and consumers. In fig. 4, free trade prices are P, with consumption points Cz and Cr. With a tariff on X, production will move to B on the tariff-ridden price line P’. Consumer 2 can now sell X at price P’ and thus his welfare increases to C;. Trade between the domestic consumers moves consumer 1 to D, and then trade with the rest of the world at price line P results in equilibrium for consumer 1 at Ci.* It is clear that the tariff has had a differential welfare effect on the two individuals, and we thus have a corollary to proposition 3. Note also that had both individuals imported X in free trade, a tariff could first reduce and then increase the welfare of consumer 2. ‘Here, and throughout, in order to avoid problems associated with changing the endowment point, we are assuming that all tariff or tax revenue is distributed, in lump-sum fashion, to the individual Irom whom it was collected. In Iig. 4 the ‘extra’ utility for consumer 1 associated with being above the domestic price line P’ through B is due to the tat$T revenue. Note that even with all the tariff revenue consumer 1 will always be worse oII with a tariK
J.R.
Melvin, Domestic
taste
diferences
Y
Fig. 4
Mundell (1960) argued that a tariff structure can be duplicated by a system of purely domestic taxes. In the present model this is not, in general, true. Consider the free trade position in fig. 4 with consumption points Cz and Ci. Assume a production subsidy to X and a consumption tax on X at the same ad valorem rate as the tariff in the previous example. The production subsidy to X will move production to i3 just as in the tariff case. The consumption tax will also increase the price of X for both individuals but, unlike the tariff, will not permit individual 2 to sell his excess supply of X at the tax-ridden price. Both individuals will still be able to trade with each other or with foreigners at the terms of trade P, and the equilibrium consumption points will be cl; and C;. Several differences between a tariff and an equal-rate tax-subsidy can be seen from fig. 4. First, it can be shown that a tariff will result in less net trade than an equal-rate tax-subsidy system. To illustrate, put point F on price line P such that C”B=BF. Then FC; is net trade with the tax-subsidy system, and DC; is net trade with a tariff. Now C; and C’; lie on the same ray from the origm because of the homogeneity of tastes. Furthermore, the line through DF is parallel to a ray from the origin going through Cl; and Cl, since C;B= BD and c’;B = BF. Thus, since the lines DF and C; C’i diverge, we have that FC’; > DC;, and trade volume is greater with the tax. And note that this conclusion does not depend on how the tariff proceeds are distributed among consumers. Suppose, for example, that all of the tariff revenue is reallocated from consumer 1 to consumer 2 in lump-sum fashion. Consumer 1 has a higher propensity to consume X than does consumer 2, and thus such a shift would reduce aggregate demand for X. With
J.R. Melvin,
Domestic
taste dilfkrences
17
unchanged production this would reduce imports (and reduce the tariff revenue to be redistributed), further increasing the difference between FC;’ and DC;. Thus, for a tariff, imports are maximized when consumer 1 receives all the tariff revenue. The welfare effects are also different, for with a tax-subsidy program both consumers lose, while with a tariff, consumer 2 is better off and consumer 1 worse off. One can, however, redistribute income to generate the same welfare effects for these two policies. This is most easily seen by beginning with the tariff case and transferring the bundle C;C; from consumer 2 to consumer 1. Consumer 1 is then at F and could trade to reach cl;. The equilibrium consumption points would be cl; and C;, the same as for the tax.g The redistribution question is therefore important in this model. Indeed, it is the redistribution of the tariffs and taxes which determine whether and to what extent the two types of policies will differ. There seems to be no reason, however, to suppose that redistributions to make the two policies equivalent will always be undertaken. Thus, in general, we have: Proposition 5. The effects of a tariff and an equal-rate tax-subsidy will not be identical when domestic consumers have qualitatively different excess demands in free trade. lo Net trade will be less with a tariff, and the welfare consequences for domestics will differ.
One final characteristic of the frictionless model of section 2 can now be noted. While a tariff will reduce trade to the essential net-trade component, a tax-subsidy system need not. In fig. 4, for example, in the frictionless model both consumers could still trade with foreigners after the imposition of the taxes. Indeed the tax-subsidy system could actually increase gross trade. It has been shown that if one wishes to restrict trade, taxes may be preferable to tariffs for the former provides a more even treatment of all consumers. The somewhat pdradoxical result that a domestic policy tool may be more appropriate in the international sector stems from the differences in the way in which taxes and tariffs affect the two different consumers in the domestic economy. With a tariff one of the consumers is, in effect, forced to trade at two different price ratios, while with a tax this does not occur. To put the matter slightly differently, with a tax the government collects the same per unit tax regardless of from whom the commodity is purchased, and thus, including the assumed redistribution of income, both individuals 9Note that this transfer has generated international trade. Indeed, any change which increases income for consumer 1 will increase the imports of commodity X. ‘OIf in free trade both individuals have excess demand for the same commodity, then the standard result that tariffs and tax-subsidies are identical will be true, at least for small tariffs. At some level, however, a tariff will eliminate trade for one individual, and higher tariffs will have the effect shown in fig. 4. Thus, as long as tastes differ, there is some tariff for which the standard result does not hold.
78
J.R. Melvin,
Domestic
taste diferenccs
actually trade at the ratio C2C, in fig. 4. With a tariff the government collects revenue on goods purchased abroad but not on domestic exchange, which allows one of the domestic consumers to gain by trading at the tariff distorted domestic prices. In effect this consumer collects the ‘tax’ rather than the government. Thus, the choice between taxes and tariffs is a choice of where we impose the distortions - between domestic consumers or between domestics and foreigners. In his survey of the theory of customs unions, Lipsey (1960) argued that there could be no general presumption that customs unions would be welfare improving, for while a customs union removes some price inequalities it creates others. The same kind of argument applies here. The switch from a tariff to a tax removes some price distortions but creates others. Some individuals will enjoy increased utility and others will have their welfare reduced. In general there can be no presumption as to which is preferred. 4. Domestic transportation costs We now suppose that domestic transportation costs are uniformly higher than transportation costs between countries.” Specifically, we assume that our two groups of domestic consumers are distant from one another but are each close to foreign markets. In Canada, for example, transportation costs between Vancouver and Toronto are much higher than they are between Vancouver and Seattle or between Toronto and New York. To simplify, we make the assumption that foreign transportation costs are zero while domestic transportation costs are finite. We also assume that both domestic regions face the same foreign prices. These two assumptions are not crucial for the results and can easily be relaxed. Some results are immediate. The indeterminacy of section 2 is eliminated, for now all trade is with foreigners, and gross trade is maximized. Crosshauling will occur if the two consumers have excess demands for different goods. Furthermore, because there is no indeterminacy, tariffs and quotas will have the standard effects within a region, but as shown below, will have quite different effects for the economy as a whole. Terms of trade changes will now increase welfare in one region and reduce it in the other, giving rise to regional disparities. Note that such disparities are entirely due to taste differences, for all consumers are still earning identical factor incomes as long as commodity prices are the same for both regions. The initial equilibrium is shown in fig. 5 with free trade prices P, and the trade vector for the economy C2C1.AC1 > AC2 and the net trade vector for the economy is AC. We now introduce a tariff, and note first that because both “See Melvin (1985) for a discussion of the erects of transportation costs and tariffs in a regional model where all consumers have identical tastes but where factor endowments direr among regions.
J.R.
Melvin,
Domestic
taste
difirences
19
Fig. 5
goods are imported, a tariff can be imposed on either or both commodities. We first consider a uniform tariff on both commodities at a rate less than the transportation cost for both goods between the two consuming groups. This amounts to a tariff on X for consumer 1 and a tariff on Y for consumer 2, and the production and consumption changes will be the usual ones. In fig. 5 production moves to B, and B, and consumption to D, and D, for consumers 1 and 2, respectively. I2 The price lines facing the two consumers (both in their role as producers and consumers) are P, and P,. Because both still trade with the rest of the world at terms of trade P, both can consume above producer price lines through B, and B2. Welfare for consumers has been reduced in both regions. Furthermore, production for the economy is inefficient, and factor prices in the two regions are now different. Note also that if a tariff had been applied only to one of the two commodities only one of the regions would have been affected. Tariffs can also change the pattern of trade in this model. As an illustration, note that with free trade the economy is a net importer of X. If a tariff is imposed on X (but not on Y) so that consumption moves to D, for region 1 and remains at C2 in region 2, there will be net imports of commodity Y If the economy was a net exporter of the commodity which used its abundant factor intensively in free trade, it will not be doing so after the tariff is imposed. I3 The same result can also be derived with tariffs on both commodities. Thus we have: ‘*It is again assumed that the tariff revenue is returned to the consumer from whom it was collected as a non-distorting, lump-sum subsidy; thus avoiding endowment changes. 13This could be considered as another possible explanation of the so-called Leontief Paradox.
80
J.R. Melvin,
Domestic
taste dij&wces
Proposition 6. If regions difer in tastes, tariffs can create regional disparities a,rld can change the pattern of net trade.
Now suppose that the tariffs are increased so that they are just equal to the transportation costs. To simplify the diagram assume this equality occurred at the tariff-ridden prices P, and P, of fig. 5. For any small reduction in transportation costs consumers find it profitable to internalize trade rather than to trade with foreigners. When this occurs the gains from foreign trade are lost, and if both consumers are able to satisfy their excess demands domestically, both will consume at points such as F, and I;, on price lines P, and P,. If domestic trade just balances, there is no longer any tariff revenue. The effect of the tariff has been to internalize trade and substitute domestic tariff revenue for domestic transport - transportation which uses resources but does not provide any utility to consumers. We have thus shown: Proposition 7. Tariffs which internalize trade in the presence of domestic transportation costs result in welfare reductions by substituing waste@1 transportation for tar$f revenues.
In general one would not expect trade between regions to balance, and in fig. 5 one individual or the other would be expected to trade with the rest of the world even after trade has been internalized. Nor is it necessary that the tariff internalize trade in both commodities. Thus, some gains from foreign trade would be expected to remain. Of course with a high enough tariff all foreign trade could be eliminated. Certain countries, of which Canada is an obvious example, have pursued a policy of high tariffs as a method of ensuring domestic markets for domestic production. In the Canadian case tariffs have been imposed on manufacturers to protect Eastern industry or, in other words, to force Western consumers to buy Eastern output. It is not clear, however, that the true costs of such protection have been appreciated. Any tariff which is successful in internalizing trade imposes on consumers not only the traditional cost associated with the distortion but also the costs identified in proposition 7. For countries such as Canada it may well be that the major cost of a tariff is the waste of resources associated with unnecessary transportation. Paradoxically, the major cost of the tariff is due to the fact that it results in domestic trade. It can also be seen that the effects of tariffs and economy-wide tax-subsidy systems are very different in this regional model. A federal tax-subsidy system must apply to all consumers and producers, and clearly there is no combination of taxes and subsidies which would produce the tariff situation of fig. 5. The production tax on Y required to generate production at B,
J.R. Melvin,
Domestic
taste
d@%rences
81
would be exactly offset by the production tax on X required to generate Bz, and similarly for the consumption side. If there were a tariff on only one commodity, any tax-subsidy system would affect producers and consumers in both regions, while the tariff does not. Of course taxes and subsidies applied at the regional level could be equivalent in their effects to a tariff or, perhaps more importantly, regional governments could use taxes and subsidies to counteract the undesirable consequences of federal tariffs. Even in this case, however, taxes and subsidies could exactly offset a tariff only if the entire tariff revenue is returned to the region in which they were generated, and there is no reason to expect this to be the case in real world situations.i4 Finally, it is interesting to think of this regional model in terms of the customs union question. Suppose the two regions are two countries, called A and B, and that the transportation costs are now called ‘tariffs’. Note that these ‘tariffs’ are pervasive and apply to all goods which move between A and B. Initially tariffs between both A and B and the rest of the world are small (or zero). A and B now decide to form a customs union. Rather than remove the ‘tariffs’ between them, however, they choose to erect common tariff barriers with the rest of the world that are so high that trade is generated between A and B over the internal ‘tariffs’. Of course such a strange customs union would never be seriously considered. This would seem to be a reasonably accurate description of the tariff policies of countries such as Canada, however.
5. Conclusions
The purpose of this paper has been to investigate the implications of taste differences among consumers within a country for several of the traditional trade propositions. To permit such an investigation, a model was developed where tastes differed among groups but where all individuals were assumed to have the same factor endowments. In a frictionless world it was shown that the volume of trade was indeterminate and that cross-hauling was to be expected. Tariffs and quotes may sometimes have no welfare consequences in such a world. When international transportation costs were introduced it was shown that terms-of-trade changes would have differential welfare effects on consumers, even though endowments were identical. In contrast to the traditional model, tariffs and a tax-subsidy system were found not to be equivalent, for both the volume of trade and the welfare effects will differ. “‘While tari!% on both goods has been assumed, note that the same analysis applies if there is a tarir on only one good, except that now one region bears the entire burden of both welfare losses. The loss will be even larger if we relax the assumption thai the tariB proceeds are returned to the region where they were collected.
82
J.R. Melvin, Domestic taste differences
With domestic transportation costs the dominant friction, cross-hauling will again occur, and tariffs were found to affect regions differently. Tariffs can also internalize trade and result in the wasteful use of resources to provide transportation. It was also shown that tariffs can change the pattern of trade. References Bhagwati, J.N. and R.A. Brecher, 1980, National welfare in’an open economy in the presence of foreign-owned factors of production, Journal of International Economics 10, 103-115. Brecher, R.A. and J.N. Bhagwati, 1981, Foreign ownership and the theory of trade and welfare, Journal of Political Economy 89,497-511. Chipman, J.S., 1965, A survey of the theory of international trade: Part 2, The neo-chtssicaf theory, Econometrica 33,685-760. Johnson, H.G., 1959, International trade, income distribution and the offer curve, Manchester School 27,241-260. Kenen, P.B., 1957, On the geometry of welfare economics, Quarterly Journal of Economics 71, 426-447. Kenen, P.B., 1959, Distribution, demand, and equilibrium in international trade: A diagrammatic analysis, Kyklos, 629-638. Lipsey, R.G., 1960, The theory of customs unions: A general survey, Economic Journal 70, 496 513. Melvin, J.R., 1985, The regional economic consequences of taritfs and domestic transportation costs, Canadian Journal of Economics 18. Mundell, R.A., 1960. The pure theory of international trade, American Economic Review 50, 67110. Rao, V.S., 1971, Tariffs, and welfare of factor owners: A normative extension of the StolperSamuelson Theorem, Journal of International Economics 1,401-415. Samuelson, P.A., 1939, The gains from International trade, Canadian Journal of Economics and Political Science V, 195-205.