Local comparative advantage: Trade costs and the pattern of trade

• Author: Alan V. Deardorff
• DOI: http://doi.org/10.1111/ijet.12025
• Published date: 01 March 2014
• Date: 01 March 2014

doi: 10.1111/ijet.12025

Local comparative advantage: Trade costs and

the pattern of trade

Alan V. Deardorff∗

When there are costs of trade, such as transport or other costs, the pattern of trade may not

be well described by the usual measures of comparative advantage, which simply compare a

country’s costs or autarky prices to those of the world. Instead, a better comparison takes into

account the costs of trade. This paper shows first, in an example, how trade patterns can vary

with costs of trade. It then provides restatements of the law of comparative advantage, first in

a Ricardian model with trade costs, then extending a 1980 result due to Deardorff and to Dixit

and Norman to include trade costs explicitly in a general framework. It uses this latter result to

derive two correlations relating trade patterns to measures of comparative advantage that take

account of both autarky prices and the costs of trade. Finally, the paper examines the solution

to a trade model with product differentiation in order to make the potential role of trade costs

more explicit, both algebraically and graphically. Withproduct differentiation either by country

or by firm, net trade in an industry, both bilaterally and globally, depends on a country’s costs

of both production and trade relative to an index of those costs for other countries.

Key wor ds comparative advantage, transportation costs, trade costs

JEL classification F11, F12

Accepted 28 October 2013

1 Introduction

Trade theory customarily explains trade by comparisons that are done globally: A country exports

a good for which its own relative cost of production is low compared to the world; or it has a

comparative advantage in goods that make intensive use of a factor that it has relatively more of

than the world.1This may be increasingly appropriate, if we believe both proponents and opponents

of globalization, who seem to see us to be moving ever closer to a fully integrated world economy

∗Ford School of Public Policy, University of Michigan, Ann Arbor,USA. Email: alandear@umich.edu

I must thank the students in my Trade Policy course who, in their papers, showed how useful it can be to look locally

at comparative advantage as they sought to fulfill their assignment of explaining the trade patterns of their various

chosen countries. I have also benefited from comments by numerous colleagues and participants at seminars at an

NBER summer institute, at a session of the AEA, and at the following colleges and universities: Bocconi,Boston College,

Columbia, Copenhagen, Emory, Jaume I (Castell´

on, Spain), Macalester, Miami, Michigan, Notre Dame, Nottingham,

Oregon, Pittsburgh, Rochester,Stockholm University, Stockholm School of Economics, and Syracuse. Finally,I thank the

anonymous referee.

1There are many sources for all of this, but an excellent one is Ethier (1984). Indeed, Ethier’s writings in trade theory,

which this issue of the journal honors, have inspired many of us both toappreciate trade theor y and to try to build on it.

In my own case, Bill has not only been an inspiration, but has frequently and selflessly made important contributions to

what passes as my own work.

International Journal of Economic Theory 10 (2014) 9–35 © IAET 9

International Journal of Economic Theory

Local comparative advantage Alan V.Deardorff

where costs of trade are negligible. But in fact, the volume of trade remains far less that it would be

if all impediments to trade were zero, and several authors have suggested that trade may be more

costly than we previously believed, based only on the obvious measurable costs of transportation.

If that is the case, then patterns of trade may be driven as much by these costs of trade as by the

global comparisons of production costs that we have usually attended to. Indeed, even production

costs may matter differently for a country’s trade when the relevant comparison is not to the world,

but only to those countries that are somehow close enough for trade with them to be most feasible.

These are the issues explored in this paper.

The most obvious cost of trade is transportation, but even this has been surprisingly neglected

in trade theory. Transport costs get only three mentions in the index of Jones and Kenen’s (1984)

Handbook, and all of these come from my chapter on empirical work, either bemoaning the absence

of transport costs in trade theory or suggesting the role that they might play if they were intro-

duced. They are, of course, not entirely absent. Samuelson (1952) included them—in his innovative

“iceberg” form—in a paper about the transfer problem, thus perhaps condemning them to be mostly

ignored thereafter. Icebergtr ansport costs havereappeared in the literature whenever circumstances

made it impossible to ignore transportation altogether, and occasionally they have been illuminat-

ing. Dornbusch et al. (1977), for example, used them creatively to explain which goods would and

would not be traded. Transport costs have also, necessarily, played a role in attempts to provide

theoretical underpinning for the gravity equation, as in Anderson (1979), Bergstrand (1985, 1989),

and Deardorff (1998). But to my knowledge, they have been at most allowed for, and never focused

on, in theoretical studies of the commodity composition of trade.

One might have thought that this neglect would be of diminishing importance as transport costs

themselves have fallen dueto technological progress. In fact, it is not clear that these costs are coming

down, as Hummels (2007) has found in his detailed study of actual costs of shipping.2And whether

they are falling or not, there is increasing evidence that the volume of trade is far smaller than we

would expect from just the observed costs of trade. It took Trefler (1995) to point this out, with his

Mystery of the Missing Trade, but perhaps we should have realized it from earlier empirical failures

of standard trade models, such as Bowen et al. (1987), or even from just looking at the data (which

many of us never did). Studies, such as Hakura (1995), Debaere (1998), and Davis and Weinstein

(2001), have had somewhat more success in explaining trade flows, in part by departing from some

of the implications of zero trade costs, such as factor price equalization. Obstfeld and Rogoff (2001)

suggested that unobserved costs of trade may account for severalof the puzzles that have bemused the

fields of both trade and international finance.3In a related vein, the evidence for “network effects”

in trade, surveyed by Rauch (2001), makes most sense (to me, at least) only if there are unobserved

costs of trade that networks serve to reduce (see Deardorff 2001).

If we believe that costs of trade may be large enough to matter for the patterns of trade, then we

need to find out how they may matter. That is the purpose of this paper. In Section 2, I will use a

simple partial-equilibrium example to show that costs of trade could cause a country to, say, export

a good that it would have been expected to import based on global comparative advantage.Then, to

try to find a more general pattern to the importance of trade costs in general equilibrium, I will turn

first to a Ricardian model in Section 3. The Ricardian model has been around so long that someone

must surely have worked out the role of trade costs within it. But if so, I have not seen it, and I

think in any case that the formulation here may be instructive. The Ricardianmodel is rather special,

2Hummels found that, while the costs of fast shipping, by air,did fall over the recent half-century, the costs of the much

more common ocean shipping rose due to fuel prices that oftenoffset the gains from improved technology.

3These puzzles include also the “border effect” found by McCallum (1995) and Helliwell(1998).

10 International Journal of Economic Theory 10 (2014) 9–35 © IAET