i4 3•4

Public Disclosure Authorized

PRO C E ED ING S OF T HE W OR L D B AN K ANN U AL CON FE

* ( I°tI
Apn
RE NC E
ON D EVE LO PM ENT ECO N O M I C S 1989

The Noncompetitive Theory of International Trade

and Trade Policy

Elhanan Helpman
Public Disclosure Authorized

This paper reviews recent theoretical developments in the analysis of trade structure
and policy. It emphasizes how understanding monopolistic (noncompetitive) market
structures and elements can help explain trade flows and the relation between trade
and growth, and can be useful in evaluating tariffs, quotas, and research and devel-
opment subsidies in noncompetitive markets. Noncompetitive trade theory identifies
testable relations that have already received empirical support in various studies. Once
their significance is recognized, it is important to take them into account in designing
policy. Policies exist that raise welfare, but simple policy prescriptions do not. Theory
helps to identify situations in which particular policies work, but under only slightly
different circumstances opposite policies may have to be implemented. Recent studies
have shown that long-run growth rates depend on an economy's structural features
and the country's trading partners. So policy can affect long-run growth-but identi-
fying useful policies requires an understanding of market structure and conduct, entry
Public Disclosure Authorized

constraints, intersectoral links, and the like. More empirical studies are needed to elicit
this information. Meanwhile, policy should be designed on a case-by-case basis and-
because good policies improve welfare only slightly-no intervention (free trade) re-
mains a good rule of thumb. All the more so when one takes into account the competitive
pressure of a free trading world system, the probability of retaliation, and the political
economy of protection.

A decade of intensive research on increasing returns to scale and market structure

has generated a host of insights about their role in international trade and trade
policy and narrowed the existing gap between theory and application. The new
approach offers explanations for a number of empirical regularities, and provides
new tools for policy analysis. Its findings underline the need for a case-by-case
approach to policy design. The vitality of this work is by no means exhausted,
Public Disclosure Authorized

and the research has recently been redirected to deal with dynamic concerns.
The new line of research began with models of monopolistic competition
designed to explain intraindustry trade. The point of departure was the obser-
vation that many industrial products are differentiated, not homogeneous. If
countries have a taste for such product differentiation-a typical example would
Elhanan Helpman is a professor of international economicrelationsat Tel AvivUniversity.He thanks
June Flandersand T. N. Srinivasanfor their comments.
© 1990 The International Bank for Reconstruction and Development/ THE WORLD BANK.

193
194 Noncompetitive Theory

be demand for variety in consumer electronics or cars-and if variety-specific

economies of scale exist in manufacturing, we may expect intraindustry trade
in differentiated products (see Balassa 1967). Variety-specific economies of scale
ensure specialization in brands; the demand for a wide spectrum of products
ensures a market for them in every country. Under these circumstances, every
country specializes in certain brands and imports brands produced by its trading
partners. This leads to intraindustry trade.
The formal theory that was developed on the basis of this insight is consistent
with three observations. First, it is consistent with the factor proportions view
of the factor content of net trade flows: that is, it predicts that a country exports
embodied factor services of inputs with which it is relatively well endowed, and
imports embodied factor services of inputs with which it is relatively poorly
endowed (see Leamer 1984 for evidence). Second, the theory explains large trade
volumes between similar countries-a well-documented phenomenon (see, for
instance, Linnemann 1966) which the traditional theory failed to explain. Third,
it explains the determinants of the composition of the volume of trade (that is,
intraindustry versus intersectoral), which the traditional theory again could not
address. The predictions of the new theory are supported by empirical findings
(see Balassa 1986, Helpman 1987). The theory applies to consumer as well as
producer goods (see Ethier 1982a and Helpman 1985).
Because space is limited, I do not review the above line of work or other
theories of trade structure based on oligopolistic competition (for a recent review,
see Helpman 1989). Instead, with an eye toward application, I survey the de-
velopments that followed the static theories of monopolistic and oligopolistic
competition as applied to trade structure. My hope is that economists who work
on applied issues of trade and development will find useful guidance in the
results, though the coverage is necessarily selective.
The next section deals with recent studies of structural issues related to dy-
namics of international trade, long-run growth, and product cycles. Section 11
reviews arguments for an activist trade policy in noncompetitive economies, first
in a static framework, which permits consideration of the role of one-sided
market power and strategic trade policy, and second in a growth context. The
paper closes with a short section of concluding comments.

1. TRADE DYNAMICS

The static model of monopolistic competition that was designed to study

intraindustry trade in a framework consistent with the factor proportions view
of intersectoral trade (as described, for example, in Helpman and Krugman
1985, chapters 7-8), has been extended by Grossman and Helpman (1988) to
a dynamic framework. In the static model, brand-specific fixed costs are often
associated with product development and design. A proper treatment of such
costs requires, however, a dynamic model in which costs are incurred before
actual manufacturing takes place, and are gradually recovered over time as the
 He1pman 195

entrepreneur collects monopoly profits. This Schumpeterian view of dynamic

competition can be combined with Chamberlin's view of monopolistic compe-
tition to shed light on trade issues such as the dynamic evolution of trade when
technology changes over time, the role of technological leadership, the role of
product imitation, and the like.
The decision to develop a new product is a central ingredient in this line of
inquiry. An entrepreneur needs to hire resources at cost C,(t) in order to design
a product. He or she then needs to estimate the future flow of profits Tr(r), T '
t, that can be derived from the ownership of the exclusive knowledge or right
to manufacture and market the product. (Naturally, product-specific monopoly
power may be lost at some future point, as I will discuss below; at this stage
assume that it lasts forever). Then the entrepreneur will choose to develop the
product if, and only if, the present value of these profits does not fall short of
the research and development (R&D) costs. If there is free entry into this line of
business and there are no indivisibilities (strictly speaking, the number of prod-
ucts is a continuum), the present value of profits just equals product development
costs in an equilibrium with active R&D. In this case, the instantaneous profit
rate 7r/c,,plus the capital gain on R&D costs (the rate of increase in cj) equals
the interest rate. This dynamic relation combines pricing of the firm on the stock
market together with absence of arbitrage between the cost of product devel-
opment and the value of the firm. It can be embodied in a complete model in
which (a) consumers use prevailing interest rates to allocate spending and saving
optimally over time; (b) full employment of resources takes account of their use
in R&D; and (c) all markets clear, which also implies equality of saving and
investment.
Grossman and Helpman (1988) have done just that in the framework of a
simple, two-country, two-sector, two-factor model with fixed coefficients of
production and no factor accumulation, so that all dynamics result from product
development. In their model one sector supplies differentiated products while
the other supplies a homogeneous product.
The Evolution of Trade
In the dynamic framework it is useful to think about capital as human capital
rather than machines and equipment, which makes it natural to suppose that
R&D is the most capital-intensive activity. Also, assume manufacturing of dif-
ferentiated products (that were developed) to be more capital intensive than
production of the homogeneous product. Finally, assume for the moment that
factor prices are the same in both countries at every instant of time (even though
factor prices change over time). Then if no country begins with a relative ad-
vantage in the number of products (a) the relatively capital-rich country develops
relatively more products; (b) the trade pattern at each point in time resembles
the pattern that emerges in the static model of trade in differential products (that
is, the capital-rich country imports the homogeneous product, it is a net exporter
of the differentiated product, and intraindustry trade exists in differentiated
196 Noncompetitive Theory

products); and (c) the volume of trade grows faster than world gross domestic
product (GDP) (a well-established phenomenon in the postwar period). The world
converges to a steady state in which product development ceases. The steady
state looks very much like an equilibrium of a static world.
In a South-North interpretation, where the North is taken to be the country
relatively rich in capital, these results suggest that in a free trade environment
the North's technological leadership lasts forever, as do its net exports of man-
ufactured differentiated products. This conclusion rests on the particular model,
which is restrictive in many ways, but it does point out a realistic mechanism
at work. More mechanisms need, however, to be considered.
The model can also be used to predict the point at which multinational cor-
porations will emerge. As investment in R&D declines and employment in the
manufacturing of differentiated products rises, the capital-labor ratio employed
in these two activities declines. For this reason, a world structure that, in the
early stages of development, permits factor price equalization without multi-
nationals may reach a point at which it can do so no longer. This happens
necessarily when the capital-labor ratio in manufacturing of differentiated prod-
ucts exceeds the endowed capital-labor ratio of the country relatively rich in
capital. Then, from the point at which total capital per worker in the differ-
entiated product sector (R&D plus manufacturing) exceeds the capital-rich coun-
try's capital-labor ratio, multinationals emerge. The degree of multinationality-
as measured by employment in subsidiaries, their volume of output, or the
number of brands produced by subsidiaries-increases over time until a steady
state is reached. The steady state resembles a static world with multinational
corporations (see Helpman 1984b).
No long-run dynamics exist in the scenario described above. Recent research
has concentrated on discovering mechanisms that generate such long-run dy-
namics. Trade and development theories that explore implications of economies
of scale have joined forces with new approaches to economic growth (for the
latter, see Romer 1986, 1988; Lucas 1988; Helpman 1988). At the heart of
these approaches are dynamic economies of scale (such as the product devel-
opment process) coupled with externalities associated with knowledge capital.
Thus, in Grossman and Helpman (1988) growth peters out because the profit
rate falls over time as more and more brands crowd the differentiated product
sector. This reduces the return on R&D until it stops being profitable. If, however,
knowledge capital serves as an input in R&D and this capital stock rises over
time as a result of experience (that is, learning by doing a la Arrow 1962), it
may counteract the effect of product crowding on the profitability of R&D and
thereby sustain product development and growth in the long run.

Multiple Equilibria
Before discussing the effects of knowledge capital on long-run growth, I would
like to pause to set it in the context of the emergence of multiple equilibria. The
 Helpman 197

tendency of product-specific learning by doing to perpetuate every initial pattern

of specialization introduces persistence into trade patterns. Krugman (1987), for
example, constructed a model with product-specific learning by doing in which
every historically determined pattern of trade and specialization lasts forever.
Under such circumstances temporary shocks-whether from technology, policy,
or other sources-have permanent effects.
The observation that temporary events have lasting effects arises from two
sources that have been widely studied. One is a case in which the long-run
equilibrium depends on initial conditions, for which Krugman's model of learn-
ing by doing provides an instance. Here, shocks that change initial conditions
extract long-run effects. The other is a case in which more than one long-run
equilibrium exists, and the economy can converge to each of them from the
same initial conditions, depending on expectations. This phenomenon has been
recognized in international trade at least since Graham's (1923) famous argu-
ment for tariff protection.
Graham envisioned a two-sector economy whose opening to international
trade may lead to resource migration from the industry in which returns to scale
are increasing, to the industry in which they are decreasing, thereby depressing
GDP so much that the usual gains from trade are outweighed. This observation
led to a heated debate between Graham and Knight (see Helpman 1984a for a
review of the debate). Graham was vindicated by Ethier (1982a), who studied
countries that have an industry with external economies of scale and perfect
competition (that is, a firm's productivity depends on aggregate output, but the
firm treats productivity as an exogenous parameter). In this type of economy a
number of trading equilibria may differ in the degree of specialization in the
increasing-returns industry. In the absence of intersectoral adjustment costs, the
instantaneous allocation of resources relies entirely on expectations about factor
rewards, and several sets of self-fulfilling expectations exist, each one leading
to a different outcome. These outcomes can be Pareto ranked (see Helpman and
Krugman 1985, chap. 3).
As an illustration, consider a two-sector economy with a single resource, say
labor, that faces constant terms of trade and a constant labor-output ratio in
the non-increasing-returns sector. The firm's perceived marginal product value
of labor in the increasing-returns sector (sector X), depends on the industry's
output level; the larger aggregate employment and output, the larger the marginal
value product. Suppose also that the perceived marginal value product equals
zero when the industry's output equals zero, and that the marginal value product
in X is larger than in the alternative use when X employs all resources. Two
self-fulfilling-expectations equilibria will then exist with complete specialization.
In one, all labor works in the constant-returns-to-scale industry and the wage
rate equals its marginal value product in that sector. Labor's marginal value
product in X equals zero, so that there are no incentives to produce in X. In
the other equilibrium, all labor works in X and the wage rate-which equals
the marginal value product in X-exceeds labor's marginal value product in the
198 Noncompetitive Theory

constant-returns industry. The country is clearly better off in the latter equilib-
rium.
Recently Krugman (1989) has extended this analysis to an economy with
adjustment costs in factor reallocation. As usual, the adjustment costs bring
about gradual intersectoral adjustment in response to economic incentives. He
finds that, given some initial conditions, the economy converges to one steady
state, while with others it converges to another. In yet other initial conditions
it may converge to either one of those possible steady states, depending on
expectations (self-fulfilling expectations are assumed throughout). In this last
case the resulting dynamics involve cycles of rising amplitude. Expectation-driven
equilibria are of course not peculiar to international trade; they play a prominent
role in other areas, such as macroeconomics (see, for example, Diamond 1982;
Shleifer 1986; Cooper and Jones 1988; Murphy, Shleifer, and Vishny 1988).
All this implies that in certain circumstances an economy's trajectory is un-
predictable, because it may follow more than one equilibrium trajectory, or that
small shifts in initial conditions may have dramatic long-run effects. In either
case it may be possible to use policies to shift initial conditions or to influence
expectations, to force the economy to follow a desired path. An appealing feature
of such policies is that often they need to be applied for only a short time. As
usual, however, they are formidably difficult to design, because the required
information is seldom available. The long-standing debate about infant industry
protection represents well those difficulties (see Baldwin 1969).
Long-Run Growth
We now return to long-run growth. Suppose that current experience with
product development reduces R&D costs to all future product developers. The
product developer has thus generated a twofold output: an appropriable blue-
print that can be used to acquire future monopoly rents, and a contribution to
knowledge capital that is not appropriable. The contribution to knowledge may
disseminate equally quickly to all future entrepreneurs, or faster to entrepreneurs
from his own country. Suppose also that the differentiated-product sector pro-
vides intermediate inputs that are used in the manufacture of final consumer
goods (as in Ethier 1982b). Each country has the technology to produce a
different consumer good, and trades in both intermediate and final goods.
Grossman and Helpman (1989a) have studied a two-country world of this
type. In their framework both countries converge to the same long-run growth
rate, even if they differ in size and sectoral productivity levels. The long-run
growth rate depends on the size of each country and the composition of demand
for their final goods. When knowledge gets disseminated at an equal speed to
both countries, the larger the country with comparative advantage in R&D and
the smaller the relative demand for the final good in which it specializes, the
faster the common growth rate. The growth rate may be increasing or decreasing
with the size of the country that has comparative disadvantage in R&D, but it
 Helpman 199

is definitely higher the larger the relative demand for the final good in which
that country specializes.
The last point identifies a mechanism of more general relevance. The larger
the relative demand for the final good of the country that has comparative
advantage in R&D, the larger the demand for its resources and the lower the
demand for resources in the other country (other things being equal). Under
these circumstances, the intermediate-product sector and the R&D sector contract
in the former and expand in the latter. Given the structure of comparative
advantage, aggregate effective employment in product development declines in
the world economy, thereby slowing growth (because the growth rate depends
on the equilibrium size of the R&D sector).
If we interpret this model in a South-North context-where the country with
comparative advantage in R&D is the North-this analysis suggests, for example,
that the South grows faster the larger the North, but that the North's growth
rate may be slowed down by a larger South. It also suggests that a shift of
demand from Northern to Southern final goods raises the world's growth rate.
So far, our discussion has relied on what may be termed "natural" comparative
advantage in R&D, which builds on endowed differences in technology. We have
seen that it is an important determinant of long-run growth (and of policy effects,
as I discuss in the next section). If, however, the diffusion of knowledge is faster
within countries than across them, then natural comparative advantage does not
fully determine a country's long-run comparative advantage overall-because
these differing learning speeds give a country that does more R&D to begin with
a lasting cost advantage. In this instance the final position of comparative ad-
vantage depends also on the relative size of the country's resource base and the
derived demand for its resources for other uses. Thus, other things being equal,
long-run comparative advantage in R&D is larger the larger the resource base
and the smaller the demand for the country's final goods.
Innovation and Initation
Comparative advantage in R&D has been prominent in discussion of North-
South trade problems. It is manifested in an extreme form in Vernon's (1966)
product cycle and its later elaborations. In this approach only the North is
capable of developing new products. Immediately after a product is developed
the North has also the cost advantage in its manufacturing, until the production
techniques are standardized. Afterward, the cost advantage-and with it pro-
duction-shift to the cheap labor region, that is, the South.
Vernon's approach was formalized by Krugman (1979; see also Dollar 1986;
Jensen and Thursby 1986, 1987), who assumed that the rate of growth of new
products g (rate of innovation) and the rate at which the South imitates products
in which the North has monopoly power p. (rate of imitation) are constant. This
specification suffices to describe the evolution of products that are manufactured
in every region without specifying additional details of economic structure. In
200 Noncompetitive Theory

the steady state, the South produces a proportion VL/(p+ g) of the available
products. By imposing on these dynamics a model of oligopolistic price com-
petition in differentiated products with labor as the only primary input, Krugman
showed that the long-run relative wage of the South is increasing in (pJg)(LNI
L,), where Ls stands for the South's labor force and LN for the North's labor
force. Hence, the South's relative wage is larger the larger the rate of imitation,
the smaller the rate of innovation, and the smaller its relative labor force.
Grossman and Helpman (1989b) have reexamined the long-run implications
of the product cycle approach in light of the fact that both the rate of innovation
and the rate of imitation result from the interaction of market forces with the
explicit decisions of Northern entrepreneurs to innovate and Southern entre-
preneurs to imitate. Imitators invest resources in learning and reversed engi-
neering in expectation of future monopoly profits, just as innovators invest
resources in R&D in expectation of future monopoly profits. But the innovators,
uncertain as to when their product will be imitated-and hence when their
monopoly profits will cease-discount profits with an interest rate that includes
a risk premium, the risk premium being equal to the rate of imitation.
In this environment the long-run rates of innovation and imitation depend on
country size and sectoral productivity levels. Innovation is faster the larger the
North or the South (with one minor exception), whereas the rate of imitation
is larger the larger the South and the smaller the North. Both regions grow faster
when they trade with each other than in autarky. Now the relative wage of the
South rises with the South's relative labor force (taking account of the endog-
enous response of innovation and imitation). This is just the opposite of Krug-
man's (1979) finding. It shows how crucial the explicit decisions to innovate
and imitate are in bringing into full play the dynamic economies of scale.
To illustrate the point, consider Grossman and Helpman's "wide gap" case.
Here the relative wage of the South is low enough for a Southern imitator to
charge his monopoly price without risking undercutting by the Northern original
innovator. In the wide gap case the South's relative wage is increasing in (,a/
g)(L, - g)/Ls - g) (using a suitable normalization). Hence, for constant g and
i this relative wage increases with the North's labor force and declines with the
South's labor force, as in Krugman. But when the effects of labor on g and p.
are taken into account, the results are reversed. That is, the indirect effects that
changes in labor have on innovation and imitation are stronger than the direct
effects.

Il. POLICY

In competitive economies two efficiency considerations may exist for trade

policy: improvement in the terms of trade and a second (or third) best improve-
ment in resource allocation in the presence of domestic distortions. Both exist
 Helpman 201

in noncompetitive environments. In fact, imperfect competition necessarily in-

volves a domestic distortion because firms do not engage in marginal cost pricing.
For both objectives various different policies might be helpful (at least from the
point of view of a single country). But can any broad policy conclusions be
drawn, such as "whenever domestic firms compete against foreign oligopolistic
firms in export markets we should subsidize their exports" or "whenever do-
mestic import competing firms face noncompetitive foreign exporters in the
domestic market we should impose import restrictions"? The answer turns out
to be negative; no policy conclusion of this sort can validly be drawn. To design
successful policies, instruments must be tailored to particular industries on the
basis of their degree of concentration, the conduct of firms, the position of
domestic firms relative to foreign, the industry's links with other sectors of the
economy, and the like. In short, to exploit imperfect competition for policy
purposes one requires detailed information about the economy. Such information
is seldom available (seeHelpman and Krugman 1989); furthermore, experiments
with actual data reveal that the potential gains to be derived from such policies
are rather small (see Helpman and Krugman 1989, chap. 8). On the other hand,
existing tariff structures go much too far in terms of protection relative to optimal
policies (see Harris and Cox 1984). The implication is that, given the current
state of knowledge, a government that engages in a deliberate welfare-increasing
policy takes significant risks; it stands to gain little but may cause significant
losses.
The profusion of cases that need to be considered is described in figure 1 for
a single market that can be either domestic or foreign (thereby immediately
doubling the number of cases). There can be perfect competition, one of two
cases of one-sided market power, or a case of two-sided market power. When
market power is one sided, it of course makes a great deal of difference whether
domestic or foreign suppliers own market power. In addition, a matrix of this
sort applies to different types of conduct: one matrix for single firms with
monopoly power, one for Cournot oligopolies (in which a small number of firms
compete in quantities), one for Bertrand oligopolies (in which a small number
of firms compete in prices), one for a cartel of a particular form (in which the
allocation of benefits among members results from a particular solution to a
bargaining problem), and so on. Then there are links with other industries that
matter-we need to know how each policy affects entry, and so on.
The task of sorting and integrating the variables to elicit results that will be
useful for policy is not as hopeless as it sounds: a number of the results described
below reveal important considerations for a successful policy and indicate the
information that will be required to make it effective.
In the first two subsections I discuss situations in which the number of firms
is constant and all firms minimize costs. This state of affairs ensures efficiency
of production (that is, output is on the transformation surface) although the
composition of output need not be efficient. If we restrict attention to homo-
202 Noncompetitive Theory

Figure 1. Considerations for Trade Policy: Competitiveness and Market Power

Foreign

Competitive Market power

Competitive Perfect competition One-sided

Home

Market powers One-sided market power Two-sided market power

Source: Author's typology.

geneous products and trade taxes only, the change in aggregate welfare can be
measured by

dU-- -m dp'` + t-dm + (p - c) dX,

where m is the vector of net imports (a negative component represents exports),
p` stands for the foreign price vector and p for the domestic price vector (for
consumers and producers), t - p - p' represents the vector of trade taxes (a
positive component represents an import tariff if the good is imported and an
export subsidy if the good is exported), c is the vector of marginal costs, and
X stands for the output vector.
The first term on the right-hand side of the equation represents the usual
terms of trade effect: a country gains when the price of its exports rises or the
price of its imports declines. The remaining two terms represent considerations
of efficient supply. The last term says that an expansion of domestic output of
goods that are priced above marginal cost raises welfare. Competitive industries
price according to marginal cost, so that their contribution to this term equals
zero. In noncompetitive sectors price exceeds marginal cost, which implies that
expansion of their output is desirable (because domestic valuations exceed supply
costs). Hence, other things being equal, policies that lead to an average expansion
of noncompetitive sectors improve welfare. A similar interpretation can be ap-
plied to the second term in relation to imports. The marginal cost of imports
equals the foreign price. If the domestic price exceeds the foreign price (as a
 HeIpmnan 203

result of a tariff or a quota, for example) an expansion of imports is desirable.

Much of the welfare analysis of various policies concerns the tradeoffs among
these three considerations.

One-Sided Market Power

Bhagwati in his famous (1965) paper analyzes a tariff in the presence of a
domestic monopolist and fixed foreign supply price. He shows that in the sit-
uation depicted in figure 2-where the foreign price p ¢ is below the prohibitive
domestic price P-gradual tariff increases (beginning from zero) that raise the
domestic price toward P induce the monopolist to expand output. In this range
he chooses output by equating price to marginal cost. Hence, the contribution
of the third term in the equation above equals zero and welfare declines, because
imports contract and the domestic price exceeds the foreign price (the contri-
bution of the first term also equals zero because the foreign price does not
change). When the domestic price reaches the prohibitive price P, imports cease
and are never renewed for further tariff increases. Further tariff increases, how-
ever, induce the monopolist to reduce output, because now he equates price
with demand until the monopoly price pMis reached. In this case the second
term in the equation equals zero (there are no imports), but welfare declines
owing to the third term, because price exceeds marginal costs and output de-
clines. This example also shows that with imperfect competition import pro-
tection can be effective even when imports equal zero. Here the effect of the
mere threat of imports is not negligible.
In this instance a tariff is more restrictive than a quota in the following sense.
Suppose we replace the tariff with a quota that equals the import volume under
the tariff. The monopolist responds by cutting back output. Quotas thus lead
to lower consumption and a higher price. The reasoning here needs to be clearly
understood. A quota reduces the elasticity of demand perceived by the monop-
olist. In its absence a price increase leads him to lose sales to consumers on
account of the downward-sloping demand curve and to importers who replace
his sales (when imports are imperfect substitutes for his output; otherwise he
loses all sales to importers). In its presence he does not lose sales to importers,
and so his effective demand curve becomes steeper. This lowers his marginal
revenue, and he responds by contracting output. The same reasoning applies to
Cournot oligopolies. It can also be used to show that a quota equal to the free
trade level of imports leads the monopolist or a Cournot oligopoly to contract
output. Hence, whereas in a competitive environment a quota at the free trade
level of imports has no effects, here it does. With monopoly power, moreover,
quotas that exceed the free trade level of imports (up to a limit) also lead to
lower consumption and a higher price. So for an oligopoly the quota leads to
a more collusive outcome.
The question of whether quotas (or quantitative restrictions) facilitate col-
lusion is of great interest. The previous analysis suggests that they do, and I
204 Noncompetitive Theory

Figure 2. Effects of Tariff Increases on Monopolistic Output

PM MC

\D

MR
6 X

Note: p price, p = foreign price, P = domestic price, PM = monopoly price, MR marginal

revenue, MC = marginal cost, D = demand, X = output.

think that this is a reasonable presumption. But there are exceptions, one of
which serves to illustrate additional considerations.
The previous analysis relied on a static environment. Recently, however, much
of oligopoly theory has been reformulated in order to allow firms to interact
repeatedly. Repetition brings in important new elements, such as the possibility
of implicit, as opposed to explicit, collusion. (Explicit collusion, in the form of
a binding agreement, is often impossible because such a contract cannot be
specified for all relevant circumstances or is illegal.)
Implicit collusion of repeatedly interacting firms may force an oligopoly to
charge a price lower than the monopoly price (see Tirole 1988, chap. 6). That
is, implicit collusion may not suffice to achieve the fully cooperative outcome
attainable if it were possible to write a binding contract, for the following
reasons.
fn order to sustain an implicit agreement it has to be in the interest of each
member: the present value of profits to be obtained from the cartel must not
fall short of the present value of profits derived by deviating from the implicit
 Helpmnan 205

agreement. It is then usually supposed that if a member deviates, the cartel falls
apart in the next period and the noncooperative equilibrium (say, Cournot) gets
established forever (this equilibrium is time-consistent in the sense that at each
point in time every firm finds it desirable to follow the specified strategy). Hence,
a potential deviator has to compare the one-period gains from choosing his best
deviant strategy when everyone else obeys the implicit agreement with the present
value of future losses that will result from the noncooperative outcome. The
comparison depends on the size of the one-period gains, on how bad the non-
cooperative outcome is relative to the cooperative outcome, and on the rate at
which future profits are discounted. Naturally, the smaller the one-period gains
from a deviation and the worse the noncooperative outcome, the less likely it
is that a deviation will pay off.
For these reasons an implicit agreement sometimes needs to specify a price
below the monopoly price in order to sustain collusion. The lower price reduces
the gains from deviation to the point at which collusion is viable, while at the
monopoly price the gains from deviation are too high to sustain collusion (be-
cause when everyone restrains output in order to achieve the monopoly price,
the deviant can make large one-period profits). Rotemberg and Saloner (forth-
coming) have shown that under those circumstances a quota at the free trade
level may restrict collusion rather than facilitate it. In their example, the quota
raises the noncooperative equilibrium profit level (which is possible, as we have
seen above, even though the quota exceeds the noncooperative import level).
This forces the cartel to reduce price in order to prevent profitable deviation.
Their example (even if not realistic) shows how important repetitive interactions
can be for policy considerations (see also Davidson 1984 on tariffs).
We turn now from import-competing markets in which domestic firms have
market power to those in which the domestic firms are competitive and foreign
suppliers have market power. Here a desirable trade policy may consist of import
subsidies rather than tariffs.
Suppose a monopolist foreign supplier who chooses a strategy that equates
marginal revenue of the import demand function to his marginal costs. Now
suppose that we impose a small tariff. If foreign supply were competitive and
upward-sloping, the tariff would have improved the terms of trade and would
have raised welfare. With the foreign supply controlled by a monopolist, there
is no guarantee that a small tariff improves the terms of trade, and the terms
of trade are the only relevant consideration. To illustrate the last point, observe
that under those circumstances the last two terms on the right-hand side of the
equation above are zero, because domestic firms price according to marginal
costs and the initial tariff rate equals zero. Hence, we only need to consider the
effect of a small tariff on the terms of trade. Now, the tariff, of say $1 per unit
imports, raises the monopolist's marginal costs of supplying the domestic market
by $1. Assume for simplicity that his tariff-exclusive marginal costs are constant.
Then the contraction of sales equals the inverse of the slope of the marginal
revenue curve, because he equates marginal revenue to marginal costs. The
206 Noncompetitive Theory

increases in the domestic price equals the contraction of sales times the slope of
the demand curve. Therefore the domestic price rises by less than $1 if, and only
if, the marginal revenue curve is steeper than the demand curve. If the domestic
price rises by less than $1, the terms of trade improve. The terms of trade worsen
when the domestic price rises by more than $1 (because the import price p-
equals p - 1). For example, when the demand curve is linear it is flatter than
the marginal revenue curve and a tariff improves the terms of trade. Conversely,
when the demand curve has a constant elasticity that exceeds 1 it is steeper than
the marginal revenue curve and a tariff worsens the terms of trade. In the latter
case an import subsidy improves the terms of trade. We have therefore a simple
condition on the relative slopes of the demand and marginal revenue curves that
determines whether a tariff or an import subsidy is desirable (see Brander and
Spencer 1984).
An important point about this type of one-sided market power is that even
in cases in which a tariff improves welfare its replacement with a quota reduces
welfare below the free trade level. This does not result from differences in the
level of domestic production. Indeed, if the quota equals the import level that
prevails under the tariff, both policies lead to the same levels of imports, domestic
production, and domestic price. The difference arises from the fact that under
the quota the foreign monopolist exploits the quantitative restriction to charge
the consumer price. Therefore instead of improving the terms of trade the quota
worsens them. Alternately, under a quota the equivalent of the tariff revenue
(which translates into quota rents under competition) accrues to the monopolist
rather than to domestic owners of import licenses (see Shibata 1968). This result
applies also to foreign oligopolies (which compete with imperfect substitutes)
as long as the quota exceeds a minimal level. Helpman and Krugman (1989,
chap. 4) show that for sufficiently small quota levels domestic owners of import
licenses collect rents, but that in the linear demand case these are never sufficient
to compensate for the initial losses (see also Krishna 1988a, 1988b). Whether
circumstances exist in which a quota can bring about a less collusive outcome
that would be preferable to free trade remains an open question.

Strategic Policy
In the presence of two-sided market power, economic policy has a strategic
value as well: it changes the terms on which domestic noncompetitive firms
interact with foreign noncompetitive firms. The best-known examples in inter-
national trade concern precommitment strategies. In particular, in situations in
which domestic firms do not have the means to precommit to a particular course
of action-even though that is desirable-the government can sometimes act
to ensure (albeit indirectly) the desired precommitment. This typically requires
the government to have the first-mover advantage-to be able to announce or
execute a reliable policy before firms complete their strategic choices.
For instance, suppose that a domestic firm competes against a foreign firm in
a third-country market. (We are concerned only with our firm's gross profits.)
 Helpmilan 207

Competition takes place in two stages. In the first stage, firms decide whether
to enter the market. This may involve the development of a product or the
settting up of a marketing network. In the second stage, the firms produce and
compete in either price or quantity. Now, suppose that the market is small, so
that when only one firm enters its second-stage profits exceed its first-stage entry
costs, and when both enter, second-stage profits fall short of entry costs in each
one of them. In this case two equilibria exist: one in which only the domestic
firm enters and the other in which only the foreign firm enters. Clearly, the
domestic firm and the home country prefer the former.
Because the two equilibria exist, the domestic government may want to force
establishment of the preferred equilibrium. The following strategic policy could
achieve this. Before the firms make their entry decisions the government provides
the domestic firm with an entry subsidy that exceeds the loss that materializes
when both firms enter. Under these circumstances the domestic firm chooses to
enter independently of the foreign firm's decision. Consequently, the foreign
firm does not enter and this is the unique equilibrium. The same can be achieved
by a government commitment to a lump-sum export or production subsidy as
long as the commitment is made before the entry decision and a mechanism is
in place to make it good. Second-best policies in the form of ad valorem export
subsidies can also be used for this purpose. Naturally, the foreign government
has an equal incentive to engage in a strategic policy, and so the outcome may
be a three-stage game in which governments choose policies in the first stage,
firms make entry decisions in the second, and production and sales take place
in the third (see Dixit and Kyle 1985).
Strategic policies do not apply exclusively to entry; they can also be used
effectively when domestic and foreign firms have established themselves in a
market. Consider an export market with one established domestic and one
foreign firm that compete in prices with imperfectly substitutable products. Let
each firm's profit maximization require a price rise in response to its rival's price
increase. In this case the domestic government can raise its firm's gross profit
level (and therefore welfare) by taxing exports (see Grossman and Eaton 1986).
This result can be shown as follows (Helpman and Krugman 1989, chap. 5):
the firm equates perceived marginal revenue to marginal costs, where perceived
marginal revenue is calculated for a fixed price of the rival. When the domestic
firm reduces price, however, the rival responds with a price reduction of his
own. Nevertheless the home firm cannot take advantage of this information as
long as both set prices simultaneously. If one could exploit this information,
one would recognize that true marginal revenue is lower than perceived marginal
revenue, because the foreign firm's price response to the home firm's price
reduction brings about an increase in home sales that is smaller than the perceived
sales increase. For this reason it is desirable to induce the home firm to charge
higher prices and limit sales. An export tax achieves just that. The government
can exploit the first-mover advantage by establishing an export taxation program
that acts as a precommitment device. Then the firms compete with the program
208 Noncompetitive Theory

in place and the outcome is higher prices for both products. Here-unlike the
entry promotion programs discussed previously-both countries gain higher
profits, because the best response of a firm leads to higher profits the higher the
rival's price. In this case a two-stage game in which both governments choose
taxation programs in the first stage and firms choose prices in the second leads
to a time-consistent equilibrium in which both countries are better off than
under free trade.
Strategic policies thus need not lead to a conflict of interest. Although in the
entry-intervention case one government's successful policy harmed the rival
country, in the export-intervention case one government's successful policy
brought a positive benefit to the other country. The inference to be drawn is
more subtle, however, than a simple distinction between entry and export pol-
icies. In the first example entry decisions were strategic substitutes (when one
firm entered, the other abstained from entering), whereas in the second prices
were strategic complements (when one firm raised its price, the other responded
with a price increase). The distinction between strategic substitutability and
complementarity is key in understanding these results. The same distinction is
also central in understanding the direction of desired policies.
These points can be demonstrated by means of an alternative example of two
established firms that compete in an export market where governments intervene
in foreign trade. But this time instead of competing in prices (a la Bertrand) the
domestic and foreign firms compete in quantities (a la Cournot). Assume-as
would be most likely-that a firm responds with an output contraction to an
output expansion of its rival, thus ensuring strategic substitutability. The critical
difference from the previous example is not the strategy space of the firms but
rather the strategic relationship. Now an export subsidy rather than an export
tax proves to be desirable (see Brander and Spencer 1985).
The argument can be made as follows (see Helpman and Krugman 1989,
chap. 5). The domestic firm chooses output that equates perceived marginal
revenue with marginal costs. It calculates perceived marginal revenue for a fixed
output of the rival. The rival, however, responds with an output decline to an
output increase of the domestic firm. Consequently, true marginal revenue ex-
ceeds perceived marginal revenue and the firm would earn higher profits if it
could precommit to a larger output level. Unfortunately it cannot, because both
firms play simultaneously. The government can improve the outcome by pro-
viding the necessary precommitment. To raise output the government should
subsidize exports. The subsidy has to be in place (or be committed to be put
into place) before the firms make their decisions. The firms can then choose
outputs recognizing the existence of the export promotion program and end up
in an equilibrium in which the domestic firm sells more and the foreign firm
sells less.
Two points need to be underlined. First, in contrast to the Bertrand case, here
export subsidies are required rather than export taxes. Second, countries face a
conflict of interest in their trade policies. When one country engages in export
 Helpman 209

promotion the other loses, because the policy-active country forces its rival to
contract output, and output contraction as a best response to the domestic firm's
output expansion leads to lower profits for the foreign firm. This conflict of
interest leads to a Prisoners' Dilemma in the policy game. For suppose that there
are two stages: governments choose their export policies in the first, and firms
choose quantities in the second. For simplicity, also assume symmetry and con-
stant marginal costs. Then in the resulting time-consistent equilibrium both
governments subsidize exports and both firms sell more than under free trade.
Observe, however, that even under free trade a Cournot duopoly produces too
much, in the sense that joint output exceeds the output level of a single mo-
nopolist so that a further output expansion reduces profits per firm. Hence, the
two countries are better off in the free trade equilibrium than in the equilibrium
with active policies. The problem is that when one country does not promote
its exports it pays the other to do so. Consequently, free trade is not an equi-
librium unless policies are coordinated (that is, governments cooperate in the
first stage).
We have seen that one can make a case for export taxation as well as export
promotion on strategic grounds, depending on circumstances. In either set of
circumstances the existence of more than one domestic firm strengthens the need
for taxation-because the policymaker cares about aggregate profits of the
exporting firms while each firm cares only about its own profit level (see Dixit
1984). Naturally, when a single domestic firm considers the effects of its price
or output decisions on perceived marginal profits it does not take into account
the effects on profits of other firms. Therefore, other things being equal, prices
are too low and output levels too high when a number of domestic firms par-
ticipate in the oligopolistic market. To offset this negative externality, an export
tax is called for. Clearly, in the Bertrand case this strengthens the need for export
taxation. In the Cournot case it conflicts with the need to subsidize exports on
strategic grounds. The net result may be the need either for lower export subsidies
or for taxation.
Entry
So far the discussion has concentrated on cases in which the number of firms
is assumed to be fixed, or more to the point, in which firms do not enter or exit
in response to policy measures. This is not, however, a safe assumption. Export
subsidies may lead to entry, whereas export taxes may lead to the exit of domestic
firms, independently of conduct. This is a significant consideration whenever
there are firm-specific increasing returns to scale. For example, when fixed entry
costs exist one must take account of the resource loss from entry of new firms
(see Helpman and Krugman 1989, chap. 5). This consideration weakens the
case for an export subsidy and strengthens the case for an export tax. In the
presence of free entry that drives to zero tax- and subsidy-inclusive profits, export
promotion damages welfare while a small export tax raises welfare (see Horst-
man and Markusen 1986).
210 Noncompetitive Theory

The last point applies to all forms of conduct. If-as has been assumed so
far for the industry under discussion-domestic firms export but do not sell in
the local market, the change in welfare equals the change in aggregate gross
profits. Conversely, aggregate gross profits equal aggregate net profits plus tax
revenue minus the subsidy bill. Free entry ensures zero net profits. Therefore
the change in welfare equals the change in net revenue. The imposition of a tax
raises revenue, and thereby welfare. The provision of a subsidy reduces revenue,
and thereby welfare.
All this suggests that if anything there is a presumption in favor of export
taxation rather than export promotion. Export promotion is desirable only when
a firm's choice variables are strategic substitutes, the number of firms is rather
small, and the scope for entry in response to export subsidies is limited.
Intersectoral Links
To evaluate the response of resource allocation to policy we need to use correct
measures of marginal costs. Much of the previous discussion relied on the as-
sumption that firms use social marginal costs in their profitability calculations.
This supposition is correct when all other sectors are competitive, but it is
typically incorrect when some are noncompetitive. For this reason policymakers
need to know the difference between true and perceived marginal costs as well
as the difference between true and perceived marginal revenue. In other words,
one cannot design a successful policy without properly taking account of inter-
sectoral links (see Dixit and Grossman 1986). For example, when true marginal
revenue in an export sector exceeds perceived marginal revenue it does not
guarantee that export promotion will increase welfare. In order to see this point,
suppose that the subsidized sector expands in response to the policy incentive
by drawing resources from another export sector in which true marginal revenue
exceeds perceived marginal revenue. If the divergence in the latter sector is large
enough, the net result will be a decline in aggregate profits.
Differentiated Products
In the presence of product differentiation, a variety effect exists along with
the terms of trade and the efficient supply effects that appear in the equation
above, that has a bearing on policy design. Before discussing it, however, I would
like to make two points.
First, the supply of many brands does not eliminate a country's market power
even when the country is very small. Gros (1987) has demonstrated this in the
following way (see Helpman and Krugman 1989, chap. 7, for a simple expo-
sition). In a one-sector, one-factor, two-country world with product differen-
tiation and a constant elasticity of substitution across brands (see Dixit and
Stiglitz 1977), output per product does not depend on country size. The reason
is that with Dixit-Stiglitz preferences the elasticity of demand does not change
with the number of products. In addition, the number of brands is proportional
 Helpman 211

to country size. Thus ad valorem trade taxes, which do not affect the elasticity
of demand, cannot change the number of brands that each country produces,
or output per brand. If they affect anything at all it must be the terms of trade.
Calculating the optimal tariff for the home country, one finds that it equals
1/(1 - s)(e - 1), where s represents the share of world spending allocated to
the home country's products and e represents the constant elasticity of demand.
The smaller the country the smaller s and the smaller the optimal tariff. But
even when the relative size of the country shrinks to zero, the optimal tariff
remains positive. For no matter how small a country is, it specializes in a range
of products in which it maintains monopoly power; the demand for a variety
is downward sloping, and even a small country can affect its terms of trade.
The second point concerns the production efficiency effect. Consider a case
in which the number of products and relative prices are constant but output per
brand can change (see Helpman and Krugman 1989, chap. 7, for a model that
ensures it). Then the imposition of import duties on brands that compete with
domestic products shifts domestic demand from foreign to domestic varieties
and shifts demand away from all varieties. Output per domestic brand may thus
expand or contract and welfare may increase or decline (see Flam and Helpman
1987; Helpman 1989).
To return to the effect of variety on welfare: other things being equal, con-
sumers prefer more variety. One can, in fact, think about a consumer price index
that is lower the larger the variety choice. If a tariff reduces this price index by
raising the available variety choice (as in Flam and Helpman 1987) or by chang-
ing the composition of products in favor of the home country at the expense of
the foreign country (as in Venables 1987), it necessarily improves home welfare.
But the increase in variety is not guaranteed (see Markusen 1988; Helpman
1989). A tariff may shift demand away from differentiated products so much
that available variety is reduced. This contraction of variety choice may bring
a decline in welfare. Conversely, in some circumstances the tariff raises available
variety and welfare (for example, see Flam and Helpman 1987; Venables 1987).
Consequently, it is not clear a priori whether small tariffs are desirable; all the
effects mentioned above have to be taken into account. Large tariffs lead to
additional welfare losses that stem from the undersupply of imports (the second
term in the equation above). Moreover, even where tariffs are desirable they
correct only indirectly the distortion that emerges from monopolistic or oligop-
olistic competition. Direct correction of the distorted price-cost margins, if fea-
sible, would be preferable.
Promotion of Growth
In a dynamic economy the static issues reviewed so far have to be augmented
by explicit consideration of the links between policy and growth. In the growth
models described in section 1, commercial policy and other forms of industrial
policy affect long-run growth rates, exerting strong influences on welfare. But
212 Noncompetitive Theory

the resulting relations are far from simple. For example, in the world studied
by Grossman and Helpman (1989a)-where both countries develop new inter-
mediate products and one of them has a comparative advantage in R&D-an
import tariff on final consumer goods slows down world growth if imposed by
the country with a comparative advantage in R&D and speeds up world growth
if imposed by the country with a comparative disadvantage in R&D. The intuition
behind this result reveals a channel of influence that is not specific to the model.
When a country imposes a tariff on imports of final goods, it shifts the com-
position of demand toward its own final goods. The expansion of the final goods
sector draws resources from the manufacturing of intermediate products and
product development. Opposite shifts in resource allocation take place in the
other country. In particular, its product development sector expands. Whether
these changes accelerate or slow down growth depends on whether the con-
traction of the R&D activity in the tariff-imposing country is smaller or larger
than the expansion of the R&D activity in the other country. The answer depends
on comparative advantage in R&D; world output of R&D declines only if the
tariff-imposing country has a comparative advantage in R&D.
In this type of world one expects R&D subsidies to speed up growth-as
indeed turns out to be so when the subsidy is provided by the country whose
R&D is relatively more efficient, or when both countries subsidize at an equal
rate. When the country with relatively less efficient R&D subsidizes product
development, however, it may lead to slower growth. The outcome depends on
structural features that cannot be spelled out in the available space.
Conversely, in the North-South model with a product cycle that was discussed
in section I (Grossman and Helpman 1989b) innovation subsidies in the North
and imitation subsidies in the South speed up growth. However, they each affect
the rate of imitation differently, and thereby the average length of the first phase
of the product cycle. Innovation subsidies reduce the rate of imitation and the
average length of the first phase, whereas imitation subsidies raise the rate of
imitation and shorten the average length of the first phase.
Grossman and Helpman (1989c) study a small-country variant of their growth
models with a focus on the consequences of various policies for welfare. The
resulting equilibrium differs from the first-best because, first, of markup pricing
in the differentiated intermediate product industry, and second, of the externality
that a product developer imposes on future product developers through his
contribution to knowledge capital. Small R&D subsidies raise welfare. Larger
subsidies accelerate growth more, but eventually reduce welfare. A small tariff
that speeds up growth may either raise or reduce welfare. But whether it speeds
up or slows down growth depends on the factor intensity of the import-com-
peting sector relative to the exporting sector and the product development ac-
tivity. Here, quotas also affect growth and welfare. They are particularly dam-
aging relative to tariffs if they induce rent seeking that uses up entrepreneurship
in which product development is relatively intensive.
 Helpinani 2173

III. CONCLUDING COMMENTS

The new theory of international trade and trade policy evidently encompasses
numerous relevant elements. Although judgments may differ as to the relative
importance of each, I believethere should be no controversy over the significance
of the package as a whole. Existing empirical evidence on trade structure (see,
for example, Havrylyshyn and Sivan 1984; Balassa 1986; and Helpman 1987)
support the new view, and "calibration" studies of policy experiments (see
Helpman and Krugman 1989, chap. 8, for a review) give quantitative support
to many of the considerations that were discussed in section 11.The most recent
studies that embody those elements in a dynamic setup should make the approach
even more useful.
One major conclusion emerges from both theory and the "calibration" studies:
there are no simple answers to many important questions. This conclusion ap-
plies with particular force to policy concerns. Proper evaluation of outcomes
requires detailed information about conduct, market structure, entry constraints,
intersectoral links, and the like; we need more empirical studies designed to
reveal this information. Such studies, as in the past, will also help to identify
weaknesses in the theory and point out directions for future research. In any
case, since (a) the information needed for a successful policy design is not avail-
able; (b) the policy recommendations are very sensitive to this information; and
(c) the "calibration" studies indicate that good policies improve welfare only
slightly; free trade remains a good rule of thumb-the more so given retaliation,
the competitive pressure of a free trading world system, and the political economy
of protection.

REFERENCES

Arrow, Kenneth J. 1962. "The Economic Implications of Learning by Doing." Review

of Economic Studies 29: 155-73.
Balassa, Bela. 1967. Trade Liberalization among Industrial Countries. New York:
McGraw-Hill.
- =. 1986. "Intra-Industry Specialization: A Cross-Country Analysis." Eutropean
Economic Review 30: 27-42.
Baldwin, Robert E. 1969. "The Case Against Infant-Industry Protection." Jouirnal of
Political Economy 77: 295-305.
Bhagwati, Jagdish N. 1965. "On the Equivalence of Tariffs and Quotas." In Robert E.
Baldwin and others, eds., Trade, Growth and the Balance of P'aymenlts: Essays in
Honor of Gottfried Haberler. Chicago: Rand McNally.
Brander, James A., and Barbara J. Spencer. 1984. "Tariff Protection and Imperfect
Competition." In Henryk Kierzkowski, ed., Monopolistic Competition and Interna-
tional Trade. Oxford, England: Blackwell.
. 1985. "Export Subsidies and Market Share Rivalry." Journal of International
Economics 18: 83-100.
214 Noncompetitive Theory

Cooper, Russell, and Andrew Jones. 1988."Coordinating Coordination Failures in

Keynesian Models." Quarterly Journal of Economics 103: 441-63.
Davidson, Carl. 1984. "Cartel Stability and Tariff Policy." Journal of International
Economics 17: 219-37.
Diamond, Peter. 1982. "Aggregate Demand Management in Search Equilibrium." Jour-
nal of Political Economy 90: 881-94.
Dixit, Avinash K. 1984. "International Trade Policy for Oligopolistic Industries." Eco-
nomic Journal (Supplement) 94: 1-16. Reprinted in Jagdish Bhagwati, ed., Interna-
tional Trade: Selected Readings. Cambridge, Mass.: MIT Press, 2d ed., 1987.
Dixit, Avinash K., and Gene M. Grossman. 1986. "Targeted Export Promotion with
Several Oligopolistic Industries." Journal of International Economics 21: 233-50.
Dixit, Avinash K., and A. S. Kyle. 1985. "The Use of Protection and Subsidies for Entry
Promotion and Deterrence." American Economic Review 75: 139-52.
Dixit, Avinash K., and Joseph E. Stiglitz. 1977. "Monopolistic Competition and Opti-
mum Product Diversity." American Economic Review 67: 297-308.
Dollar, David. 1986. "Technological Innovation, Capital Mobility, and the Product Cycle
in the North-South Trade." American Economic Review 76: 177-90.
Drazen, Allan. 1985. "State Dependence in Optimal Factor Accumulation." Quarterly
Journal of Economics 100: 357-72.
Drazen, Allan, and Nils Gottfries. 1987. "Seniority Rules and the Persistence of Un-
employment in a Dynamic Optimizing Model." Institute for International Economic
Studies Seminar Paper 387. Stockholm.
Eaton, Jonathan, and Gene M. Grossman. 1986. "Optimal Trade and Industrial Policy
under Oligopoly." Quarterly Journal of Economics 101: 383-406. Reprinted in Jag-
dish Bhagwati, ed., International Trade: Selected Readings. Cambridge, Mass.: MIT
Press, 2d ed., 1987.
Ethier, Wilfred J. 1982a. "Decreasing Costs in International Trade and Frank Graham's
Argument for Protection." Econometrica 50: 1242-68.
. 1982b. "National and International Returns to Scale in the Modern Theory of
International Trade." American Economic Review 72: 389-405.
Flam, Harry, and Elhanan Helpman. 1987. "Industrial Policy under Monopolistic Com-
petition." Journal of International Economics 22: 79-102.
Graham, Frank D. 1923. "Some Aspects of Protection Further Considered." Quarterly
Journal of Economics 37: 199-227.
Gros, Daniel. 1987. "A Note on the Optimal Tariff, Retaliation and the Welfare Loss
from Tariff Wars in a Framework with Intra-Industry Trade." Journal of International
Economics 23: 357-67.
Grossman, Gene M., and Elhanan Helpman. 1988. "Product Development and Inter-
national Trade." National Bureau of Economic Research Working Paper 2540. Cam-
bridge, Mass. Forthcoming in the Journal of Political Economy.
. 1989a. "Comparative Advantage and Long-Run Growth." National Bureau of
Economic Research Working Paper 2809. Cambridge, Mass. Forthcoming in American
Economic Review.
-. 1989b. "Endogenous Product Cycles." Foerder Institute for Economic Research,
Working Paper 10-89. Tel Aviv, Israel.
 Helpman 215

1989c. "Growth and Welfare in a Small Open Economy." Foerder Institute for
Economic Research, Working Paper 15-89. Tel Aviv, Israel.
Harris, Richard, and David Cox. 1984. Trade, Industrial Policy, and Canadian Man-
uifacturing. Toronto, Canada: Ontario Economic Council.
Havrylyshyn, Oli, and E. Civan. 1984. "Intra-Industry Trade and the State of Devel-
opment." In P. K. M. Tharakan, ed., The Economics of Intra-Industry Trade. Am-
sterdam: North-Holland.
Helpman, Elhanan. 1984a. "Increasing Returns, Imperfect Markets, and Trade Theory."
In Ronald W. Jones and Peter B. Kenen, eds., Handbook of International Economics.
Vol. 1. Amsterdam: North-Holland.
- . 1984b. "A Simple Theory of International Trade with Multinational Corpo-
rations." Journal of Political Economy 92: 451-71. Reprinted in Jagdish Bhagwati,
ed., International Trade: Selected Readings. Cambridge, Mass.: MIT Press, 2d ed.,
1987.
-. 1985. "International Trade in Differentiated Middle Products." In Donald Hague
and Karl G. Jungenfelt, eds., Structural Adjustment in Developed Open Economies.
London: Macmillan.
- . 1987. "Imperfect Competition and International Trade: Evidence from Fourteen
Industrial Countries." Journal of the Japanese and International Economies 1: 62-
81. Reprinted in Michael A. Spence and Heather A. Hazar, eds., International Com-
petitiveness. Cambridge, Mass.: Ballinger, 1988.
- 1988. "Growth, Technological Progress and Trade." Empirica [Austrian Eco-
nomic Papers] 15: 5-25.
- 1989. "Monopolistic Competition in Trade Theory." Frank Graham Memorial
Lecture, 1989. Special Paper 16, International Finance Section, Princeton University.
Princeton, N.J.
Helpman, Elhanan, and Paul R. Krugman. 1985. Market Structure and Foreign Trade.
Cambridge, Mass.: MIT Press.
-. 1989. Trade Policy and Market Structure. Cambridge, Mass.: MIT Press.
Horstman, Ignatius, and James R. Markusen. 1986. "Up the Average Cost Curve: In-
efficient Entry and the New Protectionism." Journal of International Economics 20:
225-47.
Jensen, Richard, and Marie Thursby. 1986. "A Strategic Approach to the Product Life
Cycle." Journal of International Economics 21: 269-84.
. 1987. "A Decision Theoretic Model of Innovation, Technology Transfer and
Trade." Review of Economic Studies 54: 631-49.
Krishna, Kala. 1988a. "The Case of the Vanishing Revenues: Auction Quotas with
Monopoly." Harvard University. Cambridge, Mass. Processed.
. 1988b. "The Case of the Vanishing Revenues: Auction Quotas with Oligopoly."
Harvard University. Cambridge, Mass. Processed.
Krugman, Paul R. 1979. "A Model of Innovation, Technology Transfer, and the World
Distribution of Income." Journal of Political Economy 87: 253-66.
- 1987. "The Narrow Moving Band, The Dutch Disease and the Competitive
Consequences of Mrs. Thatcher: Notes on Trade in the Presence of Dynamic Scale
Economies." Journal of Development Economics 27: 41-55.
216 Noncompetitive Theory

1989. "History vs. Expectations." Massachusetts Institute of Technology. Cam-

bridge, Mass. Processed.
Leamer, Edward. 1984. Sources of Comparative Advantages. Cambridge, Mass.: MIT
Press.
Linnemann, Hans. 1966. An Econometric Study of International Trade Flows. Amster-
dam: North-Holland.
Lucas, Robert E., Jr. 1988. "On the Mechanics of Economic Development." Journal of
Monetary Economics 22: 3-42.
Markusen, James R. 1988. "Specialized Intermediate Inputs and Derationalizing Tariffs."
London, Canada: University of Western Ontario. Processed.
Murphy, Kevin M., Andrei Shleifer, and Robert Vishny. 1988. "Industrialization and
the Big Push." National Bureau of Economic Research Working Paper 2708. Cam-
bridge, Mass.
Romer, Paul M. 1986. "Increasing Returns and Long-Run Growth." Journal of Political
Economy 94: 1002-37.
. 1988. "Endogenous Technological Change." University of Chicago. Processed.
Forthcoming in Journal of Political Economy.
Rotemberg, Julio J., and Garth Saloner. Forthcoming. "Tariffs vs. Quotas with Implicit
Collusion." Canadian Journal of Economics.
Shibata, Hirofumi. 1968. "A Note on the Equivalence of Tariffs and Quotas." American
Economic Review 58: 137-42.
Shleifer, Andrei. 1986. "Implementation Cycles." Journal of Political Economy 94:
1163-90.
Tirole, Jean. 1988. The Theory of Industrial Organization. Cambridge, Mass.: MIT Press.
Venables, Anthony J. 1987. "Trade and Trade Policy with Differentiated Products: A
Chamberlinian-Ricardian Model." Economic Journal 97: 700-17.
Vernon, Raymond. 1966. "International Investment and International Trade in the Prod-
uct Cycle." Quarterly Journal of Economics 80: 190-207.
PRO C E ED ING S OF T HE W OR L D B AN K ANN U AL CON FE RE NC E
ON D EVE LO PM ENT EC O N O M I C S 1989

COMMENT ON "THENONCOMPETITIVETHEORY OF INTERNATIONALTRADE AND

TRADE POLICY," BY HELPMAN

T. N. Srinivasan

Roughly a decade ago some perceptive trade theorists saw an intellectual

arbitraging opportunity in applying developments in industrial organization to
the theory of international trade and policy. Pioneer arbitrageurs, including
Professor Helpman, reaped handsome returns by imaginatively combining in-
sights of industrial organization and traditional factor proportions theory to
explain such stylized facts as intraindustry trade, trade between countries with
similar factor endowments, and a positive correlation between country size and
volume of trade.
Their new theory suggested a more active role for government policy in trade.
In some cases it appeared to provide an intellectually respectable economic
argument for such not-so-respectable policies as protection. Those die-hard de-
velopment economists who were loath to give up their intellectual investment
in an inward-oriented development strategy-despite mounting evidence of its
failure-latched on to the new theory in hopes of salvaging their intellectual
investment.
Professor Helpman, in the manner that we have all come to expect from him,
has given us a balanced, masterful survey of this literature without making any
exaggerated claims for it. Since he is unlikely to make an analytical error, there
is nothing that I can criticize about the internal logic of his models. I propose
instead to place the inward-oriented-based theory in the perspective of traditional
trade theory; to emphasize the extreme fragility of its conclusions and their lack
of robustness; to point out the inherent conceptual and econometric problems
that arise when you try giving the theory empirical content, or use calibration-
cum-simulation methods to evaluate empirically alternative strategic policy in-
terventions; and to argue that the fragile policy conclusions are likely to be of
limited relevance to developing countries.
To begin with, almost all the problems analyzed by new trade theory have
their counterparts in traditional theory: increasing returns, monopoly power,
intraindustry trade, multiple equilibriums, and even the possibility that countries
with access to identical technology, identical factor endowments, and identical
tastes may trade in equilibrium. Traditional theory analyzes increasing returns

T. N. Srinivasan is a professor of economics at Yale University.

C) 1990 The International Bank for Reconstruction and Development / THE WORLD BANK.

217
218 Comment

in a competitive equilibrium among atomistic firms and consumers by postu-

lating that scale economies are external to a firm but internal to an industry.
Similarly, market power exists only at the national level. Because firms do not
perceive scale economies or national market power, they cannot reflect them in
their profit maximizing decisions. A policy intervention in the form of an ap-
propriate subsidy, tax, or tariff ensures that external scale economies or markets
are reflected in the profit maximizing calculus internal to the firm. Put another
way, policy in effect makes firms do what they cannot do by themselves. Tra-
ditional theorists also recognize that one government's intervention may invite
retaliation by other governments. Three decades ago Harry Johnson explicitly
modeled a Nash tariff policy equilibrium in a two-government world. Subsequent
writers have extended his analysis to quotas.
The new theory postulates increasing returns at the firm level and hence has
to consider noncompetitive market structures and equilibria. In the small group
case, oligopolistic firms that perceive their market power will behave strategi-
cally. Nevertheless they may not be able to precommit themselves credibly to
actions that will improve their profits given other firms' reactions. Government
policy intervention in such cases achieves what the firms cannot credibly do by
themselves. Once again reaction to one government's policy intervention by other
governments has to be allowed for. The new theory explicitly analyzes the issues
of time consistency and credibility, issues that are either irrelevant or are not
raised in traditional theory. But that under these circumstances free trade is not
optimal, and that suitable intervention can improve welfare over free trade, is
not a particularly novel conclusion nor unique to the new theory.
The new theory explains intraindustry trade in terms of product differentiation
and increasing scale economies. The traditional explanation for intraindustry
trade is to trace it to one or more of three problems: aggregation over com-
modities, over space, or over time. Obviously, if the level of aggregation of
commodities in defining an industry is high, even though two countries exchange
commodities that are different at a disaggregated level, the exchange will show
up as intraindustry trade in the aggregate statistics. A large country may export
and import the same commodity because it is cheaper to export its production
at one end of the country rather than transport and sell it to the other end where
it may be cheaper to import it from abroad. Similarly, within a year a country
may export a commodity in one season and import it in another. I am not
entirely persuaded that, after careful disaggregation, much of what is shown in
statistics as intraindustry trade will be left to explain.
The possibility of multiple equilibria in models involving nonconvexities is
well-known. Let me take just two textbook examples from traditional theory.
Given scale economies or externalities at an industry level, the production pos-
sibility curve of the economy may include convex and concave stretches, and
multiple competitive equilibria are possible. What is more, under such circum-
stances it is easy to depict an equilibrium with nonzero trade between two
identical countries because each is producing at a different production point
 Srinivasan 219

even though they face the same prices. In a dynamic context, more than thirty
years ago Solow showed in his famous growth model paper that if the production
function is not concave, multiple steady state equilibria are possible given the
same saving rate-some of which are stable, and to which the economy will
converge, given the appropriate initial condition. Even an argument based on
the Solow model for "big push" in savings and investment can be found in the
development literature. Although hysteresis and multiple equilibria in the new
theory are based on different dynamics, qualitatively the results are similar.
Let me turn now to the policy implications of the new theory. Professor
Helpman points out that one of the major implications of the new theory is the
need for a case-by-case approach. This is somewhat understated. In fact the
policy conclusions are extremely fragile and unrobust to changes in the oligo-
polistic model: not only the levels at which policy instruments are set but even
their signs can be changed by changes in modeling. Take the partial equilibrium,
duopoly model in which the home and foreign firms have no domestic sales but
compete only in the third market. Under the same set of assumptions about
demands, costs, and foreign government nonintervention, if each firm has con-
sistent conjectures about the other's response to its change in sales, optimal
policy is laissez-faire. If the conjectures are Cournot, the optimal policy is an
output subsidy; if the conjectures are Bertrand, the optimal policy is a tax. This
lack of robustness points to the need in devising policy for empirical work on
the structure of competition in the industry.
Before I turn to the empirical issues, let me point out that this literature rarely
takes on board an important insight of the theory of distortions and the second
best: that a policy that is optimal in the presence of a single distortion-say, a
noncompetitive structure in the market for one good-need not be optimal when
several other distortions are present. It goes without saying that no economy in
the world, developed or developing, is characterized by a single distortion. In-
deed, traditional theory has analyzed trade policy in the context of several
distortions, especially factor market distortions.
A number of conceptual and econometric issues arise in attempting empirical
work in this area. Conceptually, many of the game-theoretic formulations of
strategic policy choice assume common knowledge among participants about
payoffs, strategic space, and the like. That is to say, each participant knows the
others' payoffs, the others know his payoffs, he knows that the others know his
payoffs, others know that he knows their payoffs, and so on ad infinitum. In
my view, common knowledge is what is called in Sanskrit Swayam Bbava or
self-being-it comes to existence by itself. There is no process by which common
knowledge can come about. It has to be viewed as an untested and untestable
hypothesis.
Leaving this thorny issue aside, econometrically speaking the Lucas critique
applies with great force in this context: if, before formulating policy, the nec-
essary parameters (other than the so-called deep ones relating to technology and
taste) are estimated, those parameters are virtually useless for policy change
220 Comment

because they are by definition dependent on existing policy. Even if the Lucas
critique is ignored, one has to recognize that the relevant market structure has
to be part of the model specification if the estimated parameters are to be used.
It is not easy to infer the strategies and responses that oligopolistic firms are
using for collecting data on market outcomes.
Another empirical problem is that what constitutes an industry for the purpose
of analysis is debatable, since the elasticities of substitution both for supply and
demand can be substantial between products produced by different industries
under any classification. What does fee entry-or for that matter, absence of
entry-mean in such a context? Once again this problem was recognized long
ago. Soon after Chamberlin published his Monopolistic Competition, Triffin
(1940) examined it in a general equilibrium context. Those who avoid full-
blown econometric estimates but use the so-called literature-based estimates of
some parameters, while choosing others to calibrate their model to reproduce
a given data set, also encounter serious problems, because there are many ways
to choose the parameters for which literature estimates are to be used and those
that are to be calibrated. And this choice can drastically alter the policy rec-
ommendations, as Kala Krishna recently showed by reexamining an earlier cal-
ibration exercise of Avinash Dixit (Krishna, Swagel, and Hogan, forthcoming).
As for the relevance of the new theory ro development, Professor Helpman
rightly noted that in neoclassical growth theory of the 1960s, the steady state
growth rate of the economy equals the natural rate of growth of labor force
and the rate of labor augmenting technical progress. But both these rates were
assumed to be exogenous, not because there was any compelling empirical evi-
dence to support the assumption, but because the economic determinants of
fertility and technical progress, theoretical and empirical, were not well estab-
lished. Considerable progress has been made since then, particularly about
changes in fertility. It is fair to say that the determinants of technical progress
are still unsettled. Despite considerable progress in formalizing Schumpeter's
theory of innovation and market structure and testing it empirically with data
from developed countries, no strong support has emerged either for or against
the Schumpeterian hypothesis.
Long ago, in the framework of a neoclassical optimal growth model of an
open economy, Bardhan (1970) introduced learning by doing a la Arrow. Be-
cause, except for this externality, Bardhan's model was of a small, open, com-
petitive economy, strategic policy intervention as such did not arise. In this
respect the new theory is richer.
But how important learning or, for that matter, scale economies are relative
to global market demand is an empirical issue. For example, Benhabib and
Jovanovic (1989) reexamined the aggregate data for the United States for the
postwar period on growth of output, labor, and capital and found that the data
were consistent with the absence of externalities and increasing returns to scale.
Anecdotal evidence suggests that learning effects are important, but rigorous
econometric studies showing economically significant learning effects are almost
 Srinivasan 221

nonexistent. The Indian passenger automobile industry began assembling cars

more than four decades ago, for example. One of the firms that began manu-
facturing cars (and still produces them) was established at roughly the same time
as Toyota. As recently as ten years ago India was producing more passenger
cars than the Republic of Korea. Yet until the entry of Suzuki was allowed
recently, the industry was stagnant. There is not much evidence of learning
associated with cumulative output in this industry in India! Korea, however, has
achieved a significant toehold in the U.S. auto market. It seems that an industry
established and nurtured by policy, heavily protected from competition from
imports and from entry by other domestic firms, is not likely to generate much
learning.
It is dangerously simplistic, because markets in many developing countries
appear to be oligopolistic, to decide that the policy conclusions of the new theory
apply. First, many of these oligopolies are creations of inappropriate public
policy and not the result of increasing returns or externalities in production.
Second, the capacity of governments to gather information, arrive at appropriate
policies, and implement them-without at the same time unleashing resource-
wasting rent seeking-is extremely limited. And the character of appropriate
policies can change drastically depending on whether rent seeking is significant
or not. Social welfare may be higher without government intervention than with
possibly inappropriate intervention. It is therefore essential to take explicit ac-
count of a country's policymaking and rent-seeking realities. The new theory
has yet to take this step, and until it does the traditional arguments for limited
or no intervention in trade will remain unchallenged.

REFERENCES

Bardhan, Pranab K. 1970. Economic Growth, Development and Foreign Trade: A Study
in Pure Theory. New York: Wiley-Interscience.
Benhabib, Jess, and Boyan Jovanovic. 1989. "Growth Accounting and Externalities."
Working Paper 80-10. New York: C. V. Starr Center for Applied Economics, New
York University.
Krishna, Kala, Phillip Swagel, and Kathleen Hogan. 1989. The Non-Optimality of Op-
timal Trade Policies: The U.S. Automobile Industry Revisited. Working Paper. Cam-
bridge, Mass.: National Bureau of Economic Research.
Triffin, Robert. 1940. Monopolistic Competition and General Equilibrium Theory. Cam-
bridge, Mass.: Harvard University Press.
PRO C E ED ING S OF THE W OR L D B AN K ANN U AL CON FE RE NC E
ON DEVELOPMENT ECONOMICS 1989

COMMENTON "THENONCOMPETITIVE TRADEAND

THEORYOF INTERNATIONAL
TRADEPOLICY,"BYHELPMAN

Nancy Barry

When Dennis de Tray asked me to participate in this session I had to ask

myself why. I've come to the conclusion that it is because of my well-known
capabilities as a troublemaker and perhaps to provide some comic relief but
certainly not because of my known competence as a theoretician of either the
old or the new trade policy. Perhaps it is also because I am one of a number of
people in this room that have been mucking about with these issues over the
last fifteen years in at least fifteen countries: issues of industry, trade, and finance.
So perhaps I can make some remarks about the relevance of the new trade
theory. So as not to disappoint Dennis, these comments will be cantankerous
and heretical and do not represent his views or those of the World Bank.
I have reviewed the new trade theory, including the paper that is being dis-
cussed today, and I find overall that the new trade theory seems to be a timid
departure from the old trade theory. Most of it appears to be tinkering with the
two-by-two-by-two model, usually one assumption at a time. As such, I do not
find the new trade theory to be particularly better at simulating realities, pre-
dicting performance, or prescribing policy.
I also find the new trade theory literature somewhat disappointing. It seems
to be a set of tremendous talents talking to each other-rather than getting closer
to the realities that are being faced in both developed and developing countries.
I would like to talk about a few of those realities because I think the purpose
of this seminar is to deal efficiently with growth and development in the Bank's
member countries.
I think it is clear that OECD countries are rapidly liberalizing everything but
trade. It is also clear that most trade is taking place among like countries. And
it is clear that a tremendous amount of game playing is taking place, with strategic
alliances being formed among the main players, who recognize how much needs
to be invested in research and development and how development costs quickly
reap the benefits. The concept of a production function is rapidly disappearing,
given rapid technology change.
The other phenomenon that seems obvious is that the real success stories of
the 1980s are Japan, the Republic of Korea, and Taiwan. We may try to rewrite
Nancy Barry is chief of the Industry Development Division in the Industry and Energy Department of
the World Bank.

© 1990 The International Bank for Reconstruction and Development / THE WORLD BANK.

223
224 Comment

that history, but it is clear these economies have several things in common. And
one of them is that they have been factor endowment makers-not factor en-
dowment takers. They have really sat down and said, "Where is the world going
to be in the year 2000, where do we want to be in that world, and what are
the pieces that we don't have that we need to get?" And that type of thinking
has been in the context of using international competitiveness, present and future,
as the litmus test. The strategies pursued place the East Asians apart from other
developing economies-not in the level of government intervention but in the
purposes of that intervention.
Now, looking at the realities of our member countries, the developing coun-
tries-be they the poorest of the poor or the next-in-line newly industrializing
countries-it seems that most countries have got themselves tied in terrible
knots-which means that a free trade model is not terribly informative, and
that most of these countries are not in a position to do strategic game playing.
In fact, it could be argued that most countries that the Bank works in have
developed comparative advantage in backing the losers. So the concept of trying
to pick winners-where you have a convergence of vested interests in the public
sector, business as usual in the private sector, and insulation from the dynamic
processes-has very real risks.
What seems to matter in these settings? Clearly, competition matters, and
progress on that front-trade liberalization, deregulation, or unraveling of all
the subsectoral incentives that create rigidities and inability to respond to
change-is urgently needed.
It is also clear that building capabilities matters. I have a hard time believing
on a commonsense basis that we can really say that learning does not matter.
What the whole East Asian experience has been about is learning, entering global
markets at the low end, moving up, and taking advantages of the learning which
comes with playing in the global market.
It is also clear that dynamic processes matter and that within those processes,
organizations and people-the created endowments-increasingly explain who
is competitive. In our research work we need to give a lot more attention to
those functions.
In my view, therefore, what we should not do over the next ten years, in
research and policy work, is try to figure out to what extent the new trade
theory or the old trade theory applies to developing countries. We should not
try to squeeze data and realities into theoretical models. We need to focus on
how to unravel all of the ties that are binding our client countries; on how to
combine these incentives for competition and competitiveness with the building
of endowments; and on how to deal with the transitional issues-that it is less
important whether you think that free trade or strategic trade is the ideal than
how you get out of the mess you are in now. And we need to focus on how
countries and companies get on the bandwagon of one thing leads to an-
other-human resource development, catalysts, direct foreign investment-how
 Barry 225

countries can be in as opposed to out of what is admittedly a very unfair global

game.
I agree that the last portion of Professor Helpman's discussion is potentially
the most interesting because what we are really dealing with in all of our countries
is market takers, technology takers, but potential endowment makers. I think
that the Bank needs to understand and deal in the "missing middle"-not in
firm size but in thinking about innovation systems, of closing the gap even
between best and worst practice within a country.
So I am advocating that in the next five years we deal much more with the
how tos-how to promote effective competition policies, capability buildup
programs, and commercial links in countries at different levels of development.
I
PRO C E ED ING S OF T HE W OR L D B AN K ANN U AL CON FE RE NC E
ON DEVELOPMENT ECONOMICS 1989

FLOOR DISCUSSION OF HELPMAN PAPER

Observing that it would be as unfair to take credit for all the developments in
strategic trade theory in the last ten years as it would be to take the blame for
them, Helpman began the discussion by responding to the panelists. He could
understand Barry's (discussant) feeling that the literature was doing too little
too late, and he would be the first to admit that the answers they had come up
with were unsatisfactory, but, he argued, they were working in the directions
she indicated. Strategic trade theorists were focusing on the very elements she
had highlighted-acquired comparative advantage, dynamic competition, learn-
ing by doing, and so on.
After concurring with Srinivasan's (discussant) point about the fragility of the
policy conclusions-indeed, this was a central point in Krugman's and his recent
book on trade policy-Helpman observed that the fragile empirical content was
not unique to the new trade theories but applied to economics in general. Con-
clusions depend on assumptions about conduct, and economists do not know
how to pinpoint precisely behavior in a particular industry. For example, with
calibration models, using the data and parameters we have, we assume one type
of conduct and adjust the remaining parameters to fit the data. If we change
our assumptions about conduct, we get a different remaining set of parameters,
and we calculate different answers to our policy experiments. Does that mean
economists should abandon their work and go home? Presumably not. It means
more work is needed. Srinivasan countered that to apply the conclusions of
strategic trade policy one needed far more information than one needed to apply
traditional trade theory.
Responding to Srinivasan's comment about the need for common knowledge,
Helpman responded that it really does not matter much. He agreed that going
through the process Srinivasan described in his panel comments, one would
discover that common knowledge is a stringent requirement, but he would argue
that without assuming common knowledge the analytical problems would be
even more severe.
Nor did Helpman consider the problem of externalities and spillovers as cited
by Srinivasan unique to trade policy. It occurs also in public finance, urban
economics, rural development, and so on. Data in recent studies, he felt, point
to spillovers and externalities explicitly related to technology. Was this enough
for the new trade theories? Probably not. But both theoretical and empirical

This session was chaired by Heba Handoussa, professor of economics, American University in Cairo.
© 1990 The International Bank for Reconstruction and Development / THE WORLD BANK.

227
228 Floor Discussion of Helpman Paper

work were lacking, and it was too early to rule all of this out of hand.
Finally, Helpman disagreed that there was nothing new in the new trade
theories. However, he also did not argue that these developments broke entirely
with tradition; he had worked hard to show continuity.
Referring to the debate between the new and old theories as a Jekyll and Hyde
phenomenon, a participant said Helpman was the Dr. Jekyll in this case because
his paper was so logical and precise. This was not true of everyone who tried
to persuade the U.S. Congress that finally we have invented a new argument for
protection and export promotion. He argued that this was not the first argument
for protection-it was simply an important extension of older arguments. In
confining his attention to noncompetitive product markets, Helpman had ig-
nored the huge literature (from development economics) on factor market im-
perfections, sector-specific minimum wages, sticky wages, generalized sticky
wages, and wage differentials-all of which apply to imperfect competition. He
felt that by concentrating only on noncompetitive product markets Helpman
understated the case for appropriate intervention. Helpman agreed that there
were many other arguments for protection-but explained that he was asked
to discuss only this particular line of research.
The participant also felt Helpman had neglected the institutional side of the
problem. In presenting the case for intervention, economists should also consider
the solid empirical and theoretical work done on commercial policy, rent-seeking,
and tariff formation. Recommendations for intervention, in other words, should
be qualified by indicating not only what they might capture, but also what
possible wrong outcomes might ensue. He particularly urged the World Bank
to view the theory in its entirety, not just compare the old and new.
Helpman was asked if he had general recommendations. On balance, did he
feel there should be a free trade orientation, with some exceptions made ac-
cording to simple decision rules? Or should we introduce a blanket policy of
import substitution and let people argue for free trade on a case-by-case basis?
The participant emphasized the importance of investigating what happens when
you actually use strategic trade policy.
Helpman responded unequivocally to the question of policy recommendation.
Krugman and he, in the last chapter of their recent book, conclude that given
current knowledge about strategic trade policy-or general trade policy built
on market imperfections-the best bet is not to intervene, for two reasons. First,
the policy conclusions of the new theories are still fragile. Second, empirical or
semlempirical studies that evaluate the consequences of optimal strategic trade
policy have come up with small potential gains from those policies in the frame-
work of calibration models using existing parametric estimates. In other words,
one stood to gain little, but if one made a serious mistake, one could lose a lot.
Pursuing the question further, a participant asked Helpman to be as specific
as he could about the sorts of policy conclusions he thought this literature might
yield when all the empirical work he called for was done, ten or fifteen years
from now. For example, one sort of policy conclusion might be that government
 Floor Discussion of Helpman Paper 229

should have a discrete, ever-changing policy for each significant industry (as had
already happened with civil aviation). A second policy conclusion might be that
one needs a general sort of policy rule but that the way it operates should change.
Citing antidumping measures as a policy system designed largely to deal with
strategic trade concerns, he asked how Helpman might design a strategic policy
in the future to deal with the perceived problem of dumping. To this specific
query about antidumping measures, Helpman responded that in his view the
current models that support such measures are far more fragile than the others,
because all the arguments rely on market segmentation supposedly brought about
by the behavior of firms, and Helpman did not believe this was a reasonable
description.
Another participant asked what process Helpman could imagine that would
give both the rule by which you could decide on an intervention and at the same
time insulate it from the political process. Many countries are persuaded to
lower tariffs, but when they try, those who are adversely affected by the tariff
protection find five thousand good reasons not to lower it.
Concurring with most of the views expressed by previous participants, Sri-
nivasan felt that it was important to set up institutions and transparent rules
that were as politically unmanipulable as possible, rather than deal with trade
policy on a case-by-case, discretionary basis. For example, he considered
strengthening multilateral trade arrangements through the Uruguay Round ne-
gotiations more important than any gains from unilateral actions of the Super
301 type introduced by the recent U.S. trade legislation, for which the new trade
theory seemed to be providing a rationale. Generally, Srinivasan found that this
literature assumed the oligopolistic or noncompetitive market structure as given,
and then looked to see how a nation could benefit from the structure-rather
than question whether the international structure itself could be changed. He
regretted the tendency in U.S. trade policy to relax antitrust laws, permit mergers
and acquisitions, and promote what earlier would have been considered
noncompetitive behavior-in order to give U.S. firms competitive advantage in
the rest of the world. It seemed to Srinivasan that one should, on the contrary,
extend the antitrust laws globally rather than relax them domestically to gain
perceived advantages in international competition.
A participant said he would like to raise a very specific question about the
allocation of quotas. In his discussion of tariffs and quotas in the context of
the new theories, Helpman had seemed to draw conclusions regardless of how
quotas were allocated. It seemed to the participant that the method of allocating
quotas would affect the degree of competition. The Japanese, for example, assign
their beef quotas to the producers' cooperatives, which have monopoly power
in Japan.
Another participant pleaded with trade theorists and empirical economists to
use less technical jargon and make their results more transparent, so that policy
and country analysts could have an effective policy dialogue with developing
countries on the new developments in trade policy. He then asked the panel if,
230 Floor Discussion of Helpman Paper

in line with what Singh had said in his keynote address, new capital-intensive
and skill-intensive technologies and the extensive use of nontariff barriers and
strategic alliances had rendered obsolete the operational notion of comparative
advantage, even for such traditionally labor-intensive products as textiles. How
should we apply the notion of "new comparative advantage"? Can we replace
the "old" notion of comparative advantage with something else, or should
economists simply fly by the seat of their pants in discussing trade policy?
Barry responded that technological change, particularly in electronics, is mak-
ing traditional concepts of comparative advantage increasingly obsolete. In ma-
ture industries (automotive, textiles and garments) and in the "high-tech" in-
dustries, it is changing the entire definition of a product and dramatically
increasing the possible combinations of factor use and the importance of the
transformation of process technology itself. Scale economies no longer matter
so much on the shop floor. Technological change requires a different kind of
organization within the company, of everything from sourcing to distribution.
The whole system is a product, and the issues of technology accumulation and
learning make it ill-suited to modeling.
Helpman agreed that economists must look beyond traditional comparative
advantage. The new trade policy literature, he felt, does exactly that. He con-
curred with a point made earlier about rent-seeking behavior, which he thought
was even more important for Schumpeterian dynamic competition and strategic
trade.
In response to the broad question about the overall policy message of strategic
trade theories raised by several participants, Helpman said that the more forceful
global competition is, the less room it leaves for single-economy trade policy.
On average, this works in everyone's interest, because it typically reduces market
power in every country-and therefore reduces the price-cost margin globally.
This is good from an international point of view, so economists should try not
to segment economies and isolate them from international markets. At this stage
it seems reasonable to recommend a global approach, with global competition
and little intervention. In any international system, however, a prisoner's di-
lemma situation could arise-where it is good for everybody to adopt one policy,
but if everybody does, it pays one country to deviate. Therefore, it may well be
that a good case for intervention could be made in some isolated industries-
in which research and development are important and there are externalities
within the industry, or across industries, or over time. The intervention need
not be in trade policy, however. With research and development (R&D) com-
petition, for example, the best policy might be to subsidize R&D or provide
venture capital.