Professional Documents
Culture Documents
To cite this article: Professor Aditya Bhattacharjea (2002) Infant Industry Protection Revisited,
International Economic Journal, 16:3, 115-133, DOI: 10.1080/10168730200000024
Download by: [Indian Institute of Management - Indore] Date: 11 December 2017, At: 13:28
INTERNATIONAL ECONOMIC JOURNAL
Volume 16, Number 3, Autumn 2002
ADITYABHATTACHARJEA*
Delhi School of Economics
1. INTRODUCTION
imposes an anti-dumping (AD) duty equal to the difference between the prices that
the firm charges in the home and foreign markets. Crucially, this duty is assumed
to be both credibly committed4 and also sufficient to ensure the viability of the
home firm. More recently, Herander and Kamp (1999) have deployed the
Milgrom-Roberts model in a setting that is the mirror image of the one I have in
mind: they model a domestic incumbent who attempts to deter a foreign entrant by
signalling low costs in the first period. They examine the effects of different
levels of a tariff that remains committed for both periods. In neither of these last
two papers is there any claim that the tariff policy is optimal for the market
structure of that period. In the model to be developed below, the optimality of
the tariff gives it credibility, and I demonstrate (rather than assume) that it is
adequate to guarantee the profitability of the entrant in a situation that would
otherwise be conducive to entry-deterrence by the incumbent.
In other related work, Kolev and Prusa (1999a) have an optimizing government
confronting a foreign monopolist with private cost information, but no threat of
entry. Here the monopolist raises its first-period price to signal higher costs and
elicit a lower tariff. Kolev and Prusa (1999b) introduce a domestic rival, and
interpret the foreign firm's price-enhancing (output-restricting) signal as a
voluntary export restraint aimed at moderating the home government's optimal
tariff, which they interpret as an anti-dumping duty. The home firm learns the
foreign cost independently, so the signal is again directed exclusively at the home
government. In contrast, I investigate the strategic use of private cost
information when both the tariff and the home firm's entry decision are
conditioned on the foreign cost. This introduces a tension into the foreign firm's
problem: it needs to signal higher costs to the government in order to secure a
lower tariff, but lower costs to the home firm in order to deter its entry. Section 4
resolves this conflict, as well as the theoretical problems with Hamilton's
argument.
In order to derive the optimal tariff in the post-entry situation, I begin with the
standard international duopoly (SID) model which has been widely used in the i
4 ~ assumes,
e not unreasonably, that AD procedures make the threat of the duty credible.
However, an AD action depends on demonstrable injury to a domestic firm,which is not yet in
existence in the entry-deterrence story of our model.
INFANT INDUSTRY PROTECTION REVISITED 119
0= U(x, y) + z, tvhere
I
Here, x denotes the consumption of the home-produced variety and y the foreign
variety of the good produced under imperfect competition. z is the consumption
of a numeraire good produced and sold competitively at an exogenous world price.
All parameters are strictly positive, with k representing the degree of
substitutability between home and foreign varieties. Assume
and
On the supply side, the two firms have constant per-unit costs c, (< a,) and c,
(< u,), the latter defined to include transport costs, over the relevant range of.
outputs. The foreign firm's cost level is private information, but the distribution
120 ADITYA BHATTACHARJEA
Assume to begin with that the two firms know each other's cost level, but the
government does not know the foreign cost.6 With firms competing in quantities,
the equilibrium quantities for a given tariff level can be easily obtained as:
where H = 4b,b, - k 2 is strictly positive, from (2). From (5) and (6), social
welfare can be written
so that
'~amiltonhad actually suggested using the tariff revenue to subsidize domestic production,
offsetting the rise in domestic prices that would otherwise hurt consumers. Dixit (1988)
formally established that a tariff combined with a production subsidy to the domestic firm
maximizes home welfare. However, this becomes quite intractable with unknown costs. Also,
for reasons spelt out in more detail in Bhattacharjea (!995), subsidies are impractical in this
context.
6 ~ h iassumption
s is partially relaxed in section 4, where the home firm is also uninformed of
the foreign cost before entry.
INFANT INDUSTRY PROTECTION REVISITED 121
dW ( p , -ex)-+t-+y
--
dt "
dt dt " 21 I--
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
Five of the seven parameters of the model have dropped out in this formula. This
yields
PROPOSITION 1. The optimal tar@ formula in the SID model with unknown
costs depends only on the expectation of foreign costs, and is independent of both
domestic costs and the degree of substitutability between home and foreign
varieties.
PROPOSITION 2: The optimal tar@ in the SID model with asymmetric cost
information is the same as for a foreign monopolist with unknown costs.
These propositions are specific to the SID model, and do not survive the
introduction of non-linear demand and costs, or more firms, or conjectures other
122 ADITYA BHATTACHARJEA
than ourn not.' Although they do not therefore solve the information problems
inherent in the strategic trade literature, they do achieve what physicists might call
"weak unification" of a commonly-used subset of models in that literature. The
next two sections assess the social desirability of domestic entry by making
welfare comparisons between foreign monopoly and duopoly, utilizing this
invariance of the optimal tariff formula.
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
7 ~ h tariff
e formula (12) and Propositions 1 and 2 extend results derived in Bhattacharjea
(1995) to the case where foreign costs are unknown to the home government.
INFANT INDUSTRY PROTECTION REVISITED 123
I
I
which is strictly positive, by (2).
(iii) It can also be shown that the function Q (c,) has two roots, the smaller of
which is c, = c , ~ , ~ .The latter can be seen directly by noting that as c, tends to
C~mzn
which gives the comer solution, the duopoly outcome tends to the foreign
monopoly outcome, and domestic profits disappear. Consequently, from (13), R
tends to zero. If the larger of the two roots is less than q, then R < 0 for larger
values of c,. Although the implied restriction on the parameters required for this
possibility is too complicated to have any easy interpretation, it can be shown that
R < 0 for values of c, close to Cy with E(c,) close to c , ~ , ~ .
From these three properties we can conclude that the graph of R (c,) is an
inverted U, and if at all R is negative, it can only be at values of c, > E(c,). This
gives us
That is, contrary to what one might expect, domestic entry reduces welfare
only if the foreign firm is relatively ineficient. Although entry is always more
124 ADITYA BHATTACHARJEA
profitable when the realized foreign cost is higher, (15) shows that it necessarily
results in a welfare gain if the tariff is based on actual costs. Therefore, the
possibility of a welfare loss arises only from a tariff based on expected costs.
The reason is that if the foreign firm has above-average costs, then a tariff set on
the basis of (12) is higher than optimal, attenuating the welfare gain even as it
promotes entry.
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
The framework employed thus far has not allowed the credibility issue posed
in the Introduction to be incorporated into the analysis, nor has the foreign firm
been allowed to use its private cost information strategically. Now consider the
possibility that the domestic firm is also uninformed of the cost level of the
foreign firm, which can then attempt to deter domestic entry by signaling lower
costs via limit pricing, as in Milgrom and Roberts (1982). The additional
complication in the present setting is that lower costs attract a higher tariff, from
(12). In the two Kolev and Pmsa papers (1999a, 1999b) which investigate this
without the possibility of domestic entry, the foreign firm has an incentive to
signal higher costs to elicit a lower tariff. The Milgrom-Roberts and Kolev-
Prusa stories individually make for complicated models that ultimately rely on
numerical simulations to demonstrate the existence and welfare properties of
signaling equilibria, even with linear demands and constant costs. However,
combining the two stories actually turns out to be relatively simple. I show that
the necessary conditions for a limit-pricing equilibrium become mutually
incompatible when the home government revises its tariff on the basis of the cost
information that the incumbent needs to signal to deter entry. Consequently, no
such equilibrium exists, and one need not enter into a discussion of the out-of-
equilibrium beliefs that are needed to support the signaling equilibria, and the
refinements that are needed to eliminate multiplicity. This complicated and
controversial terrain is bypassed by the result to be derived below.'
First, substituting the optimal tariff formula (12) into (7) and (8) and taking
expectations gives expressions for expected quantities as functions of expected
foreign cost. We can then obtain an expression for domestic profits as a function
of E(c,), and using the linearity properties again,
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
since the expression in square brackets is positive for an interior solution. Thus,
it seems that the foreign firm still has an incentive to signal lower costs to deter
entry by the domestic firm, even though lower costs invite a higher tariff. This is
because from (12), adjustment of the optimal tariff offsets only one-third of the
perceived reduction of the foreign cost level.9 However, such signaling requires
setting a price that is lower than the one which maximizes static monopoly profit,
and this sacrifice must be weighed against the gain from deterring entry.
In order to investigate this, consider now a two-period setting, with the foreign
firm's costs initially known only to itself. The expectation of the distribution
from which it is drawn, as well as domestic costs and all demand parameters,
remain common knowledge. In period 1, the foreign firm is a monopolist, and
chooses a quantity to export to the home market, where some non-prohibitive
tariff tl is in effect. Observing this choice of quantity, the home firm and
government update their assessments of the foreign cost level. In period 2, the
home firm decides on whether to enter, and then the government imposes a revised
tariff. If entry occurs, the firms compete as a Cournot duopoly in period 2,
otherwise the foreign firm remains a monopolist. In each period, the tariff is
committed before the output decision of the firm(s).1 Assume, as in Milgrom
and Roberts (1982), that the incumbent's costs are revealed if entry takes place.
This means that any duopoly interaction takes place under symmetric information.
For the present, I work with a stripped-down homogeneous product version of
the model of the earlier sections, so that henceforth a, = a, = a, and b, = b, = b = k.
(It will be shown that the main result carries over to the case of differentiated
products.) Suppose foreign costs are known to be drawn from a distribution that
simplified Milgrom-Roberts model similar to the one in this paper. Even with adaptive
learning, convergence to the predicted equilibrium occurs very slowly, if at all.
' ~ i x i t(1988) obtained this result as a special case of his demonstration that foreign export
subsidies should be only partially offset by a domestic countervailing duty. He did not
however emphasize that an optimal tariff like (12) must already be in place for the partial
countervailing result to hold.
'1n the terminology of Neary and Leahy (2000), the government is assumed to be able to
commit intra-temporally (within a period), but not inter-temporally.
126 ADITYA BHATTACHARJEA
yields a cost level that is high (cH) with probability h and low (cL)with probability
1-h. Normalize cL = 0. Also, let the two realizations of foreign costs straddle
the (known) domestic cost level. As before, relative cost levels are constrained
to ensure interior solutions to the duopoly in all states of nature, should entry occur.
Under free trade, the home firm's duopoly operating profits are assumed to cover
its entry costs if the foreign firm is high cost, but not if it is low cost. Thus,
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
where the superscript refers to the foreign cost type and the subscript x to the
domestic firm.
I shall not explain the various incentive-compatibility constraints that yield the
equilibria of the Milgrom-Roberts model with two types of incumbent, since they
are well-known from Tirole (1988); see also Herander and Kamp (1999). I focus
initially on a separating equilibrium, in which the foreign firm's behavior in period
1 enables the home firm and government to infer its type, so that h is updated to 0
or 1 for period 2. No-one is fooled, and the second-period duopoly outcome is
then described by the homogeneous-product versions of (7) and (8), with type-
contingent full information tariffs tH or k given by substituting cHor cLfor E(c,) in
(12). The home firm will not enter if the resulting profits do not cover its entry
costs.
Consider two of the necessary conditions for a separating equilibrium with the
low-cost type signaling its cost level by charging a price lower than its static
profit-maximizing price, modifying them to take into account the government's
tariff response. First (compare eqn. 9.11 in Tirole),
where M represents monopoly profits, which always pertain to the foreign firm, 3
its duopoly profits, and superscripts its cost type as before. Period 2 profits are
discounted by a factor of S The left side of this inequality represents the profits
of a type H which deters entry by mimicking the monopoly pricing behavior of a
type L in period 1, even though it attracts a higher tariff tL in period 2. The right
side represents its profits if it were to maximize its first-period monopoly profits,
revealing high costs and thereby inviting a duopoly in the second period. If (20)
is satisfied, then type L must deviate below its first-period profit maximizing price
INFANT INDUSTRY PROTECTION REVISITED 127
Technically, (20) and (21) are both necessary for an entry-deterring perfect
Bayesian equilibrium in which the low-cost type must charge a price below the
price that maximizes its first-period monopoly profit, and all agents' strategies are
optimal given their beliefs, which are consistent in a Bayesian sense with the
actions observed under those strategies. I show in the Appendix that (20) and
(21) are mutually incompatible. The intuition behind this result is as follows.
(20) implies a minimal market size a 2 3cH relative to costs for entry-deterrence
to be worthwhile, or alternatively a maximal spread between the incumbent's two
cost types (recall that cL has been normalized at zero) relative to market size so
that it is not too costly for a type H incumbent to mimic a type L. However, this
requirement also increases the magnitude of the optimal tariff, k = aI3. With a
dynamically consistent tariff policy, it turns out that the condition necessary to
make signaling worthwhile also makes tariff-assisted entry profitable. Signaling
low costs is therefore pointless.
It is important to be clear about what exactly has been proved here. Demand
may of course be too low relative to domestic costs to justify entry even with the
higher tariff, so entry is "blockaded" and limit-pricing is unnecessary. It may on
the other hand be high enough to encourage entry even if the government remains
passive. What has been demonstrated above applies to an intermediate situation.
If market size is large enough to make signaling worthwhile for the incumbent,
then it is also large enough for the entrant to cover its fixed costs, aided by the
higher optimal tariff that signaling elicits. Further, with entry inevitable, and
costs being revealed after entry, the government can calibrate its period 2 tariff
precisely to the foreign cost level. Thus, a low cost incumbent also gains nothing
by signaling high costs in period 1 to attract a lower tariff. Since neither type
gains from pretending to be the other, and signaling involves a first-period
sacrifice of profits, both types can do no better than to produce their myopic
monopoly profit-maximizing quantities in the first period.
Before completing the description of the equilibrium and its welfare properties,
I show that the market size restriction that ruled out an entry-deterring separating
equilibrium also precludes a pooling equilibrium. The latter would require both
incumbent types to sell the same output in period 1, and the uninformed entrant,
unable to update its information, would then make its decision on the basis of the
prior E(c,.). If it enters, the foreign firm reveals its costs, and the government
imposes the appropriate type-contingent tariff. If not, t* is imposed in the second
128 ADITYA BHATTACHARJEA
period. The usual pooling equilibrium that survives the intuitive criterion is one
in which both types pool at the monopoly price of Type L. This requires a
condition resembling (20), but with t* rather than tL in the second term:
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
Since t* < tL,this change necessarily raises the value of MH in the second term of
(20), so if (20) is satisfied, so is (20a). Instead of (21), we now have
However, it has already been shown that if (20) holds, then (21) does not, and
thus (21a) cannot hold either. It cannot hold even if the government does not
learn the entrant's costs after entry, so that (21a) is conditioned on the tariff t*
from (12), based on the unchanged prior E(c,) > cL. This is because by (17),
expected domestic profits incorporating the optimal tariff are rising in E(c,).
Thus, if (20) holds, (21) is violated, and so is (21a), whether the government
updates its prior or not. Since entry cannot be deterred, the incumbent can again
do no better than sell the type-contingent monopoly quantity in the first period.
The candidate pooling equilibrium is thereby destroyed.
The only sequentially rational equilibrium, then, is what is sometimes called a
no-distortion separating equilibrium. The two incumbent types produce their
static profit-maximizing output levels in period 1, from which the government and
home firm correctly infer the true type. In period 1 the uninformed government
can therefore impose t, = t * , the average tariff (12) based on such non-distorting
behavior and the expectation of costs. (The Appendix also shows that condition
(20) is independent of the first-period tariff.) In period 2, with cost information
revealed, the government imposes the type-contingent tariff, and the home firm
enters. From the homogeneous-product version of (15), privately profitable entry
under such a tariff is always welfare-improving when tl is also based on known
costs; it must be so a fortiori if t, = t* is suboptimal because costs are unknown
in period 1. ( W, in (13) is lower, and therefore R is higher.)
These results carry over to the case of differentiated products. By Proposition
1, the anticipated period 2 optimal tariff is the same for differentiated and
homogeneous products. For any given level of tariff and home and foreign costs,
greater differentiation enables the domestic firm to earn higher profits, so if entry
was profitable with homogeneous products, it will be even more profitable with
INFANT INDUSTRY PROTECTION REVISITED 129
is welfare-improving.
1 5. CONCLUSIONS
In the course of exhuming Alexander Hamilton's two century old argument for
protection, this paper has established some intermediate results of independent
interest. It has derived an optimal tariff formula for a foreign monopolist whose
costs are unknown to the home government, and shown that it remains unchanged
in the presence of a domestic Cournot competitor, independent of the degree of
product differentiation and domestic costs. It has also shown that entry of such a
domestic firm under this tariff regime will be socially inefficient only if the
foreign firm is relatively high-cost. Finally, it has restated Hamilton's argument
in modern terms, and shown that anticipation of a dynamically consistent tariff,
based on the government's updating of its information, does indeed "fortify
adventurers against the dread of such combinations ... demonstrating that they
must in the end prove fruitless". It has also shown that domestic entry under
such conditions is always welfare-improving."
Note that Hamilton actually wrote of "combinations", which have been
interpreted here as monopolies. Bagwell and Ramey (1991) showed that
allowing for multiple incumbents (as opposed to multiple types of a single
incumbent) in the Milgrom-Roberts model makes coordinated limit-pricing
impossible, and separation occurs at non-distorted oligopoly prices. However,
Martin (1995) modified this result, showing that coordinated limit-pricing can
occur if the incumbents are allowed to have different costs. (Multiple potential
entrants can obviously be accommodated in this model, since only the most
efficient one need be deterred.)
The highly simplified linear demand and cost structures used in this paper were
inevitable (and even they created algebraic difficulties). However, it is true that
the partial equilibrium setting does not permit questions of income distribution and
intersectoral resource allocation to be raised. It is worth recalling that in the
century after Hamilton's Report, the policy of industrial protection adopted by the
II
What could not be examined, due to the prohibitively complicated algebra, is whether a
policy of government non-intervention could be even better. This possibility arises because
limit-pricing by the foreign firm benefits consumers in the first period, and this might outweigh
the loss of domestic profits and consumer surplus arising from the preservation of foreign
monopoly in the second period.
130 ADITYA BHATTACHARJEA
United States led to recurring political conflicts between different interest groups,
including an outcry against the so-called "Tariff of Abominations" in 1828 by the I
agricultural exporters of the American South. The intention of this paper was
only to set Hamilton's argument on acceptable microfoundations, not on a pedestal.
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
Substituting the optimal full-information tariffs for the second period, tH = (a-
cH)/3 and tL = a n , into the standard profit-maximizing expressions for prices,
output and profits under monopoly and duopoly, (20) becomes
Solving this as an equation, and noting that all the terms are strictly convex in a,
this implies
for the positive square roots. (The quadratic form under the square root sign can be
shown to be positive.) Notice that tl drops out in (A2). For reasons that will
become clear in a moment, subtract 3cHh r n both sides of (A2) and simplifl, to get
We want to show that the right side of this is non-negative. Suppose not. Then 1
Both sides being positive, squaring preserves the inequality, and yields (after some
I simplification)
Downloaded by [Indian Institute of Management - Indore] at 13:28 11 December 2017
The expression on the left is a quadratic form in cH and c, that can be shown (by
calculating eigenvalues) to be positive definite for 6 ~ ( 0 , 1 ) , or positive
semidefinite for 6 = 1, contradicting our supposition. Thus, ultimately (20)
implies
1 Now consider
The first term in the numerator is non-negative, from (A6). As for the second
term, since we have assumed that the home firm is competitive with the high-cost
foreign firm in the post-entry duopoly under free trade, the numerator of the
expression for its C:ournot output (a + cH - 2c,) > 0. Adding six times this to
(A6) gives 7a + 3cH- 12cx>0. Thus, fi-om (A7), nf (tL)> nXH (0). But the
latter is greater than F, from (19). This contradicts (21). QED.
It can be shown that (A6) also suffices to ensure that all the monopoly and
duopoly profit expressions involved in (20) and (21) are positive at the relevant
optimal tariffs.
REFERENCES
Bagwell, Kyle and Ramey, Gareth, "Oligopoly Limit Pricing," Rand Journal of
Economics, Summer 1991, 155-172.
Bhattacharjea, Aditya, "Strategic Tariffs and Endogenous Market Structures:
Trade and Industrial Policies under Imperfect Competition," Journal of
Development Economics, August 1995,287-312.
Brainard, S. Lael and Martimort, David, "Strategic Trade Policy Design with
Asymmetric Information and Public Contracts," Review of Economic Studies,
January 1996, 81-105.
and Martimort, David, "Strategic Trade Policy with Incompletely
132 ADITYA BHATTACHARJEA
Tirole, Jean, The Theory of Industrial Organization, Cambridge: MIT Press, 1988.
Wright, Donald J., "Incentives, Protection and Time Consistency," Canadian
Journal of Economics, November 1995,929-938.