Vous êtes sur la page 1sur 36

14.

581 MIT PhD International Trade


Lecture 25: Trade Policy (Theory)
Dave Donaldson
Spring 2011
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Trade Policy Literature
A Brief Overview
Key questions:
1
Why are countries protectionist? Can protectionism ever be optimal? Can
we explain how trade policies vary across countries, industries, and time?
2
How should trade agreements be designed? Can we explain the main
institutional features of actual trade agreements (e.g. WTO, NAFTA, EU)?
In order to shed light on these questions, one needs to take a stand on the:
1
Economic environment: What is the market structure? Are there distortions,
e.g. unemployment or pollution?
2
Political environment: What is the objective function that governments aim
to maximize, e.g. social welfare, welfare of the median voter, political support?
What are the trade policy instruments, e.g. import taris, quotas, product
standards? Are trade policy instruments the only instruments available?
3
Constraints on the set of feasible contracts: Do trade agreements need to
be self-enforcing? How costly is it "to complete contracts?
This Lecture
We will restrict ourselves to economic environments such that:
1
All markets are perfectly competitive.
2
There are no distortions.
We will be focusing on two questions:
1
How should taris vary across countries and industries?
2
What is the rationale for trade agreements?
Our goal will be to contrast the implications of two assumptions:
1
Governments only care about welfare [the Traditional economic approach.]
2
Governments are politically motivated [the Political Economy apprach,
where we will focus in particular on the work of Grossman and Helpman
(1994).]
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Economic Environment
Consider a world economy with 2 countries, c = 1, 2.
There are two goods, i = 1, 2, both produced under perfect competition.
Good 2 is used as the numeraire, p
w
2
= 1.
Notations:
p
c
= p
c
1
/p
c
2
is the relative price in country c.
p
w
= p
w
1
/p
w
2
is the world (i.e. pre-tax) relative price.
d
c
i
(p
c
, p
w
) is the demand for good i in country c.
y
c
i
(p
c
) is supply of good i in country c
Economic Environment (Cont.)
WLOG, country 1 (resp. 2) is a natural importer of good 1 (resp. 2):
m
1
1
_
p
1
, p
w
_
= d
1
1
(p
1
, p
w
) y
1
1
_
p
1
_
> 0
m
2
2
_
p
2
, p
w
_
= d
2
2
(p
2
, p
w
) y
2
2
_
p
2
_
> 0
x
1
2
_
p
1
, p
w
_
= y
1
2
(p
1
) d
1
2
_
p
1
, p
w
_
> 0
x
2
1
_
p
2
, p
w
_
= y
2
1
(p
2
) d
2
1
_
p
2
, p
w
_
> 0
Trade is balanced:
p
w
m
1
1
_
p
1
, p
w
_
= x
1
2
_
p
1
, p
w
_
m
2
2
_
p
2
, p
w
_
= p
w
x
2
1
_
p
2
, p
w
_
Market clearing for good 2 requires:
x
1
2
_
p
1
, p
w
_
= m
2
2
_
p
2
, p
w
_
(1)
Political Environment
Policy instruments
Both governments can impose an ad-valorem tari t
c
on their imports
p
c
c
= (1 +t
c
) p
w
c
p
c
c
= p
w
c
Taris create a wedge between the world and local prices which implies
p
1
=
_
1 +t
1
_
p
w
(2)
p
2
= p
w
/
_
1 +t
2
_
(3)
Comments:
If the only taxes are import taris, then local prices faced by consumers and
producers are the same, as implicitly assumed in our previous slides.
Equations (1)-(3) implicitly dene p
w
= p
w
_
t
1
, t
2
_
and p
c
= p
c
(t
c
, p
w
)
Political Environment
Governments objective function
Both governments are welfare-maximizers. They simultaneously set t
c
in
order to maximize utility of representative agent:
max
t
c
V
c
(p
c
, I
c
) = V
c
[p
c
, R
c
(p
c
) +T
c
(p
c
, p
w
)] (4)
where:
R
c
(p
c
) = max
y
p
c
1
y
1
+p
c
2
y
2
[y feasible
T
c
(p
c
, p
w
) = t
c
p
w
c
m
c
c
(p
c
, p
w
) =
_ _
p
1
p
w
_
m
1
1
_
p
1
, p
w
_
if c = 1
_
p
w
/p
2
1
_
m
2
2
_
p
2
, p
w
_
if c = 2
p
1
, p
2
, p
w
satisfy Equations (1) (3)
Unilateral Taris
Proposition 1 For both countries, unilateral (Nash) taris satisfy
t
c
=
1

c
, where
c
=
d ln x
c
d ln p
w
Proof:
1
For expositional purposes we focus on country 1. FOC =
_
V
1
p
V
1
I
_
_
dp
1
dt
1
_
+
_
dR
1
dp
1
__
dp
1
dt
1
_
+
_
dp
1
dt
1

p
w
t
1
_
m
1
1
_
p
1
, p
w
_
+t
1
p
w
dm
1
1
_
p
1
, p
w
_
dt
1
= 0
2
Roys identity =
V
1
p
V
1
I
= d
1
1
(p
1
, p
w
)
3
Perfect competition =
dR
1
dp
1
= y
1
1
(p
1
, p
w
)
4
1+2+3 = t
1
=
_
ln p
w
t
1
_
/
_
d ln m
1
1
(p
1
,p
w
)
dt
1
_
5
4 + market clearing, m
1
1
_
p
1
, p
w
_
= x
2
1
_
p
2
, p
w
_
= t
1
= 1/
2
How Should Taris Vary Across Countries (and Industries)?
Proposition 1 oers a simple positive theory of tari formation:
Taris= inverse of the elasticity of foreign export supply.
This is true whether or not the other government is imposing its Nash tari.
Though other governments tari does aect elasticity of foreign export supply.
In the case of a small open economy,
p
w
t
1
= 0 =
2
= +
A small open economy never has an incentive to impose a tari.
According to traditional economic approach, import taris are intimately
related to countries market power.
It is a countrys ability to improve its terms-of-trade that leads to strictly
positive taris.
Potential concerns:
1
Do we really believe that governments maximize welfare?
2
How many countries are large enough to aect their terms-of-trade?
3
Do trade negotiators really care about their terms-of-trade?
Are Unilateral Taris Pareto-Ecient?
Following Bagwell and Staiger (1999), we introduce
W
c
(p
c
, p
w
) = V
c
[p
c
, R
c
(p
c
) +T
c
(p
c
, p
w
)]
Dierentiating the previous expression we obtain
dW
c
=
_
W
c
p
c
_
dp
c
dt
c
_
+W
c
p
w
_
p
w
t
c
__
dt
c
+W
c
p
w
_
p
w
t
c
_
dt
c
The slope of the iso-welfare curves can thus be expressed as
_
dt
1
dt
2
_
dW
1
=0
=
W
1
p
w
_
p
w
t
2
_
W
1
p
1
_
dp
1
dt
1
_
+W
1
p
w
_
p
w
t
1
_
(5)
_
dt
1
dt
2
_
dW
2
=0
=
W
2
p
2
_
dp
2
dt
2
_
+W
2
p
w
_
p
w
t
2
_
W
2
p
w
_
p
w
t
1
_
(6)
Are Unilateral Taris Pareto-Ecient?
Proposition 2 If countries are large, unilateral taris are not
Pareto-ecient.
Proof:
1
By denition, unilateral (Nash) taris satisfy
W
c
p
c
_
dp
c
dt
c
_
+W
c
p
w
_
p
w
t
c
_
= 0,
2
If
_
p
w
t
1
_
and
_
p
w
t
2
_
,= 0, 1+ (5) and (6) =
_
dt
1
dt
2
_
dW
1
=0
= + ,= 0 =
_
dt
1
dt
2
_
dW
2
=0
3
Proposition 2 directly derives from 2 and the fact that Pareto-eciency
requires
_
dt
1
dt
2
_
dW
1
=0
=
_
dt
1
dt
2
_
dW
2
=0
Are Unilateral Taris Pareto-Ecient?
Graphical analysis (Johnson 1953-54)
N corresponds to the unilateral (Nash) taris
E-E corresponds to the contract curve
If countries are too asymmetric, free trade may not be on contract curve
What is the Source of the Ineciency?
The only source of the ineciency is the terms-of-trade externality.
Formally, suppose that governments were to set their taris ignoring their
ability to aect world prices:
W
1
p
1
= W
2
p
2
= 0
Then Equations (5) and (6) immediately imply
_
dt
1
dt
2
_
dW
1
=0
=
_
p
w
t
2
___
p
w
t
1
_
=
_
dt
1
dt
2
_
dW
1
=0
Intuition:
In this case, both countries act like small open economies.
As a result, t
1
= t
2
= 0, which is ecient from a world standpoint.
Question:
How much does this rely on the fact that governments maximize welfare?
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Economic Environment
Endowment economy
We consider a simplied version of Grossman and Helpman (1994).
To abstract from terms-of-trade considerations, we consider a small open
economy.
There are n + 1 goods, i = 0, 1, ..., n, produced under perfect competition:
Good 0 is a freely-traded numeraire with domestic and world price equal to 1.
p
w
i
and p
i
denote the world and domestic price of good i , respectively.
Agents are endowed with one unit of good 0 and one unit of another good
i ,= 0.
We refer to an agent endowed with good i as an i -agent.

i
denote the share of i -agents in the population.
Total number of agents is normalized to 1
Economic Environment (Cont.)
Quasi-linear preferences
All agents have the same quasi-linear preferences:
U = x
0
+

n
i =1
u
i
(x
i
)
Indirect utility function of i -agent is therefore given by:
V
i
(p) = 1 +p
i
+t (p) +s (p)
where:
t (p) = governments transfer [to be specied]
s (p) =

n
i =1
u
i
(d
i
(p
i
))

n
i =1
p
i
(d
i
(p
i
))
Comments:
Original GH (1994) model is a specic-factor model (not endowment model as
here).
Given quasi-linear preferences, this is de facto a partial equilibrium model
Political Environment
Policy instruments
For all goods i = 1, ..., n, the government can impose an ad-valorem import
tari/export subsidy t
i
p
i
= (1 +t
i
) p
w
i
We treat p =(p
i
)
i =1,...,n
as the policy variables of our government.
The associated government revenues are given by:
t (p) =

n
i =1
(p
i
p
w
i
) m
i
(p
i
) =

n
i =1
(p
i
p
w
i
) [d
i
(p
i
)
i
]
Revenues are uniformly distributed to the population so that t (p) is also
equal to the governments transfer, as assumed before.
Political Environment
Lobbies
An exogenous set L of sectors/agents is politically organized.
We refer to a group of agents that is politically organized as a lobby.
Each lobby i chooses a schedule of contribution C
i
() : (R
+
)
n
R
+
in
order to maximize the total welfare of its members net of the contribution:
max
C
i
()

i
V
i
_
p
0
_
C
i
_
p
0
_
subject to: p
0
= arg max
p
G(p)
where G() is the objective function of the government [to be specied], and

i
relates to the size/importance of lobby i .
Political Environment
Government
Conditional on the contribution schedules announced by the lobbies, the
government chooses the vector of domestic prices in order to maximize a
weighted sum of contributions and social welfare:
max
p
G(p) =

i L
C
i
(p) +aW (p)
where:
W (p) =

n
i =1

i
V
i
(p) and a _ 0
Comments:
GH (1994) model has the structure of common agency problem: many
principals (ie lobbies) share a common agent (ie the government).
We can therefore use Bernheim and Whinstons (1986) results on menu
auctions.
Equilibrium Contributions
We denote by
_
_
C
0
i
_
i L
, p
0
_
the SPNE of the previous game.
We restrict ourselves to interior equilibria with dierentiable equilibrium
contribution schedules.
Whenever we say in any SPNE, we really mean in any interior SPNE where
C
0
is dierentiable.
Lemma 1 In any SPNE, contribution schedules are locally truthful:
\C
0
i
_
p
0
_
=
i
\V
i
_
p
0
_
Proof:
1
p
0
optimal for the government =

i L
\C
0
i
_
p
0
_
+a\W
_
p
0
_
= 0.
2
C
0
i
() optimal for lobby i =

i
\V
i
_
p
0
_
C
i
_
p
0
_
+

i
/
L
\C
0
i
/
_
p
0
_
+a\W
_
p
0
_
= 0.
3
1+2 = \C
0
i
_
p
0
_
=
i
\V
i
_
p
0
_
.
Equilibrium Trade Policies
Lemma 2 In any SPNE, domestic prices satisfy:

n
i =1

i
(I
i
+a) \V
i
_
p
0
_
= 0,
where I
i
= 1 if i is politically organized and I
i
= 0 otherwise.
Proof:
1
p
0
optimal for the government =

i L
\C
0
i
_
p
0
_
+a\W
_
p
0
_
= 0.
2
1 + Lemma 1 =

i L

i
\V
i
_
p
0
_
+a\W
_
p
0
_
= 0.
3
Lemma 2 directly derives from this observation and the denition of W
_
p
0
_
.
Comment:
In GH (1994), everything is as if governments were maximizing a social welfare
function that weighs dierent members of society dierently.
Equilibrium Trade Policies (Cont.)
Proposition 2 In any SPNE, trade policies satisfy
t
0
i
1 +t
0
i
=
I
i

L
a +
L
_
z
0
i
e
0
i
_
for i = 1, ..., n, (7)
where
L
=

i
/
L

i
/ , z
0
i
=
i
/m
i
, and e
0
i
= d ln m
_
p
0
i
_
/d ln p
0
i
.
Proof:
1
Roys identity denition of V
i
_
p
0
_
=
V
i
/
_
p
0
_
p
i
= (
i
/
i

i
) +
_
p
0
i
p
w
i
_
m
/
_
p
0
i
_
where
ii
/ = 1 if i = i
/
and
ii
/ = 0 otherwise.
2
1 + Lemma 2 = for all i
/
= 1, ..., n,

n
i
/
=1

i
/ (I
i
/ +a)
_

i
/
i

i
+
_
p
0
i
p
w
i
_
m
/
_
p
0
i
__
= 0
3
2 + denition of
L
=

i
/
L

i
/ =
(I
i

L
)
i
+
_
p
0
i
p
w
i
_
m
/
_
p
0
i
_
(
L
+a) = 0
Equilibrium Trade Policies (Cont.)
Proof (Cont.):
4. 3 + t
0
i
=
_
p
0
i
p
w
i
_
/p
w
i
=
t
0
i
=
I
i

L
a +
L
_


i
p
w
i
m
/
_
p
0
i
_
_
=
I
i

L
a +
L
_

z
i
m
_
p
0
i
/
_
p
w
i
m
/
_
p
0
i
/
_
_
5. Equation (7) directly derives from 4 and the denition of z
0
i
and e
0
i
How Should Taris Vary Across Industries (and Countries)?
GHs (1994) basic insights
According to Proposition 2:
1
Protection only arises if some sectors lobby, but others dont: if
L
= 0 or 1,
then t
0
i
= 0 for all i = 1, ..., n.
2
Only organized sectors receive protection (they manage to increase price of the
good they produce and decrease the price of the good they consume).
3
Protection decreases with the import demand elasticity e
0
(which increases the
deadweight loss).
4
Protection increases with the ratio of domestic output to imports (which
increases the benet to the lobby and reduces the cost to society).
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Are Unilateral Taris Ecient?
In the case of a small open economy, which is the case considered by GH
(1994), the answer is trivially, yes.
GH (1995) extend the previous analysis to the case of two large countries:
In this situation, unilateral taris are not Pareto-ecient.
Terms-of-trade changes may aect other countries, and so, provide a rationale
for trade agreements.
As we mention before, the interesting question, however, is:
Do political-economy motives provide a rationale for trade agreements above
and beyond correcting the terms-of-trade externality?
Bagwell and Staigers (1999) answer is, no.
Terms-of-Trade Externality Revisited
Bagwell and Staiger (1999)
Political-economy motives aect preferences, W
c
(p
c
, p
w
), over domestic
and world prices.
For example, in GH (1994), a small open economy may not choose free trade.
However, at a theoretical level, if we can still write governments objective
function as W
c
(p
c
, p
w
), then the only source of the ineciency has to be
the terms-of-trade externality:
Nothing in the rst part of this lecture relied on
W
c
(p
c
, p
w
) = V
c
[p
c
, R
c
(p
c
) +T
c
(p
c
, p
w
)]!
Intuitively, starting from a situation where W
c
p
c (p
c
, p
w
) = 0 for all c, the
only rst-order eect of a tari change has to be the change in p
w
.
Since this is a pure terms-of-trade eect, and hence an income eect, it
cannot aect world welfare.
Reciprocity in the WTO
Bagwell and Staiger (1999)
Using the previous insight, one can rationalize the principle of reciprocity
within the WTO.
Reciprocity = Mutual changes in trade policy such that changes in the value
of each countrys imports are equal to changes in the value of its exports.
Formally, a change in taris t
1
= t
1/
t
1
and t
2
= t
2/
t
2
is reciprocal
if:
p
w
_
m
1
1
_
p
1/
, p
w/
_
m
1
1
_
p
1
, p
w
__
=
_
x
1
2
_
p
1/
, p
w/
_
x
1
2
_
p
1
, p
w
__
Using trade balance, this can be rearranged as:
_
p
w/
p
w
_
m
1
1
_
p
1/
, p
w/
_
= 0 = p
w/
= p
w
Hence mutual changes in trade policy that satisfy the principle of reciprocity
leave the world price unchanged, which eliminates the source of ineciency.
Todays Plan
1
Introduction to the study of trade policy.
2
The traditional economic approach (welfare-maximizing trade policy).
1 Johnson (1954) via Bagwell and Staiger (1999).
3
The political-economy approach (government not necessarily
welfare-maximizing).
1 Grossman and Helpman (1994).
4
Rationalizing trade agreements.
1 Bagwell and Staiger (1999).
5
Other topics in the study of trade policy.
Strategic Trade Policy
Strategic trade policy was a ver active area of research in the 1980s.
Objective:
Normative analysis of trade policy under many forms of imperfect
competition (not just monopolistic competition, as weve covered extensively
in this course).
Classics:
1
Brander and Spencer (1985): export subsidies may be the optimal way to shift
prots away from foreigners and towards domestic rms (in a Cournot
duopoly).
2
Grossman and Eaton (1986): optimal policy crucially depends strongly on the
details of the model (e.g. Cournot vs. Bertrand).
Strategic Trade Policy (Cont.)
Recently, a few papers have revisited the implication of imperfect competition
for trade agreements. In particular, does imperfect competition provide a new
rationale for trade agreements?
Ossa (2011) uses a monopolistic competition model as in Krugman (1980).
Recall that this model features a purely domestic ineciency if there is more
than one industry, due to the home market eect. Hence Ossa (2011) says,
yes.
Bagwell and Staiger (2009) argue that this result obtains in Ossa (2011)
because that paper restricts government trade policy to only include import
taris. Hence they say, no.
From an empirical standpoint:
Can we gure out which assumptions about market structure t best a given
industry? If so, why would Grossman and Eaton (1986) be a problem?
There have of course been huge advances in empirical IO (eg estimation in
settings with strategic interactions) since 1986.
Why Do Governments Use Trade Policy Instruments?
Most papers analyzing trade policy start from ad-hoc restriction on the set of
instruments (e.g. taris, quotas, export subsidies, no production subsidies).
Conditional on this ad-hoc restriction, paper then explains why trade policy
may look the way it does and what its consequences may be.
But why would governments use inecient instruments in the rst place?
In developing countries, this may be the best feasible way to raise revenues
(Gordon and Li, 2005).
Inecient methods may be used to reduce total redistribution (Dixit,
Grossman and Helpman,1997).
Inecient redistribution may be used as a commitment device (Acemoglu and
Robinson, 2001).
Understanding the WTO
What are the implications of the self-enforcing nature of trade agreements?
Bagwell and Staiger (1990), Maggi (1996).
What is the rationale for trade agreements in the presence of NTBs?
Bagwell and Staiger (2001) consider the case of product standards (and
conclude that, once again, only the terms-of-trade externality matters).
How can we rationalize simple rigid rules (e.g. an upper bound on taris)
within the WTO?
Amador and Bagwell (2010), Horn, Maggi, and Staiger (2010).
Quantitatively, how large are the gains from the WTO?