Firm Behavior and Market Access in a Free
Trade Area with Rules of Origin
Jiandong Ju
University of Oklahoma
Kala Krishna1
Pennsylvania State University and NBER
April 2002
1
Contact author: Kala Krishna, Department of Economics, Pennsylvania State
University, University Park, PA16802; Telephone: 814-865-1106; FAX: 814-863-4775;
E-mail: kmk4@psu.edu. Krishna gratefully acknowledges support from the National
Science Foundation under Grants SBR-9320825 and SBR-9725019. We are grateful to
Philip Levy and Rob Reed for comments.
Abstract
We study a Free Trade Area with Rules of Origin and show that there are two
distinct regimes. In one regime, all Þrms behave in the same manner, while in
the other, they do not. Making ROO tighter has opposite effects in the two
regimes and there is a regime switch when ROO become restrictive enough. As
ROO become stricter, imports into the F T A of the intermediate good Þrst fall
and then rise while imports of the Þnal good as well as the price of the domestic
input, Þrst rise and then fall.
1
Introduction
There has recently been a surge of interest in Preferential Trading Arrangements
or P T As. Much of this is linked to the creation of the European Union (EU ),
the North American Free Trade Area (NAF T A), and the proposed Free Trade
Area of the Americas (F T AA).
In a Free Trade Area (F T A) tariff rates among members are zero, although
tariff rates set by members on non members are not necessarily equalized. As
such, tariffs on the same good are likely to differ among member countries.
Given this difference in tariffs, and in the absence of transport costs, what
prevents trade in a product from going through the country with the lowest tariff
on it and then being shipped within the F T A? The answer is Rules of Origin or
ROO. A good is eligible for zero tariffs in the F T A only if it originates there.
ROO specify conditions which have to be met for such origin to be granted. To
comply with ROO, a fraction, α, of the variable cost, must come from F T A
(domestic) made inputs. How do such rules affect Þrm behavior, market access,
and the welfare of the various parties in an F T A? This is the question addressed
in this paper. We consider the short run where the number of Þrms is given,
but where Þrms can choose how to use their production capacity.
We show that there are two distinct regimes, the homogeneous and heterogenous regimes. In the homogeneous regime all Þrms choose to meet the
ROO. In the heterogenous regime Þrms are indifferent between meeting the
ROO and not meeting it as their proÞts are the same in either case. Hence,
some Þrms could choose one option while others choose the other. This gives
rise to heterogeneous behavior ex post, even when Þrms are identical ex ante.
The homogeneous regime occurs when the ROO is not very restrictive. Once
the ROO is restrictive enough, the heterogeneous regime prevails. The effect of
1
more restrictive ROO on trade ßows of Þnal and intermediate goods is exactly
opposite in the two regimes and ßips at the switching point. Thus, as shown
below, imports from the rest of the world of the intermediate good Þrst fall and
then rise while imports of the Þnal good as well as the price of the domestic
input, Þrst rise and then fall. Such effects can put Þnal and intermediate good
producers at opposite sides of the table in pushing for stricter ROO.
There is a large literature on ROO in law, see for example Vermulst (1994).
In economics, a few of the notable papers are Rosellon (1994), Schiff (1997),
Falvey and Reed (1998) and Krishna and Krueger (1995).1 While it is tempting
to think of F T As as liberalizing, they are often not. ROO are in themselves
hidden protection: they create what looks like tariffs on imported intermediate
inputs and affect the price of domestically made inputs as well. As ROO are
negotiated industry by industry, there is enormous scope for well organized
industries to essentially insulate themselves from the effects of the F T A by
devising suitable ROO. For this reason, there is a lot of lobbying by industries
and employment of trade lawyers in the process of deÞning the ROO for their
industry. Almost 200 pages of the draft NAF T A agreement itself had to do
with deÞning these ROO!
There are large differences in the effects of an F T A with and without ROO.
In the absence of ROO, an F T A results in large changes in trade ßows as trade
seeks the lowest tariff entry point into the F T A. Goods are then trans-shipped to
their Þnal destination in the F T A. Of course this results in large tariff revenue
transfer effects as this “trade deßection” transfers tariff revenue from a good to
the country with the lowest tariff entry point.2 As pointed out by Richardson
1 This paper also contains a summary of some of the related literature in law and in international trade.
2 See McGillivray (2000) who argues that this is exactly what happened in the period 17771789, the Atricles of Confederation Period, when British merchants landed their goods in the
state in America offering the most favorable terms.
2
(1995) this can result in a race to the bottom in setting tariffs. However, in the
presence of ROO, this trans-shipment is not possible. Nevertheless, there is still
some trade deßection possible. By shipping domestic production, which meets
the ROO, to its F T A partners and meeting domestic demand via imports, the
low tariff country can still attract trade to its ports.
Of course, the presence of ROO tends to attract investment in order to
circumvent ROO. Thus, F T As with ROO lead to large investment ßows, while
those without ROO result in large trade ßows! This point is made in Krishna
and Krueger (1995) in a model with constant returns to scale. Such a model
may represent the long run quite well, but is unlikely to be of much use in the
short run. It is these short run effects, which are likely to differ from the long
run ones, that we focus on.
There are many aspects of F T As that we neglect in this paper. Issues of
why they might be created, who lobbies for and against them, whether they
are stepping stones to global integration or impediments to it,3 which kinds of
F T As are good and which kinds bad, are all outside the scope of this paper.
The reader is directed to Krueger (1997) for a thoughtful discussion of such
issues, to Bhagwati and Panagariya (1996) for a biting criticism of much of the
work in the area, to Richardson (1995) for a discussion of the effects on tariffs of
an F T A without ROO, and to the work of Bagwell and Staiger (1998) for some
fascinating ideas on developing rules which ensure that F T As, when formed, do
not result in negative externalities for non members.
3 Using a median voter model, Philip Levy (1997), shows that in a Heckscher-Ohlin setting
a bilateral agreement cannot supplant multilateral free trade, but bilateralisim can undermine
support for future multilateral liberalization when there are variety beneÞts from trade. Pravin
Krishna (1998), looks at similar questions using a Cournot model of oligopoly to look at the
effect on Þrm proÞts.
3
2
The Model
We want to consider the effects of an F T A. The simplest framework in which
this can be done is a three country world. Let the three countries be called
A, B, and C. Countries A and B form an F T A, excluding C which can be
thought of as the rest of the world. There are two goods, a Þnal good x and
an intermediate good (input) z in addition to a numeraire consumption good.4
Prior to the F T A, both A and B import the Þnal and intermediate goods from
C. Of course, the numeraire good is exported by them to C so that trade
balances. There is no trade between A and B prior to the F T A. However,
trade between A and B will occur in both goods due to the F T A.
Both A and B are small countries so that they take the world price as given.
The world price for x is denoted by pw
x and the world price for z is denoted by
i
pw
z . Country i has tariffs tj on good j. These are assumed to be speciÞc tariffs.
Let pi0
j be the price of good j in country i, at time t = 0 ( t = 0 before the F T A
w
i
and t = 1 after it). pi0
j = pj + tj , for i = A, B and j = x, z, as both goods are
imported.
When A and B form an F T A, their tariffs on each others products are set
at zero, while those on imports from the rest of the world are unchanged. Let
B
the tariff on x set by A exceed that set by B, that is, tA
x > tx . Despite this,
the rest of the world cannot access A’s market via simple trans-shipment from
B due to rules of origin which have to be met to obtain preferential treatment.
For this reason, prices in A and B need not be the same even after the F T A.
We will also assume throughout that the tariff on z set by A is less than that
B
set by B, that is, tA
z < tz .
4 In what follows we choose to make simplifying assumptions wherever possible rather
than exploring each and every possibility that might arise. Our basic point, that Þnal and
intermediate goods are affected in opposite directions by ROO and that there are two regimes
with very different comparative statics, is, we believe a robust one.
4
We assume that the ROO in the intermediate goods is automatically met
when the input is made in the F T A but can not be met if the input is made
w
A
abroad.5 The price of the imported intermediate after the F T A, pA1
z = pz + tz
w
B
and pB1
z = pz +tz , is thus unchanged by the F T A while the domestically made
intermediate input price, denoted by p1z , is equalized in the two countries after
the F T A, though it need not equal the price of the imported intermediate input.
Although the imported and internally made inputs are physically identical, they
are treated as different by users. This is because the ROO in the Þnal good
requires a minimum cost share to arise within the F T A. Imported intermediate
inputs cannot contribute to this cost share while F T A made inputs can. As
a result, even if the price of F T A made inputs exceeds that of the imported
inputs, they may still be used in order to meet the ROO.
What about the price of the Þnal good after the F T A? Final goods can come
from C directly or from an F T A member if the ROO is met. It is not possible
for the price of the Þnal good to rise after the F T A due to the possibility of
importing directly from C. There will be no incentive for Þrms in country A to
B
export to country B because tA
x > tx . However, Þrms in country B may wish
to export to A. As A has higher prices than B, it may pay for B to export to
A even if there are additional costs in meeting the ROO. Since only Þrms in C
export to B the price of x in B stays unchanged. In addition, we assume that
B cannot meet all of A0 s demand so that the price in A cannot fall either.6
In what follows we will examine the problem in the short run where there
is no investment, but where Þrms can use their existing capital as they wish
to either export to A and meet the costly ROO or to sell at home at a lower
price and at a lower cost. The restrictiveness of the ROO is indexed by α. In
Section 3 we consider the problem facing Þrms at given prices. In Section 4
5 This
6 In
removes complications arising from trans-shipment of intermediate goods in the FTA.
A1
B1
B0
other words, pA0
x = px > px = px .
5
we consider the determination of the price of the domestic intermediate input.
The more Þrms in B who choose to export to A, the greater the demand for
the F T A made intermediate input since these are the only ones who demand
it if its price exceeds that of the imported input. The greater the demand for
the F T A input, the higher is its price, p1z , and hence the lower is the proÞt of
the Þrms using it in order to export to A free of tariffs. We show that if the
ROO are not too severe, then the domestic input price is low and all Þrms in B
choose to export to A. However, with all Þrms in B exporting to A , an increase
in the restrictiveness of the ROO shifts out the demand for the domestic input
and raises its price. Thus, once the ROO reaches α1 , Þrms in B are equally
well off meeting the ROO and selling in A and not meeting it and selling at
home! Increases in α beyond α1 have a different effect. In order to keep Þrms
indifferent between exporting and selling at home, p1z must fall, not rise, as
α rises! This reversal drives our model and yields some unusual comparative
statics results which we explore in later sections.
In Section 5 we consider the implications of our analysis for market access.
We show that market access depends on the restrictiveness of ROO. More restrictive ROO on the Þnal good Þrst raises and then lowers imports of the
Þnal good, but Þrst lowers then raises imports of the intermediate good. Their
turning point is common so that imports of the Þnal good are maximized and
imports of the intermediate good are minimized at a common level of restrictiveness of the rules of origin. In Section 6, we offer some concluding remarks
and alternative interpretations of our model. We show that our model can be
reinterpreted to show that more restrictive ROO on the Þnal good Þrst improve
and then harm the fortunes of labor.
6
3
The Firm’s Problem
Let F i (Z, k) denote the constant returns to scale (CRS) production function
for the Þnal good in country i, i = A, B, where k denotes the capital used, and
Z the intermediate good used by the Þrm. In the short run, being considered
here, k is given so we suppress it from here on. As Þrms in A do not wish to
export to B, they do not care about the source of the intermediate good. They
use the cheapest source for the input and equate the marginal value product of
the input to its price. The supply of a typical Þrm in A after the F T A is thus
A1
denoted by xA (pA1
x , pz ). It is, as usual in such problems, positively related to
A1
pA1
x and negatively related to pz .
To comply with the ROO, a fraction, α, of the variable cost share, must
come from F T A (domestic) made inputs or
7
p1z z
≥ α.
∗
p1z z + pB1
z z
(1)
where z denotes the domestic made input and z ∗ denotes the imported input
used by a Þrm.
Firms in B will either meet the ROO and export to A or not meet the ROO
and sell to their home market, depending on what is more proÞtable.8 Let
Z B (px , min(p1z , pB1
z )) be the total input demand function for a given Þrm in B
in the absence of any restrictions as only the cheaper input is used.
then all Þrms will
What is the effect of ROO? For a given px , if p1z ≤ pB1
z
weakly prefer the domestic input and the ROO are automatically met. The
input demand is as if there were no ROO. If p1z > pB1
z then Þrms would prefer
to use the imported input but may choose to use the domestic one in order to
7 We use this form here but using the physical form for the ROO would not make a difference
as shown in Ju and Krishna (2002).
8 If a Þrm in B exports to A without meeting the ROO, it will be charged by a tariff tA .
x
B1
Hence, the Þrm will only receive the world price pw
x which is less than the home price px , by
doing so. Therefore, no Þrm in B has an incentive to export to A without meeting the ROO.
7
qualify for a higher output price.
If p1z > pB1
z and the ROO are not met, the Þrm will maximize
B ∗
B1 ∗
pB1
x F (z , k) − pz z .
¢
¡
B1
B1
is a constant as pB1
The maximized value of proÞts, Π̄ pB1
x , pz
x and pz are
B1
Þxed. The demand for inputs and the outputs by such a Þrm are z̄(pB1
x , pz )
B1
and x̄(pB1
x , pz ) respectively.
If ROO are met the Þrm maximizes
B
∗
1
B1 ∗
pA1
x F (z + z , k) − pz z − pz z
subject to (1). The restriction is met exactly, which gives
z∗ =
p1z (1 − α)
z.
αpB1
z
(2)
¸
p1z (1 − α) + αpB1
z
z,
αpB1
z
(3)
In turn, this yields
Z = z + z∗ =
·
Using (2) we get,
∗
p1z z + pB1
z z
p1 z
p1 (1 − α)
z= z
= p1z z + z
α
¸ α
·
p1z pB1
z
=
(z + z ∗ )
p1z (1 − α) + αpB1
z
where the second equality comes from inverting (3). This gives proÞts for a
given Þrm as
1
A1 B
1
Π(pA1
x , φ(α, pz )) = px F (Z, k) − φ(α, pz )Z.
Note that φ(α, p1z ) =
p1z pB1
z
p1z (1−α)+αpB1
z
(4)
plays the same role as min(p1z , pB1
z ) did in
the absence of a ROO. It is the virtual price of the input when the ROO are
8
1
met. Thus, total input demand is Z(pA1
x , φ(α, pz )). Inverting (3) gives demand
for the domestic input to be
1
1
A1
1
z(pA1
x , pz , α) = ψ(α, pz )Z(px , φ(α, pz ))
where ψ(α, p1z ) =
αpB1
z
.
p1z (1−α)+αpB1
z
It is easy to verify that φ(α, p1z ) is increasing9
in α and p1z and that ψ(α, p1z ) is increasing in α but decreasing10 in p1z . As
1
1
1
Z(pA1
x , φ(α, pz )) is decreasing in φ(α, pz ), an increase in pz reduces both com1
1
A1
1
A1 1
ponents of z(pA1
x , pz , α), ψ(α, pz ) and Z(px , φ(α, pz )), so that z(px , pz , α) is
decreasing in p1z .
4
Equilibrium Conditions
What does the demand for domestic inputs in B look like? Clearly this depends
on whether Þrms choose to meet the ROO or not. The demand for domestic
1
11
is
inputs is depicted in Figure 1. The total demand curve, nB Z(pA1
x , φ(α, pz )),
depicted by the line AB in Figure 1. If p1z < pB1
z then the ROO are automatically
met and there is no demand for imported inputs. All Þrms in B buy only the
domestic input and the demand for the domestic input equals that for total
inputs at these prices which corresponds to the segment HB. If p1z = pB1
z ,
Þrms do not care whether they overfulÞll the ROO or not and the demand for
domestic input corresponds to the segment F H.
If p1z > pB1
z , but not by a lot, we are in the homogeneous regime. As
1
B1 B1
Π(pA1
x , φ(α, pz )) > Π(px , pz ), all Þrms in B will be better off meeting the ROO
1
and selling to A. The total demand for domestic inputs will be nB z(pA1
x , pz , α)
which is downward sloping and corresponds to the segment DF . As p1z rises
9 An increase in α reduces the denominator of φ(.), raising φ(.). Divide the numerator and
denominator in φ(.) by p1z and note that an increase in p1z reduces the denominator, thereby
raising φ(.).
10 As p1 > pB1 , an increase in α raises the numerator and reduces the denominator of ψ(.)
z
z
therby raising ψ(α, p1z ). An increase in p1z raises the denominator and so reduces ψ(α, p1z ).
11 nB denotes the Þxed number of Þrms in B.
9
Figure 1: Equilibrium of FTA Made Input
p1z
A
p̃1z (α0 )
p̃1z (α00 )
p̃1z (α1 )
p̃1z (α01 )
pB1
z
0
S
@
@
@
@
@
@
@
@
G
D
@
@
@
@
@
@
@
@
F
@H
@
@
@
@
@
@
B
10
Quantity of Input
there comes a point, at p1z = p̃1z (α) where
1
B1 B1
Π(pA1
x , φ(α, p̃z )) = Π(px , pz ).
(5)
and Þrms are indifferent between meeting the ROO and not doing so and some
Þrms meet the ROO while others do not. At this point we are in the heterogeneous regime. As a result, total domestic input demand is anywhere from zero
1
to nB z(pA1
x , p̃z (α), α), which corresponds to GD, so that total input demand
is horizontal at p̃1z (α). If p1z rises further, then all Þrms prefer not meeting the
ROO, and domestic inputs are not demanded at all, which corresponds to the
segment AG. The total demand for the domestic input is thus given by the dark
line AGDF HB in Figure 1.
Of course, p1z is determined by the intersection of demand and supply for
domestic input which we turn to below. As a Þrst step we examine the behavior
of domestic input demand in the two regimes. We show in the next lemma that
the horizontal segment at p̃1z (α) shifts down as α rises.
Lemma 1
dp̃1z (α)
dα
< 0.
Proof. From (5) it follows that p̃1z (α) is decreasing in α since φ(α, p1z ) is
increasing in α and p1z and φ(α, p1z ) must be kept constant for (5) to remain
true.
Now we show that the downward sloping segment shifts out as α rises unless
the derived demand for the input is very elastic.
Lemma 2
1
d[z(pA1
x ,pz ,α)]
dα
> 0 if B <
1
θ
where B is the elasticity of Z(pA1
x , φ(.))
with respect to φ deÞned as a positive number and θ =
11
α(p1z −pB1
z )
p1z
< 1.
Proof. Note that z(.) = ψ(.)Z(.) so that
¤
£
1
¡
¢
d z(pA1
d [ψ(.)Z(.)]
x , pz , α)
A1
=
= Z pA1
x , φ(.) ψ α (.) + ψ(.)Zφ (px , φ(.))φα (.)
dα
dα
¢
¡ 1
pz − pB1
1
z
+ ψ(.)Zφ (.)φ(.) 1
= Z(.)φ(.) 1
[pz (1 − α) + αpB1
[pz (1 − α) + αpB1
z ]
z ]
Z(.)φ(.)
(1 − Bθ) > 0
=
[p1z (1 − α) + αpB1
z ]
if B < 1θ .
Now we can turn to the determination of p1z in equilibrium and the effects
of a more severe ROO. Before doing so however, we will show that in the
heterogeneous regime, Þrms in B who export to A produce less than Þrms who
sell domestically.
Lemma 3 In the heterogeneous regime, the output of Þrms in B who meet the
B1 B1
ROO, x(pA1
x , φ(.)), is less than the output of those who do not, x̄(px , pz ).
Proof. ProÞts are quasi convex in prices and increasing in px but decreasing
in pz . Hence iso proÞt contours are upward sloping. ProÞts are zero at the origin.
If input prices remain at zero while the output price becomes positive, proÞts
are unbounded. Thus, iso proÞt contours all emanate from the origin in (px ,
1
pz ) space. Firms who meet the ROO have proÞts Π(pA1
x , φ(α, pz )) while those
B1
who do not have proÞts of Π̄(pB1
x , pz ). Since these proÞts are equal, the points
¤
£ A1
¤
£
B1
lie on the same upward sloping iso proÞt contour
px , φ(α, p1z ) and pB1
x , pz
£ A1
¤
¤
£
B1
and px , φ(α, p1z ) lies above and to the right of pB1
. By quasi convexity
x , pz
¤
£ A1
¤
£
B1
yield
of proÞts, all points on the line connecting px , φ(α, p1z ) to pB1
x , pz
lower proÞts than either of them. Hence this iso proÞt contour is as depicted
£
¤
1
in Figure 2. The line connecting the origin to pA1
x , φ(α, pz ) is steeper than the
¤
B1
£
B1
. Thus, pφA1 > ppzB1 . Recall that proÞt
line connecting the origin to pB1
x , pz
x
x
maximization equates the marginal products of the input to
φ
pA1
x
and
pB1
z
pB1
x
when
the ROO is met and not met respectively. As a result, total input usage, and
12
pz
φ(α, p1z )
pB1
z
Figure 2: Iso ProÞt Contour
©©
©
©©
©
©© ³³
©
³³
©©
³³
³
©
©³³
©
³³
©
³
³
©
pB1
pA1
x
x
px
hence output, of a Þrm meeting the ROO is less than that of a Þrm not meeting
the ROO since production function F (.) is concave.
Proposition 4 For 0 ≤ α < α0 an increase in α has no effect on the equilibrium. For α0 ≤ α < α1 , all Þrms in B choose to meet the ROO. In this, so
called homogeneous regime, assuming that the derived demand for the input is
not too elastic, an increase in α raises usage and price of the F T A made input
while reducing total input usage and hence total output produced by Þrms in B.
In contrast, when α1 ≤ α ≤ 1, some Þrms in B choose to meet the ROO while
others choose not. In this, so called heterogeneous regime, an increase in α reduces usage and the price of F T A made input but raises total output produced
13
by Þrms in B.
Proof. At very low levels of α, very little of the domestic input is needed
to meet the ROO so that the intersection of demand and supply occurs at
p1z = pB1
z . As α rises, the downward sloping segment of demand shifts out till at
α = α0 it intersects supply at the lower kink as depicted in Figure 1. As α rises
above α0 , the intersection of demand and supply occurs along the downward
sloping part of domestic input demand and we are in the homogeneous regime.
As α rises, as shown in Lemma 2, the downward sloping segment shifts out. As
a result, the equilibrium level of p1z and the usage of domestic inputs increase.
The increases in α and p1z raise the virtual price φ(.), therefore, the total input
usage and total output fall. Finally, at α = α1 , the intersection of demand
and supply occurs at the upper kink as depicted in Figure 1. This puts us in
the heterogeneous regime. As α rises above α1 , the intersection of demand and
supply occurs at p̃1z (α). As shown in Lemma 1, an increase in α reduces p̃1z (α)
as well as the total usage of the domestic input, but as we will show, raises the
usage of total inputs.
ProÞts from not meeting the ROO are Þxed and as a result, so are the proÞts
of meeting it in this regime. Hence, φ(.) is Þxed so that total input usage,
1
Z(pA1
x , φ(α, pz )), and output per Þrm remains unchanged as α rises. Domestic
1
input demand per Þrm, z(pA1
x , pz , α)) = ψ(.)Z(.) increases in this heterogeneous
regime as ψ(.) is increasing in α while Z(.) remains unchanged. However, total
domestic input usage falls as α rises. This can only occur if the number of Þrms
exporting to B, nB
A , falls. Now Lemma 3 shows that Þrms in B who sell to A
produce less than Þrms who do not. Therefore, the total output of Þrms in B
increases as α rises since we know that the number of Þrms selling to A falls as
α rises and these Þrms are replaced by Þrms selling to B who produce more.
14
5
Market Access Effects
Much of the concern in the popular press regarding preferential trading arrangements has been with the implications of such arrangements on market access.
Will arrangements like EU or N AF T A lead to an increase in trade among
member countries and a reduction in trade between the area and the rest of the
world? Here we have a few comments to offer.
First, the F T A with ROO creates trade between A and B in our simple
model. A begins to import the Þnal good from B after the F T A. It also imports
the intermediate good. In fact, if α > α0 , B imports S A (.), the entire supply
of the domestic input in A. Second, effects in both Þnal and intermediate good
market need to be considered as they can work in opposite directions. We show
below, for example, that market access in Þnal goods is improved, while in
intermediate goods it is reduced when the ROO is more than α1 . Finally, it is
important to look at the effect on the F T A as a whole, as done here, as well
as on each member. This is because compensating ßows occur in other member
countries in response to changes in ßows in one member country.12
5.1
The Path of Imports
We now look at the level and change in imports of Þnal and intermediate goods
= φ(.). It is
as the ROO becomes more restrictive. For α ≤ α0 , p1z = pB1
z
costless to meet the ROO and the effect of the F T A is just to let Þrms in B sell
B1
at a higher price, pA1
x not px . Thus, their output rises and so does their input
use. As a result, B imports more of the Þnal good, and A imports less, but the
imports of the F T A as a whole of the Þnal good falls as Þrms in B produce
more when they sell to A. As these Þrms use more inputs in doing so and as
12 For example, if A imports from B due to the FTA as above, then B must import more
from the rest of the world to meet its own demand which was formerly met by domestic
production.
15
the price of inputs is unchanged by the F T A, their imports into B and in the
F T A as a whole must rise. Moreover, making the ROO more restrictive has no
effect so that imports are unaffected by changes in α in this region.13
For α0 ≤ α ≤ α1 (the homogeneous regime), the effect on imports of an
increase in α comes only from the behavior of Þrms in B who sell to A. As α
rises, so does φ(.). This increase in their virtual input price reduces their total
input use and thus their output. This in turn raises imports of the Þnal good.
However, as an increase in α shifts the downward sloping part of domestic input
demand outwards, it raises the use of the domestic input. Since total input use
falls, imports of the intermediate good fall as well.
For α1 ≤ α ≤ 1, not all Þrms in B prefer to produce for A. Some Þrms
choose to produce for the domestic market. Recall that in this case p1z falls as
α rises and reduces the use of domestic inputs. To see what happens to imports
of the intermediate good we need to see what happens to output of the Þnal
good. If this rises, then total input use must rise as well. Since domestic input
use has been shown to fall, this means that imports of the intermediate input
must rise.
What happens to output as α rises? The only Þrms who are affected are the
Þrms in B who choose to meet the ROO. As shown in Lemma 3, these Þrms
make less output than Þrms who do not export to A. Moreover, as shown in
Proposition 4, their number falls as α rises so that total output increases as
α rises and as a result, imports of the Þnal good fall and we are done. The
imports of intermediate goods rise while the imports of the Þnal good fall with
an increase in α.
Our results are depicted in Figure 3. The lines I F and I I trace out the
13 This is the case considered in Bhagwati and Panagariya (1996) as they assume the supply
curve of Þrms is unaffected by ROO. They point out the possibility of trade revenue transfers
due to trade deßection.
16
imports of the Þnal and intermediate goods as a function of α. I F 0 and I I0 are
horizontal lines which depict their pre F T A levels. Note that I F lies below I F 0
at α = 0 reßecting the fact that imports of the Þnal good fall when an F T A is
created with α = 0 as explained above. As argued above, imports of the Þnal
good are unchanged as α rises up to α = α0 . From α = α0 to α = α1 they rise
and then after α = α1 , they fall. Their level in the region where α > α1 must
lie above the pre F T A level. This is because in this region, the Þrms in B who
export to A produce less than Þrms who do not! As a result, imports of the
Þnal good in the heterogeneous regime must rise due to the F T A, no matter
what the level of the ROO.
Similarly, I I lies above I I0 at α = 0 reßecting the fact that imports of the
intermediate good rise when an F T A with α = 0 is created. As argued above,
imports of the intermediate good are unchanged as α rises up to α = α0 . From
α = α0 to α = α1 , they fall and then after α = α1 , they rise. However, their level
for very high α remains below that prior to the F T A. The reason is simple. In
the heterogeneous regime, imports of the intermediate good must lie below the
pre F T A level since even at high α, some Þrms in B export to A and these Þrms
produce less of the Þnal good and hence use less of the intermediate input in
total. As the price of domestic inputs is higher than before the F T A, domestic
supply has risen, and as demand has fallen, imports must fall.
One way to interpret these results is to note that the F T A seems to result
in trade diversion as Þrms in B export to A so that A0 s imports from the rest
of the world fall. However, as these Þrms in B no longer produce for B, B 0 s
imports rise by what these Þrms would have produced had they not produced
for export to A. The trade diversion dominates trade substitution for α < α0 as
Þrms in B who export to A produce more than before the F T A and use more
inputs doing so. As a result, their total imports of the Þnal good fall, while
17
their imports of the intermediate good rise due to the F T A.
Figure 3 also shows that the intermediate good market is the most protected
(in the sense that both imports reach a minimum and the price of F T A made
input reaches its maximum) when α = α1 . In sharp contrast, the Þnal good
market is most open (in the sense that the imports reach their maximum) at
the same point, that where α = α1 .
5.2
Winners and Losers From an F T A with ROO
In this section we look at two things. First, we look at what happens to the
welfare of various groups who are affected by an F T A. Second, we identify the
most preferred point of producers of Þnal and intermediate goods and show that
these points differ so that conßict between them is likely.
First, note that since the price of the Þnal good is independent of the creation
of an F T A or the level of ROO, welfare is made up of producer surplus and tariff
revenue only. Only the proÞts of the Þnal goods producers in B are affected by
an F T A. Since they choose to sell to A their proÞts must rise from doing so.
Intermediate good suppliers are also affected. Any increase in their price raises
their producer surplus. Moreover, since an F T A causes more imports of the
Þnal good to enter through B, B’s tariff revenues from the Þnal good rise while
those of A fall. Revenue from intermediate good imports tends to rise in A as
Þrms in A use only imported inputs, while it tends to fall in B as all the F T A
made inputs are used by the Þrms in B. Thus, the net welfare effects in the
two countries are ambiguous: some components rise while others fall. In some
special cases a clear direction is possible. For example, if there is no tariff on
intermediates in both countries then B gains from an F T A as its Þrms in Þnal
and intermediate sectors gain and its tariff revenue rises.
More important is the fact that intermediate goods producers are best off
18
Figure 3: Imports of Final and Intermediate Goods
pB1
z
p1z (α)
©``
©
```
©
```
©
```
©
```
©
`
1
©
pz (α) ©
0
1
```
```
¡
¡
```
```
¡
``
`
¡
¡
¡
¡
¡
I F (α)
IF 0
0
1
I I (α)
I I0
0
c
c
c
c
(
((
c
((((
(
(
c
((
c
c((((
α0
α1
19
1
at α = α1 as this is where their product price and hence rents are maximized.
However, Þnal goods producers in B gain the most when they have access to
A’s markets at the least cost, that is at any α ≤ α0 . Thus these two groups are
likely to lobby for different levels of ROO.
6
Conclusion
Much of the concern felt about regional trading areas has been that they will
exclude non member countries from their markets. Much of this discussion, see
for example Krugman (1991) and counter arguments by Bond and Syropoulos
(1996) has been couched in terms of the greater market power, and hence higher
optimal tariffs of larger trade blocs. This is not the direction we follow for two
reasons. First, tariffs on most manufactured goods are bound by concessions
made in successive rounds of negotiations under the auspices of GAT T. While
the use of Non Tariff Barriers (N T Bs) provide a way to circumvent these limits,
they often cause unanticipated side effects and stand to be disallowed if too
blatantly abused. Moreover, there is little evidence that this avenue has been
important in practice. Second, we believe that the avenue we focus on, ROO, are
less well understood, do end up providing reason for concern, and are extremely
important in practice.
In this paper we have shown that F T As can be far from free, especially
with ROO. More restrictive ROO in the Þnal good market can initially serve to
restrict access to the intermediate good market and improve access in the Þnal
good market. However, beyond a certain point, they must begin to improve
access to the intermediate good market and reduce access to the Þnal good
market!
We believe that the model developed here has an internal structure that
makes it suitable for a number of other purposes. An interesting and provocative
20
result of the model is the non monotonic effect of an increase in restrictiveness
of the ROO on the price of the intermediate input made in the F T A. If, for
example, this input was made by labor alone using a unit of labor to make it,
the wage would be equal to the price of the domestically made input and would
imply that labor would favor restrictive, but not too restrictive ROO.
Another application might be to study market access requirements. Consider, for example a market access requirement in auto parts on Japanese Þrms
linked to preferential tariffs on autos in the following manner. If a Japanese
Þrm buys sufficient U.S. auto parts for a minimum cost share of U.S. parts to
be met, then it qualiÞes for a low tariff to the U.S. If it does not, then it faces a
penalty tariff. Such a situation is analogous to the one studied above. For a low
enough cost share, all Japanese Þrms will meet the requirement and sell to the
U.S. meeting their own demand through imports. As the cost share rises, the
price of U.S. parts will rise but all Japanese Þrms will still export to the U.S.
At some point Japanese Þrms will be indifferent between exporting and serving
their own market. From this point onward the price of the U.S. inputs will fall
as the cost share increases, with changes in the number of Japanese Þrms who
serve the U.S. market acting to equalize the proÞts from these two options. Our
results above suggest that there is a level of this cost share which simultaneously maximizes exports of the intermediate good and minimizes imports of the
Þnal good. Other examples of such restrictions could be taxes (or rewards if the
target is met) if certain kinds of workers, be they welfare recipients or domestic
workers, do not make up a given fraction of the work force or certain kinds of
fuels do not make up a given proportion of energy consumption.
21
References
[1] Bagwell, Kyle and Robert Staiger (1998). “The Simple Economics of Labor
Standards and the GATT,” NBER Working Paper No. 6604.
[2] Bhagwati, Jagdish and Arvind Panagariya (1996). “Preferential Trading
Areas and Multilateralism: Strangers, Friends or Foes?” pp. 1-78 in The
Economics of Preferential Trade Agreements, AEI Press, Washington D.C.
[3] Bond, Eric and Constantinos Syropoulos (1996). “The Size of Trading
Blocs: Market Power and World Welfare Effects,” Journal of International
Economics, Vol. 40, No.3-4, pp. 411-437.
[4] Falvey, Rod and Geoff Reed (1998). “Economic Effects of Rules of Origin,”
Review of World Economics, Vol. 134, No. 2, pp. 209-229.
[5] Ju, Jiandong and Kala Krishna (2002). “Regulations, Regime Switches
and Non-Monotonicity when Non-Compliance is an Option:
An
Application to Content Protection and Preference,” (see webpage
http://econ.la.psu.edu/˜kkrishna/index.html).
[6] Kowalczyk, Carsten (1990). “Welfare and Customs Unions,” Dartmouth
College Working Paper.
[7] Krishna, Pravin (1998). “Regionalism and Multilateralism: A Political
Economy Approach,” Quarterly Journal of Economics, Vol. CXIII, pp.
227-252.
[8] Krishna, Kala and Anne O. Krueger (1995). “Implementing Free Trade
Areas: Rules of Origin and Hidden Protection,” pp. 149-187 in New Directions in Trade Theory, A Deardorff, J. Levinsohn and R. Stern (Eds.),
University of Michigan Press, 1995.
22
[9] Krishna, Kala (1991). “Openness: A Conceptual Approach,” Mimeo.
[10] Krueger, Anne O. (1995). “Free Trade Agreements Versus Customs
Unions,” Journal of Development Economics, Vol. 54, pp. 169-187.
[11] Krugman, Paul (1991). “Is Bilateralism Bad?” pp. 9-23 in International
Trade and Trade Policy, E. Helpman and A. Razin (Eds.), MIT Press:
Cambridge and London.
[12] Levy, Philip (1997). “A Political-Economic Analysis of Free-Trade Agreements,” American Economic Review, Vol. 87, No. 4, pp. 506-519.
[13] McGillivray, Fiona (2000). “Bargaining Leverage for Smaller States: SouthSouth Customs Unions,” Mimeo, Yale University.
[14] Richardson, Martin (1995). “Tariff Revenue Competition in a Free Trade
Area,” European Economic Review, Vol. 39, No. 7, pp. 1429-1437.
[15] Rosellon, Juan (1994). “Rules of Origin and Welfare Analysis,” Mimeo.
[16] Schiff, Maurice (1997). “Small is Beautiful: Preferential Trade Agreements
and the Impact of Country Size, Market Share, Efficiency, and Trade,”
Journal of Economic Integration, Vol. 12, No. 3, pp. 359-387.
[17] Vermulst, E. (1994). “Rules of Origin in International Trade: A Comparative Study,” in Studies in International Trade Policy, P. Waer and J.
Bourgeois (Eds.), University of Michigan Press: Ann Arbor.
[18] Viner, J. (1950). The Customs Union Issue, Carnegie Endowment for International Peace, New York.
23