Can import tariffs deter outward FDI ?
David Collie (*) and Hylke Vandenbussche (**)
(*) Cardiff Business School, UK
(**) Catholic University of Leuven, Belgium
CEPR affiliate, London
August 2003
Abstract
In this paper we analyze a country’s optimal trade policy when its labor market is unionized and firms are
footloose. We show that an important objective for governments to use import protection is to prevent their
domestic multinationals to go to a non-unionized location abroad and to serve their country from a distance. A
domestic government will set a positive tariff to dissuade its multinational from engaging in outward FDI when
the additional profits it repatriates, do not compensate for the loss of domestic union rent. To put it differently,
we show that when the domestic labor market is unionized, trade liberalisation between countries with similar
wage levels is likely to result in domestic welfare losses as a result of outward FDI. Only when wage differences
between countries are large enough, can outward FDI improve domestic welfare and optimal tariffs will be zero.
JEL codes: L13, F23
Key words: FDI, monopoly union, trade policy, Cournot competition
Address for correspondence: Hylke Vandenbussche, Faculty of Economics, University of Leuven, Naamsestraat
69, 3000 Leuven, Belgium, [email protected], T: + 32 (0) 16 32 69 20; F: + 32 (0) 16
32 67 32
This paper has benefited from comments at a CEPR-TMR network on FDI in London 1999, Center-Tilburg
1999, UCL Louvain-la-neuve 1998 and IUI, Stockholm School of Economics 2001, Nottingham University
2001 and EARIE-Dublin 2001. We particularly thank Ana Falzoni, Reinhilde Veugelers, Zhihao Yu, Harry
Huizinga, Claude d’Aspremont and Jozef Konings.
1. Introduction
In this paper we document theoretically that countries with unionized labor markets and footloose
multinationals clearly have an interest in using trade policy to deter their firms from engaging in outward FDI to
serve their market from a distance.
This paper aims to show that incentives for trade policy against imports from abroad can be substantially
altered in the case of footloose firms. With fixed locations, foreign rent-extraction primarily motivates the use
and the magnitude of import tariffs by a home government (Brander & Spencer, 1981). In the case of footloose
firms, the use of tariffs will show to be no longer evident, because it may also affect home firms locating abroad.
The outward FDI we are considering in this paper is FDI that aims to serve relevant product markets from a
distance, attracted by differences in labor cost and/or productivity between locations.
For this purpose we develop a simple partial equilibrium duopoly model 1, consisting of two footloose firms,
one domestic and one foreign. The firms sell a homogenous product 2 and engage in output (Cournot)
competition in the domestic market but can locate either in the domestic or in the foreign country. Focusing on
the domestic product market is similar to assuming that the foreign product market size is relatively small which
greatly facilitates the analysis.3 This implies that the outward FDI with the domestic firm moving abroad, is not
of the ‘market potential type’ but rather of the ‘production cost’ type. We assume the domestic country to be
unionised with a monopoly union setting the wage for the imperfectly competitive sector. The domestic wage
level is endogenous depending upon the number of firms that locate in the domestic market. The wage in the
foreign country is exogenous and can be higher or lower than the domestic wage level. The inward FDI with the
foreign firm locating in the domestic country, is of the ‘market access type’. The motive behind this last type of
FDI is to be close to consumers and to save on tariff costs.4 While we consider both the case of outward FDI and
inward FDI, the novelty of our analysis lies in the analysis of the outward FDI case. The inward FDI case will
also be discussed but is not different from the traditional tariff-jumping argument.
We first study the firms’ location pattern in the case of free trade. From the welfare effects it will become
clear that FDI, both inward and outward is not always in the domestic country’s interest. Trade liberalisation can
result in a strong welfare decrease for the domestic country with the unionized labor market as a result of
outward FDI. This result has recently also been pointed out by Lommerud et al. (2002). But while that paper
1
Although under current WTO rules it is increasingly difficult for a country or trade bloc to unilaterally impose
a tariff, the use of alternative protection in the form of antidumping measures for example continues to be
possible.
2
The homogeneous product assumption is mainly to simplify and not particularly important for the results. In the
duopoly setting we assume, even in the case of differentiated products our main results would hold.
3
Haufler and Wooton (1999) consider a model with two countries of unequal size competing for the location of
one multinational by means of tax competition. They assume wages in both countries to be equal. In our model
we have one government with a tariff policy and two multinationals of different nationality. We focus on the
unionisation of factor markets instead.
4
The distinction between the production cost motive for FDI and the market access motive for FDI is well
documented in Lankes and Venables (1996). Their conjecture was that transition countries will continue to grow
1
analyses the welfare effects of an exogenous reduction in transport costs, our paper considers how optimal
domestic trade policy can be used to affect the location of FDI and the welfare of the domestic country. We
allow for that by including an additional stage to the model where the domestic government sets an import tariff
to maximize domestic welfare as the sum of domestic producer surplus, consumer surplus and tariff revenue 5.
Similarly to Lommerud et al. (2002) we find that for a unionised country, trade liberalisation can reduce welfare.
But then we take the analysis one step further and identify the optimal tariff set by the domestic government for
the purpose of dissuading firms from moving abroad. We show the tariff to be positive and welfare improving
when foreign countries wage levels are close to domestic levels. Hence we identify an instance where an import
tariff is welfare improving but for a reason quite different than the traditional ‘foreign rent-extracting’ motive as
shown under Brander and Spencer (1981). In our paper, the tariff is welfare improving because it deters outward
FDI and it prevents union rent from being lost.
We qualify this result further by showing that once you allow for large differences in wages between the
home and the foreign country (with the foreign location having the lower wage), that outward FDI becomes
welfare improving for the unionized domestic country and the optimal tariff policy in that case is zero. Or in
other words, in a world of footloose firms and unionised labor markets, domestic countries are more inclined to
set positive tariffs against foreign countries with relatively similar wage levels, while they would prefer free trade
with low wage foreign countries. The underlying reason is that when a domestic firm moves to a country with
only slightly lower wages, this leaves the domestic country worse off. The domestic firm’s extra profits from
moving abroad, do not make up for the loss in domestic union rent as a result of the jobs and wages lost
domestically. In contrast, when the domestic firm moves to a very low wage country, the extra profits repatriated
home will be much larger and therefore compensate for the loss in union rent, leaving the domestic country
better off after the outward FDI. Hence, trade liberalisation only lowers home welfare when domestic firms
would move to countries with small wage differences. Trade liberalisation improves home welfare if it
encourages firms to move to countries with large wage differences compared to the home country.
These results only apply under unionization of the domestic labor market. In the absence of unionization, the
domestic government has no incentive to intervene with firms’ private location decisions and the optimal tariff
will be zero. This non-unionization scenario will be discussed further in section 5.
Another result derived in our analysis is that the optimal tariff schedule is very different for footloose firms
compared to firms with fixed locations. We find that in an integrated world where firms can move freely, optimal
tariffs are bound to be lower than under fixed firms’ locations. And finally, in contrast to common wisdom, our
findings suggest that a negative correlation between the import tariff and the amount of inward FDI is possible.
Empirical work has already indicated that the correlation between FDI and tariffs need not be a positive one
(Blonigen 1998, Belderbos, 1997) but this has not received a lot of attention in the theoretical literature thusfar.
Our model gives a theoretical explanation as to why this may be and this will be discussed more in detail in
section 5 of the paper.
as hosts of vertical production mainly because of their geographical proximity to western Europe and the
production cost difference.
2
The rest of the paper is organised as follows. The next section discusses the model. Session 3 looks at the
results under free trade. In section 4 we introduce the domestic trade policy. Section 5 discusses the overall
results and their robustness and section 6 concludes.
2. The Model
In a two-country model, the home market is supplied with a homogeneous product by a Cournot duopoly
consisting of a home firm and a foreign firm. The home firm, labelled as firm one, is owned by shareholders in
the home country while the foreign firm, labelled as firm two, is owned by shareholders in the foreign country.
Each firm has to decide where to locate production. The wage for the imperfectly competitive sector in the home
country is set by a monopoly union that maximises union rent, while the labor market in the foreign country is
perfectly competitive. 6,7 All consumption of the oligopolistic product occurs in the home market where the
market price, P, is given by the linear inverse demand function: P      q1  q2  where q1 is the output of
firm one and q2 is the output of firm two. Total production in the home country is q H while total production in
the foreign market is qF so qH  qF  q1  q2 . The labor required to produce one unit of the oligopolistic
product is a H in the home country and aF in the foreign country. The wage set by the monopoly union in the
home country is wH while the competitive wage is wH in the home country and wF in the foreign country.
Hence, the unit labor cost of the oligopolistic industry is H  aH wH in the home country and F  aF wF in the
foreign country.8 Unit labor costs in the two counties may differ due to differences in productivity and/or wage
rates.
Although our model has a strong partial equilibrium flavour, we implicitly assume the existence of a
perfectly competitive numeraire sector that pays the competitive wage rate. The numeraire sector in the domestic
country is assumed to absorb labor freed up in the imperfectly competitive sector at the competitive wage rate
that is equal to one. This assumption implies that there is no unemployment in our model. 9 Trade in the
numeraire good ensures that trade between the two countries is balanced.
In terms of the labor market, we assume that domestic workers in the oligopolistic sector belong to a
monopoly union that sets the wage rate to maximise union rent while foreign workers are not unionised and are
paid the competitive wage rate.
6
Booth (1995) argues that unions are more likely to bargain over wages than over employment. That is why in
our model union only engage in wage setting, the employment decision is left to the firm.
7
Papers by Leahy and Montagna (1999) and Zhao (1998) have shown that the degree of centralisation at which
wage bargaining takes place may affect the incentives for FDI. In order to avoid the issue of centralisation in the
case of bargaining we assume a monopoly union. Since we look at countries with asymmetric labor markets
institutions, the monopoly union amplifies the labor market asymmetry.
8
Although we do not explicitly model transport costs, they are implicitly present in the analysis since the foreign
unit labor cost could include transport costs without affecting the analysis.
9
Unemployment could easily be incorporated in the model. In the absence of a numeraire sector, the outside
wage for workers in the imperfectly competitive sector would be zero instead of one.
3
As usual in the literature on trade under imperfect competition, we assume quasi-linear preferences for the
domestic consumers so that income effects are zero and hence consumer surplus is a valid welfare measure.
Relaxing this assumption and allowing for income effects could lessen or strengthen the results we get but would
not change the direction of the results. This assumption implies that domestic workers’ surplus and shareholders
profits do not affect total demand. This is equivalent to assuming that the number of domestic workers and
shareholders in the oligopolistic industry is small compared to the total number of consumers.
The structure of the model is as follows: in stage one, the home government sets a tariff to maximise the
social welfare of the home country including home firm’s profits, consumer surplus and tariff revenue. Then, in
the second stage, the two firms each decide either to locate in their own country or to locate production in the
other country. Wherever each firm decides to locate, both the home and the foreign country involve an identical
fixed cost which we will not explicitly take on board in the analysis but which secures that production will only
take place in one location (either the home country or the foreign country). After the firms have committed to
their locations, in the third stage of the model, the monopoly union in the home country sets its wage to
maximise its union rent.10
In the final stage of the game, the two firms compete as Cournot duopolists in the relevant product market,
taking as given the tariff (t), the union wage ( wH ) in the home country and the fixed cost (F) which is equal in
both the home and the foreign country. As costs depend upon location, the profits of the two firms will depend
upon where the firms are located. Hence, the profits of the two firms are:
 P  H  q1  F
 P  F  t  q1  F
1  
 P  F  t  q2  F
2  
 P  H  q2  F
if firm 1 locates in the home country
if firm 1 locates in the foreign country
(1)
if firm 2 locates in the foreign country
if firm 2 locates to the home country
Each firm independently and simultaneously sets its output to maximise its profits, and this yields the firstorder conditions for profit maximisation. Since demand is assumed to be linear, it is straightforward to solve for
the Cournot equilibrium outputs and market price in the four possible outcomes of the final stage of the model:
10
A paper by Zhao (1995) has reversed the location and the union stage used in this paper. By allowing firms to
choose location after the union sets wages, they show that ‘the threat of relocation’ lowers the wages bargained
over in the union stage.
4
Equation (2)
Firm 1 locates in home (H) country
Firm 1 locates in foreign (F) country
Firm 2 locates in the
q1HF    2H  F  t  3
q1FF    F  t  3
foreign (F) country
q2HF    H  2F  2t  3
q2FF    F  t  3
P HF    H  F  t  3
P FF    2F  2t  3
Firm 2 locates in the home
q1HH    H  3
q1FH    H  2F  2t  3
(H) country
q2HH    H  3
q2FH    2H  F  t  3
P HH    2H  3
P FH    H  F  t  3
At the third stage of the game, the monopoly union in the home country sets its wage to maximise union rent
given the locations of the firms. Union rent is equal to the total wage premium paid to union members:
U   wH  wH  lH , where employment in the oligopolistic sector is lH  aH qH . This means we can write union
rent as U  H  H  qH . The competitive unit labor cost in the home country is H  aH wH . When both firms
are located in the home country, the monopoly union is in a strong position and wages will exceed the
competitive wage. Whereas if no firms locate in the home country then it has no monopoly power whatsoever
and the domestic wage is the competitive wage of the numeraire sector which we normalise to 1. Maximising
union rent with respect to the unit labor cost yields the following first-order condition:
q
U
 qH  H  H  H  0
H
H
(3)
Since the Cournot equilibrium outputs (2) are linear functions of the unit labor cost it is straightforward to
solve this first-order condition in the four possible outcomes for the monopoly unit labor cost in the home
country:
Equation (4)
Firm 1 locates in home country (H)
Firm 1 locates in foreign country (F)
Firm 2 locates in the foreign
HHF    2H  F  t  4
 HFF  H
HHH    H  2
HFH    2H  F  t  4
country (F)
Firm 2 locates in the home
country (H)
Having solved for the Cournot equilibrium outputs, market price, and the monopoly unit labor cost, (2) and
(4), it is now possible to solve for the maximised profits of the two firms in the four possible outcomes:
5
Equation (5)
Firm 2 locates in
the foreign country
(F)
Firm 2 locates in
the home country
(H)
Firm 1 locates in home country (H)
1HF 
 2HF 
  2H  F  t 
36
2
F
 5  2H  7F  7t 
 2HH 
   H 
36
F
 2FF 
2
36
   H 
1FF 
2
144
1HH 
Firm 1 locates in foreign country (F)
F
1FH 
F
 2FH 
2
  2F  t 
2
F
9
   F  t 
9
2
F
 5  2H  7F  7t 
2
144
  2H  F  t 
36
F
2
F
The equilibria in the top left cell and the bottom right cell of equation (5) are very similar in terms of
payoffs. In both equilibria there is one firm in each market. The only difference is that in the top left cell, each
firm locates in its own market, which we label the NO FDI equilibrium, whereas in the bottom right cell their
positions are swapped and the home firm locates in the foreign market while the foreign firm locates in the home
market. Technically speaking there is nothing in the model that prevents the equilibrium in the bottom right cell
to prevail, but more intangible aspects like information asymmetries between countries make the NO FDI
equilibrium with each firm in its own market arguably more likely. In any case, in what follows, we want to
focus our discussion on the welfare effects of OUTWARD and INWARD FDI compared to the NO FDI situation,
where the home firm is located at home and the foreign firm is located in the foreign market. We want to use the
NO FDI equilibrium as a benchmark and than look at what happens to welfare under OUTWARD FDI (top right
cell), where the home firm moves abroad to serve the home market from a distance and the INWARD FDI
equilibrium (bottom left cell), where the foreign firm locates inside the home market. The equilibrium in the
bottom right cell is of less interest to us and will not be discussed in what follows.
In the next section we will examine the free trade conditions under which the NO FDI equilibrium occurs
and compare it to the conditions for OUTWARD FDI and for INWARD FDI to occur. In section 4 we will see
how domestic trade policy alters the incentives for firms to engage in FDI.
3. Free Trade
In this section we analyse firms’ private location decisions under free trade (t=0). From equation (5), we can
see that OUTWARD FDI is a Nash equilibrium if 1FF  1HF and  2FF   2FH but, since 1FF   2FF and
1HF   2FH , these two inequalities are equivalent. Hence, OUTWARD FDI is a Nash equilibrium if unit labor
costs in the foreign country are sufficiently low: F  A    2H  3 . Let us label the foreign wage cost below
which the domestic firm wants to move abroad under free trade as A. Foreign unit wage cost in A in fact
corresponds to the monopoly unit labor cost in the home country.
6
NO FDI is a Nash equilibrium if 1HF  1FF and  2HF   2HH ; hence, NO FDI is a Nash equilibrium if A
 F  B   3  4H  7 . Let us label this other critical foreign wage cost as B. We can than say that for
foreign wage unit costs lying between A and B, neither of the firms wants to leave their market.
INWARD FDI is a Nash equilibrium if 1HH  1FH and  2HH   2HF but, since  HH   2HH and
1FH   2HF , these two inequalities are equivalent; hence, INWARD FDI is a Nash equilibrium if unit labor costs
in the foreign country are sufficiently high, exceeding the critical foreign wage cost B ( F  B).
Proposition 1: OUTWARD FDI by the home firm occurs if F  A;
NO FDI occurs if A  F  B and
INWARD FDI by the foreign firm occurs if F  B.
Proposition 1 indicates that that there are two critical values of F , the foreign unit wage cost which we labelled
A and B where A < B. For foreign wage costs below A, both firms want to be in the low cost foreign location and
serve the domestic market from a distance. This is the OUTWARD FDI equilibrium. For foreign wage costs
between A and B, each firm wants to locate in its own market Or in other words, the incentives for the domestic
firm to move abroad stop to exist for this range of foreign wage costs. This is the NO FDI regime. And finally
for relatively high foreign wage costs beyond B, both firms want to locate inside the domestic market. This is the
INWARD FDI equilibrium. With INWARD FDI, the foreign firm realises that if it locates in the home country
then the monopoly unit labor cost in the home country will increase as INWARD FDI increases the power of the
monopoly union. Therefore, INWARD FDI will not occur when the monopoly unit labor cost in the home
country is just less than the foreign unit labor cost. Only when the monopoly unit labor cost in the home country
is significantly lower than the foreign unit labor cost will the foreign firm have an incentive to locate there. Thus,
there is a range of foreign wage costs where NO FDI occurs.
Let us now analyse the effects of OUTWARD and INWARD FDI on total domestic welfare and the welfare
components under free trade when t  0 and illustrate this graphically. In our description of welfare and its
components we will use the NO FDI regime that holds between foreign unit wage cost A and B, as a point of
reference. The jumps in variables that occur at the regime switches A and B will be interpreted as the effects of
INWARD and OUTWARD FDI. The analytical results that are presented are entirely general for the particular
functional form of linear demand, but in order illustrate the results we will draw figures with the foreign unit
wage cost on the horizontal axis that use particular parameter values (   10 ,   1 , H  1 , aH  aF  1 , and
F  0 ). The corresponding domestic wage level in each of the equilibria described above can be found by
looking at the corresponding scenarios listed in equation 4.
7
We start with total domestic welfare under free trade (FT) which is illustrated in figure 1. Domestic welfare
is the sum of domestic consumer surplus, profits of firm one and union rents WH  V ( P)  1  U . A discussion
of each of these welfare components follows below. First we turn to total domestic welfare which can be shown
to be:
 2    F  2 9   F

2
2
  7  2H  5F  5   2 H   F 


F
288
72 

    2 4  F
H 

WHFT
if F  A
if A  F  B
(6)
if F  B
The bold lines in Figure 1 show welfare of the home country as a function of the foreign unit labor cost
under free trade. In general, home welfare is continuous and strictly decreasing in the foreign unit labor cost until
point A. At that point home welfare is discontinuous and jumps up from 4   H  27   F under
2
OUTWARD FDI to 2   H  9   F under NO FDI. At point B home welfare again jumps up from
2
263   H  1176  F under NO FDI to   H  4   F under INWARD FDI.
2
2
Let us now take NO FDI as a point of reference. We can than say that OUTWARD FDI in the proximity of A
reduces home welfare compared to the NO FDI scenario while INWARD FDI increases home welfare. From the
discussion of the welfare components below we will see that the reduction in home welfare at the switching point
A, is solely caused by the reduction in union rents. Or in other words, when the footloose domestic firm decides
to locate abroad, the domestic country is worse off because the profits repatriated from the foreign location do
not outweigh the loss in union rent that results from the jobs lost in the domestic country with the decision of the
footloose firm to locate elsewhere.
However, when foreign wages are small enough, that is well below A , OUTWARD FDI can improve
domestic welfare. This can be seen in Figure 1 for levels of the foreign wage close to the origin. The reason is
that when domestic firms move to very low wage countries, the profits repatriated by domestic firms will now
more than compensate for the union rent lost. Therefore we expect domestic government intervention to differ
depending on the wage level of the country the domestic firm wants to move to. The next section will show that
the home government will want to affect the home firm’s private incentive to locate abroad whenever the
efficiency gain (lower wages) from locating abroad is lower than the union rent lost.
The dotted lines in figure 1 indicate the changes in domestic welfare in the case of optimal domestic trade
policy, which will be discussed in more detail in section 4.
Let us now look at the various components of domestic welfare starting with domestic profits that are
illustrated in figure 2. The profits of the home firm under free trade (FT) for the three FDI regimes can be shown
to be:
8

FT
1
  F 2 9  F

2
   2H  F  36  F

2
  H  36  F
if F  A
if A  F  B
(7)
if F  B
The bold lines in Figure 2 show us the profits of the domestic firm as a function of the foreign labor cost in
the case of free trade for the particular parameter values. While home profits are decreasing in the foreign wage
cost up to point A, they are continuous at point A where 1FT  4   H  81  F . Hence, OUTWARD FDI
2
has no effect on the profits of the home firm. In the other regime switch that takes place at B that divides the NO
FDI regime from the INWARD FDI regime, there is a discontinuity in the home firm’s profits which can be seen
in figure 1. This suggests that when the foreign firm jumps into the domestic market, the profits of the home firm
fall from 25   H  441  F under NO FDI down to   H  36  F under INWARD FDI. The reason
2
2
is that the home wage rate with INWARD FDI is higher than in the NO FDI regime when there is only one firm
in the home country, which lowers home profits.
The dotted lines in figure 2 indicate how the domestic firm’s profits change in the presence of an optimal
domestic trade policy which will be discussed more in detail in section 4.
Consumer surplus in the domestic country under free trade (FT) in the three FDI regimes can be shown to
be:
V
FT
 2    F  2 9 

2
  7  2H  5F  288

2
2   H  36
if F  A
if A  F  B
(8)
if F  B
Consumer surplus as a function of the foreign unit labor cost for the particular parameter values is
continuous at point A where V FT  8   H  81 . Hence in comparison to the NO FDI regime, OUTWARD
2
FDI has no effect on market price or consumer surplus. There is a discontinuity at point B where consumer
surplus drops from 289   H  3528 under NO FDI down to   H  18 under INWARD FDI. As
2
2
explained above, INWARD FDI increases the home wage level which in turn results in higher domestic price
level. This explains the reduction in home consumer welfare going from the NO FDI to the INWARD FDI
regime.
Union rent in the home country under free trade can be shown to be:
U FT
0

2

   2H  F  24

2

  H  6
if F  A
if A  F  B
(9)
if F  B
The bold lines in Figure 3 shows the home union rents as a function of the foreign labor cost under free
trade for the particular parameter values. In the OUTWARD FDI regime, none of the footloose firms are located
in the domestic market and union rents in the imperfectly competitive domestic sector are zero. At point A,
9
beyond which the NO FDI regime starts with the domestic firm located in the domestic market, union rent jumps
up from 0 under OUTWARD FDI up to 2   H  27  under NO FDI. Figure 3 shows a discontinuity at
2
point A. At point B, beyond which also the foreign firm locates in the domestic market, Union rent jumps up
further from 25   H  294 under NO FDI to   H  6 under INWARD FDI. If we now take the NO
2
2
FDI regime as our reference point, we can say that going from a situation where two countries each have their
own firm located in their market, OUTWARD FDI by the domestic firm to the foreign location reduces union
rent in the domestic country. In contrast, going from a NO FDI situation where each country has its own firm
within their national borders to a situation where the foreign firm engages in INWARD FDI in the domestic
country, increases union rent. The domestic firm locates production facilities abroad when unit labor costs
abroad are lower than in the home country. This OUTWARD FDI results in a loss of domestic union rent. It will
become clear in the next section that the home government will want to affect the home firm’s private incentive
to locate abroad whenever the efficiency gain (lower wages) from locating abroad is lower than the union rent
lost.
The dotted lines in figure 3 indicate the changes in union welfare under the optimal domestic trade policy
that will be discussed in more detail in section 4.
The profits of the foreign firm under free trade can be shown to be:
 2FT
  F 2 9  F
if F  A

2

  5  2H  7F  144  F if A  F  B

2
  H  36  F
if F  B


(10)
In general, the profits of the foreign firm are continuous at point A where  2FT  4   H  81  F and
2
at point B where  2FT    H  36  F . Hence, in comparison to the NO FDI benchmark, both OUTWARD
2
and INWARD FDI have no effect on the profits of the foreign firm.
Welfare in the foreign country under free trade is just equal to the profits of the foreign firm. Aggregate
world welfare is therefore given by  FT  WHFT   2FT . Since the profits of firm two are continuous, the jumps in
world welfare are due to the jumps in home welfare. World welfare falls with OUTWARD FDI and increases
with INWARD FDI.
The results of this section for OUTWARD FDI under free trade are summarised in the following proposition:
Proposition 2: OUTWARD FDI under Free Trade, reduces union rent and
consequently reduces the welfare of the home country and aggregate world welfare.
10
Since OUTWARD FDI occurs when the foreign unit labor cost is just equal to the monopoly unit labor cost
in the home country (point A), it has no effect on the marginal cost of firm one so does not affect the profits of
firm one or the market price. Hence, the only effect is to eliminate the union rent and thereby to reduce the
welfare of the home country.11 Aggregate world welfare falls because the actual costs of producing in the home
country are lower than in the foreign country but the union wage drives firm one to locate production in the
foreign country.
The results of this section for INWARD FDI under free trade are summarised in the following proposition:
Proposition 3: INWARD FDI by the home firm increases union rent, reduces the profits of
the home firm and consumer surplus. INWARD FDI by the foreign firm increases the
welfare of the home country and aggregate world welfare.
4. Optimum Tariff
The private location decisions of firms under free trade are not always in the interest of the domestic country
as measured by the sum of profits of the home firm, consumer surplus and union rents. Proposition 2 clearly
indicated that under free trade, outward FDI results in a domestic welfare decrease, as illustrated in figure 3.
Now, we turn our attention to the first stage of the model where we assume the home country to pursue an
optimal trade policy. Consider the optimum tariff of the home country assuming that the foreign country
passively pursues a policy of free trade. Since the tariff will affect the firms’ location decisions, the optimum
policy will be rather complex and will involve comparisons of welfare in different regimes. The situation is best
described in figure 4 where every possible value of the tariff (t) is shown on the vertical axis against the foreign
unit labor cost ( F ) on the horizontal axis. The bold line segments in figure 4 show the optimal trade policy by
the domestic government. The derivation of these bold line segments will be discussed below.
For now concentrate on the thin full lines shown in figure 4, labelled OO and II. In the (t, F )-space these
lines divide the regions where different FDI regimes rule. This as opposed to the line segments in the free trade
case. In the region below the OO curve, which has a slope of minus one, there will be OUTWARD FDI,
F  A  t , while in the region above the II curve, which also has a slope of minus one, there will be INWARD
FDI, F  B  t . In the region in between the lines OO and II there will be NO FDI. And, this holds for the
range of foreign wages lying in between A  t  F  B  t .
11
Unions are not the only mechanism that can drive this result. Other distortions in the labor market like
efficiency wages which result in wages levels above the marginal product of labor would also lead to the result
described in proposition 2.
11
In the region with OUTWARD FDI (below OO), both firms are located in the foreign country so imports are
equal to qF  q1  q2  Q . With OUTWARD FDI, the rent of the home monopoly union is equal to zero since
there is no production in the home country. Hence, the welfare of the home country is given by the sum of
consumer surplus, the profits of the home firm, and the tariff revenue:
WHFF  V  P   1FF  tQ
(11)
Maximising the welfare of the home country with respect to the tariff yields the first-order condition:
WHFF
q
P
Q
 q2
  P  F  1  t
Q  0
t
t
t
t
(12)
Using the result from (2) to solve this first-order condition yields the optimal tariff t FF  0 . Hence when the
foreign unit labor cost is low, the optimal policy for the home government is to set a zero tariff and to allow
OUTWARD FDI to occur. The optimal policy of a zero tariff for low foreign unit labor costs is indicated in
figure 4 by the bold line segment between the origin and point C (see below) along the horizontal axis.12 The
reason why the domestic government allows the domestic firm to locate in the low wage market abroad is that
while union rent is lost, the efficiency gain to the domestic firm in terms of higher profits outweighs this loss.
However, as the foreign unit labor cost increases and we move along the horizontal axis in figure 4, the
efficiency gain to the home firm from locating abroad is reduced. We will refer to the critical value of foreign
unit labor cost where repatriated profits exactly outweigh the loss in union rent as C. Hence up to

 
 
F  C  38  4 82   4 82  16 H
22 , optimal tariff policy is zero. At this level of foreign unit labor
cost, the home firm still has a private incentive to locate abroad (C < A) but location abroad is not in the home
country’s interest. Therefore, the home government will set a positive tariff to make sure that it is no longer
profitable for the home firm to locate in the foreign country.
In order to prevent OUTWARD FDI beyond point C, the domestic government will have to set a tariff higher
than that given by the line OO. Or in other words the domestic government needs to set a tariff that is high
enough to secure the NO FDI equilibrium in order to dissuade the home firm from moving abroad. To derive the
optimal tariff policy, we first consider what is the home welfare in the interior of the region with NO FDI, (in
between OO and II). The home welfare now includes home union rent and is given by:
WHHF  V  P   1HF  U HF  tq2  V  P    P  H  q1  tq2
12
(13)
This result is in part assumption specific. It depends on the linearity of the inverse demand curves and the fact
that there is one domestic and one foreign firm. It can be shown that when demand is convex, the optimal tariff is
positive while for concave demand, the optimal tariff is negative (equivalent to an import subsidy). The size of
the optimal tariff also depends on the number of domestic versus the number of foreign firms that are located
abroad. In a model with more than two firms, the size and sign of the optimal tariff depends on the relative
number of foreign firms. However, despite the assumption specific nature of the zero tariff result here, it remains
true that the optimal tariff will always be lowest for low foreign wage countries.
12
where 1HF  U HF   P  H  q1 . Maximising the welfare of the home country with respect to the tariff yields the
first-order condition:
WHHF
q
P
Q
 q2
  P  F  1  t
Q  0
t
t
t
t
(14)
Thus, solving this first-order condition using equations (2) to (5) yields the optimal tariff prevailing in the
NO FDI area lying in between the lines OO and II:
t HF 
1
15  2H 13F 
41
In figure 4, this optimal tariff in the NO FDI region between the lines OO and II is given by the line TT.
This line corresponds with the optimal rent extracting tariff derived by Brander and Spencer (1981) when firms
have fixed locations. In figure 4, the NO FDI area is the only area where firms do not have an incentive to move.
Hence, the part of the line TT that lies in the NO FDI region is the optimal tariff for values of the foreign unit
labor cost greater than C. The optimal tariff is indicated with a bold line segment along the TT-line in figure 4.
However, for values of the foreign unit labor cost greater than D, F  D   9  89H  98 , the optimal
tariff changes. Beyond the foreign wage indicated by D, the TT line enters the INWARD FDI regime. This means
that should the domestic country continue to use the Brander and Spencer (1981) tariff, the foreign firm would
engage in tariff jumping and would locate in the domestic country. The question is whether this results in optimal
domestic welfare. In order to answer that we now have to compare domestic welfare under NO FDI with welfare
under INWARD FDI. It turns out that for values of the foreign unit cost between the foreign wage level D and E
in figure 4, D  F  E where D   9  89H  98 and E   53  143H  196 , the NO FDI regime with a
tariff that just prevents the foreign firm from engaging in INWARD FDI is optimal. This is achieved through a
tariff just  below the line II. Again this is indicated in figure 4 by a bold line segment for foreign wages lying
between D and E, just below the II-line. By setting this tariff the home country continues to keep the domestic
firm inside its borders and at the same time ensures that the foreign firm does not want to tariff jump. This leads
us to conclude that for this range of the foreign unit costs the home country is better off continuing to extract rent
from the foreign firm through a tariff than through higher union rents if the foreign firm had jumped in the
domestic market.
However, once the foreign unit cost level exceeds a critical value E, F  E, the optimum tariff changes.
For values of the foreign unit cost greater than E, it is optimal for the home government to set a tariff level above
the level given by the II curve. For this range of foreign unit costs, the foreign firm will jump into the home
country and the home union can extract more rent from the foreign firm than a tariff can if the foreign firm had
remained in the foreign country. This is why the home government decides to set a tariff which is high enough to
alter the foreign firm’s private incentives compared to free trade. By choosing any tariff above line II, the foreign
firm will want to locate in the home country. In Figure 4 we have assumed an ‘efficient’ home government that
sets a tariff just  above the II -line, which is indicated by a bold line segment between foreign wage levels E and
B. Point B is the level of foreign wage cost beyond which the foreign firm wants to tariff jump in the case of free
13
trade. Hence there is no more need for the domestic government to set a positive tariff beyond point B since
foreign firms will tariff jump anyway. Hence the optimal tariff turns zero which is indicated in figure 4 by the
bold line along the horizontal axis for all wages higher than B. The optimal tariff policy derived above can be
summarised in the following proposition:
Proposition 4: The optimum tariff is:
t  0 for F  C;
t  15  2H  13F  41
for C  F  D;
t   3  4H  7F  7  
for D  F  E;
t   3  4H  7F  7
for E < F  B and
t =0
for F  B
In order to illustrate the welfare effects of the optimal trade policy by the domestic government we have
indicated the changes to domestic welfare, domestic profits and union rents in figure 1, 2 and 3 respectively. The
dotted lines show the changes compared to free trade. From figure 1 we see that between foreign wage cost C
and B, domestic welfare under optimal trade policy is higher than under free trade. The sharp drop in welfare as
a result of OUTWARD FDI under free trade has been avoided. The reason is that the import tariff by keeping the
domestic firm inside the home borders avoids the loss in union rent. In addition, the tariff revenue results in
higher welfare levels in the NO FDI regime than was previously the case. The results can also be reversed.
Going from a situation of optimal trade policy to free trade we can say that trade liberalisation is bound to lower
domestic welfare, not just because of the tariff revenue that is lost but because OUTWARD FDI by domestic
firms will result in the loss of union rents.
From figure 2 we see that between foreign wage cost C and B, domestic profits are lower than what they were
under free trade. The reason for this is twofold. Between C and E, the home firm can no longer realise its private
incentive to move abroad. It is forced by the import tariff to stay in the NO FDI regime which implies lower
profits than when it could have produced at cheaper costs abroad. Beyond foreign unit wage cost E, the tariff is
set in order to invite the foreign firm to locate in the home market which triggers the INWARD FDI regime.
There again home profits are lower than in the free trade case because INWARD FDI raises home wages. In
figure 3 we see that union rents, in contrast to home profits, are always higher under the optimal trade policy
between C and B. The reason is that the government keeps the domestic firm inside the national borders, where
under free trade it was more profitable to move abroad up to point A. Between A and B under free trade the NO
14
FDI regime ruled with just one firm located in the home country. But under domestic trade policy, the INWARD
FDI regime rules between A and B. With two firms located in the home market, the union is better off.
5. Discussion of Results
The results we have obtained are largely driven by the presence of unions in the domestic country. In the
absence of a union, the results would be far less interesting. Consider what would happen then. First of all the
NO FDI regime would disappear. The reason is that firms in the industry deciding on location under free trade
will now simply compare the exogenous wage level at home with the wage level abroad and go where it is
cheapest. In other words as soon as the domestic wage would be lower than the foreign wage both firms will
locate in the home market (INWARD FDI), and conversely when the foreign wage is lower than the domestic
wage, both firms will locate abroad (OUTWARD FDI).
Optimal trade policy would not change this result since what is good for the home firm in terms of lower wage
costs and higher profits is good for the home country as a whole. Therefore, the domestic government would not
interfere with firms’ private location decisions.
The incentive for government intervention in our model is driven by the presence of a domestic union. We
have shown that with a union present, a trade-off arises in the case of OUTWARD FDI. On the one hand the
domestic firm will still repatriate higher profits from abroad if it decides to leave the country but on the other
hand there is a negative social externality for the domestic country namely the loss of jobs and wages. As long as
the benefits from OUTWARD FDI for the country as a whole do not outweigh the costs, the home government
will use an import tariff to dissuade the home firm from moving abroad. Our model shows that such a tariff
raises home welfare.
A description of the model’s main results follow. When a domestic firm is considering location in a very
low wage country (up to C in figure 4), our model suggests that this firm is likely to face free trade when
shipping its goods back to the home market. The domestic government allows it to locate abroad because the
additional profits it earns abroad outweighs the union rent lost in the domestic country. Empirically this implies
that FDI from the ‘North’ towards the ‘South’ should be observed. As long as the wage difference between the
foreign location and the home country is large enough, the home government does not have an incentive to affect
that location behaviour.
For an intermediate range of foreign wage levels (between C and E in figure 4), the domestic government
opts for a positive tariff that actually serves three different purposes. On the one hand, the tariff on imports is
meant to prevent the domestic firm from locating abroad and serving the domestic market from a distance. In
addition, the tariff is aimed at rent-extraction from the foreign firm. And at the same time the tariff is set low
enough in order not to trigger tariff jumping by the foreign firm. For foreign wage costs between C and E, the
increase of the domestic firm’s profits from locating abroad can not overcome the loss in union rent.
Our model is an illustration of a situation where the domestic firm looses from domestic import protection
while the union gains. This may seem counterintuitive at first since protection against imports is often regarded
15
in the interest of domestic producers (see for example Hilman and Ursprung, 1993). But here we show that when
the domestic firm is a footloose firm, this need not be the case. Arguably, our model could be used to understand
why the opening of Central-Europe did not initially induce many West European firms to locate assembly
activities there. The many antidumping duties in place on imports from Central-Europe may explain the less than
expected location of EU manufacturing activities there. While wage differences were in principle large enough
for EU firms to set up production in Central-Europe and to serve the EU form a distance, the use of EU
antidumping tariffs may have prevented that outward FDI of West-European firms. The same argument may
apply to Mexico for US firms before NAFTA.
Third, the results in the previous section also indicate that for imports from relatively high wage countries
(wage levels above E in figure 4), the optimal domestic trade policy is to set a tariff such that tariff-jumping by
the foreign firm into the domestic market will occur. The reason is that by attracting foreign investors, union
rents will go up. Both wages and employment in the domestic country will increase as a result of INWARD FDI.
Empirically, the positive effect of investment flows on wages has been confirmed amongst others by Aitken,
Harrison and Lipsey (1995).
From a foreign wage of E onwards, foreign unit costs are relatively similar to home unit costs. This implies
that the cost difference between the home and the foreign firm becomes too small for rent-extraction through an
import tariff to be interesting. A tariff to encourage tariff-jumping is more interesting because the increase in
union rent as a result of INWARD FDI is now higher than the amount of rent the domestic government could
have extracted through an import tariff. This confirms that the high import tariffs and duties by the US and EU
against a country like Japan, may have been set in order to attract inward Japanese FDI both in the EU and US
(Veugelers and Vandenbussche, 1999).
Our findings may also explain why a positive relationship between tariff levels and tariff-jumping FDI is
difficult to establish empirically (Blonigen (1998); Belderbos (1997)). Our model shows that a negative
relationship between tariffs and inward FDI is possible. We find that high tariffs will occur in the regime with
NO FDI (see figure 4, for foreign unit costs between C and D) whereas low tariffs will occur in the regime with
INWARD FDI (see figure 4, for foreign unit costs higher than E). Therefore our model suggests that a negative
correlation between tariffs and inward FDI may exist. Note also that for foreign wage costs above E, the ‘threat’
of protection is sufficient to trigger inward FDI. Actual protection in this case will never be observed.
Nevertheless, it is the threat of the import tariff that persuades the foreign firm to locate in the home market. This
suggests that empirical studies trying to establish a link between protection and FDI should use indicators for the
‘threat’ of protection or a measure of a ‘government’s willingness’ to take protection, rather than actual tariffs or
duties.
According to the traditional view, FDI and Trade are substitutes since the relocation of production facilities
abroad was generally thought to correspond with a reduction in the trade flows. Our findings suggest however,
that trade and FDI may be complements since when outward FDI occurs in our model, trade follows FDI. After
the home firm has located abroad a trade flow arises that previously was not there. Recent empirical evidence
also goes in the direction that FDI need not necessarily reduce trade flows (Pain and Wakelin, 1998).
16
6. Conclusions
In this paper we argue that country level differences in labor market institutions can result in free trade location
patterns that are detrimental for domestic welfare, compared to a situation with fixed firm locations. This induces
domestic governments to use import tariffs as a means to affect firms’ private location decisions and to increase
domestic welfare compared to free trade. This paper shows that the presence of unions in the domestic country
is what drives the domestic government to use optimal trade policy. The reason is that if a domestic firm decides
to move abroad to serve the domestic market from a distance, this decision results in the loss of domestic jobs
and wages. Only when the repatriated profits from moving abroad are large enough to compensate the domestic
country for the loss in union rent, the home government will not intervene. But in many cases moving abroad,
while profitable for the domestic firm involved, reduces welfare for the domestic country as a whole. In that
case, an import tariff used by the home government to dissuade its firm from moving abroad raises home
welfare. The tariff motive discussed in this paper is very different from the ‘foreign rent-extracting motive’
discussed by Brander and Spencer (1981).
In the absence of unions, governments would not intervene, because in that case what is good for the home firm
in terms of lower wage costs and higher profits, is good for the home country as a whole.
We have derived the optimal tariff schedule for all wage levels of the foreign country and find the optimal
domestic tariff to be very different from the one under fixed locations. In an integrated world with footloose
firms, tariffs are always lower or equal to the tariff policy in the absence of relocation possibilities. And finally
we have shown that tariffs and FDI can be negatively correlated. This leads to a better understanding of some of
the empirical work that has failed to find a positive relationship between tariffs and FDI.
One final remark is in place namely that our analysis should not be seen as a stance for trade policy
intervention. By indicating the unilateral incentives that countries have to deviate from free trade we show that a
continued effort in multilateral talks at the level of the WTO is called for to refrain countries from pursuing tariff
policies.
17
Figure 1
Domestic
welfare
A  (  2 H ) / 3
B  (3  4 H ) / 7


C  (38  4 82 )  (4 82  16) H / 22
D  (9  89 H ) / 98
E  (53  143 H ) / 196
Welfare
under
optimal trade policy
C
E
A
B
Foreign wage
Welfare under
free trade
18
Figure 2
Profits of
home firm
A  (  2 H ) / 3
B  (3  4 H ) / 7


C  (38  4 82 )  (4 82  16) H / 22
D  (9  89 H ) / 98
E  (53  143 H ) / 196
Profits of home
firm under free
trade
Profits of
home
firm
with optimal
trade policy
C
E
A
B
Foreign wage
19
Figure 3
Union
rent
A  (  2 H ) / 3
B  (3  4 H ) / 7


C  (38  4 82 )  (4 82  16) H / 22
Union rent
with optimal
trade policy
D  (9  89 H ) / 98
E  (53  143 H ) / 196
Union rent
under free
trade
C
E
A
B
Foreign wage
20
Figure 4
Tariff
A  (  2 H ) / 3
I
B  (3  4 H ) / 7

Inward FDI

C  (38  4 82 )  (4 82  16) H / 22
O
D  (9  89 H ) / 98
E  (53  143 H ) / 196
T
Optimal
rent-extracting
tariff
T
Outward FDI
O
C
D
E
A
I
B
Foreign wage
21
References
B. Aitken, A. Harrison and R. Lipsey (1995),”Wages and Foreign ownership: a comparative study of Mexico,
Venezuela and the US, NBER WP n° 5102.
R Barrell and N. Pain (1996),”An econometric analysis of US FDI”, Review of Economics and Statistics, 78(2).
P.P. Barros and X.M.Giralt (1999),”On the Effects of Antidumping legislation”, Regional Science and Urban
Economics, vol. 29, issue 1, pp 53-72.
R. Belderbos (1997), “Antidumping and Tariff Jumping Japanese firms’ DFI in the EU and US”,
Weltwirtschaftliches Archiv 133 (3), p. 419-457.
B. A. Blonigen and R.C. Feenstra (1997),”Protectionist Threats and FDI” in R.C. Feenstra (ed.), Effects of US
Trade Protection and Promotion Policies, University of Chicago Press for NBER.
A. Booth (1995), Trade Unions, Cambridge University Press.
J. Brander and B. Spencer (1981),”Tariffs and the Extraction of monopoly rents under potential entry”,
Canadian Journal of Economics, n° 3, pp.374-389.
J. Bughin and S. Vannini (1995),”Strategic Direct Investment under unionised oligopoly”, International Journal
of Industrial Organisation,13, 127-145.
T. Cordella and I. Grilo (1998),”Social dumping and relocation: is there a case for imposing a social clause ?”,
CEPR discussion paper n°1931.
S. Dowrick (1989),”Union-oligopoly bargaining”, Economic Journal, 99, pp 1123-1142.
J.E. Driffill and F. Van der Ploeg (1993),”Monopoly Unions and the liberalization with imperfect competition in
goods and labor markets”, Scandinavian Journal of Economics, 95, pp. 249-255.
R. Feenstra and G. Hanson (1996),”Globalization, Outsourcing and Wage Inequality”, American Economic
Review, 86 (2), pp. 240-245.
J. Haaland and I. Wooton (1998),”International competition for MNE investment”, paper presented at CEPR
workshop in Rotterdam June 1998.
A.L. Hillman and H.W. Ursprung (1993),”Multinational Firms, Political Competition and International Trade
Policy”, International Economic Review, vol. 34, n° 2, May.
I. Horstmann and J. Markusen (1992),”Endogenous Market Structure in international trade (natura facit
saltum)”, Journal of International Economics, 32, pp. 109-129
H. Huizinga (1992), “International Market Integration and Union Wage Bargaining”, Scandinavian Journal of
Economics, 95 (2), pp. 249-255.
P. Krugman and A. Venables (1995),”Globalization and the Inequality of Nations”, Quarterly Journal of
Economics, November, pp. 857-80.
H.P. Lankes and A. J. Venables (1996),”FDI in economic transition: the changing pattern of investments”,
Economics of Transition, vol. 4 (2), pp. 331-347.
D. Leahy and C. Montagna (1999), “Unionisation and Foreign Direct Investment”, CEPR-discussion paper n°
2260, October.
K.E. Lommerud, F. Meland and L. Sorgard (2003),”Unionised Oligopoly, Trade Liberalisation and Location
Choice”, CESIfo Working paper.
22
J. Markusen (1995),”The boundaries of MNE’s and the theory of international trade”, Journal of Economic
Perspectives, 9, pp.169-189.
J.Markusen and A. Venables (1998),”Multinational firms and the new trade theories”, Journal of International
Economics, 46(2), pp. 183-203
C. Mezzetti and E. Dinopoulos (1991), “Domestic Unionization and Import Competition”, Journal of
International Economics 31, pp.79-100.
M. Motta (1992), “Multinational Firms and the tariff jumping argument”, European Economic Review, 36, pp.
1557-1571.
R. Naylor (1999),”Union wage strategies and international trade”, Economic Journal, 109, 102-125.
A. Oswald (1985),”The economic Theory of Trade Unions”, Scandinavian Journal of Economics, pp. 161193.
N. Pain and K. Wakelin (1998), Export performance and the role of FDI, The Manchester School, pp. 5-16.
R. Rowthorn (1992), Intra-industry trade and investment under oligopoly; the role of market size, Economic
Journal, pp. 35-69.
A. Smith (1987),”Strategic Investment, Multinational Corporations and Trade Policy”, European Economic
Review, 31, pp. 89-96.
J.R. Sorensen (1993),”Integration of Product Markets when labor markets are unionised”, Recherches
Economiques de Louvain, 59, pp. 485-502.
R. Veugelers and H. Vandenbussche (1999),”European Antidumping Policy and the profitability of National and
International collusion”, European Economic Review, January 43/1, pp. 1-28.
D. Wheeler and A. Moody (1992), “International Investment Location decisions. The case of US firms”, Journal
of International Economics, 33, pp. 57-76.
L. Zhao (1998), “The impact of FDI on wages and employment”, Oxford Economic Papers, 50(2), April, pp.
284-301.
23