Department of Economics Working Paper Series
Do Labor Issues Matter in the Determination of U.S. Trade Pol-
icy? An Empirical Reevaluation
Xenia Matschke
University of Connecticut
Shane M. Sherlund
Federal Reserve Board
Working Paper 2004-36
November 2004
341 Mansfield Road, Unit 1063
Storrs, CT 06269–1063
Phone: (860) 486–3022
Fax: (860) 486–4463
http://www.econ.uconn.edu/
Abstract
Some recent empirical studies, motivated by Grossman and Helpman’s (1994)
”protection for sale” model, suggest that very few factors (none of them labor-
related) determine trade protection. This paper reexamines the roles that labor
issues play in the determination of trade policy. We introduce collective bargain-
ing, differences in labor mobility across industries, and trade union lobbying into
the protection-for-sale model and show that the equilibrium protection rate in our
model depends upon these labor market variables. In particular, our model pre-
dicts that trade protection is structurally higher than in the original protection-
for-sale model if the trade union of a sector lobbies but capital owners do not,
because union workers collect part of the protection rents; equilibrium protection
is lower if capital owners lobby but the trade union does not, because part of the
protection rents is dissipated to workers. Using data from U.S. manufacturing,
we find that collective bargaining, differences in labor mobility across industries,
and trade union lobbying indeed play important roles in the determination of U.S.
trade policy.
Journal of Economic Literature Classification: F13, F16
Keywords: Trade protection, protection for sale, labor market
We thank the editor David Card and two anonymous referees for helpful com-
ments that greatly improved this paper. We also thank Scott Taylor for his valu-
able comments and advice and Kishore Gawande and Daniel Trefler for providing
data. We are grateful to Bruce Hansen, Bob Staiger, and Gautam Tripathi for for
their helpful comments and suggestions and to Ellen Dykes and Jonas Robison for
their help with editing the language and grammar. Xenia Matschke gratefully ac-
knowledges financial support from a Deutsche Forschungsgemeinschaft research
fellowship. This paper represents the views of the authors and does not necessarily
represent the views of the Federal Reserve System, its members, or its staff.
2 MATSCHKE AND SHERLUND
1. Introduction
Lobbyists for trade and other industrial policies represent different interest groups in
society. In particular, distinguishing betwe en labor and capital lobbies is common. Labor
interests, usually represented by trade unions, often lobby for trade protection. For example,
U.S. trade unions strongly opposed NAFTA in the 1990s because of fears that freer trade
would decrease domestic employment and wage leve ls. Further, according to Baldwin (1985)
and Baldwin and Magee (2000), trade union contributions are positively correlated with the
probability that a U.S. congressman votes against trade liberalization.
The “protection for sale” model of Grossman and Helpman (1994), however, suggests
that very few factors — none of them labor-related — determine trade protection. In the
protection-for-sale model, wages are fixed and equal across industries, and there is full
employment. Only capital owners are allowed to lobby for trade policy, but even if workers,
too, were allowed to lobby, they would want import subsidies in order to benefit from lower
product prices. Hence, the GH model cannot explain why trade unions lobby for trade
protection so as to secure higher wage and employment levels.
The GH model is also at odds with the older empirical trade protection literature
(Rodrik, 1995, provides an overview) that finds that labor market considerations are an
important trade policy determinant. However, more-recent empirical studies (for example,
Goldberg and Maggi, 1999; Gawande and Bandyopadhyay, 2000; Eicher and Osang, 2002)
find strong support for the protection-for-sale model. Some of these studies also test whether
labor market variables have additional explanatory power and find them to be statistically
insignificant.
This paper shows that the conclusion that labor market variables do not influence
trade protection is misleading. The earlier papers that estimate the GH model employ the
nonlinear form of protection suggested by that model for estimation.
1
But since the GH
model has nothing to say about labor market variables, the inclusion of these variables
in empirical studies thus far has been ad hoc. The main contribution of this paper is to
show that, once labor market variables have been appropriately controlled for, statistical
methods strongly reject the null hypothesis that labor market variables are irrelevant to
trade protection.
To this end, we construct a model in the same spirit as GH but relax assumptions
about the labor market. In particular, we allow for (1) industry-specific trade unions that
bargain with capital owners over union wages and employment, (2) differences in lab or
mobility across industries, and (3) active lobbying by trade unions. Our model predicts
1
For example, in the GH model, import protection decreases with import penetration ratio and imp or t
demand elasticity when capital owners lobby, but it increases with these two variables when capital owners
do not lobby.
3
that trade protection is structurally higher than in the GH model if the trade union of
an industry lobbies but capital owners do not, because union workers collect part of the
protection rents. However, equilibrium protection is lower if capital owners lobby but the
trade union does not, because workers receive part of the protection rents. Moreover, as long
as trade protection increases the wages of at least some non-unionized workers, equilibrium
protection is lower than in the GH model even if both the capital owners and the trade
union of an industry lobby. In contrast to the protection-for-sale model without trade
union activity, the equilibrium protection rate in our model depends upon sectoral wage
and employment elasticities that, in turn, vary according to the mobility of workers across
industries.
We test our model predictions using 1983 data from U.S. manufacturing. Since our
framework nests the GH model, we can test the statistical validity of the GH restrictions.
Our major finding is that we can reject the GH model in favor of our labor-augmented
model. Consistent with our theory, we find that, compared with the GH predictions, trade
protection is indeed higher when trade unions lobby and capital owners do not, but lower
when capital owners lobby. Not only does trade protection vary according to whether capi-
tal owners of an industry lobby, but it also depends on trade union activity and differences
in labor mobility across industries. Moreover, our labor-augmented model delivers con-
siderably lower estimates of the percentage of lobbies in the population and the weight of
contributions in the governmental welfare function, both of which had been found to be
unreasonably high in previous tests of the GH model.
The remainder of this paper is organized as follows. In Section 2, we derive the
equilibrium tariffs for industries with mobile and immobile labor when trade unions bargain
with firms over wages and employment and are also allowed to lobby for trade protection.
We present the econometric model and its predictions in Section 3. In Section 4, we describe
the data, and we proceed with estimation and testing in Section 5. Finally, in Section 6,
we conclude the paper and make suggestions for future research.
2. The Model
2.1. Model Basics. In the following, we augment the GH model to allow for labor market
considerations. Consider a small country with n + 1 industries, each producing a single
good. The country has fixed endowments of labor, L, and industry-specific capital, K
i
,
where i = 1, . . . , n. Each worker (capital owner) inelastically supplies one unit of labor
(industry-specific capital).
On the consumption side, all individuals, h = 1, . . . , H, have identical quasilinear
preferences of the form U
h
= x
h
0
+
P
n
i=1
u(x
h
i
), where x
h
i
denotes h’s consumption of good
i and u is strictly concave and increasing in x
h
i
. If each individual has enough income to
4 MATSCHKE AND SHERLUND
consume all goods, quasilinearity of preferences ensures that demand of a good i = 1, . . . , n
depends only on its own price.
Let I = {0, 1, 2, . . . , n} denote the set of all industries. The numeraire industry,
i = 0, uses only labor for production according to F
0
= L
0
. The world price of good
0 is fixed at ¯p
0
, and no trade barriers are imposed on it. Each non-numeraire industry,
i = 1, . . . , n, consists of two sectors — A, which is unionized, and B, which is non-unionized
— with identical production functions. Firms in these industries employ three production
factors: capital, labor , and the numeraire good 0 as an intermediate input.
2
Each unit of the
final good i requires a fixed (but differing with i) amount of good 0 (Leontief technology).
To keep notation simple, we denote the price of the amount of good 0 required for one unit
of good i by q
i
and then write (p
i
− q
i
)F
i
(K
i
, L
i
) as value added. Unlike the intermediate
good 0, capital and labor are substitutable in the production function. Capital employed
in the sectors of any non-numeraire industry — namely, K
iA
in the unionized sector A and
K
iB
= K
i
− K
iA
in the non-unionized sector B — is immobile. In contrast, labor may
or may not be mobile across industries as discussed in the next paragraph. The reduced
production function F
i
(K
i
, L
i
) is linearly homogeneous and weakly concave, where F
i
LL
< 0,
F
i
KK
< 0, and F
i
KL
> 0.
3
We allow for differences in the interindustry mobility of labor. For simplicity, we
assume that non-union workers are either completely mobile between certain industries or
that they cannot exit their industry and workers from other industries cannot enter. If
industry i ∈ I
M
, its labor pool potentially consists of all laborers in the mobile subset of
industries. If we assume that 0 ∈ I
M
, the competitive wage must be w
i
= ¯p
0
for i ∈ I
M
.
Union workers may switch industries if i ∈ I
M
; however, they cannot be employed in the
unionized sectors of industries other than i itself.
4
If industry i ∈ I
I
, where I
M
∪ I
I
= I
and I
M
∩ I
I
= ∅, industry i’s workers are immobile and can work only in industry i.
In the unionized sector A, the capital owners bargain with the i-specific trade union,
which has N
i
members, over wages and employment.
5
In the non-unionized sector B,
2
To keep the analysis focused on the influence of labor issues on trade protection, the modeling of
intermediate goods as inputs is kept as simple as possible. We introduce them only to take into account that
firms and unions bargain over value added, not the entire value of shipments. Without this adjustment, we
would substantially and systematically underestimate union bargaining strength.
3
We omit the intermediate input as an argument in the production function. Because good 0 is a Leontief
input, we must adjust the amount of good 0 prop or tionally with F
i
when capital or labor inputs vary.
4
This assumption maintains the de facto partial equilibrium structure of the GH model, which would be
destroyed if the industry-specific trade unions also had to take into account that their members might find
employment in unionized sectors elsewhere.
5
Assuming bargaining over both wages and employment (efficient bargaining) restricts the effect of union-
firm bargaining to redistributive issues. Efficient bargaining seems a justifiable assumption because empirical
tests between this model and the competing right-to-manage model either have been inconclusive or have
