Is Trade Policy for Sale? Congressional Voting on Recent Trade Bills
Author(s): Robert E. Baldwin and Christopher S. Magee
Source: Public Choice, Vol. 105, No. 1/2 (2000), pp. 79-101
Published by: Springer
Stable URL: http://www.jstor.org/stable/30026171
Accessed: 24-08-2017 12:42 UTC
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Public Choice
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
Public Choice 105: 79-101, 2000.
79
© 2000 Kluwer Academic Publishers. Printed in the Netherlands.
Is trade policy for sale? Congressional voting on recent trade bills
ROBERT E. BALDWIN' & CHRISTOPHER S. MAGEE2*
I Department of Economics, University of Wisconsin, Madison, WI 53706-1393, E-mail:
rebaldwi @facstaff wisc.edu; 2Department of Economics, Bard College, Annandale-OnHudson,
NY 12504, E-mail: magee@bard.edu
Accepted 31 March 1999
Abstract. This paper examines voting by U.S. Representatives on the North American Free
Trade Agreement, the Uruguay Round Agreement, and most-favored nation status for China.
Using recent political economy models of trade policy to formulate an empirical specification
of congressional voting behavior, we find evidence that campaign contributions influenced
legislators' votes on the NAFTA and Uruguay Round bills. Labor group contributions were
associated with votes against freer trade while business contributions were associated with
votes in favor of freer trade. Economic conditions in each member's district as well as the
broad policy views of the legislators also affected representatives' voting decisions.
"... The bigger contributions you accept, the more expectations some
people have that they have a call on their government for something in
return".1
Senator Joseph Lieberman, October 22, 1997
1. Introduction
Conventional wisdom suggests that interest groups are buying somethin
when they contribute to a politician's campaign. These interest groups mu
be giving money to influence either the outcome of the election or the pol
decisions made by elected officials. Senator Lieberman's statement highligh
the second possibility - that campaign contributions allow interest groups
affect policy outcomes. Theoretical models in the economics and politica
science literature suggest that this quid pro quo aspect of contributions m
play an important role in the determination of public policy (Magee, Broc
and Young, 1989; Grossman and Helpman, 1994; Austen-Smith, 1995).
This paper attempts to determine the importance of campaign contribu
tions and other factors affecting voting behavior in the House of Representatives
on three important trade-policy bills that came before the United Sta
* We are grateful to the Institute for International Economics for research support for t
paper.
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
80
Congress in 1993-94: the North American Free Trade Agreement (NAFTA),
the General Agreement on Tariffs and Trade (GATT) Uruguay Round agreement,
and most favored nation (MFN) status for China. Understanding the
economic, social and political factors that influenced the voting decisions
made by members of Congress on these measures is particularly important
in light of the failure of Congress in 1997 and 1998 to approve fast-track
negotiating authority for an expansion of NAFTA. Our analysis also provides
a test of traditional hypotheses about congressional voting behavior.
The following section briefly outlines some of the analytical models developed
by economists and political scientists in trying to understand how
members of Congress and various political pressure groups interact to determine
voting outcomes. In Section 3, we provide a brief legislative history
of each of the three trade bills we examine. Section 4 outlines the political
economy framework on which our econometric model is based while Section
5 includes the specification of the empirical model and a description of
the data used in the tests. The results of the statistical analysis are presented
in Section 6, and the final section concludes.
2. Models of political behavior
In the public choice literature, policy outcomes are endogenous. Policies
are determined by the interactions between elected officials, who are suppliers
of particular public policies, and organized interest groups, who are
demanders of such policies. Interest groups provide the campaign funds that
public officials need to stress the merits of their candidacies to imperfectly
informed voters. In exchange, politicians provide public policies that raise the
economic rents earned by the interest groups. These rent-seeking activities
are constrained by increased political opposition from individuals and firms
whose welfare is reduced by the policy actions.
In one well-known version of endogenous policy modeling, elected officials
weigh the increased political support they receive from pursuing policies
beneficial to a particular industry against lost support from other industries
and from consumers. In the end, government officials implement the policy
that maximizes their political support (see Hillman, 1989).2
Two main models of trade policy formation attempt to reduce the "black
box" elements of the political support function approach. In Magee, Brock,
and Young (1989), political candidates set their policy positions and then
interest groups contribute funds in order to improve the probability that their
preferred candidate will be elected. In Grossman and Helpman (1994), interest
groups offer the government a menu of political contributions that each
group is willing to pay for a variety of trade policies chosen. The governThis
content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
81
ment policy maker chooses a particular set of contributions and associated
policies to maximize his or her welfare function. In both models, politicians
are essentially selling trade policy in exchange for campaign contributions.
Political scientists have long viewed some public policies as determined
by competing domestic interest groups interacting with government officials,
as E.E. Schattschneider's (1935) classic study of the Tariff Act of 1930 illustrates.
They model the policy-making process in broader political terms
than economists, however, and give greater importance to the institutional
framework within which the decision process takes place.3
The orthodox view of political scientists has been that political contributions
buy access to legislators but only influence their voting behavior
under special circumstances (Smith 1995: 91-97; Morton and Cameron 1992:
80-83; Austin-Smith and Wright 1992 and 1994). According to this interpretation,
members of Congress generally respond favorably only to information
supplied by lobbyists whose basic ideological positions are consistent with
their own. Nevertheless, some political scientists maintain that contributions
can influence voting behavior if the public visibility of the issue is low or the
issue is specialized.
Austin-Smith (1991, 1995), Austin-Smith and Wright (1992, 1994), and
Ainsworth and Sened (1993) have argued that the conventional view underestimates
the importance of campaign contributions and lobbying in influencing
voting behavior and have formally modeled the manner in which this influence
takes place. Their models exploit the idea that legislators possess
incomplete information about the consequences of a particular policy for their
electorate. By strategically providing information that reduces elected officials'
uncertainty, lobbyists can sometimes gain by altering legislators' voting
behavior. Smith (1984) also suggests that other elected officials or lobbyists
are able to influence a legislator's voting behavior with arguments about how
the measure being considered affects the individual's personal goals.
As the preceding description of political economy models indicates, there
are significant differences in the manner in which economists and political
scientists analyze political decision-making. In most economic models, campaign
contributions from organized interest groups play the major role in
accounting for the existence of public policies that reduce social welfare. In
the Grossman-Helpman model, for example, the socially desired outcome
occurs if politicians do not value campaign contributions. In contrast, the
mainstream view of political scientists is that contributions influence voting
behavior only through increased access to a legislator, and they reject the notion
that campaign contributions buy political favors directly (Austin-Smith,
1991).
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
82
Both economists and political scientists agree, however, that constituent
interests are important determinants of a legislator's vote. The politicalsupport
and campaign contribution approaches of economists include the
concept that elected officials consider the interests of constituents in determining
their voting behavior, since to do otherwise leads to defeat at the polls.
Most political scientists also regard constituent interests as a major factor
shaping voting behavior because of the power of constituents at the ballot box
(Kingdon, 1973 and Arnold, 1990). Political scientists in addition postulate
that voting behavior is affected by such factors as pressure from political
leaders in the legislative and executive branches, as well as by personal ideology.
As Kau and Rubin (1982: 31-35) point out, some of these factors can
be interpreted as indicating the manner in which legislators gauge constituent
interest rather than representing independent forces shaping voting behavior.
Among previous empirical studies that have included campaign contributions
as a variable in analyzing the determinants of congressional voting
on recent trade bills are Kahane (1996), Steagall and Jennings (1996), BoxSteffensmeier,
Arnold and Zorn (1997), Holian, Krebs and Walsh (1997)
and Uslander (1998). These studies all focus on the NAFTA vote and find
that labor contributions are negatively correlated with a favorable vote on
this measure, while business contributions are positively correlated with a
favorable vote. A problem with these studies is that they assume campaign
contributions to be an exogenous variable rather than, as Chappell (1982)
points out, an endogenous variable. Labor organizations, for instance, were
likely to donate extra funds in 1992 to representatives they knew would oppose
NAFTA. Thus, campaign contributions are correlated with the residual
in the voting equation and the coefficient estimates in a single-equation model
are biased. We solve this problem by simultaneously estimating the probability
of voting for each trade bill and the amount of campaign contributions
received.
3. NAFTA, GATT, and MFN for China: A brief legislative history
Congressional approval in 1993 and 1994 of measures establishing a free
trade area for the United States, Canada and Mexico, implementing the
multilateral agreements reached in the Uruguay Round and continuing most
favored nation status for China represented major accomplishments for those
favoring greater trade liberalization. Their approval was remarkable in that
interest groups that were exploiting the appeal of "fair" trade to push for
protectionist policies seemed to be gaining political dominance in the 1980s
and early 1990s.
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
83
3.1. NAFTA
Of the three measures we examine, NAFTA was clearly the most contro
sial. Destler (1995: 217) states that NAFTA "set off the most prominent
contentious domestic debate on trade since the Smoot-Hawley Tariff Ac
1930". The intensity of the debate was surprising because half of Mex
imports were already entering the United States duty free and the rema
faced an average tariff rate of only 4 percent. In addition, there was alr
the lack of significant investment barriers between the two countries, and t
Mexican economy was small compared to that of the United States (M
ico's GDP was only 1/25th of that of the United States). Consistent w
these facts, most general equilibrium analyses predicted the impact of
agreement would be very modest, with U.S. average wages estimated to
from between .1 and .3 percent and aggregate U.S. employment to inc
by between .03 and .08 percent (International Trade Commission 1993,
2). These models predicted appreciable job losses in only a few industr
such as apparel, household appliances, sugar and ceramics.
Several factors explain why the NAFTA debate was more intense
widespread than seems justified by its predicted economic impact. The
important reason was the decision by organized labor to oppose congress
approval in the strongest terms. In doing so, most labor unions were
vinced that the adverse employment effects would be much more widesp
than economists had predicted. They also feared that approval of NAF
would lead to similar agreements with other low wage countries and resu
a flood of imports, exerting strong downward pressures on U.S. employ
and wages.
Concerns expressed by various environmental groups over the polluting
activities of the many maquiladoras near the U.S.-Mexican border further
focused attention on NAFTA. Not only were these groups apprehensive about
a rise in pollutants along the border as production in Mexico expanded, but
they were also fearful that a rush of U.S. firms to Mexico would encourage
American legislators to weaken U.S. environmental laws.
The initial position of President Clinton toward NAFTA also helps account
for the heated nature of the debate. The agreement had been negotiated by
the Bush administration and was formally signed in December 1992. Candidate
Clinton endorsed the basic text but called for side agreements that
would strengthen organizing rights and safety standards for Mexican workers,
tighten environmental standards in Mexico and guard against surges of
imports. Once in office, President Clinton did not press for rapid negotiation
of the side agreements, and thus they were not completed and approved until
August of 1993. The White House finally launched a vigorous effort to gain
support for the agreement in early fall. Traditional supporters of a liberal
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
84
trade policy such as the business community, editorial writers and academics
were organized in a more coherent manner, while opposition from most environmental
groups ceased due to the provisions of the side agreement on the
environment. In addition, meetings between President Clinton and undecided
House members at which the President made concessions or granted them
unrelated favors appear to have been effective in gaining votes for NAFTA.
The success of these various efforts is indicated by the fact that, far short
in September of the votes needed to gain House approval,4 the administration
had closed to within ten votes of a majority a week before the final vote.
The House approved the North American Free Trade Agreement by a vote
of 234 to 200 on November 17, 1993. Only 40 percent of House Democrats
(102 members) voted in favor of the measure, however, whereas 75 percent
of the Republicans (132 members) supported NAFTA. The Senate approved
the pact 61 to 38 three days later.
3.2. GATT
Though it received far less national media attention than the vote on N
the House and Senate legislation in November and December of 1994
plementing the Uruguay Round agreements is likely to have a greater
on the welfare of U.S. citizens than the NAFTA. Among the major a
plishments of the agreements were: (i) the reduction of export subsid
import quotas in the agricultural sector and the phasing out of the Multif
Arrangement; (ii) the extension of GATT rules to trade in services; (i
negotiation of a multilateral agreement protecting intellectual property r
(iv) the liberalization of trade-related investment measures; (v) the
tion of average tariff levels by about one-third and the ban of volu
export restraints; (vi) the strengthening of dispute settlement procedures
(vii) the establishment of the World Trade Organization (WTO) to pro
framework for administering trade agreements.5
Unlike the NAFTA, President Clinton did not inherit an alreadyinternational
agreement on the GATT Uruguay Round. Although th
administration made a concerted effort to complete the negotiations,
had started in 1986, the effort failed and the Clinton administration was
with the tasks of completing the international negotiations and gettin
gress to approve the resulting agreements. After seven years of negot
the appointments of a new Director General of the GATT (Peter Suthe
a new chief negotiator for the European Union (Leon Brittan) and a n
Trade Representative (Mickey Kantor) seemed to revitalize the negot
in 1993, and they were completed in December.
The congressional approval process continued for another year. Obje
raised by some members of Congress concerning a loss of U.S. sover
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
85
under the WTO and the high level of R&D subsidies permitted under the
agreement were two of the main reasons for the delay. Environmentalists,
who had split over NAFTA approval, were united in their disapproval of the
Uruguay Round agreement, whereas organized labor, which had vigorously
fought against NAFTA, criticized the agreement but did not oppose it as
strenuously. Ross Perot and consumer advocate Ralph Nader opposed approval
of the agreements, as they had NAFTA. Nonetheless, the implementing
legislation easily passed in the House of Representatives on November 29,
1994 by a 288 to 146 vote and in the Senate on December 1, 1994 by a 76
to 24 vote. A majority of both parties in the House and Senate approved the
measure.
3.3. MFNfor China
As part of the efforts to improve U.S.-Chinese relations an
China's move in the late 1970s toward a more market-orie
President Carter first granted most favored nation status t
under authority set forth in the Trade Act of 1974. This Act p
President can extend MFN treatment on an annual basis, pr
does not vote to disapprove such action. Until the brutal act
against Chinese dissidents in Tienanmen Square, there was
sional resistance to the continuance of this trade policy tow
then, however, many members of Congress have sought to
of MFN status to improvements in the Chinese government
behavior. In 1993, President Clinton secured a one-year exte
an executive order tying the 1994 extension to improved h
formance by the Chinese government. Under strong politic
the business community, however, he renewed MFN for Ch
though he explicitly recognized that the Chinese governme
the progress in human rights called for in his executive order.
maintaining close contact with the Chinese through trade w
overall U.S. goals than the withdrawal of MFN status. While
dissatisfaction with this policy position, the proposal to disa
Clinton's renewal recommendation was easily defeated in the
of 320 to 106.
4. Political economy framework
The political economy framework we adopt draws on the modeling efforts of
both economists and political scientists. Members of Congress are assumed
to behave in a manner designed to maximize their chances of reelection.
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
86
In doing so, they are influenced by what they perceive to be the impact of
the legislation on their constituents as well as by the wishes and information
provided by the major contributors to their campaigns.
Two main trade models provide divergent predictions about which groups
in the U.S. will support trade liberalization. In the Heckscher-Ohlin trade
model, relatively scarce factors of production lose economically from international
trade while relatively abundant factors gain. With perfect mobility
between sectors, the industry in which a factor is employed does not affect
this prediction. Since the United States is relatively scarce in less skilled
labor, the model suggests that legislators will be more likely to oppose
NAFTA, GATT and MFN for China the higher the proportion of less educated
individuals and the lower the per-capita income in their districts. Because
labor unions represent mainly blue-collar workers, a higher proportion of
union members in a district also increases the likelihood the representative
will oppose the trade bills.
The Ricardo-Viner trade model, on the other hand, assumes that the services
of some productive factors are completely or partly industry-specific. A
natural resource or particular type of physical capital may be suitable for use
only in a single industry or a few industries, for example, and workers may
acquire sector-specific skills. The implication is that individuals' attitudes toward
trade liberalization depend on the industry in which they are employed
rather than on their factor status. We include detailed data on employment
by industry within each congressional district to determine the importance of
specific industries in shaping representatives' voting behavior.6
Ideological considerations are likely to be important in determining how
legislators vote for two reasons. Since most voters view agreements such
as NAFTA as having little direct impact on their real income levels, their
broad political, social and economic concerns are not overwhelmed by the
personal economic implications of the agreements. The political leanings of
each constituency, reflected in the broad policy stances of their legislators,
will affect the position each legislator adopts. Politicians may also have an
interest in defining themselves as conservative, liberal, strong on national
defense issues, or allies of labor in order to appeal to constituencies that are
important to their re-election efforts.
Campaign contributions may affect legislators' voting patterns, either directly
as economists postulate, or through increased access and strategically
provided information as the political science literature suggests. Because we
do not observe access to legislators, we are unable to distinguish between
these two views. We divide campaign contributions into those from PACs
representing labor unions and from PACs representing business interests. The
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
87
Heckscher-Ohlin model suggests that labor unions will oppose free trade
while business groups will support it.
As political scientists have documented, the campaign funds that interest
groups give to a member of Congress depend partly on the representative's
influence in the legislative process. Members of important committees will
receive greater contributions from interest groups that are affected by the
committee actions. Business groups, for example, are likely to provide large
contributions to members of the House Ways and Means Committee, which
deals with taxes, whereas labor unions are likely to target members of the
Education and Labor Committee. Political action committees also tend to
support legislators with a record of voting in their interest.
5. Econometric specification and data
The political economy framework described above suggests that a legislator
voting behavior is influenced by various constituency characteristics and b
the magnitude of campaign contributions from different interest groups. The
campaign funds received by a representative depend on the legislator's polic
positions and the influence that he or she wields within the government. A
Chappell (1982) and Stratmann (1991) point out, voting behavior should no
be estimated using only a single equation because contributions are endogenous.
In addition, legislators who take a policy stance on one of the trade
bills are likely to vote similarly on the other bills. Thus, the residuals in th
voting equations for NAFTA, GATT and MFN for China may be correlated
Consequently, we analyze the voting on all three legislative initiatives and
the campaign contributions received by each representative from labor and
business groups simultaneously by the method of full information maximu
likelihood (FIML). The system of equations we estimate for the House
Representatives is:
Votenafta =
F(A'X + A((Labor contributions) + Ag (Business contributions) + en
Votegatt =
F(B'X + Bt (Labor contributions) + B (Business contributions) + Eg
Votemfn =
F(C'X + Ct (Labor contributions) + C (Business contributions) + em
. D' Y if D' Y- o<7 > 0
Labor contributions = D'Y if D'Y - l <
O if D'Y - oae1 < 0
Business contributions = E'Z + Eb
where F is the cumulative standard normal distribution, X is a
constituency variables that influence members voting behavior,
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
88
Z are vectors of variables that determine campaign contributions received by
legislators from labor and business PACs, respectively.
In order to test the endogeneity of the contribution variables, we run a
Hausman test on labor and business contributions in each of the three voting
equations. In the NAFTA equation, the Hausman test rejected the null hypothesis
(at the 5% significance level) that business contributions are exogenous
but failed to reject the hypothesis that labor contributions are exogenous. In
the GATT equation, the Hausman test did not reject the exogeneity hypothesis
for either contribution variable, and only the labor contribution variable was
found to be endogenous in the China MFN equation. Given the mixed results
from these tests and the theoretical justifications for doing so, we estimate the
system of equations above, treating campaign contributions as endogenous.7
The means and descriptions of each variable are presented in Table 1. The
roll call votes of members of the House on the GATT, NAFTA, and MFN
for China measures are available in the Congressional Quarterly Almanac. A
vote in favor of approval is assigned a value of one, while a vote against a bill
is assigned a zero. Campaign contributions from labor PACs are estimated
as a Tobit equation since the dependent variable is censored at zero. Because
there were so few representatives (16) who received no contributions from
business groups, the business contribution equation is assumed to be linear.
Data on campaign contributions to members of Congress are taken from
Makinson and Goldstein (1994). The contribution data used in this study are
the total contributions received by each representative in the 1992 election
from political action committees that Makinson and Goldstein identify as
representing either labor or business interests.
Congressional district characteristics such as the proportion of individuals
over 25 with no high school degree, the fraction with a high school diploma
but no college degree, the level of per capita income, the unemployment
rate, and the proportion of the population of Hispanic origin are available in
the Census publication Population and Housing Characteristics for Congressional
Districts of the 103rd Congress. We include the percentage Hispanic
variable only in the NAFTA equation. The proportion of private sector workers
in each district who are unionized is from Box-Steffenmeier, Arnold, and
Zorn (1997) and was provided by the authors.
Employment in congressional districts by two-, three-, and four-digit manufacturing
industries is estimated from data collected at the county level in
the 1993 County Business Patterns. If a county contains more than one congressional
district within its borders, the number of workers from an industry
who are in each district is estimated by using the fraction of the county's
population (in 1990) residing in each district. Data on population size by
congressional district and county are reported in Congressional Districts in
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
89
Table 1. Means and descriptions of variables
Variable Description Mean
NAFTA = 1 if representative voted for NAFTA 0.54
GATT Uruguay Round = 1 if representative voted for GATT 0.66
MFN China 1993 =1 if representative voted to approve MFN for 0.75
China 1993
Ways and Means Committee =1 if member of Ways and Means Committee 0.09
1993
Labor Committee = 1 if member Education and Work Force Com- 0.09
mittee
Terms in Office House terms served (including 1993-94) 5.0
Labor Contributions Labor group contributions 1991-92, thousands
of $
Business Contributions Business group contributions 1991-92, thou- 152.2
sands of $
No High School Degree 1990 Fraction of population (25+) without a high 0.25
school degree
No College Degree 1990 Fraction of pop. (25+) with HS degree, 0.49
without college degree
Per-capita Income 1990 District per-capita income (thousands of $) 14.4
Unionization Rate 1991-92 fraction of private sector workers uni- 0.12
onized
Unemployment Rate 1990 District unemployment rate (percent) 6.4
Percent Hispanic 1990 Fraction of population of Hispanic origin 8.8
(percent)
Export ratio Employment in export industries/employment in 1.37
import industries
ACU Rating 1993-94 Rating by American Conservative 46.7
Union (out of 100)
AFL-CIO Rating 1993-94 Rating by the AFL-CIO (out of 100) 58.7
NSI Rating 1993-94 National security index rating by 60.2
American Security Council
COC Rating 1993-94 Rating by the Chamber of Commerce 59.5
(out of 100)
LCV Rating 1993-94 Rating by League of Conservation 51.4
Voters (out of 100)
Democrat = 1 if Democrat 0.59
Agriculture Thousands of persons employed in agriculture, 7.18
1992
Food 1993 Employment in SIC 20 / total employment 0.017
Tobacco 1993 Employment in SIC 21 / total employment 0.000
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
90
Table 1. Continued
Variable Description Mean
Textiles 1993 Employment in SIC 22 / total employment 0.001
Apparel 1993 Employment in SIC 23 / total employment 0.011
Lumber 1993 Employment in SIC 24 / total employment 0.008
Furniture 1993 Employment in SIC 25 / total employment 0.005
Paper 1993 Employment in SIC 26 / total employment 0.007
Printing 1993 Employment in SIC 27 / total employment 0.015
Chemicals 1993 Employment in SIC 28 / total employment 0.009
Petroleum 1993 Employment in SIC 29 / total employment 0.001
Rubber products 1993 Employment in SIC 30 / total employment 0.001
Leather 1993 Employment in SIC 31 / total employment 0.001
Stone, clay, glass 1993 Employment in SIC 32 / total employment 0.0
Primary metals 1993 Employment in SIC 33 / total employment 0.007
Fabricated metals 1993 Employment in SIC 34 / total employment 0.015
Machinery, except electrical 1993 Employment in SIC 35 / total employment 0.018
Electronic equipment 1993 Employment in SIC 36 / total employment 0.015
Transportation equipment 1993 Employment in SIC 37 / total employment 0.017
Instruments 1993 Employment in SIC 38 / total employment 0.009
Miscellaneous 1993 Employment in SIC 39 / total employment 0.004
the 1990s (1993). The industry results reported in this paper are based only
on 2-digit SIC industries. We have also examined 3-digit SIC industries whe
studies predicted that they would be strongly affected by either the GATT
agreement or NAFTA, but we found that 3-digit industry variables rarely had
sizable impacts on voting decisions.8 Employment data for the agricultural
sector comes from the 1992 Census ofAgriculture.
Specific industries predicted to be significantly harmed by NAFTA include
apparel, ceramic tiles, household appliances, sugar, citrus fruit, and veget
ables. The computing sector is usually cited as the main employment gainer
The sectors expected to be harmed by the GATT agreement, according to
industry representative or studies by economists, are dairy, textiles, appare
lumber and wood products, paper, footwear, steel, and motor vehicles. Sectors
expected to gain and thus support the Uruguay Round agreements include
most agricultural industries, chemicals, nonelectrical machinery, computin
equipment, and instruments. Little formal analysis of the sectoral impact o
withdrawing MFN for China exists but we expect labor-intensive sectors
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
91
such as apparel and footwear to favor withdrawal and skill-intensive, high
technology industries to support the continuation of China's MFN status.
To investigate the influence of a district's trade with other countries on
members' voting behavior, we create a variable (Export ratio) indicating the
dependence of the district on export relative to import-competing jobs. We divide
manufacturing industries (at the 4-digit SIC level) into net importing and
net exporting sectors and then sum up total district employment in exporting
and in importing industries. Because trade liberalization tends to lower the
domestic prices of import-competing goods and raise the domestic prices of
export goods, the Ricardo-Viner trade model suggests representatives will be
more likely to support trade liberalization the greater the number of workers
in export industries relative to import industries. Since the proportion
of capital relative to unskilled labor tends to be greater in export industries
than import-competing sectors, the Heckscher-Ohlin model provides a similar
prediction.
Information about representatives' ideology and other characteristics
comes from a variety of sources. Interest groups rate representatives on the
percentage of times the legislator voted in the group's interest. We combined
the two yearly ratings for the 103rd Congress (1993-94) into one variable for
each interest group and interpret these ratings as indicators of the broad policy
views of a majority of a legislator's constituents. The American Federation of
Labor, Congress of Industrial Organization (AFL-CIO) ratings measures how
closely each politician is aligned with labor interests, while the Chamber of
Commerce (COC) ratings indicates how closely he or she is tied to business
interests. The American Conservative Union (ACU) rating measures the
conservative leanings of each member of Congress, the League of Conservation
Voters (LCV) ranks members on environmental votes, and the American
Security Council provides a national security index (NSI) measuring how
consistently the representatives vote in favor of strong national defense. Since
the AFL-CIO, COC and the ACU included how the members voted on one or
both of the NAFTA and GATT trade bills in their ratings, we recalculated the
ideological ratings excluding those two bills. A dummy variable (Democrat)
is used to indicate the legislator's political party, where a one signifies that he
or she is a Democrat. Data on the number of terms in office and membership
in House committees are from Duncan (1994).
6. Empirical results
Table 2 presents the results of estimating the five equation empirical model
simultaneously by the method of full information maximum likelihood.9 The
first part of Table 2 shows the coefficient estimates for the labor and business
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
92
contribution equations while the second part presents the estimates of the
equations for the House votes on NAFTA, GATT, and MFN for China in
1993. Because the equations are nonlinear, the coefficients are adjusted to
show the effect of a unit increase in the variable (above its mean) on the probability
of a favorable vote or on the expected campaign contributions received
( , where Y is the dependent variable and X the right-hand side variable).
There were 417 observations in the estimation, and the model predicts about
79-82% of the votes correctly on each bill.
The contribution equation estimates are consistent with both an electoral
and a policy-influencing motivation for campaign contributions. If political
action committees attempt to improve the election chances of like-minded
candidates, labor groups will contribute mostly to candidates highly rated by
the AFL-CIO while business groups will contribute to those rated highly by
the Chamber of Commerce. Table 2 indicates that these predictions are true
for the 1992 elections. If PAC's try to affect policy decisions directly by their
campaign funds, they will target contributions at influential legislators. Our
estimates show that labor contributions are higher for members of the House
Committee on Education and Labor while business contributions are higher
for members of the House Ways and Means Committee. Both business and
labor groups, ceteris paribus, gave greater campaign contributions to Democrats,
the majority party in 1993-94, while business groups gave more to
representatives with longer tenure in the House.
The voting equation estimates in Table 2 show that labor and business
contributions significantly affected legislators' decisions on both the NAFTA
and GATT bills. The labor contributions coefficient is statistically significant
at the 1% level in both equations while the coefficient on business contributions
are significant at the 5% and 10% levels, respectively, in the two
equations. In the NAFTA vote, a $1,000 increase in a member's contributions
from labor groups beyond the mean level reduced the probability of voting
for the agreement by 0.52 percentage points, whereas a $1,000 addition to
contributions from business PAC's increased the probability of voting to approve
the agreement by 0.12 percentage points. The comparable marginal
effects of $1,000 on the probability of voting for GATT were 0.27 and 0.05
percentage points. These are rather large impacts on voting probabilities considering
that the standard deviation of labor contributions is $61,000 and that
of business contributions is $123,000 in our data set. Consistent with the fact
that China's most-favored nation status was unlikely to have a large impact
on interest group welfare, neither labor nor business contributions had a statistically
significant effect on voting on the MFN bill. The coefficients on the
two contribution variables are jointly significant (measured by the likelihood
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
93
Table 2. Coefficient estimates of empirical model
Labor contributions equation
Variables Adjusted
coefficients
Constant -56.6434
AFL-CIO rating 1.1438***
Labor committee 32.0348***
Terms in office -0.8721
Democrat 17.5141
Sigma (scale parameter 38.4257
Business contributions equation
Variables Adjusted
coefficients
Constant -4.9062
COC rating 1.2419***
Ways and means committee 83.9211***
Terms in office 9.4320***
Democrat 47.0700
Voting equations Adjusted Adjusted Adjusted
variables coefficients coefficients coefficients
NAFTA GATT MFN93
Constant 5.0047** 8.7638** 1.2114
Labor contributions -0.0052*** -0.0027*** 0.0002
Business contributions 0.0012** 0.0005* 0.0001
ACU rating -0.0290*** -0.0285*** -0.0119***
AFL-CIO rating -0.0174*** -0.0114*** -0.0060***
NSI rating 0.0071*** 0.0093*** 0.0017*
COC rating 0.0047 0.0040** 0.0021
LCV rating -0.0016 -0.0009 -0.0048***
Percent hispanic 0.0093***
Union -3.0821*** -0.1409 0.0562
Democrat -0.1438 -0.1375 0.1695*
Export ratio 0.2771*** 0.1524*** 0.0482**
No high school degree -2.3108*** 0.7321 0.2300
HS, no college degree 0.1755 -0.2236 0.6019
Unemployment rate 0.0423 -0.0288** 0.0328**
Per-capita income -0.0121 0.0039 0.0082
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
94
Table 2. Continued
Voting equations Adjusted Adjusted Adjusted
variables coefficients coefficients coefficients
NAFTA GATT MFN93
Agriculture -0.0041 0.0015 -0.0038
Food 1.1442 -2.1937* 5.0206**
Tobacco 16.0501 11.2703 -4.6853
Textiles 0.7690 -4.2681*** -2.6220**
Apparel 2.6896 2.2154 -1.2546
Lumber 1.8370 3.1938* -0.9865
Furniture 7.7908 1.7984 1.0480
Paper 1.9281 -3.3008 2.4059
Printing 7.6666 1.2000 -2.0042
Chemicals -9.8419*** -1.7165 -0.4784
Petroleum -12.1218 1.5548 -0.0889
Rubber products 3.7262 -2.8579 -3.2974
Leather 1.5563 0.0024 -1.1740
Stone, clay, glass 4.5491 0.6880 -1.4141
Primary metals -4.8087 0.6024 -1.2935
Fabricated metals 8.0815* 2.4093 3.4824
Machinery, except electrical -5.8261 2.4228 -1.0788
Electronic equipment 5.1445 -3.6337** 1.0421
Transportation equipment 2.1863 2.2588** -0.4670
Instruments -3.1073 0.7116 0.3154
Miscellaneous 8.4961 6.9679 6.4391
% predicted correctly 79.3% 81.8% 81.5%
Number of observations 417
Log likelihood -5212.68
*indicates that the coefficient estimates are statistically significant a
**indicates that the coefficient estimates are statistically significant
***indicates that the coefficient estimates are statistically significant
ratio test) at the 1% level in the NAFTA and GATT votin
insignificant in the MFN equation.
Organized labor groups' stronger opposition to NA
is revealed most clearly in the coefficients on union rep
presence of low-skilled workers. Increased unionism
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
95
workforce greatly reduced the likelihood of a vote in favor of NAFTA but
had no significant effect on the probability of a vote for GATT. Thus, where
labor's anti-NAFTA campaign seems to have paid off in the House is through
lobbying pressures exerted on representatives by union members and less
educated workers.
The negative impact of union strength and lower-skilled workers on
support for trade liberalization is consistent with the predictions of the
Heckscher-Ohlin model. Consistent with both the Heckscher-Ohlin and
Ricardo-Viner models, a favorable vote on NAFTA, GATT and MFN was
more likely for a higher ratio of workers involved in export-oriented to
import-competing industries. A high unemployment rate in a member's district
reduced the probability of a vote for the GATT agreement but raised the
probability of voting for MFN for China. In addition, an affirmative vote for
NAFTA was more likely the larger the proportion of Hispanics in a member's
district. The coefficients on the general economic characteristics of the district
(education variables, export employment relative to import employment,
unemployment rates, and per-capita incomes) are jointly significant at the
1% level for GATT and NAFTA and at the 5% level for MFN for China in
likelihood-ratio tests.
A district's employment in most industries had little impact on legislators'
voting. Large employment in the textiles industry significantly reduced the
likelihood of voting for the GATT and MFN bills, but (consistent with the
fact that the industry did not oppose NAFTA) did not affect the NAFTA vote.
Few of the other industries predicted by various studies either to gain or lose
appreciably from NAFTA or GATT, however, had coefficient estimates that
are statistically significant with the expected sign.'0 Despite the relatively few
significant coefficients among the industry variables, it would be a mistake to
omit them. Likelihood-ratio tests show that the coefficients on the industry
variables are jointly significant at the 1% level in the three voting equations.
This result suggests that previous estimates from voting studies, which often
exclude industry characteristics, are flawed because of omitted variable bias
if industry employment is correlated with the variables such as contributions,
per-capita income, union strength, or ideological ratings included in the study.
Ideological variables, as measured by legislators' past voting records,
were particularly important determinants of congressional voting. Congress
members ranked highly by the AFL-CIO tended to oppose NAFTA, GATT
and MFN for China, while those ranked highly on the National Security Index
tended to favor these measures. A high Chamber of Commerce rating was
positively correlated with approval of GATT, but not significantly correlated
with the probabilities of voting for NAFTA or MFN for China. Environmental
concerns did not play a major role in legislators' voting decisions on NAFTA
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
96
or GATT, but "green" legislators were less likely to approve of MFN for
China. Surprisingly, a higher rating by the American Conservative Union
indicates that a legislator was more likely to vote against each of the three bills
despite the traditional view of conservatives as free traders. Perhaps, given
the various other variables in the equation measuring free-trade leanings, this
ideological rating is picking up the concerns on the part of conservatives
about a loss of sovereignty with such agreements as NAFTA and GATT.
In order to investigate the economic significance of the impact of campaign
contributions on voting probabilities, we perform a number of counterfactual
simulations. First, we used the coefficient estimates from our model
and the values of the variables for each representative to predict his or her
probability of voting in favor of the bill. The sum of all representatives'
probabilities of voting for each bill yields the predicted number of favorable
votes. We then recalculate each representative's probability of voting for the
bills under three counterfactuals. First, we held all other variables at their
actual levels but set contributions from labor groups to each representative
equal to zero. For the second simulation, we set business contributions to
each representative equal to zero, and finally we set both business and labor
contributions to zero. The sum of all representatives' probabilities of voting
for each bill in the three cases reveals the model's predicted number of votes
if no labor contributions were made, if no business contributions were made,
and if no campaign contributions of either kind were made.
The results of these simulations, shown in Table 3, reveal the large effects
that labor and business contributions had on the total predicted number of
votes for NAFTA and the GATT agreement. First, without labor contributions,
the model predicts that 67 more representatives would have voted in
favor of NAFTA while 57 more would have voted in favor of the GATT agreement.
In the absence of business contributions, about 41 fewer representatives
would have voted for NAFTA and 35 fewer for GATT. Since NAFTA passed
the House by only 17 votes, the model predicts that the agreement would not
have been approved had there been no campaign contributions from business
PAC's. The net effect of contributions as a whole was to reduce the number
of votes for NAFTA by 25 votes and to reduce the number of votes for the
GATT agreement by about 32 votes.
The large impact of contributions on predicted votes in our estimates suggests
that either trade policy is for sale or that money buys access, which
interest groups are able to use effectively to influence legislators' decisions.
What is the price an interest group must pay to sway one vote in the House of
Representatives? Our definition of the price of one extra vote is the increase
in money spent by the interest group (divided equally among all representatives)
necessary to raise or lower the expected number of votes on each trade
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
97
Table 3. Counterfactual predictions of the model
NAFTA GATT MFN93
Number of votes we can predict 426 424 419
Total number of favorable votes 229 283 313
Predicted number of favorable votes 227 285 317
Predicted favorable votes with no labor contributions 294 342 313
Predicted favorable votes with no business contributions 186 250 307
Predicted favorable votes with no contributions 252 317 302
Predicted effect of labor contributions on number of votes -67 -57 4
Predicted effect of business contributions on number of votes 41 35 10
Predicted effect of total contributions on number of votes -25 -32 15
bill by one. We have sufficient data to predict the probability of
vote on NAFTA for 426 representatives. Our model predicts t
these will vote in favor of the bill. In order to reduce the expecte
226, labor groups would have had to spend $825 more per represen
about $352,00 more. Business groups would have needed to increa
contributions by $3,717 per representative, or about $1,583,000.
for labor to sway one expected vote against the GATT Uruguay R
was $738 per representative, or about $313,000. The comparable p
business groups on the GATT bill was $4,412 per representative,
$1,871,000. Interest groups could, of course, influence the voting o
a lower cost than these numbers indicate by targeting their contr
representatives where money has a large marginal effect on the p
of voting for each bill. Since we are examining contributions that
at least one to two years before the bills were voted on, however,
targeting of this kind would require exceptional foresight. The hi
business groups needed to pay to sway one vote is likely due to the
some industries opposed the trade bills, while others supported th
unions were much more united in their opposition to the trade lib
measures.
Political action committees were not the only groups lobb
ives to secure votes on the trade bills. President Clinton als
the oval office to ensure NAFTA's passage. Based on in
public records, Grayson (1995, ch. 9, Table 10) provides
entatives who allegedly obtained special benefits from t
return for supporting NAFTA. Using the model in Tabl
predicted probability of a vote for NAFTA by these me
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
98
Table 4. Correlations between residuals
Enafta Egatt 8mfn93 8labcon 8buscon
enafta 1
egatt 0.3720 1
(o.oooo)
Emfn93 0.0946 0.1833 1
(0.0529) (0.0002)
Elabcon 0.0671 0.1950 -0.0935 1
(0.1671) (0.0001) (0.0558)
Ebuscon -0.0224 0.0026 0.0023 0.36171
(0.6447) (0.9579) (0.9628) (0.0000)
of any special favors. Of the 46 representatives for wh
data to make a prediction, we estimate that 35 would h
without any special consideration. Thus President Clin
garnered 11 extra votes through the concessions he m
Those votes were an important gain considering NAFT
was only 17 votes. Three representatives were particu
timate that J.J. Pickle of Texas, Clay Shaw of Florid
of Washington, all of whom received concessions, had
probability of voting for NAFTA without the favors gran
Table 4 reports the correlations between the residu
equations as well as the p-values for the hypothesis t
equal zero (in parentheses). A p-value below 0.05 indic
is significantly different from zero at the 5% level. Excep
the labor contribution and GATT voting equations, the
ence that the residuals in the voting equations are signific
those in the contribution equations. This result suggest
were not giving more money to candidates who, for un
likely to vote in the group's favored manner on the trade
The residuals in all three voting equations are stron
each other, however, indicating that legislators have a pro
against free trade that is not captured in any of the five
include in the regression. With correlated residuals, si
of the voting equations is preferable to running the regre
cause it reduces the variance of the estimated coefficients. Labor and business
contributions are also positively correlated. This result is consistent with the
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
99
hypothesis that both types of interest groups are trying to curry favor with a
few influential representatives.
7. Conclusions
The major conclusion of this study is that political contributions to
ors by organized labor and business groups significantly affected th
outcome on two (NAFTA and GATT) of the three trade bills analyz
estimate that labor contributions or access to legislators gained throu
contributions resulted in 67 extra votes against NAFTA and 57 ext
against the GATT Uruguay Round bill. Contributions from busines
resulted in 41 extra votes in favor of NAFTA and 35 extra votes for the GATT
bill. This last result is particularly interesting because it suggests that NAFTA
would have failed if business groups had made no contributions to representatives.
We estimate the price for labor groups to sway one vote against
NAFTA and GATT to be about $352,000 and $313,000 respectively.
Private interest groups were not the only parties attempting to influence
representatives by providing them with special benefits. We estimate that
President Clinton secured about 11 House votes in favor of NAFTA by
granting concessions to individual legislators. While campaign contributions
and presidential favors were important, they were not the only significant
factors determining how representatives voted on the trade bills. Factor-status
variables that are suggested by the Heckscher-Ohlin model such as the proportion
of less educated workers and the degree of unionization significantly
affected the votes on NAFTA. A district's employment in export-oriented
versus import-competing industries also played a large role in whether or
not a representative voted for trade liberalization in each case. Few of the
variables indicating the proportion of employment in specific industries had
large impacts on congressional voting. The broad policy views of legislators,
as measured by their ratings by interest groups, however, were very important
determinants of representatives' decisions on all three votes. We interpret
these various results as evidence that legislators are responding on trade legislation
to the economic and social concerns of their constituencies as well
as to the wishes of their major contributors.
Notes
1. Quoted in the New York Times, October 23, 1997.
2. For a systematic review of the political economy models of economists, see Rodrik
(1995).
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
100
3. See Morton and Cameron (1992), Smith (1995), and Bender and Lott (1996) for recent
surveys of this literature.
4. A USA-NAFTA poll of the House of Representatives on September 20, 1993 found that
190 representatives opposed approving NAFTA while 161 supported it. (Destler, 1995:
225)
5. See Preeg (1995) for an excellent discussion of the Uruguay Round negotiations.
6. Kaempfer and Marks ((1993) find that Heckscher-Ohlin factor variables are strongly
correlated with voting on the 1991 fast-track bill but that Ricardo-Viner sector-specific
variables are only weakly and inconsistently linked with voting by members of Congress.
7. Assuming that labor and business contributions are exogenous and running singleequation
estimates on each voting bill yields qualitatively similar results.
8. The main studies from which we take predictions about the impact of the NAFTA and
GATT agreements are USITC (May 1992), USITC (January 1993), and USITC (June
1994).
9. We have also estimated the determinants of Senate voting behavior. The results were
qualitatively similar to the House of Representatives, although the smaller number of
observations limited the variables we could include in the system of equations.
10. Omitting the ratio of the number of workers employed in export industries in each congressional
district to the number of workers employed in import-competing industries
(Export ratio) does not change this finding.
References
Ainsworth, S. and Sened, I. (1993). The role of lobbyists: Entrepreneurs with two audiences.
American Journal of Political Science 37: 834-866.
Arnold, R.D. (1990). The logic of congressional action. New Haven: Yale University Press.
Austen-Smith, D. (1991). Rational consumers and irrational voters: A review essay on Black
hole tariffs and endogenous policy theory by Stephen Magee, William Brock and Leslie
Young, Cambridge University Press, 1989. Economics and Politics 3: 73-92.
Austen-Smith, D. and Wright, J. (1992). Competitive lobbying for a legislator's vote. Social
Choice and Welfare 9: 229-257.
Austen-Smith, D. and Wright, J. (1994). Counteractive lobbying. American Journal of
Political Science 38: 25-44.
Austen-Smith, D. (1995). Campaign contributions and access. American Political Scien
Review 89: 566-581.
Bender, B. and Lott, J. (1996). Legislator voting and shirking: A critical review of the
literature. Public Choice 87: 67-100.
Box-Steffenmeier, J., Arnold, L. and Zorn, J. (1997). The strategic timing of position taking
in Congress: A study of the North American Free Trade Agreement. American Political
Science Review 91: 324-338.
Chappell, H. (1982). Campaign contributions and congressional voting: A simultaneous
probit-tobit model. Review of Economics and Statistics 64: 77-83.
Congressional Districts in the 1990's: A portrait of America. (1993). Washington, D.C.:
Congressional Quarterly Inc.
Duncan, P. (Ed.) Politics in America: 1994. Congressional Quarterly Press: Washington D.C.
Destler, I.M. (1995). American trade politics. Third Edition. Washington. D.C.: Institute for
International Economics.
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms
101
Grayson, G. (1995). The North American Free Trade Agreement. New York: University Press
of America.
Grossman, G. and Helpman, E. (1994). Protection for sale. American Economic Review 84:
833-850.
Hillman, A. (1982). Declining industries and political support protectionism. America
Economic Review 72: 1180-1187.
Holian, D., Krebs, T. and Walsh, M. (1997). Constituency opinion, Ross Perot, and roll-ca
behavior in the U.S. House: The case of the NAFTA. Legislative Studies Quarterly 22
169-392.
Kahane, L. (1996). Congressional voting patterns on NAFTA: An empirical analysis. American
Journal of Economics and Sociology 55: 395-409.
Kau, J. and Rubin, P. (1982). Congressmen, constituents, and contributors: Determinants of
roll call voting in the House of Representatives. Boston: Marinus Nijhoff Publishing.
Kaempfer, W. and Marks, S. (1993), The expected effects of trade liberalization: Evidence
from US congressional action on fast-track authority. World Economy 16: 725-740.
Kingdon, J. (1973). Congressmen's voting decisions. Ann Arbor: University of Michigan
Press.
Magee, S., Brock, W. and Young, L. (1989). Black hole tariffs and endogenous policy theory.
Cambridge University Press: Cambridge.
Makinson, L. and Goldstein, J. (1994). Open secrets: The encyclopedia of congressional
money and politics. Congressional Quarterly Inc.: Washington D.C.
Morton, R. and Cameron, C. (1992). Elections and the theory of campaign contributions: A
survey and critical analysis. Economics and Politics 4: 79-108.
Preeg, E. (1995). Traders in a brave new world. Chicago: University of Chicago Press.
Rodrik, D. (1995). Political economy of trade policy. In: G.M. Grossman and K. Rogoff(Eds.),
Handbook of international &tonomics, 1457-1494. Amsterdam: North-Holland.
Schattschneider, E.E. (1935). Politics, pressures and the tariff. Englewood Cliffs, N.J.:
Prentice Hall.
Smith, R. (1984). Advocacy, interpretation, and influence in the U.S. Congress. The American
Political Science Review 78: 44-63.
Smith, R. (1995). Interest group influence in the U.S. Congress. Legislative Studies Quarterly
20: 89-139.
Steagall, J. and K. Jennings (1996). Unions, PAC contributions, and the NAFTA vote. J
of Labor Research 17: 515-521.
Stratmann, T. (1991). What do campaign contributions buy? Decipherering causal effec
money and votes. Southern Economic Journal 57: 606-620.
United States International Trade Commission (1992). Economy-wide modeling of th
nomic implications of a FTA with Mexico and a NAFTA with Canada and Me
Washington, D.C.: USITC Publication 2516.
United States International Trade Commission (1993). Potential impact on the U.S. ec
and selected industries of the North American Free Trade Agreement. Washington,
USITC Publication 2596.
United States International Trade Commission (1994). Potential impact on the U.S. e
nomy and industries of the GATT Uruguay Round Agreements. Washington, D.C.: USIT
Publication 2790.
Uslaner, E. (1998). Let the chips fall where they may? Executive and constituency influences
on congressional voting behavior on NAFTA. Legislative Studies Quarterly 23: 347-371.
This content downloaded from 147.251.185.127 on Thu, 24 Aug 2017 12:42:40 UTC
All use subject to http://about.jstor.org/terms