Economics Letters 154 (2017) 31–34

Contents lists available at ScienceDirect

Economics Letters journal homepage: www.elsevier.com/locate/ecolet

Do local exports impact congressional voting on free trade agreements? Michael Malcolm West Chester University, United States

highlights • Legislator votes on trade deals are influenced by exports to the subject country. • This paper explores legislative votes on more recent trade deals. • This is the first paper to examine local exports to the subject country.

article

info

Article history: Received 25 October 2016 Received in revised form 27 January 2017 Accepted 8 February 2017 Available online 15 February 2017 JEL classification: D72 F13 F51

a b s t r a c t The United States is currently a party to 14 free trade agreements, which cover 20 countries. This paper explores a new aspect of economic-interest models of legislative support for trade openness, using updated data on more recent trade agreements. I show that House and Senate members are more likely to vote in favor of trade agreements when their states have higher exports to the country that is the subject of the agreement. © 2017 Elsevier B.V. All rights reserved.

Keywords: Free trade agreements Political economy Protectionism

1. Introduction The United States exported $2.26 trillion of goods and services abroad in 2015, which represented 12.5% of GDP. Nevertheless, despite the importance of trade in the US economy, significant barriers to trade with much of the rest of the world continue to persist. In an effort to promote trade, the United States is currently a party to 14 free trade agreements (FTA), which cover 20 countries. Twelve of these agreements are bilateral, while NAFTA and CAFTA cover multiple trading partners.1 Table 1 shows all active free trade agreements. Negotiations have been underway for the Transatlantic Trade and Investment Partnership with the European Union and for the Trans-Pacific Partnership with a number of South American and Pacific Rim countries, the latter apparently stalled indefinitely. Each agreement is unique and they are complex, typically covering a whole host of issues, but all the agreements aim to reduce tariffs and other trade barriers. E-mail address: [email protected]. 1 All information on trade agreements is from the Office of the United States Trade Representative. http://dx.doi.org/10.1016/j.econlet.2017.02.018 0165-1765/© 2017 Elsevier B.V. All rights reserved.

While trade agreements are negotiated by the executive branch, the House and Senate vote to approve them. Voting on trade agreements is an interesting political economy question because it does not obey traditional partisan fissures. Table 1 shows the percentage of affirmative votes among Democrats and Republicans for all 14 active free trade agreements.2 Overall, Republicans are more inclined to support FTA’s (even when the president is a Democrat), but Republican support is not universal and a sizeable number of Democrats do support them. There is a small body of literature that studies the determinants of congressional voting on trade deals. Roughly, these studies fall into two categories. Some studies try to explain free trade votes by appealing to economic models that predict a pattern of winners and losers from free trade, while others attribute voting directly to lobbying and exertion of pressure by special interest groups. Of course, the two may not independent. Special interest group pressure may be a function of economic considerations of constituents. 2 Congressional voting data are taken directly from roll calls on the House and Senate websites.

32

M. Malcolm / Economics Letters 154 (2017) 31–34

The literature on the impact of constituents’ varied economic interests on congressional voting has focused mostly on capital versus labor interests and on labor of varying skill level. However, one consideration that seems to have been missed is the impact of exports to the country that is the subject of the agreement. Even low-skill workers might experience a welfare improvement from trade liberalization if their jobs, or many jobs in the local economy, are supported by exports. While Baldwin and Magee (2000) find that overall exports are a significant determinant of congressional FTA voting, nobody has yet looked at the impact of exports specifically to the country that is the subject of the agreement. This paper attempts to fill the gap. In doing so, we also add to the literature by updating earlier research and examining more recent trade agreements that cover a broader range of partners. While there is a substantial literature on NAFTA, there is very little on the many trade agreements that have followed. The main result is that House and Senate members are more likely to vote for FTA’s when they represent states with large exports to countries that are the subject of the agreements. The result holds overall and for both Democrats and Republicans separately.

votes, respectively. Wang et al. (2013) also find that business and labor contributions were important in explaining congressional voting on Permanent Normal Trade Relations with China – much more important than Hecksher–Ohlin type considerations of constituent welfare. Gilbert and Oladi (2012) find that agricultural contributions were particularly salient with respect to the same agreement. Matschke and Sherlund (2006) find that labor lobbying raises trade barriers while lobbying by capital owners reduces them. If we think about the association in the other direction, Beaulieu and Magee (2004) show that labor groups favor representatives who back protectionism and capital owners support representatives who back trade liberalization.4 The analysis in this paper falls more along the lines of economic considerations, although of course lobbying influences may intermediate the impact of economic welfare on politicians’ positions. In addition to exploring a new and economically relevant determinant of free trade voting, this paper also contributes to the literature generally by leveraging more data and examining a wider swath of trade agreements. 3. Data

2. Related literature The literature on determinants of congressional voting on FTA’s can be roughly sorted into economic models where legislators represent their constituents’ economic interests and political economy models where legislators are rent-seekers who respond to lobbying and donations. For the first category of studies, the Heckscher–Ohlin model and the Stolper–Samuelson theorem assert that relatively abundant factors of production gain from free trade while relatively scarce factors of production lose. For the United States, this implies that relatively scarce low-skill labor will suffer a welfare decline from trade openness (Baldwin and Magee, 2000). It is worth noting that the result is not universally accepted. The most common criticism is that the welfare implications may be industry-specific if factors are not perfectly mobile across industries. Davis and Mishra (2007) argue that the Stolper–Samuelson result does not hold up empirically for the case of Latin America. Despite theoretical ambiguity, organized labor vociferously opposes free trade agreements, and a number of studies present evidence that legislators respond to Stolper–Samuelson considerations in their votes on trade openness. Beaulieu (2002) finds that district-level relative skill endowment is correlated with House voting on NAFTA; results are mixed on the industry-specific hypothesis. Conconi et al. (2012) find that representatives from more skill-abundant districts are more inclined to support both trade and immigration liberalization. Going back in history, Epstein (2014) finds that voting on the 1913 Underwood tariff was strongly associated with state-level relative factor endowments.3 The basic result seems to hold internationally. Dutt and Mitra (2002) find that inequality is associated with stronger trade barriers in capitalrich economies. At a micro level, across many countries, individuals who are more educated and who work in industries not susceptible to import competition are more inclined to support free trade (Mayda and Rodrik, 2005). A second strain of literature examines the impact of lobbying and special interests on congressional trade votes. Baldwin and Magee (2000) find for NAFTA that labor and business campaign contributions were significant predictors of negative and positive 3 Facchini et al. (2013) argue that there is an important interaction effect with media coverage, and that heightened exposure makes legislators more responsive to their citizens’ preferences on immigration. In a similar vein, Ito (2015) finds for Japan that legislators are more inclined to support trade liberalization when they do not face close races.

The data cover every active Free Trade Agreement to which the US is a party with the exception of NAFTA, which has been well-studied and which was signed a decade prior to any of the other FTA’s used in the study. The dataset includes both the House and the Senate, and each observation is an individual legislator’s recorded vote on an FTA. Unanimous votes passed without a roll call are excluded, as are abstentions. Combined, the sample size is 5558 legislator votes on 11 different FTA’s. Table 1 presents summary statistics on congressional voting for each FTA, broken down by chamber and by party. The dependent variable is binary: equal to 1 if the legislator voted for the FTA and equal to 0 if the legislator voted against the FTA. These data were obtained from House and Senate voting records. The primary independent variable of interest is the exports of the legislator’s home state to the country that is the subject of the FTA.5 These data are available for each year since 1999 from the International Trade Administration of the US Department of Commerce.6 We examine both the total volume of relevant exports and the share of relevant exports in the state’s GDP. The control variables are the same economic controls as in Baldwin and Magee (2000) and that have become standard in the literature: income per capita, the unemployment rate, union membership, percentage of population with a high school diploma and percentage of population with a bachelor’s degree, again measured at the level of the state that the voting legislator represents. 4. Results Table 2 shows the results of a probit regression of a legislator’s FTA vote on exports to the subject country and on the other controls described in the previous section. Marginal effects evaluated at the mean are given in brackets. Local exports to the FTA subject country are a significant determinant of a legislator’s vote on the FTA. Each 0.1% increase in the share of such exports in the state’s GDP is associated with a 0.01-unit increase in the probability of an affirmative vote. With respect to volume, each additional $1 billion in relevant exports is associated with a 0.01-unit increase in the probability of an 4 However, Magee (2002) argues that the free-rider problem can make financing such lobbying operations difficult in practice. 5 In the case of the Central American Free Trade Agreement (CAFTA), exports to all countries involved are included. 6 Unfortunately, import data are only available from 2011 onwards.

M. Malcolm / Economics Letters 154 (2017) 31–34

33

Table 1 Active US free trade agreements. Agreement

Passed senate

House vote

H: % R support

H: % D support

Senate vote

S: % R support

S: %D support

Australia FTA Bahrain FTA CAFTA-D Chile FTA Columbia FTA Israel FTA Jordan FTA KORUS FTA Morocco FTA NAFTA Oman FTA Panama TPA Peru TPA Singapore FTA

07/04 12/05 07/05 07/03 10/11 05/85 09/01 10/11 07/04 11/93 09/06 10/11 12/07 07/03

314/109 327/95 217/215 270/156 262/167 Unanimous Unanimous 278/151 323/99 234/200 221/205 300/129 285/132 272/155

89% 94% 88% 88% 96% – – 91% 92% 75% 88% 98% 92% 88%

58% 59% 7% 37% 16% – – 31% 60% 40% 11% 35% 48% 37%

80/16 Unanimous 55/45 65/32 66/33 Unanimous Unanimous 83/15 Unanimous 61/38 62/32 77/22 77/18 66/32

96% – 80% 86% 96% – – 98% – 74% 91% 100% 98% 86%

69% – 24% 48% 41% – – 73% – 51% 28% 59% 65% 48%

Notes: CAFTA includes Costa Rica, El Salvador, Guatemala, Honduras, Nicaragua and the Dominican Republic. KORUS is the trade agreement with South Korea. NAFTA includes Canada and Mexico. Table 2 Determinants of a legislator’s vote on free trade agreement. Share: State exports to FTA country as % of state’s GDP All

Dem *

Export share

0.2727 (0.1435) [0.0987]

Volume: State exports to FTA country in billions of dollars Rep

**

0.5174 (0.2066) [0.1982]

All

Dem

Rep

0.0336** (0.0164) [0.0122]

0.0215 (0.0226) [0.0082]

0.0771** (0.0393) [0.0119]

0.0988 (0.2649) [0.0153]

Export volume 0.1229*** (0.0046) [0.0044]

0.0023 (0.0070) [0.0009]

0.0070 (0.0081) [0.0011]

0.0119*** (0.0046) [0.0043]

0.0028 (0.0070) [0.0011]

0.0048 (0.0081) [0.0007]

−0.0006

−0.0509***

−0.0451***

(0.0137) [−0.0195]

0.0504*** (0.0192) [0.0078]

−0.0020

(0.0094) [−0.0002]

(0.0095) [−0.0007]

(0.0139) [−0.0173]

0.0399** (0.0191) [0.0061]

−0.0333***

−0.0222***

−0.0234***

(0.0052) [−0.0085]

0.0093 (0.0071) [0.0014]

−0.0336***

(0.0037) [−0.0120]

(0.0036) [−0.0122]

(0.0051) [−0.0090]

0.0160*** (0.0059) [0.0058]

−0.0348*** (0.0087) [−0.0133]

0.0067 (0.0110) [0.0010]

0.0188*** (0.0060) [0.0068]

−0.0322***

% high school diploma

−0.0392*** (0.0069) [−0.0142]

0.0180* (0.0105) [0.0069]

−0.0101 (0.0127) [−0.0016]

−0.0403*** (0.0069) [−0.0146]

0.0163 (0.0105) [0.0063]

−0.0115

% bachelor’s degree

Senate

0.1156** (0.0508) [0.0411]

0.4447*** (0.0701) [0.1744]

0.0685 (0.0968) [0.0103]

0.1202** (0.0508) [0.0427]

0.4493*** (0.0700) [0.1762]

0.0730 (0.0967) [0.0109]

Constant

−0.0534 (0.4979)

2.5638*** (0.7357)

0.3170 (0.9216)

−0.2237 (0.5051)

2.3690*** (0.7451)

0.0145 (0.9302)

Observations

5558

2616

2923

5558

2616

2923

Income per capita

Unemployment rate

Union membership

(0.0089) [−0.0123]

0.0106 (0.0070) [0.0016] 0.0122 (0.0113) [0.0019] (0.0126) [−0.0018]

Notes: Probit regression used throughout. Income per capita in thousands of real US dollars, at state level. Unemployment rate in percent, at state level. Union membership is percentage of state’s labor force that belongs to a labor union. % high school diploma and % bachelors degree measure percentage of state’s population over age 25 with the respective credential. Senate equals one for senate votes. Standard errors appear in parentheses. Marginal effects in brackets, evaluated at the mean. * Significance at 10%. ** ***

Significance at 5%. Significance at 1%.

affirmative vote. The former result is significant at the 10% level and the latter at the 5% level.7 If we break out the results by party, the effect of local export share on FTA voting is significant for Democrats but not for Republicans. On the other hand, the effect of 7 Results were not much different if exports were lagged a couple of years prior to the vote, or averaged over a few prior years, as serial correlation is quite strong. Also, the export variables both have a long right tail. We ran all regressions trimming the top 5% of the export variable to ensure that the results were not unduly influenced by outliers. The signs of the relevant coefficients matched and the p-values were similar.

local export volume on FTA voting is significant for Republicans but not for Democrats. 5. Conclusion Congressional voting on NAFTA has been thoroughly examined, both from an economic perspective (legislators as representatives) and from a political/lobbying perspective (legislators as rentseekers). However, research on voting determinants for the 12 free trade agreements that followed NAFTA has been relatively scant. This paper updates the literature by capitalizing on more

34

M. Malcolm / Economics Letters 154 (2017) 31–34

recent free trade agreements with a diverse array of countries and by exploiting variation across states in exports to the subject countries. While previous literature considered total exports as a determinant of FTA voting, it perhaps addresses the question of economic interests more directly to examine the impact of exports specifically to the FTA-subject country/ies. In this paper, we have shown that these exports do matter and that they are associated with affirmative votes on trade agreements. The result thus provides additional evidence in the context of the literature that legislators respond to economic interests in their districts in deciding how to vote on free trade agreements. Of course, the mechanism remains an open question and the result is consistent with both constituent-interest models and with rentseeking models of legislative behavior.8 Literature on legislative voting patterns continues to explore the interplay between these constituent-interest and rent-seeking models of legislative behavior. Support for free trade agreements is an interesting case, especially inasmuch as it transcends typical partisan fissures. As policymakers continue to pursue new trade agreements like the TPP, rallying political support among legislators is likely to be a major hurdle. Understanding their motivations in voting is an important question. References Baldwin, R.E., Magee, C.S., 2000. Is trade policy for sale? Congressional voting on recent trade bills. Public Choice 105 (1–2), 79–101.

8 While it is speculative, the results broken out by party could be insightful in this context and merit further exploration. One potential explanation for the responsiveness of Republican legislators to export volume versus share could be greater susceptibility to large-dollar lobbying influences, e.g. a large donor whose business engages in a high volume of trade with an FTA-subject nation, even if the trade represents a relatively small share of state GDP. Work on differences in Republican and Democratic-oriented lobbying is extensive. See Bertrand, Bombardini and Trebbi (2014) for some recent work in economics documenting a stronger dominance of large-dollar lobbying on the Republican side.

Beaulieu, E., 2002. The Stolper–Samuelson theorem faces congress. Rev. Int. Econ. 10 (2), 343–360. Beaulieu, E., Magee, C., 2004. Four simple tests of campaign contributions and trade policy preferences. Econ. Polit. 16 (2), 163–187. Conconi, P., Facchini, G., Steinhardt, M.F., Zanardi, M., 2012. The Political Economy of Trade and Migration: Evidence from the US Congress. Social Science Research Network. Working paper. Davis, D.R., Mishra, P., 2007. Stolper-Samuelson is dead: And other crimes of both theory and data. In: Globalization and Poverty. University of Chicago Press, pp. 87–108. Dutt, P., Mitra, D., 2002. Endogenous trade policy through majority voting: an empirical investigation. J. Int. Econ. 58 (1), 107–133. Epstein, J.S., 2014. Congressional Voting on the Underwood Tariff of 1913: An Econometric Analysis (Senior thesis), Brandeis University. Facchini, G., Frattini, T., Signorotto, C., 2013. Mind what your voters read: Media exposure and international economic policy making. Centro Studi Luca d’Agliano Development Studies Working Paper, (358). Gilbert, J., Oladi, R., 2012. Net campaign contributions, agricultural interests, and votes on liberalizing trade with China. Public Choice 150 (3–4), 745–769. Ito, B., 2015. Does electoral competition affect politicians’ trade policy preferences? Evidence from Japan. Public Choice 165 (3–4), 239–261. Magee, C., 2002. Endogenous trade policy and lobby formation: an application to the free-rider problem. J. Int. Econ. 57 (2), 449–471. Matschke, X., Sherlund, S.M., 2006. Do labor issues matter in the determination of US trade policy? An empirical reevaluation. Amer. Econ. Rev. 96 (1), 405–421. Mayda, A.M., Rodrik, D., 2005. Why are some people (and countries) more protectionist than others? Eur. Econ. Rev. 49 (6), 1393–1430. Wang, X., Li, K., Xie, S., Hou, J., 2013. How is US trade policy towards China determined? A political economic analysis illustrated by voting outcome of the PNTR bill. China Econ. Rev. 27, 25–36.