Trade protectionism and electoral outcome.

• Author: Yu, Miaojie

In the 2004 U.S. elections, the dispute on trade policy was one of the important issues between the Democrats and the Republicans. The Democratic Party emphasized that the U.S. government should link multinational trade agreements to labor and environmental issues. Conversely, the Republican Party advocated free trade and pledged to make the expansion of trade a consistent priority. This dispute over trade policy raises an interesting question: Does a party's tariff platform affect its electoral outcome? It is recognized that, in the late 20th century, the Democratic Party was labeled as a promoter of protectionism (Magee, Brock, and Young 1989). Simultaneously, the Democrats lost their longtime majority in the House of Representative with the election of the 104th Congress in 1994. Hence, it is natural to ask: Did the high Democratic tariff platform translate into its falling vote share?

Previous studies have recognized that campaign contributions and lobbies affect trade policy. The related theoretical models on the political economy of trade policy have two different approaches. The first one is the median-voter approach, developed by Mayer (1984), Mayer and Li (1994), and Magee, Brock, and Young (1989).

According to the first two studies, trade policy is determined by majority voting, while political parties maximize their probability of winning the election by choosing a tariff platform. Similarly, Magee, Brock, and Young (1989) considered a political model of endogenous policy formation. Under their framework, two parties set their trade platforms and then interest groups contribute funds in order to increase the probability that their preferred party will be elected.

The second theoretical approach is called political support. This approach rationalizes that tariffs are granted in response to demands made by special interest groups such as industries and unions. The government then weighs the increased political support it generates from pursuing policies beneficial to a particular industry against the support it loses from firms and consumers (Hillman 1989). Grossman and Helpman (1994) developed this idea and found the optimal tariff formation for the government. Models with the political-support approach explicitly present the relationship between campaign contributions and optimal tariffs, yet they do so at the expense of ignoring political competition between parties.

On the empirical side, previous works have focused mostly on the factors that determine congressional roll-call vote patterns on American trade policy. These works include those of Baldwin (1985), Irwin (1996), and Beaulieu (2002). The Stolper-Samuelson theorem implies that trade policy is independent of industry and depends only on the type of factor ownership. Based on this prediction, Baldwin (1985) published the pioneering work that examined the determinants of congressional vote patterns on trade legislation for the Tokyo Round of GATE Shortly thereafter, Irwin (1996) investigated this issue using data from the British general election of 1923, where trade policy was the primary issue. Both papers suggested that stronger labor unions are likely to be associated with a higher level of trade protectionism.

Two more findings are important in understanding the vote pattern from the political science perspective. First, labor unions are a strong electoral base for the Democratic Party. In addition, the Democrats also maintain a strong influence on groups such as African-American voters and metropolitan residents. Second, campaign contributions play an important role in the voting pattern. For example, Snyder (1990) found evidence of a positive relation between a political candidate's probability of winning and the amount of money the candidate collects from campaign contributors. Levitt (1994), meanwhile, found that challenger spending is marginally more productive than incumbent spending. Baldwin and Magee (2002) also found evidence that voting behavior on recent trade bills can be explained by campaign contributions from special interest groups. In particular, financial contributions from labor groups were associated with votes against free trade, while contributions from business groups were associated with votes against protectionism.

However, whether or not a party's ex ante tariff platform has any effect on its congressional vote pattern still deserves careful study. To address this question, we first consider a theoretical model with two parties (the Democratic Party and the Republican Party) in this article. The Democratic Party is assumed to be a protectionist party and has traditionally shown a preference for high tariffs. Each party has some ideological voters in the election. Therefore, the aim of each party is to maximize its vote share within the non-ideological voters by choosing a tariff platform. By doing so, campaign contributions are collected from decided non-ideological voters and such funding influences the vote of some of the undecided non-ideological voters. Therefore, a high tariff platform has two opposite effects on a party's (say, the Democratic Party's) vote share. On the one hand, a high tariff platform implies more contributions to the Democratic candidates which, therefore, could lead to a larger vote share. On the other hand, such a high tariff harms the relative capital-abundant (i.e., wealthy) owners according to the Stolper-Samuelson theorem. Therefore, some of these owners (i.e., those relatively wealthy voters) could switch their positions and vote against the Democratic Party. The net effect of the Democratic tariff platform on its vote share still remains an empirical question.

In this article, I estimate the impact of the Democratic tariff platform on its electoral outcome using congressional district data on election to the House of Representatives from the years 1982 to 1994. I control for many other financial, historical, and social factors that could affect the voting outcome. Aside from this, I also address the endogeneity issue using the instrumental variable (IV) approach. The estimation results clearly suggest three points. First, a higher Democratic tariff platform results in a higher Democratic vote share by increasing campaign contributions. Second, a higher tariff platform may also lead to a lower Democratic vote share due to the lost support from its decided non-ideological voters. These two findings are consistent with the theoretical predictions. Finally, the net effect of the Democratic tariff platform on its election outcome is shown to be significantly negative. Indeed, a high tariff platform cannot help the Democrats win more seats in House elections, though it does help them collect more campaign funding from protected industries and labor unions.

The Theoretical Model

Consider a 8x2x2x2 model: three types of voters, two parties, two goods, and two factors. A capital-abundant country produces a capital-intensive commodity and a labor-intensive commodity using capital and labor. Two political parties--the Democratic Party (D) and the Republican Party (R)---compete for office via their trade policies. Parties are partially ideological in the sense that the Democratic Party historically prefers protectionism whereas the Republican Party inclines toward a pro-trade policy. Correspondingly, each party maintains some ideological supporters. Hence, I model [τ.sup.D] >[τ.sup.R], where [τ.sup.D] is the Democratic tariff platform while [τ.sup.R] is the Republican one. I assume that each party also tries to maximize its vote share within the non-ideological voters to win the election by choosing tariff platforms. (1)

There are three types of voters in the model: decided, undecided, and ideological voters. Decided voters prefer a particular party while undecided voters do not. One example is that some voters may spend more money on imported goods and therefore care more about the import tariff. For this reason, they have the inclination to make campaign contributions to their preferred party and get benefits from their preferred trade policy if their preferred party wins the election. (2)

In contrast, undecided voters do not have a prior preference for a specific party, and their position is easily swayed by political advertisement. Finally, ideological voters may have some ex ante ideological preference for a specific party, but their position cannot be affected by either a party's tariff platform or its political advertisement.

Decided voters j∈ {1,... J} share the same additively separable quasi-linear preferences and maximize their utilities, [U.sub.j] = [Y.sub.1j] + u([y.sub.2j]), subject to their budget constraints [y.sub.1j] + [py.sub.2j] = [E.sub.j] where p is the relative import price, [y.sub.1] is a capital-intensive good, [y.sub.2] is a labor-intensive good, and [E.sub.j] is voter j's income. Each voter has the same labor endowment, which is normalized as a unit but with different capital endowments ([K.sub.j]). The import tariff income (T) is also redistributed with a lump sum subsidy. Thus, voter j's indirect utility function is:

(1) [υ.sub.j](p) = CS(p) + w(p)+r(p) x [K.sub.j] + T(p)/[bar.L],

where CS(p) is the consumer surplus, w is wage, r is capital return, and [bar.L] is the aggregate labor endowment for the whole labor force.

A voter would support the Democratic Party if and only if that voter's indirect utility under a Demoeratic tariff platform is higher than that under a Republican tariff platform. Otherwise, the voter would vote for the Republican Party. With the Metzler paradox ruled out, the import tariff (τ) is assumed to be a specific one, p = τ + [p.sup.w], where [p.sup.w] denotes the world price. This implies that the set of decided non-ideological voters for the Democratic Party, [Ω.sub.D], is {[K.sub.j]|[υ.sub.j] ([τ.sup.D]) - [υ.sub.j]([τ.sup.R])>0}. From (1), we have:

(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

The supporting...