Domestic Politics and the CET

Research on the structure of the CET attributes little importance to the political influence of domestic economic interests. Instead, the CET is often seen as the result of negotiations based on expert knowledge and technical issues. Botto and Quiliconi (Reference Botto and Quiliconi2010) argue that private lobbies did not take an interest in the CET until the negotiations were well under way and that negotiators received little input from businesses. Only those sectors with the greatest lobbying power activated their contacts among negotiators and ministerial officials (Botto and Quiliconi Reference Botto and Quiliconi2010). Bianculli and Botto (Reference Bianculli and Botto2009) show that insufficient private sector input and the dearth of adequate information hampered Argentine negotiators. Lacking clear technical foundations for their policy positions, Argentine negotiators allowed Brazil to impose its preferred CET (Bianculli and Botto Reference Bianculli and Botto2009). The Brazilian government actively isolated itself from societal interests (Botto and Quiliconi Reference Botto and Quiliconi2010). The creation of the Brazilian Board of Foreign Trade (CAMEX) insulated bureaucrats from private interests and legislators seeking to influence the structure of the CET (Santos Reference Santos2003).

Industrial and constituent interests were nonetheless incorporated into the structure of the CET. Although the CET strongly reflects Brazilian industrial interests, there is no systematic evidence that Brazil unilaterally imposed a CET on the other members, and there is evidence that the CET took into account private sector interests (Olarreaga et al. Reference Olarreaga, Soloaga and Winters1999). Business interests may have been excluded from directly participating in the CET negotiations, but members of the business community were regularly consulted throughout the negotiations (Gardini Reference Gardini2010). As early as 1992, the Argentine Industrial Union consulted with the Argentine Ministry of Foreign Affairs on the negotiations, while the Brazilian government faced pressure from domestic industrial groups (Hirst et al. Reference Hirst, Bezchinsk and Castellana1994). Legislators from Brazil’s southern state, seeking to protect agricultural producers, took an active interest in both intra- and interregional trade liberalization (Hirst Reference Hirst1996).

Beyond the formation of the CET, it is clear that domestic interests influence Mercosur’s trade policy. Domestic economic interests influenced intraregional conflicts and Argentina’s unilateral trade policy (Gómez-Mera Reference Gómez-Mera2009, Reference Gómez-Mera2013). Santana (Reference Santana2001), de Castro Neves (Reference De Castro Neves2003), and da Conceição-Heldt (Reference Da Conceição-Heldt2013) demonstrate the influence of domestic interests on Brazil’s negotiating position on trade with third parties.

The Protection of Subnational Economic Interests in the CET

International negotiations reflect the constraints placed on negotiators by the interests of voters. What negotiators seek during negotiations and what they are willing to give up depends on the constellation of interests waiting at home (Putnam Reference Putnam1988). Negotiations result from a dynamic process in which negotiators forge agreements that satisfy the interests of the executive, the interests of at least a majority of legislators, and the interests of their counterparts. This forces negotiated agreements to reflect the policy preferences of legislators even though they may not have directly participated in the formation of policy (Milner and Rosendorff Reference Milner and Rosendorff1996).

In order to generate domestic support for an agreement, governments commonly use protection or exceptions in agreements to appease industries and constituents (Grossman and Helpman Reference Grossman and Helpman1995). Even when they are excluded from the negotiations because the agreement must be within the winset of all veto players, the ratification and implementation phases give legislators significant influence over the content of international agreements (Martin Reference Martin2000).

We expect negotiators to anticipate the interests of partisan veto players and shape their policy proposals accordingly (Milner and Rosendorff Reference Milner and Rosendorff1996). This is not because negotiators are fully rational and informed, but because previous interactions allow bureaucrats to anticipate support and opposition. The relative stability of subnational economic interests means that subnational interests may not have to lobby for protection if they are ready to act if their interests are threatened. As such, even without directly participating in negotiations or actively lobbying for protection, the interests of subnationally important industries can shape the domestic winset.

Although the executive branch is responsible for negotiations, legislators’ ability to stall ratification or to hold hostage other executive policy initiatives allows subnational interests to shape trade policy. According to former Brazilian president José Sarney, policy proposals are always sensitive to subnational interests because the government “must have the goodwill of legislators” to advance policy; this gives subnational economic interests “an implicit degree of representation” in trade policy (Sarney Reference Sarney2004). Ambassador Paulo Flecha de Lima, who created the Department of Commerce in the Brazilian Foreign Ministry, could look at any trade policy proposal and know exactly which legislators would pressure the government for changes in order to protect their state (Flecha de Lima Reference Flecha de Lima2004). When a policy proposal fails to account for politically important subnational interests, it is “just the opening of negotiations” with representatives (Sarney Reference Sarney2004). According to former ambassador to Mercosur José Botafogo, Brazil’s policy positions within Mercosur were limited by the political reality that local economic interests imposed on the government (Botafogo Reference Botafogo2004).

The policies advanced by Argentina’s Ministry of Economy and its negotiators also accounted for subnational economic interests. Since the list of politically sensitive products from Argentina’s provinces is well known, policymakers always tried to balance the interests of the provinces and national policy objectives (Makuc Reference Makuc2004). Of course, the provinces never left the defense of their interests to chance, and representatives regularly pressured the government (Ochoa Reference Ochoa2004). When important subnational industries are threatened, legislators and governors pressure government officials for protection (Mayoral Reference Mayoral2004). Even before the negotiation of the CET, informal channels of political influence were well established and active (Makuc Reference Makuc2004).

In the case of the CET negotiations, Argentina’s provincial governments had a great deal of informal communication with the Ministry of the Economy. Provincial representatives regularly used their influence to secure protection for their provincial economies in the structure of Mercosur (Mayoral Reference Mayoral2004). Depending on the issue at stake, the governor, a senator, a deputy, or some other provincial representative would ensure that the Ministry of Economy was taking into account their province’s interests (Makuc Reference Makuc2004; Mayoral Reference Mayoral2004). Of course, the long history of trade conflicts in the region meant that provincial interests were well known to negotiators (Ochoa Reference Ochoa2004). As early as 1991, many of the issues that would later emerge in the CET negotiations were incorporated into their negotiating positions (Ochoa Reference Ochoa2004). As such, important aspects of the domestic winset were in place before the CET negotiations.

Subnational Interests and the National Policy Agenda

In Argentina and Brazil, regionalized parties and nonprogrammatic party-voter linkages allow subnational interests to hold considerable influence over policy outcomes (Samuels Reference Samuels2003; Lodola Reference Lodola2010; Borges Reference Borges2011).Footnote ¹ In Argentina, what Gibson and Suárez-Cao (Reference Gibson and Suárez-Cao2010) refer to as a federalized party system results in national governing coalitions that are little more than alliances between provincial parties (Calvo and Escolar Reference Calvo and Escolar2005). Although Brazilian parties have begun to nationalize, the country’s party system remains regionalized (Borges and Lloyd Reference Borges and Lloyd2016). During the period under study, legislative politics in Brazil were in the hands of legislators controlled by their local parties (Ribeiro Reference Ribeiro2013).

In both Argentina and Brazil, the regionalized nature of parties means that presidents depend on the support of national legislators interested in providing goods to their immediate constituents. Regionalization weakens party leaders, makes subnational issues a central focus of legislative careers, and provides the means and impetus for legislators to break with their national party (Gibson Reference Gibson1997; Ames 2001; Jones et al. Reference Jones, Saiegh, Spiller and Tommasi2002; Samuels Reference Samuels2003). Brazilian presidents must build multiparty coalitions, which often requires granting preferential treatment to overrepresented peripheral states (González and Mamone Reference González and Mamone2015). In order to govern, Argentine presidents must maintain unity between a “metropolitan” and a “peripheral” electoral coalition, which forces policies to reflect the interests of both groups (Gibson Reference Gibson1997). This allows a single or a group of provinces (states) to veto policies (Mainwaring Reference Mainwaring1991; Jones et al. Reference Jones, Saiegh, Spiller and Tommasi2002; Samuels Reference Samuels2003).

Although the executive in both countries retains significant control over economic policy, small groups of legislators have been able to influence economic policy in favor of subnational interests (Eaton Reference Eaton2002; Samuels Reference Samuels2003). Electoral considerations often cause Argentine legislators to break with their national party in order to defend the interests of relatively important industries in their provinces (Eaton Reference Eaton2001; Murillo and Pinto Reference Murillo and Pinto2014; Pezzola Reference Pezzola2017). Eaton (Reference Eaton2001) and Pezzola (Reference Pezzola2017) show that small groups of legislators can override the policy proposals of the Argentine executive when such proposals conflict with the interests of important economic sectors in their province. This suggests that the interests of relatively important industries at the subnational level can create veto points based on the interests of what Tsebelis (Reference Tsebelis2002) calls the partisan veto player. Because the creation of the CET required legislative ratification, the presence of partisan veto players representing subnational economic interests shaped the domestic winset of the CET.

Providing for subnational interests does not always require legislators to break with their national party, but it does change how parties organize themselves. The need to cater to subnational interests can produce a logic of legislative activity and organization that goes beyond simple pork barrel politics and serial logrolling, resulting in a legislature built around what Gilligan and Krehbiel (Reference Gilligan and Krehbiel1987) call “gains from exchange.” Party majorities form to collectively maximize the benefit of controlling government while allowing legislators to satisfy the parochial demands of their constituents. Calvo and Leiras (Reference Calvo and Leiras2012) argue that legislative party blocs exist in Argentina to coordinate the provision of goods to subnational interests. Calvo (Reference Calvo2014) provides evidence of gains from exchange in the Argentine legislature. Although scholars have not used the gains from exchange model to explain legislative activity in Brazil, Alston and Mueller (Reference Alston and Mueller2005) and Lyne (Reference Lyne2008) show that Brazilian presidents and party leaders coordinate the exchange of fiscal transfers, as well as of localized public and club goods (“pork”), for policy and to maintain legislative coalitions. As such, the Brazilian legislature seems to operate under a similar organizational logic.

Legislators’ Endogenous Trade Policy Preferences

Private interests clearly influence trade policy outcomes, either because interest groups “capture” the policymakers (Schattschneider Reference Schattschneider1935) or because political actors manipulate trade policy in order to generate political support by providing protection of specific industries (Magee et al. Reference Magee, Brock and Young1989). This article argues that Mercosur’s CET reflects the preferences of subnational economic interests in Argentina and Brazil. This is because protecting subnationally important industries allows legislators to provide direct benefits to their jurisdictions and secure needed electoral support.

The need to generate political support within their jurisdictions should make the policy preferences of legislators endogenous to the interests of their constituents (Moore et al. Reference Moore, Powell and Reeves2013). Endogeneity may stem from the fact that voters, seeking to protect their assets, prefer politicians who protect locally important industries (Blonigen Reference Blonigen2011), because the interests of voters are directly linked to where they are employed (Busch and Reinhardt Reference Busch and Reinhardt2000) or because voters care about substantive representation (Ansolabehere et al. Reference Ansolabehere, Leblanc and Snyder2012). Even if voters do not clearly voice their preferences, their interests may still have “quiet influence” over policy (Bailey Reference Bailey2001). Aware of the sensitivities of industries in their jurisdiction, legislators anticipate those preferences. The result is trade policies that safeguard important sources of local employment and production.

Scholars of Southern Cone politics widely agree that constituents evaluate politicians and parties based on the direct benefits they receive. Electoral institutions and party politics tie legislators to parochial interests and encourage them to cultivate electoral support through the direct provision of pork to their constituents, rather than through programmatic national-level policy platforms (Mainwaring Reference Mainwaring1991; Ames Reference Ames1995; Gibson Reference Gibson1997; Jones et al. Reference Jones, Saiegh, Spiller and Tommasi2002; Lodola Reference Lodola2010; Calvo and Leiras Reference Calvo and Leiras2012; Zarazaga Reference Zarazaga2014). Although pork is a riskier means of cultivating support than clientelism, its lower cost means that politicians will probably use a mixture of clientelism, pork, and public goods to secure votes (Magaloni et al. Reference Magaloni, Díaz-Cayeros and Estevez2007).Footnote ²

We normally think of pork as being provided to constituents through government spending, subsidies, or tax breaks; however, the diffuse costs and direct benefits of trade protection make it a politically efficient means of doing that (Goodhart Reference Goodhart2014). Since vote buying is a common phenomenon in Argentina and Brazil and relies on egotropic voters, it makes sense that voters will reward politicians who protect important industries where they live. This is especially true for industries that are relatively important in specific jurisdictions, since their protection is highly salient and visible to voters.

Because the aggregate costs created by trade protection are not reflected in how egotropic constituents vote, political competitiveness induces legislators to seek protection across a multitude of industries, despite the costs to the national economy (Lyne Reference Lyne2015). This can result in legislators’ exchanging protection for important industries in their province (state) for the protection of other subnationally important industries in other jurisdictions. In the case of Argentina, Pezzola (Reference Pezzola2017) provides evidence of this for unilateral trade policies, and Eaton (Reference Eaton2001) and Murillo and Pinto (Reference Murillo and Pinto2014) find that legislators are willing to break with their national party to protect industries important to their constituents.

Whether and how much protection an industry receives depends on multiple factors. The political influence of an industry is a critical element in any model of the endogenous determinants of trade policy. Yet policymakers must strike a balance between protecting politically influential industries and the welfare effects of their policies (Grossman and Helpman Reference Grossman and Helpman1994). The structure of the CET, therefore, required balancing the need to generate legislative support by protecting specific industries with the terms-of-trade effects of protection. Protection also depends on whether an industry seeks or needs protection (Grossman and Helpman Reference Grossman and Helpman1994), which, in turn, depends on the degree of its need for protection, its participation in intraindustrial trade (IIT), and the degree of import penetration it faces. Legislators’ trade policy preferences and their political support function can, therefore, be understood as based on three factors: the political influence of the industry, the welfare effects of said protection, and whether the industry needs and seeks protection (Grossman and Helpman Reference Grossman and Helpman1994).

We expect the interests of subnationally important industries in Argentina and Brazil to wield significant political influence over legislators and to shape the viable winsets of the CET. As a result, all else equal, subnationally important industries should be more likely to have benefited from protection in Mercosur’s CET than industries lacking importance in the jurisdictions that they inhabit.

This does not mean that nationally important industries do not influence trade policy or receive protection; they do. Important national industries are almost always relatively important in at least one subnational constituency. This research seeks to understand whether an industry’s political influence stems from its absolute size or from its role in the geographically specific jurisdictions it inhabits. If political influence stems only from an industry’s role in the national economy, only large industries should benefit from preferential treatment. However, if political influence is derived from the industry’s role at the subnational level, depending on their subnational importance, large, medium, and small industries may all receive preferential treatment.

Protecting an industry almost always generates adverse welfare effects. Protecting a product with lower import demand elasticity generates a greater dead-weight loss for society and increases the political costs of protection. For this reason, a lower import demand elasticity is normally associated with less protection. However, if the structure of the CET was constrained by the policy preferences of legislators protecting parochial interests, no association may exist between the aggregate welfare costs of protection and whether an industry receives protection. Economic heterogeneity across provinces (states) makes the factors influencing the welfare of subnational constituencies heterogeneous. While less expensive imports may increase the relative income of most citizens, if trade liberalization makes domestic producers uncompetitive, some people in the subnational economies where those items are produced may lose their livelihoods. Since legislators in regionalized, nonprogrammatic parties are mainly concerned with the interests of their immediate constituents, their sensitivity to aggregate welfare should erode. Under these conditions, we should expect no relationship between the import demand elasticity of a product and protectionism.

The prime source of demands for protection is the ability of producers to compete with imports. More competitive industries are less likely to demand protection, since they do not greatly benefit from it (Gilligan Reference Gilligan1997; Levy Reference Levy1997). Competition induces domestic firms to supply domestic demand at world prices. On the other hand, industries lacking a comparative advantage are more likely to seek protection. Therefore, the degree of tariff protection an industry receives from the CET should be directly related to the extraregional revealed competitive advantage (RCA_p) of the industry’s products.

Industries engaged in intraindustrial trade (IIT) have long been seen as noncompeting (Levy Reference Levy1997); however, recent scholarship on IIT finds that it may induce protection. Since firms participating in IIT produce a limited number of specific products, lobbying essentially becomes a private good (Gilligan Reference Gilligan1997). This increases the likelihood of lobbying by eliminating free riders. Moreover, firms in markets characterized by IIT may be more sensitized to the costs and benefits of trade liberalization, making them more likely to take action.

Madeira (Reference Madeira2013) finds that IIT generates shifts in political coalitions, reducing the benefits of industrywide associations and inducing firms to lobby individually. Bombardini and Trebbi (Reference Bombardini and Trebbi2012) argue that firms producing highly differentiated goods should benefit from lobbying individually, since this allows for policy outcomes specifically tailored to their interests. If we assume that individual firms will be more likely to lobby legislators because of the difficulty of penetrating the executive branch (Magee et al. Reference Magee, Brock and Young1989), then we may expect that their interests will be well represented in political systems like those of Argentina and Brazil, where political competition forces politicians to provide direct benefits to their constituents. As such, industries characterized by greater IIT should to be more likely to receive protection.

The received wisdom in the empirical literature on protectionism is that the likelihood of protection increases with greater import penetration (Lee and Swagel Reference Lee and Swagel1997; Thede Reference Thede2005). The logic is that industries and employees facing high or increasing import competition have more to gain from protection, and lobby policymakers harder to gain protection. Greater import penetration may also indicate that the industry is in decline, further incentivizing producers and employees to fight for protection. As such, extraregional import penetration at the product level should have a positive association with protection.

Models of the endogenous trade policy formation often incorporate the average wage paid by an industry. In studies of industrialized economies, scholars argue that high-wage industries are less likely to lobby for protection (Finger and Harrison Reference Finger and Harrison1994). The higher an industry’s average wage, the smaller the role of capital in the value of production, which decreases the potential losses for owners of sector-specific capital from foreign competition. Wages are also seen as a proxy for a sector’s competitiveness. However, in developing countries, the opposite may hold for an industry’s ability to demand tariffs. Higher-wage earners in the industrial sector tend to be better organized and to wield greater political clout. Given the differences in the expected relationship between wage levels and protection, it is unclear what association should exist between average wages and the structure of the CET.

The Statistical Model, Variable Specifications, and Data

During the negotiation of Mercosur’s CET, negotiators were caught between the preferences of national legislators seeking to protect their constituencies and the desire to reach a successful and beneficial agreement. This presented Argentine, Brazilian, Paraguayan, and Uruguayan negotiators with the challenging task of reconciling divergent domestic interests with their goal of establishing a viable CET.

To assess how domestic interests influenced the structure of Mercosur’s CET, we could assume that each country had equal weight during the negotiations. Mercosur formally gave equal weight to all members during the negotiation of the CET; however, this formal status does not translate into equal weight during negotiations. Smaller countries had far more to lose from a breakdown in negotiations, giving Brazil, and to a lesser extent Argentina, greater leverage. Granting equal influence during negotiations also implies equal interest by all countries, across all products and industries. Given significant differences in the makeup of each country’s economy, it is reasonable to assume that each government would defer to others in some areas and seek to influence the tariffs on products of interest at home.

Cadot et al. (Reference Cadot, De Melo and Olarreaga1996) suggest that the negotiated CET reflects the weighted sum of members’ politically optimal tariffs. Olarreaga et al. (Reference Olarreaga, Soloaga and Winters1999) build on this idea and show that Mercosur’s CET is best explained by the production-weighted sum of the political economy variables in member countries. Calfat et al. (Reference Calfat, Flores and Ganame2000) argue that the weighted sum of each country’s variables provides better explanatory power and seems more realistic. The weighted sum approach is also used to explain the structure of the European Union’s CET (Ehrlich Reference Ehrlich2009; Bandyopadhyay et al. Reference Bandyopadhyay, Lahiri and Roy2011). Therefore, the following equation is used to model the endogenous sources of trade protection within the 1995 CET:

$CE{T_{p:i}} = {\beta _0} + {\beta _k}\mathop {\rm{\Sigma }}\limits_{c = A}^U \theta _i^cPV_{p:i,k}^c + {U_{0i}} + {e_{p:i}}$

where i refers to the industry, CET_p:i is the common external tariff of product p in industry i, and β_k is a vector of the estimated coefficients of the k explanatory variables. The vector $\theta _i^c$ is the share of country c in the total production of sector i, and $PV_{p:i,k}^c$ is the matrix of the k explanatory variables for country c, where p:i indicates product p and the related industry i.Footnote ⁴ U _0i is the industry-dependent intercept. Data for the explanatory variables are from 1994 or the closest year to 1995 for which data are available. Industrial data are from 1993. Data sources can be found at the end of the article.

Mercosur’s CET covers more than nine thousand tariff lines, and trade data are also available at the 8-digit level of Mercosur’s Common Nomenclature. Unfortunately, industrial data are available only at a far lower level of desegregation. Argentina’s Censo Nacional Económico 1994 reports industrial data disaggregated into 166 sectors. Brazil’s Pesquisa Industrial Annual disaggregates industrial data into 61 sectors for Brazil’s five regions.Footnote ⁵ Paraguay and Uruguay report data at a similar level of disaggregation, but do not report subnational data.

The difference in levels of aggregation presents two options: analysis at the industrial level or analysis at the level of trade and tariff data. Analysis at the industry level implies an important loss of trade and tariff information. Moreover, variation of tariff levels within industrial sectors suggests that explaining trade policy requires running the statistical analysis at the tariff level to account for extra- and intraindustry differences. For this reason, a multilevel two-sided tobit model is used.Footnote ⁶

Variable Specifications

The seven statistical models employed to estimate the endogenous determinates of the CET use eight variables as proxies for the three aspects of legislators’ political support function (political influence, welfare effects, and demands for protection) and as alternative explanations for protection. These models incorporate legislative interests only as explanatory variables, since estimating models incorporating the interests of the executive would require establishing arbitrary weights for the interests of each and is beyond the scope of this research.

The political influence of an industry is usually measured using its absolute size in the national economy. Scholars assume that larger industries have greater political influence because they have more resources at their disposal to effectively lobby politicians and back the electoral campaigns of officials who support their interests (Milner Reference Milner1988; Grossman and Helpman Reference Grossman and Helpman1994; Lee and Swagel Reference Lee and Swagel1997).

While there is broad empirical support for the argument that larger industries receive more protection, there are numerous reasons to doubt that the absolute size of an industry has a direct relationship to the trade policy preference of legislators. Instead, industries important to the national economy receive preferential treatment because their size makes them important to the constituents of specific subnational jurisdictions. Receiving preferential treatment depends on gaining the attention and support of elected officials. Caves (Reference Caves1976) argues that the support of elected officials depends on the relative importance of an industry in their constituencies. For legislators, this is the industry’s relative importance in the economy of their province or state. Whether this is because legislators vote with the economic interests of their constituents (Arce et al. Reference Arce, Koopman and Tsigas2008), because they fear electoral repercussions from negative policy outcomes (Bailey Reference Bailey2001), or because they are sensitive to the interests of import-competing firms in their jurisdiction (Baldwin Reference Baldwin1985), there is good reason to expect that the interests of relatively important industries at the subnational level shape the policy preferences of legislators. Moreover, as the previous discussion has demonstrated, we have clear reasons to believe that subnational economic interests influence Argentine and Brazilian legislators.

Following Pezzola Reference Pezzola2013, the relative Subnational Economic Importance_i of an industry across the provinces (states) that it inhabits measures the influence of subnational economic interests on the policy preferences of legislators.

$Subnational\,Economic\,Importanc{e_i} = \mathop {\rm{\Sigma }}\limits_k {{Productio{n_{i,k}}} \over {GD{P_k}}}/n$

where Production_i,k is the value of production of industry i in jurisdiction k in U.S. dollars (USD) in 1993, GDP_k is the GDP of jurisdiction k in tens of thousands of U.S. dollars, and n is the number of jurisdictions.Footnote ⁷

To take into account the welfare effects of protecting an industry, the Import Demand Elasticity_p for each product in 1994, estimated by Kee et al. (Reference Kee, Nicita and Olarreaga2008), is used to measure how changes in the terms of trade would influence aggregate welfare.

Whether an industry lobbies for protection plays a key role in the trade policy preferences of legislators. The prime source of demands for trade protection is producers’ ability to compete with imports. More competitive industries are less likely to lobby for protection, and tariff levels should be directly related to the extraregional Revealed Competitive Advantage_p (RCA_p) of the industry’s products. The RCA_p for each product p in 1994 is estimated as

$RC{A_p} = \left( {X_{p,c}^*/{X_{p,m{s^\prime }}}} \right)/\left( {X_c^*/{X_{m{s^\prime }}}} \right)$

where $X_{p,c}^*$ is extraregional exports of product p by country c, X_p,ms′ is the total exports of product p by non-Mercosur countries, $X_c^*$ is country c’s total extraregional exports, and X_ms′ is total exports by non-Mercosur countries.

If legislative partisan veto players shaped the domestic winset of the CET, products characterized by higher levels of IIT should have received more protection. This is because IIT induces firms to lobby individually, and individual firms are more likely to lobby the legislators who represent them. To evaluate this assertion, the Grubel–Lloyd index is used to measure extraregional IIT in 1994 at the product level.

$II{T_p} = 1 - {{\left| {X_p^* - M_p^*} \right|} \over {X_p^* + M_p^*}}$

where $X_p^*$ and $M_p^*$ are extraregional exports and imports of product p.

The level of extraregional Import Penetration_p at the product level in 1993 and the average Wages_i paid by an industry are also used as proxies for the likelihood that an industry will effectively seek protection. $Import\,Penetratio{n_p} = M_{p:i}^*/Productio{n_i}$, where $M_{p:i}^*$ is imports of product p of industry i from non-Mercosur countries in 1993 and Production_i is the total value of Production_i of the industry (thousands of USD) in 1993. Average Wages_i = total wages_i / employees_i, where total wages_i is all wages paid by industry i (thousands of USD) and employees_i is the total employment by the industry in 1993.

The absolute size of an industry and its concentration across jurisdictions are used as alternative indicators of an industry’s political influence. The importance of an industry in the national economy is measured using the total value of Production_i of the industry in 1993. Following Busch and Reinhardt Reference Busch and Reinhardt1999, the Herfindahl index of industrial employment across jurisdictions is used to estimate the Political Concentration_i of an industry. The index of Political Concentration_i is the sum of the squared ratio of an industry’s employment in each jurisdiction and its total number of employees in 1993: Σ_k (employees_i,k / employees_i)². High levels of Political Concentration_i indicate that the industry is concentrated in a few jurisdictions, which should facilitate collective action but limits the number of legislators who represent it.

Empirical Results

The empirical results provide strong support for the argument that Mercosur’s CET was influenced by the economic structure of Argentina’s and Brazil’s provinces (states). The subnational importance of an industry is positively associated with protection. The results also indicate that policymakers were indifferent to the aggregate welfare effects of the CET as measured by Import Demand Elasticity_p, suggesting that the CET reflects parochial interests. Products with higher levels of IIT also received greater protection. As expected, competitive industries received less protection, and higher levels of Import Penetration are associated with greater protection. Average Wages_i have no association with protection.

Table 1 presents the estimated coefficients of seven models of the endogenous determinates of Mercosur’s CET.Footnote ⁸ The first model estimates the influence of subnationally important industries on the structure of the CET and provides evidence that subnational constituent interests influence the structure of the CET. To evaluate whether the political influence of an industry stems from its size in the national economy rather than its Subnational Economic Importance_i, three national-level models are estimated using only an industry’s absolute size as a producer (Production_i), its Political Concentration_i, and the interaction between Production_i and Political Concentration_i to measure an industry’s political influence.

Table 1 Endogenous Formation of Mercosur’s CET

^* p < 0.05

^** p < 0.01

^*** p < 0.001

Standard errors in parentheses.

Note: Estimated with xttobit in STATA 12.

Three “full” models include indicators of political clout at both the subnational and national levels. Estimating the national-level and full models allows us to compare the explanatory power of Subnational Economic Importance_i against more traditional measures of political clout. Making this comparison helps determine whether the political influence attributed to industries stems from the importance of the industry in the national economy or, as argued by this research, that a large part of the political influence normally ascribed to nationally important industries stems from their relative importance in the political jurisdictions they inhabit.

In all the models where Subnational Economic Importance_i is present, it has a positive relationship with protection. The logged values of Subnational Economic Importance_i range from 1.19 to 5.49, which means, ceteris paribus, that the industry with the greatest subnational clout has, on average, an estimated tariff 18.8 percentage points higher than the industry of least subnational importance. Since the CET was bound between 0 percent and 20 percent, the Subnational Economic Importance_i not only has a statistically significant association with protection, it has a clear substantive importance. This provides strong evidence that an industry’s relative importance in the provinces (states) it inhabits was more likely to obtain protection.

To better understand the role that subnational economic interests play in determining trade policy outcomes, it is worthwhile to examine one of the many industries that received more protection than we would expect, based on its importance in the national economy (Production_i). For instance, model 2, which uses only Production_i as a measure of political influence, appreciably underpredicts the level of protection received by producers of TVs, radios, and other communication equipment, an industry with low values of Production_i. This is because the composition of the jurisdictions where production of communication equipment is located gives the industry greater political clout than its absolute size (Production_i) would suggest. A sizable portion of the industry is located in the economically less robust states of northern Brazil, which gives the industry a high value of Subnational Economic Importance_i in Brazil. Portions of the Argentine industry are also located in the smaller economies of Salta and Río Negro, increasing its Subnational Economic Importance_i to Argentine legislators.

Since the policy preferences of legislators are endogenous to those of their constituents, we can assume that legislators from northern Brazil and from Salta and Río Negro sought to protect this industry. For this reason, the subnational model (model 1) closely predicts the level of protection the industry received. The industry’s subnational importance gave legislators from northern Brazil and from Salta and Río Negro a clear stake in defending the industry. The economic interests of their constituents influenced the composition of the domestic winsets of Argentina and Brazil. As a result, negotiators should have hesitated before advancing a CET that jeopardized the interests of producers of communications equipment.

Models of the endogenous formation of trade policy typically use an industry’s absolute size (importance) in the economy (Production_i) to measure an industry’s political clout. To compare the explanatory power of an industry’s Subnational Economic Importance_i with that of the industry’s importance in the national economy, several models were estimated incorporating the absolute size of an industry (Production_i) as an explanatory variable. In Models 2–4, which exclude Subnational Economic Importance_i, the coefficient for Protection is positive and significant, which fits with the expectations of traditional models of trade protection. However, in the full models (models 5–7), the importance of the industry in the national economy (Production _i) is not statistically significant, but the estimated coefficient of Subnational Economic Importance_i remains statistically significant and substantively large. This indicates that the subnational importance of an industry provides a better indicator of an industry’s political influence than its absolute size.

This conclusion is buttressed by the maximum likelihood tests of model 5 against the subnational and national-level models. The addition of Production to the subnational model (model 5 vs. model 1) does not significantly increase the fit of the model, but the addition of Subnational Economic Importance_i to the national-level model increases the overall fit (model 5 vs. model 2). We can also compare model 1 and model 2 using their Bayesian information criterion (BIC). The smaller BIC of model 1 indicates that the subnational model more accurately represents the process behind the formation of the CET.

We may be tempted to assume that the explanatory power of Subnational Economic Importance_i stems from a few nationally important industries with subnational bases and that influence really comes from an industry’s role in the national economy. If this were the case, we would expect Production_i to be statistically significant in models 5–7 and the maximum likelihood test to indicate that adding Production_i to the subnational model improved the model’s fit. Moreover, if the influence of Subnational Economic Importance_i stemmed only from a few nationally important industries with subnational bases, the cross-validation errors of model 1 would not be lower than those that take into account an industry’s role in the national economy. The lower cross-validation errors of model 1 compared to model 2 indicate that subnational interests better explain protection over a wide range of products and industries.

The statistical significance of Subnational Economic Importance_i and the superior fit of model 1 are not the only evidence that subnational economic interests influenced the structure of the CET. When subnational economic interests influence policy outcomes, policies are less likely to consider aggregate national welfare, and the terms-of-trade effects of protection should not influence the structure of the CET. As expected, across all models, the coefficients for Import Demand Elasticity_p are statistically insignificant. This provides evidence that the structure of the CET was not significantly influenced by aggregate welfare effects and suggests that the CET catered to parochial interests.

Across all models, products characterized by higher levels of IIT benefited from greater protection. The incentives for firms to form and support industrywide organizations are contradicted by IIT. Lacking the size and organization to penetrate the executive branch, individual firms tend to seek representation through their legislators. When policymakers cater to geographically specific constituencies, individual firms and industries that are important at the subnational level are more likely and better able to lobby for protection, which increases the likelihood of protection for protected goods characterized by higher levels of IIT. As such, the positive and significant coefficients of ITT_p suggest that subnational economic interests shaped Mercosur’s CET.

Conclusions

The creation, maintenance, and evolution of Mercosur have largely been seen as a top-down process dominated by presidents. This research argues that subnational economic interests directly influenced the structure of Mercosur’s CET by shaping the domestic winsets of member countries. Protection was granted to industries of relative subnational importance because the welfare of those industries influenced the policy preferences of the legislators who ratified Mercosur. The structure of the CET was clearly influenced by domestic interests that had deep roots in the subnational economies where voters live and work.

The quantitative analysis of Mercosur’s CET indicates that subnational economic interests influenced the structure of the CET. The statistically insignificant coefficients for Import Demand Elasticity indicate that policymakers were indifferent to aggregate welfare effects of the CET, suggesting a need to cater to subnational economic interests. The positive association between IIT and protection suggests that firms often lobbied individually and, given the difficulty for individual firms to penetrate the executive branch, that they used their access to legislators to influence trade policy. The statistical and substantive significance of Subnational Economic Importance_i indicates that the structure of subnational economies influences which industries received protection and that relatively important industries at the subnational level have greater political clout.

These results not only provide a clearer understanding of the role of domestic interests in the creation of Mercosur, they also inform the evolution of the agreement. There is no reason to believe that policymakers’ sensitivity to subnational economic interests began or ended with the negotiation of the CET. It makes sense that subnational interests also influenced Mercosur’s negotiations with other countries and economic blocs. Legislators’ need to provide direct benefits to their geographically specific constituencies provides subnational economic interests with significant influence over the evolution of Mercosur and has helped generate and maintain a particularly shallow form of integration.

The decision to create Mercosur may have been a reaction by presidents to changes in the international environment and domestic macroeconomic conditions, but it would be incorrect to assume that the same factors that led to its creation also determined its structure and evolution. Mercosur’s failure to evolve and expand clearly stems from domestic factors (Malamud Reference Malamud2013). The lack of support for further integration may stem directly from the entrenchment of subnational economic interests behind the CET. Legislators seeking to protect their constituents may hinder executives’ ability to form the political coalitions necessary for further liberalization.

While the aggregate welfare of member countries may benefit from the establishment of a true customs union, further liberalization, greater integration with the rest of Latin America, or the successful negotiations with the EU, powerful and entrenched subnational interests probably present insurmountable hurdles in attaining these goals. Sensitivity to the economic interests of specific geographic constituencies significantly restricts the flexibility of negotiated outcomes by allowing small economic groups to influence and hold hostage negotiations because of their importance to subnational representatives.

In countries like Argentina and Brazil, as well as India, Mexico, and the Philippines, where the interests of subnational constituencies play a significant role in politics, we cannot fully understand trade policy without a clear understanding of how the constellation of subnational economic interests influences the political feasibility of policies. As long as electoral politics are dominated by regionalized, nonprogrammatic political parties elected in territorially specific geographic constituencies, we should not expect the influence of these interests to dissipate.