nueva página del texto (beta)
Inglés (pdf)
Artículo en XML
Referencias del artículo
Traducción automática
Enviar artículo por email
Citado por SciELO 
Similares en
SciELO 
Permalink
1. Introduction
Since the formation of the North American Free Trade Agreement, participation in regional trade agreements (FTAs) has significantly altered how developing countries interact with the global economy. Mexico has been at the forefront of this trend, using FTAs as part of its development strategy. Although significant research has been conducted into why governments pursue and sign FTAs, few scholars have sought to understand the factors that shape the final structure of agreements signed by developing countries.
The ability to exclude or delay the liberalization of specific products and industries makes FTAs easier to negotiate and more feasible tan agreements lacking these escape valves. Notwithstanding this greater flexibility, domestic resistance to trade liberalization protracts many FTA negotiations (Manger, 2009; Nicolas, 2008).
Much like the negotiationsbof the Trans-Pacific Partnership (TPP), the negotiation of the Japan-Mexico FTA1 stalled several times due to the challenges of negotiating market access. In both cases, Japanese reticence to liberalize agricultural imports received significant media and scholarly attention. However, Japanese agriculture was not the only obstacle to trade liberalization. The political forces behind market Access negotiations went far beyond a few specific industries or products. A better understanding of the Japan-Mexico FTA can, therefore, help us understand the obstacles faced by negotiators as well as the final structure of both agreements.
Lobbying by societal forces clearly plays a role in the politics of FTA formation; otherwise, we would not see variation in the speed of liberalization across products and industries. Tariff reduction schedules (TRSs) provide an important mechanism for cultivating political support and for saving otherwise politically unfeasible agreements (Grossman and Helpman 1995). This makes TRSs a common feature of FTAs. Despite significant research into why states choose to negotiate and form FTAs, it remains unclear why specific industries are able to resist liberalization or how societal interests influence market access within FTAs. The analysis of Mexico's TRS within the Japan-Mexico FTA finds that subnational economic interests shaped the agreement's structure. The relative importance of industries within Mexico's states directly influenced whether an industry would benefit from preferential treatment.
Building on the work of Pezzola (2013), this article investigates the influence of subnational economic interests on the structure of Mexico's TRS towards Japan. Taking into account the size and interests of industries as well as the economic interests of subnational constituencies, this research demonstrates that the economic interests of territorially specific jurisdictions played a fundamental role in determining which industries Mexican politicians chose to protect. This finding makes two important contributions to our understanding of the formation of FTAs. First, the paper provides evidence that subnational economic interests influenced the negotiation and structure of the Japan-Mexico FTA by influencing the shape of the winning coalition in Mexico. This moves us away from the common three phase sequential analysis of FTAs: a) governments decide to negotiate; b) governments negotiate; c) governments seek political support to ratify the treaty. Second, the results indicate that, when legislators respond to territorially specific constituencies, the economic interests of these constituencies influence which industries receive preferential treatment.
The interests of firms and industries clearly play a role within FTA negotiations, but the factors that give specific firms or industries political clout depends on the institutional setting and economic geography of the country. Just as the scholarship on multilevel government shows that business interests may use subnational channels to influence trade policy, this research shows that territorial representation influences trade policy outcomes by giving voice to subnational economic interests. As uch, this article provides insight into the factors that influence the trade preferences and negotiating position of countries with strong systems of territorial representation, where "Janus-faced" governments react to geographically specific interests (e. g. Argentina, Australia, Brazil, India, Malaysia, Mexico, and the United States).
Although it is widely accepted within the trade policy literature on the United States that constituent interests influence policy outcomes (see Arce, Koopman and Tsigas, 2008; Fordham and McKeown, 2003), almost no research has sought to understand how the economic interests of territorially specific voters influence trade policy in developing countries. Understanding when and how subnational interests influence the trade policy of developing countries is especially important if we wish to understand the factors shaping the international economy. Since the 1990s, the number of FTAs between developing countries has increased six-fold and those between developed and developing countries has increased four-fold. Trade involving developing countries is also occupying a greater share of international transactions.
Although countervailing lobbying clearly influences which industries politicians protect (Manger, 2005; Solís, 2010), taking these forces into account is beyond the scope of this article. Doing so would require modeling the multilevel interactive negotiation process. Before beginning to taking into account the influence of countervailing foreign lobbies on a country' trade policy, it seems logical to begin by understanding how domestic interests influence protection, all things else being equal. As Putnam (1988) points out, while this type of decomposition may not be descriptively accurate, the expectation effects generated by domestic interests should have a direct influence over negotiated outcomes.
The rest of this article is divided into four sections. The next section discusses the literature on endogenous trade policy formation, with specific reference to Mexico, and presents the central hypotheses. This is followed by a discussion and description of the statistical models and variables used. A presentation of the empirical results follows along with a brief conclusion.
2. Literature and Theoretical Predictions
General Considerations
The literature on the Japan-Mexico FTA focuses almost entirely on the role of Japan. Authors have sought to explain Japan's move away from regionalism and towards bilateralism as well as Japan's decision to pursue cross-regional FTAs (Manger, 2005; Solís and Katada, 2007). The role of Japanese business in supporting the FTA has also been studied (Solís, 2010; Yoshimatsu, 2006). Others have investigated how resistance to liberalizing Japanese agricultural imports threatened negotiations (Mercurio, 2009; Solís and Katada, 2007; Yoshimatsu, 2006). However, no research has looked into the factors that influenced which industries Mexico sought to protect within the agreement.
Although the initiative of bureaucrats (Solís and Katada, 2007), strategic interests (Ravenhill, 2003), the interests of Japanese foreign direct investment (Manger, 2005), and a desire to protect exporters (Manger, 2009; Solís and Katada, 2007) help explain the proliferation of Japanese participation in FTAs, they cannot explain why specific industries received preferential treatment within the Japan-Mexico FTA.
Which industries receive preferential treatment depends on how political institutions filter private interests and the dispersion of economic activity across political jurisdictions. Political institutions clearly influence which groups policymakers privilege through trade policy (Garrett and Lange, 1996; McGillivray, 2004). We know that the strength of a country's party systems (Grossman and Helpman, 2005), the number of veto-points (Mansfield and Milner, 2012), and the strength and autonomy of the executive branch (Thacker, 2000) influence the susceptibility of policymakers to constituent interests. The number of access points to the policymaking process also favors the influence of particularistic interests (Ehrlich, 2007).
The influence of the private sector on Mexican trade policy has mainly been credited to the country's largest firms through their participation in peak associations. In 1987, President de la Madrid organized the Economic Solidarity Pact (PSE) among government, business, and labor to help stabilize the economy and advance previously ineffective liberalization efforts. This public-private pact provided critical support for the government's liberal trade policies, but did not extend far beyond the representation of the largest firms (Kaufman, Bazdresch and Heredia, 1994). In the 1990s, President Salinas reinforced the state-industry coalition by inviting large industrial groups to present their position on the negotiations taking place with the U. S. and Canada (Thacker, 1999). These interactions between the government and industry took place at the highest levels and usually first passed through the peak organization Business Coordinating Council (CCE), which created the spin-off Foreign Trade Business Coordinating Council (COECE) to funnel input and information directly into FTA negotiations. Although the CCE and COECE nominally represent all business interests, they overwhelmingly favor those of the largest firms and industries (Thacker, 2000; Shadlen, 2000; Tirado and Luna, 1995). Constructed to advance specific policy projects these alliances became institutionalized and give specific elements within the private sectors privileged access to the policymaking process and trade negotiations (Hogenboom, 2014; Kaufman et al., 1994; Thacker, 2000).
The Role of Subnational Economic Interests
Only focusing on the influence of peak associations and big business neglects the changing reality of Mexican politics. Until the political reforms of the 1990s, avenues of political pressure usually open to economic interests in other countries were closed in Mexico (Scheider, 2002; Shadlen, 2000). However, a series of reforms between 1989 and 1996 increased electoral competition and has made policymakers sensitive to societal pressures. Elected officials are no longer unresponsive to the interests of territorially based voters. Mexican legislators still depend upon national party executive committees for campaign financing and career advancement, but legislators and parties as a whole have become attentive to territorially based constituents (Cantú and Desposato, 2012; Langston, 2010; Lehoucq et al., 2005). The interests and preferences of Mexico's governors can also strongly influence legislative conduct and policymaking (Langston, 2010; Rosas and Langston, 2011). Although Kerevel (2015b) argues that theses authors overstate the influence of governors on the political careers of legislators, Mexican legislators do provide particularistic benefits to their constituents (Kerevel, 2015a).
Political institutions influence which interests policymakers support, but this is only one side of the interaction between policymakers and their constituents. Political institutions influence whether politicians react to the interests of specific political jurisdictions, but they cannot tell us what those interests are. To identify the interests of constituents we must look to the local economies where voters live and work.
Legislators from different jurisdictions should have different interests when it comes to which specific products or industries receive protection. The policy preferences of legislators are endogenous to the interests of their constituents (Chappell, 1982; Moore, Neff Powell and Reeves, 2013). As a result, the policy preferences of legislators and how they vote on trade policy is directly related to the trade sensitivity of their jurisdictions (Arce et al., 2008; Baldwin and Magee, 2000; Fordham and McKeown, 2003). This is either because subnational economic interests have a selection effect on legislators (Bombardini and Trebbi, 2011), an influence effect on legislators (Stratmann, 1991), or because of the power if constituent interests at the ballot box (Arnold, 1992). Either way, the specific distribution of economic activity across political jurisdictions should influence the policy preferences of legislators.
The degree of trade protection received by an industry depends on three factors: the political influence of the industry, the welfare effects that protection would have on society, and whether an industry seeks protection. The political influence of an industry is a critical element in any model of the endogenous determinants of trade protection. Essentially policymakers strike a balance between the interests and political influence of industries and the welfare costs of protection to consumers (Grossman and Helpman, 1994; Hillman, 1989). The terms-of-trade effects of protection influence aggregate national welfare and, therefore, the costs to politicians of protecting an industry. The competitiveness of an industry as well as the degree of import penetration and the degree of intra-industrial trade (IIT) that it faces directly influence the likelihood that an industry will seek protection.
Typically the political clout of an industry is thought to be directly proportional to the size of the industry. This is because larger industries should have more resources to contribute to politicians and because they represent more voters. However, in the case of Mexico, this article argues that we should expect policymakers to be more sensitive to the interests of industries of greater relative importance within the political jurisdictions that they inhabit and that the absolute size of an industry is not directly related to its political influence.
Since both individuals and firms perceive the consequences of trade policy within the context of their location, a growing literature investigates how subnational interests influence policy outcomes. The specific constellation of the local economy may either provide firms with allies or may drown out their petitions for protection. Once we begin to examine the types of policies that politicians propose or support, it is clear that the specific interests of politicians' constituencies influence their preferences and policy outcomes. Representatives tend to advance the interests of relatively important industries within their jurisdictions (Finger and Harrison, 1994). Grossman and Helpman (2005) find that, when legislators are tied to particular geographic jurisdictions, differences in the distribution of industries across jurisdictions induce protection of relatively important industries within jurisdictions. There is also clear evidence, at least in the United States, that subnational trade sensitivity influences how legislators vote on trade policy (Arce et al., 2008; Baldwin and Magee, 2000; Fordham and McKeown, 2003) and that voters respond to substantive representation (Ansolabehere and Jones, 2010). The perceived quality of information provided by locally important industries should also grant greater political leverage to subnationally important industries (Bennedsen and Feldmann, 2002).
Since the mid-1990s, the Mexican Congress has taken a more active role in international trade negotiations. Although negotiations remain the purview of the Executive, Congress now pays greater attention to negotiations and the ratification process gives legislators significant influence of the shape of Mexico's FTAs. Legislative ratification of agreements forces the Executive to negotiation agreements that resemble the preferred policy outcome of legislators (Milner and Rosendorff, 1996). Even if the legislature did not have to formally ratify FTAs it could still influence the structure of the agreement (Martin, 2000). This suggests that Mexico's TRSs should grant preferential treatment to important industries within the economies of Mexico's states.
Hypothesis 1: Industries of greater relative subnational importance are more likely to benefit from preferential treatment (greater protection) within Mexico's TRS.
The Supply of Protection
Before protecting an industry, politicians should evaluate the aggregate welfare effects of protection against the political support that it generates. Grossman and Helpman (1994) argue that the welfare effects generated by changes in the terms-of-trade directly influence the likelihood of a product receiving protection. Protecting sectors facing high foreign export elasticity generates a greater deadweight loss to society as a whole, increasing the political cost of restricting imports (Feenstra, 2004). All else being equal, politicians prefer to protect industries where the deadweight loss is lower (Hillman, 1989). However, this suggests that national welfare, not the interests of specific constituencies, play a fundamental role in forming politician's policy preferences. Given that the political careers of Mexican legislators depend to a significant extent on the support of their constituents and governors, their preferences should mainly reflect the interests of specific geographic constituencies and the structure of TRSs policies should be indifferent to aggregate welfare losses.
Hypothesis 2: The foreign export elasticity faced by products should have no significant relationship with protection.
Demands for Protection
Whether an industry seeks protection plays an important role in trade policy outcomes. Just as politicians face costs for protecting an industry, seekingv protection imposes costs on interested parties. Faced with finite resources to lobby policymakers, industries will use their resources where they have the greatest utility, which may include seeking changes in other policy arenas. Hence, it is necessary to account for the utility of lobbying for trade protection to understand which industries receive protection.
It has long been argued that firms engaged in intra-industry trade (IIT) are non-competing and, therefore, do not seek protection. This is because represents an evolution towards a high degree of specialization (Balassa, 1967). Specialization limits a firm's ability to diversify into the products of other firms because specialized firms enjoy a much higher degree of comparative advantage. This is thought to reduce the number of potential losers from liberalization and makes it more difficult for import competing sectors to successfully lobby for protection. Fear of retaliation from producers of substitute goods may also limit the likelihood of producers to seek protection (Marvel and Ray, 1987). Export oriented firms may use norms of reciprocity to limit protection and ensure tariff reductions in the partner country (Rhodes, 1989). Unfortunately, limited evidence exists to support the position that IIT lowers the likelihood of specific products receiving protection.
Although countries that participate in higher volumes of IIT may tend to have lower overall trade barriers, there is no reason to believe that firms that produce products characterized by high levels of IIT do not have incentives to seek protection. If specialization limits product diversification, the monopolistic nature of IIT causes lobbying to essentially becomes a private good, which increases the likelihood of lobbying by eliminating free-riders (Gilligan, 1997). Moreover, when electoral institutions encourage politicians to cater to specific geographic constituencies, the returns to lobbying are higher and IIT should have a stronger impact on policy outcomes (Kono, 2009). Bombardini and Trebbi (2012) argue that firms are more likely to lobby individually when their products are characterized by a high degree of differentiation. Since individual firms should tend to lobby individual legislators (Magee, Brock and Young, 1989), they should be more likely to benefit from protection in political systems, like Mexico's, where politicians serve territorially based constituents.
Hypothesis 3: Products exhibiting higher degrees of intra-industry trade are more likely to receive protection within Mexico's TRS.
The level of import penetration faced by an industry has become a standard component of models of endogenous trade policy formation. While economists disagree over the impact that import penetration has on the likelihood that an industry will seek protection, there is growing consensus that lower levels of import penetration increase demands for protection (Grossman and Helpman, 1994; Lee and Swagel, 1997; Maggi and Rodríguez-Clare, 2000). The intuition is that if domestic output is relatively large compared to the volume of imports, owners of specific factors have more to gain from an increase in domestic prices.
Hypothesis 4: Lower import penetration increases the likelihood of an industry receiving protection.
Countries, over the long-run, tend to specialize in and export the products where they hold a comparative advantage and import other products. However, specialization assumes costless intra-industry mobility of productive factors. While we would not expect highly competitive industries to seek protection, sectors lacking a comparative advantage are by nature importcompeting and receive more protection (Lee and Swagel, 1997; McGillivray, 1997). Industries with lower levels of comparative advantage may also be in decline, which tends to increase their likelihood of receiving protection (Ray and Marvel, 1984) and, since firms facing greater competition tend to have better information about the consequences of trade liberalization, they should be more likely to seek protection (Fernandez and Rodrik, 1991).
Hypothesis 5: Mexican products with a comparative advantage over Japanese exports should be less likely to receive protection within the Japan-Mexico FTA.
Alternative Explanations
Given the political influence of subnational constituents in Mexico, this article argues that the political clout of industries stems from their role in the electoral jurisdictions that they inhabit. Although there are numerous reasons to doubt that an industry's absolute size has a direct relationship with protection (Caves, 1976), the literature on trade protection has long argued that larger industries are more likely to receive preferential treatment. Size provides an inherent advantage that allows industries to penetrate governments because they are assumed to have more resources to lobby politicians and support the electoral campaigns of 'friendly' officials (Grossman and Helpman, 1994; Magee et al., 1989). There is also broad empirical support for the argument that larger industries receive more protection (Gawande and Bandyopadhyay, 2000; Pincus, 1975). This suggests that larger industries are more likely to receive trade protection.
Since receiving preferential treatment depends on an industry's ability to attract the attention of policymakers, many studies have employed the interaction between the size of an industry and its political concentration to measure its political clout. Although the interaction cannot directly measure the relative subnational importance of an industry, it has been employed as a proxy for subnational importance as well as the organizational ability of the industry. For all but the largest industries, it is thought that a geographically dispersed industries will have more success gaining political support, because they are present in multiple jurisdictions (Caves, 1976; Busch and Reinhardt, 1999; McGillivray, 1997).2
3. Measures and Models
To examine whether Mexico's TRS towards Japan exhibits a higher likelihood of concessions to important subnational economic interests, the hypotheses discussed above have been integrated into several multi-level ordinal logit models. If subnational economic interests influence Mexico's policymakers, then we would expect more important subnational industries to receive preferential treatment within the TRS.
Observed Dependent Variables
This paper seeks to understand the influence of societal interests on the likelihood that an industry will benefit from trade protection within Mexico's TRS toward Japan. To do this, Mexico's TRS has been coded into five ordinal categories. Mexico's TRS has fourteen basic categories, but for reasons of statistical estimation and simplicity, these have been reduced into five categories:
Immediate liberalization.
Liberalization within 4 or 5 years.
Liberalization within 6 to 11 years.
Late starts to liberalization: no liberalization before 6 years and at least 10 year to full liberalization.
Exempt from liberalization. Products with quotas have been excluded from the study given the difficulty of classifying how much protection different quota structures grant.
Observed Independent Variables
To measure the relative subnational importance of an industry across different political jurisdictions, an indicator of subnational importance must be employed. If we dispense with the assumption of a homogenous spatial distribution of industrial activity, the political concentration and absolute size of an industry by themselves tell us very little about the relative subnational importance of an industry with specific geographic constituencies. Although a relatively large industry at the national level must play an important, but not necessarily the most important, role in at least one subnational jurisdiction, we cannot say anything else about its relative economic importance within subnational jurisdictions without accounting for the geographic distribution of other industries. Moreover, as Caves (1976) points out, firms in a country's industrial heartland must compete with the voices of other industries for the attention of policymakers.
If different industries are located in different regions, as is the case in Mexico, then the absolute size and political concentration of an industry cannot tell us anything about its relative importance (Pezzola forthcoming). Therefore, the political influence of an industry and its ability to secure protection depends on the specific economic geography of each political jurisdiction that it inhabits. This means we must measure the importance of each industry relative to all other economic activity within the jurisdictions it inhabits. For this purpose, the variable Subnational Production is calculated by aggregating the relative importance of each industry within the economy of each of Mexico's states and the federal district. By dividing the value of production of an industry in each jurisdiction by the total size of the jurisdiction's economy, we measure its relative importance to voters and politicians within the jurisdiction. Aggregating these values across jurisdictions provides a measure of the relative subnational importance of the industry to voters across Mexico.3
Using the importance of an industry as a producer should not be interpreted to mean that politicians do not care about employment; after all, large producers tend to also be large employers. Rather, because large industries tend to have other politically salient resources at their disposal, an industry's value of production represents a broader measure of political clout.
Calculating the relative subnational importance of an industry at the state level has been done for two reasons. First, Mexico does not report highly disaggregated industrial data below the state level, making it difficult to discriminate between industries within Mexico's single member congressional jurisdictions. Second, measuring the subnational importance of an industry at the state level seems to better fit the political reality of Mexico. The Senate ratifies all trade agreements negotiated by the Executive and, since the approval of the Foreign Trade Act in 1993, the Senate has begun to take a more active role in trade negotiations. The variable Subnational Production also measures the political importance of industries to Mexico's governors and state level political parties, which have significant influence over legislator. This does not mean that the measure does not also serve as a good proxy for the importance of an industry to the federal deputies elected in Mexico's 300 single member districts. If an industry is important within a state's economy it is likely to be important to numerous single-member districts within the state, even if the industry is not located in the district.
The aggregate welfare effects of protection are captures through the export elasticity of foreign suppliers. Obtaining an accurate measure of export elasticity for any country is extremely difficult, especially at a high level of product disaggregation. Following Olarreaga, Soloaga and Winters (1999), Mexico's market share of world imports is used as a proxy for foreign export elasticity. Although imperfect, this proxy can be easily calculated at all levels of product disaggregation and does not rely on problematic estimates of the elasticity of world export supply. Therefore, Mexico's Import Sharep replaces export elasticity.
The level of IIT is calculated using the Gruber-Lloyd Index. The index of IIT is calculated both at the industry level (ITTi) and at the product level (ITTp). While norms of reciprocity may deter protectionism by industries characterized by high levels of ITTi, the need of politicians to cater to specific geographic constituencies provide incentives for politicians to protect firms that produce products with high levels of ITTp.
The rest of the variables are calculated as follows. The degree of import penetration of an industry (Import Penetrationi) is equal to the value of imports associated with an industry divided by the industry's total value of production. The comparative advantage of a product is calculated using the Balassa (1965) index of revealed comparative advantage at the product level (RCAp). The overall size of an industry is measured using its total value of production (Productioni). Following Busch and Reinhardt (1999), the Political Concentrationi of an industry is calculated using a Herfindahl index of employment.
Statistical Models
To evaluate how an industry's characteristics influence its political clout and, therefore, the structure of the trade agreement, the paper estimates a series of multi-level ordinal logit models. The first model only measures an industry's political clout though the variable Subnational Production. In order to compare the influence of political clout measured at the subnational level and at the national level, the second model only includes Productioni as an indicator of political clout. The next two models include Productioni and the Political Concentrationi of an industry as well as their interaction. There is no reason to believe that interested parties cannot pursue parallel lobbying efforts or that some will prefer to flex their political influence only at the subnational or national level. For this reason, three 'full' models are estimated that include both subnational and national level measures of political influence. All models include the variables Import Sharep, ITTi, ITTp, RCAp, and Import Penetrationi, in order to capture the welfare effects of protection and the likelihood that an industry will seek protection.
The political influence of an industry is marginally, not absolutely, related to its importance. There is no reason to believe that an extra million dollar of production has the same influence on the political clout of a small industry as it does on a large industry. To account for the decreasing marginal benefits of size, log(Productioni) and log(Subnational Productioni) are used in the statistical models. To take into account the decreasing marginal influence of relative comparative advantage, which includes zeros, the statistical models use the inverse hyperbolic sine of RCAp.
4. Findings: Protection from Japanese imports
The results presented below clearly demonstrate that subnational economic interests influenced the structure of Mexico's TRS towards Japan. Table 1 presents the results, first, of the 'subnational' model (Model 1) that includes the variables for relative subnational importance, welfare effects and tendency of industries to lobby for protection. Models 2-4 measure an industry's political clout using its absolute size and political concentration, instead of its relative subnantional importance. Models 5-7 examine the influence of an industry's subnational as well as national level importance.
Ordered logit model with random intercepts estimated using gllamm in Stata 12. Cluster adjusted standard errors in parentheses. Bold entries significant at the 0.05 level or better.
Table 1 Determinates of the Structure of Mexico's Tariff Reduction Schedule Towards Japan
Hypothesis 1 holds that probability of receiving protection increases with the relative subnational importance of an industry. In support of this claim the variable Subnational Production has a substantively large and statistically significant relationship with greater protection. Even when indicators of importance at the national level are taken into account (Models 5-7), the coefficient of Subnational Production remains substantively large and statistically significant. In all cases, when compared to the national level indicators (Production and Political Concentration), the subnational importance of an industry provides a better indicator of which industries received protection.4
The estimated coefficients of Subnational Production indicate that the subnational importance of an industry has a strong positive association with the likelihood of protection. Unfortunately, the estimated coefficients of logit models lack the natural interpretation of a linear regression. By taking the exponential of the coefficient of Subnational Production we obtain the change in the odds ratio associated with a change in the value of Subnational Production. In the case of Model 1, the associated change in the odds-ratio is 4.7. When indicators of both subnational and national level importance are taken into consideration (Models 5-7), the change in the odds-ratio associated with Subnational Production is over 8.
The positive association between the subnational importance of an industry and protection can also be illustrated by calculating the change in probability of protection associated with the change between two specific values of Subnational Production. Based on Model 1, when all other values are set to their means, the likelihood of an industry with a mean level of Subnational Production (4.28) receiving at least six years of protection is 45.8% (See Graphic 1). The estimated likelihood of an industry with a median value of Subnational Production (4.35) is 48.5%. In other words, a difference of 0.07 (1.6% of the mean value) in the value of Subnational Production is associated with an increase of 2.7% in the likelihood of receiving at least six years of protection. If we compare the difference between the mean value and the value of the third quartile of Subnational Production (4.28 vs. 4.92), the increase in the likelihood of receiving protection is 23.3% (45.8% vs. 69.1%).
Thus, the core hypothesis of the article finds strong support. Industries of relative subnational importance were more likely to have received protection within Mexico's TRS with Japan. Even when we take into account the role of an industry within the national economy and it political concentration, subnationally important industries are more likely to be favored. Comparing likelihoods of the 'national' level models (Models 2-4) with the 'full' models (Models 5-7) provides evidence of a strong and positive association between Subnational Production and an industry's political clout. The likelihood ratio tests of these models (Model 2 vs. Model 5, Model 3 vs. Model 6, and Model 4 vs. Model 7) indicate that the presence of Subnational Production significantly increases the goodness of fi t of the models and, therefore, the amount of variation across products and industries explained by the models (see Table 1).
Relying on likelihood ratio tests may result in models that over fit the data, since it does not introduce a strong penalty for additional parameters. The Bayesian information Criterion (BIC)5 significantly penalizes each additional parameter, reducing the risk of over fi tting the data. The lower values for Models 5-7 in comparison to Models 2-4 (a diff erence greater than ten) off ers strong evidence that Subnational Production provides a good measure of an industry's political clout even in the presence of national level indicators (Raftery, 1995).
As mentioned above, when the absolute size of an industry (Production) as well as its Political Concentration and the interaction between Production and Political Concentration are taken into consideration along with Subnational Production (Models 5-7), the subnational importance of an industry remains a positive and significant indicator of protection. Interestingly, when the importance of an industry within the national economy is also taken into account, the marginal influence of Subnational Production increases. Based on these models, there is a difference of close to 10% (versus 5.9% in Model 1) in the likelihood of an industry receiving at least six years of protection between the mean (4.28) and median (4.35) level of Subnational Production. This increase in the estimated influence of the subnational importance of an industry can also be seen in the larger estimated coefficient for Subnational Production in Models 5-7.
When both the subnational and national importance of an industry are taken into account, the estimated association between Production and protection becomes negative. This does not mean that larger industries necessarily receive less protection. Many large industries at the national level have a high degree of relative subnational importance in at least one jurisdiction, many large industries receive preferential treatment. Rather, the change in sign of the estimated coefficients between 'national' level models and the 'full' models coupled with the significantly higher estimated likelihoods of the 'full' models suggests that the absence of Subnational Production induces an omitted variable bias to the 'national' level models and the estimated coefficient of Production assumes part of the positive effect associated with Subnational Production. This provides further evidence that the political clout of both large and small industries in Mexico stems from their role within the subnational economies that they inhabit and that the absolute size of an industry, in and of itself, does not grant preferential treatment.
One might be tempted to argue that the positive association of Subnational Production in the absence of Production stems from the positive relationship between Production and protection. If this were the case, the estimated coefficient for Production should not become negative when Subnational Production is also taken into consideration. Moreover, we would expect the national level models (Models 2-4) to prove superior to Model 1, which is not the case.
Even though the 'subnational' (Model 1) and 'national' level models (Models 2-4) are not nested, we can use their BICs for comparison. The higher BICs of the 'national' level models provide strong evidence in favor of Model 1. The difference of more than 10 suggests that Model 1 better represents the 'true' model that generated the data (Raftery, 1995).
When we compare the 'subnational' model with the 'full' models, we also find evidence of the superiority of the 'subnational' model. Models 5-7 have lower log-likelihoods and the log-likelihood ratio test provides evidence that the addition of Production, Political Concentration, and their interaction increase the fit of the models (see Table 1). However, the higher BICs of the 'full' models indicate that the increased complexity of the 'full' models is unwarranted without a strong theoretical motivation. Given that both Subnational Production and Production are used to measure the political clout of industries, it is not clear that using Production as an additional proxy for political clout significantly add to our understanding of Mexican trade policy. On the other hand, these models are useful in that they demonstrate that Subnational Production has a positive association with protection, even when national level characteristics of an industry are also taken into consideration. The comparison of the 'full' and 'national' level models also demonstrates that taking into account the subnational importance of an industry significantly increases our ability to model the likelihood that an industry will receive protection.
Overall, the findings above fit with the argument that subnational political interests directly influenced Mexican trade policy towards Japan. Industries with greater subnational importance were more likely to benefit from Mexico's TRS. This relationship holds even when national level indicators of political clout are taken into account. However, to fully understand how subnational economic interests influence the policy-making process as well as the role of other national level factors, we must also carefully investigate the implications generated by the other variables included in the models.
The estimated coefficients of Import Share in all models suggest that Mexican policymakers were, at least partially, concerned with the aggregate welfare effects of the TRS. The coefficients of Import Share are negative and statistically significant. Although Hypothesis 2 suggests that politicians seeking to support local economic interests would be indifferent to aggregate welfare, it would seem that the aggregate welfare effects of the TRS were taken into account. This does not nullify the more general finding that subnational economic interests influence which industries receive protection; rather, it simply suggests that policymakers, while seeking to support local economic interests, may have also taken into account the more general welfare effects of the TRS. This result fits with the expectation that, all else being equal, politicians prefer to protect industries where the deadweight loss is lower. The sensitivity to national welfare may also indicate that national party leaders or the Executive also influence the formation of the TRS, since they must concern themselves with much broader constituencies. This result also fits with our understanding of policymaking in the presence of a veto-player (e. g. the Mexican Congress). The Executive who negotiates and proposes trade agreements seeks, among other things, to increase national welfare while still proposing an agreement that will satisfy the parochial interests of legislators and of the governors that swing their votes.
As expected, IITp has a positive and statistically significant association with protection. This coincides with the hypothesis that subnational economic interests influenced trade policy formation. As geographically-specific prívate interests have more success in penetrating the policy-making process, the influence of intra-industry trade on policy should increase. The dependence of politicians on geographically-specific constituencies also increases the sensitivity of politicians to the interests of individual firms. Since firms are more likely to lobby individually in the presence of a high degree of product differentiation, the positive coefficient of IITp makes sense. Moreover, if firms prefer to lobby individually, lobbying associated with the interests of the whole industry should lose influence over policy outcomes as evidenced by the statistical insignificance of IITi across all models.
The estimated coefficients for Import Penetration provide some evidence that producers did not seek to use trade barriers to divert imports and increase their domestic market share. Across all the models the coefficients for Import Penetration are negative, but not statistically significant.
The RCA of a product has a negative association with protection. It makes sense that competitive products do not need protection. Producers of competitive products could have lobbied for greater protection in order to temporarily secure higher domestic prices, but faced with limited resources with which to lobby policymakers securing temporary protection may not have been an efficient use of resources.
5. Conclusion
This article sets out to examine the relationship between the economic interests of geographically specific constituencies and the structure of the Japan-Mexico FTA. It argues that Mexican politicians used the TRS to appease subnationally important economic interests and gain political support for the agreement. By directly measuring subnational economic interests, strong evidence emerges that the interests of territorially based constituencies influence Mexican policymakers. Although direct vocal opposition to the agreement may have been limited to a few industries, subnationally important industries were able to leverage their political clout to receive preferential treatment. While the liberalization of Japanese agricultural imports may have played an important role in stalling negotiations, it is doubtful that negotiations would have succeeded if Mexican politicians could not have used the TRS to Foster support among politically important import competing industries.
These findings do not mean that larger industries lack political clout. Large national industries almost always play a significant role in at least one subnational jurisdiction. This paper argues that the importance of industries where voters live and work plays a critical role in determining their political clout and, therefore, the structure and viability of trade agreements. If we seek to understand support for and resistance to greater liberalization of trade, we must begin to think locally about political clout. Doing so, takes an important step in helping us understand the negotiation process of FTAs. We know that TRSs are negotiated and pay a critical role in obtaining domestic support for the ratification and implementation of FTAs.
Clearly TRSs pay an important role in obtaining domestic support for the ratification and implementation of FTAs; however, scholarship has tended to either neglect the role of societal interest on the negotiation and structure of FTAs or has focused on the influence of peak business associations on the negotiations. In the case of the Japan-Mexico FTA, subnational economic interests clearly influenced the structure of the agreement and its adoption. The results of this study provide some of the first evidence that societal interests in a developing country influenced the structure of its FTAs. Subnational economic interests not only have a voice over the adoption of FTAs, but also over the structure of the agreement.
These findings provide an important step in helping us understand the negotiation process of FTAs, especially when they include countries where political institutions and political competition provide territorially based constituents with political influence. There is little reason to believe that the influence of subnational economic interests have been limited to a single FTA or a small subset of states. Inside and outside of Latin America, political institutions make politicians sensitive to territorially based constituent interests; however, these interests have tended to be ignored when evaluating the formation of trade policy in developing countries. Scholars, policymakers, and negotiators cannot ignore subnational interest in countries like Mexico even when peak business associations play a highly visible role in trade negotiations.
Even though the media often highlights conflicts surrounding the liberalization of specific products during trade negotiations, the conflicts surrounding the sector should not distract from the broader political process behind trade policy formation and how societal interests shape policy preferences. Focusing on specific sectors and peak organizations may obscure the political forces that have shaped and slowed trade negotiations. To varying degrees across developing countries, politicians are beholden to territorially based voters. Therefore, if we wish to fully understand the politics of trade negotiations we must take into account the importance that industries play within political jurisdictions.
