Partisanship and Export Performance: The Conditional Effect of Left-Wing Government on Latin America’s Exportation

Abstract

Are there noticeable differences among political parties over the countries’ export performance? Are left-wing parties more protectionist in free trade than right-wing parties in Latin American countries, where Import Substitution Industrialization remains a legacy? This paper argues that partisanship matters, but the effects on export are more heterogeneous than previous research has acknowledged. Contrary to conventional wisdom, the left-wing governments favor exportation, and they export more instead of being protectionist when domestic industries have underperformed. Cross-country data of 15 Latin American and Caribbean countries from 1995 to 2019 supports our prediction of the conditional effect, leading to a variation in their export performance. We also test such a conditional effect on the subdivision of trade categories and trading partners, to assess more accurately how exactly this effect plays out. These findings have implications for the studies of trade and for Sino-Latin American relations in a changing global context.

1 Introduction

Do political parties have an impact on trade performance? Since international trade has predictable distributive effects, politicians are likely to be concerned about trade policy in their efforts to win elections. If so, parties should promote a certain kind of trade that reflects their constituents’ interests. By looking at who benefits, suffers, or remains unaffected by trade, we understand how trade shapes distributional conflicts and policy outcomes in today’s hyper-globalized economies. However, although there is vast literature on the political economy of trade, few have addressed the role of political parties. Instead, some claim the convergence of parties’ composition, predicting that industry-level variables rather than partisanship, better explain the supply and demand of trade policy.¹ Another opinion is that interest groups affect trade policies to benefit themselves or their causes, rendering partisanship irrelevant.² In the International Relation realm, partisanship is considered unimportant because they believe it is a country’s international position that determines its trade performance, such as a hegemon versus a small country as the Hegemonic stability theory indicates,³ or a central versus a peripheral country in the Dependency school.⁴ Cleavages should exist across countries but not within them.

According to the Stolper-Samuelson model of trade, parties should be unified internally on trade and divided among each other along class lines. This holds true but the reality is more complicated. When an individual, a firm, or a country enters international trade, it benefits from specialization and efficient resource allocation. These benefits have consequences, are path-dependent, and framed in a specific time and space. In Latin America, during the 1960s, the left-wing governments, characterized by populism, had close relations with urban workers who demanded import tax to protect domestic industries. Since the mid-1990s, rightist governments in Latin America have adhered to a strict market orientation. Class conflict over trade does not exist in a vacuum, but in “embedded liberalism” and lasts for a certain period where political agencies have a role to play. As Karl Polanyi refers to a Double Movement, along with the process of a laissez-faire economy, there arises a demand for social protection since economies are always embedded with in societies.⁵ What is less clear is exactly how political agencies matter. Studies of fiscal and monetary policy show that such party differentiation occurs as left-wing parties prefer policies that increase government spending and induce growth, while right-wing governments favor policies that induce lower spending, balanced budgets, and lower inflation.⁶ This may happen to trade policies as well.

Besides, the conventional wisdom of factor and sector models of trade could not be fully understood if the tendency of global production patterns is neglected. Baldwin asserts a second round of unbundling globalization which shifts the locus from sectors to stages of production.⁷ This requires a focus on supply chains. Krugman also acknowledges that differences between countries are one reason for trade, but that much trade occurs among similar countries. He finds that it is not the differences between countries, not the opportunity cost, not the availability of the factors, but the increasing return to scale is the fundamental reason for international trade.

There are few cross-national studies with attention paid to the causality of partisanship and trade in the developing countries. Simmons shows that for a group of countries during the inter-war period, changes in tariffs were affected strongly by partisanship, with left-wing parties in Parliament favoring reductions in tariff barriers.⁸ Rogowski loosely associates parties with factors of production.⁹ Using the Stolper-Samuelson model, he predicts that if land and capital are abundant in a country while labor is scarce, a left–right party division exists with the left in favor of protection and the right for free trade.

We explore these issues by first identifying the hypotheses that partisanship is correlated with trade performance. Based on a panel data of 15 Latin American countries from 1995 to 2019, partisanship influence has been found, but showing the opposite of what conventional wisdom predicts. Surprisingly, the left-wing governments are the ones favoring exportation in Latin America during the neoliberal and post-neoliberalism period. Holding many other factors constant, partisanship matters, and is conditional on the countries’ manufacturing industries’ performance. Left-wing governments take more free trade stances when domestic manufacturing value added as a percentage of GDP is low. Then we test whether such conditional effects remain with the subdivision on trade categories and trading partners. The paper uses this differentiated export performance by the category and the trading partner to assess more accurately whether partisanship matters and how exactly this effect plays out.

The next section gives a background of Latin American parties’ trade preferences as well as a review of trade theories based on which our hypotheses is built. The third section introduces the variables and the fourth presents the statistical results. In Sect. 5, we give Kirchner’s two terms in government as case studies and we conclude in the last section.

2 Background and Hypotheses

Since the late 1980s and early 1990s, nearly all Latin American countries abandoned Import Substitution Industrialization (ISI) and adopted orthodox neoliberal reforms. The swing to a market orientation seemed consistent with the ideology historically held by many rightist parties in Latin America. The Conservative Party in Chile under President Jorge Alessandri in 1958–1964 typified the view held by many of the Latin American right-wing parties during the post-World War II era, preferring the market instead of state intervention. The fact that many rightist parties took office in the late 1980s and early 1990s and played central roles in initiating neoliberal reforms, further bolstered their connection with market reforms.¹⁰ Wiesehomeier and Benoit asked experts to assess the policy positions of different political parties and party leaders in 18 Latin American countries.¹¹ Experts placed parties and presidents on a 1–20 scales, with higher scores on economic issues reflecting greater backing for the market. The survey results indicate that rightist parties and their leaders in all 18 countries score above 10 in support of free trade and globalization.

Other studies offer a new argument on the interaction between politics and various types of export growth. Richardson found a new populism combines broad-based benefits for urban workers with export promotion in Argentina.¹² A coalition that focuses on commodity exporting, rather than the restriction of exports, was central to the new populist coalition’s sustainability. To be specific, the Kirchner alliance relied on an undervalued exchange rate with wage-goods subsidies. Although the role of export revenue in sustaining populism and leftism in Latin America has received little attention, there is a general acceptance that in Venezuela, President Hugo Chávez used oil to fund his political projects.¹³ There are also researchers who contend that natural resource endowments influence policies. Countries that derive much of their revenues from natural resources favor statist policies, as the abundant proceeds earned from resource exports give political leaders added financial flexibility.¹⁴

Using data from UNCTAD, we see the manufacturing exportation from 10 Latin American countries accounts for only 3.89% in the world’s total exportation of manufactured goods, while the primary products exportation of the 10 countries accounts for about 10 percent of the world’s total exportation of primary goods (see Table 1). Then we look at the percentage of the 10 countries exportation in China’s importation. We see Latin American countries experience a growth in the percentage of both primary and manufactured goods (see Table 2).

Table 1 The export share by sector of 10 latin American countries as a percentage of the world’s total exports

Table 2 The export share by sector of 10 latin American countries as a percentage of China’s total imports

Taking a look at the average annual growth rate of 10 Latin American countries from 2000 to 2018 to several destinations, the growth rate of bilateral trade of most Latin American countries with China increases rapidly in both primary and manufactured exportation than that with developed countries and other developing countries (see Table 3).

Table 3 Average annual growth rate of exports (2000–2018)

The boom in industries and commodities tied to production for the global market, along with the collapse of most firms geared toward ISI since the late 1980s, has generated greater export-oriented influence in domestic labor markets in Latin America.¹⁵ Export-oriented groups are successful in lobbying politicians, who in turn, support more extensive, market-oriented trade reform.¹⁶ This leads to our first hypothesis:

H1: All other things being equal, in the neoliberalism and post-neoliberalism period, Latin American countries under a left-wing government export more, especially primary products.

Although ISI in most Latin American countries came to an end with the debt crisis, important social and economic legacies endured. It was also a decidedly political project, resting on important coalitions among urban industrialists, organized industrial labor, and the urban middle class,¹⁷ thus it is to be expected that they have political legacies as well. ISI political bargains typically worked at the expense of rural sectors.¹⁸ Many scholars note the remaining existence of association in the strength and the orientation of domestic manufacturing interest and organized labor across Latin America. Where the ISI’s legacy most remained, we expect a larger coalition of protectionism for domestic manufacturing industries.

Another body of relevant literature concerns the resource curse on exportation. If the abundance of oil, mineral, and agricultural products leads to poor economic performance,¹⁹ we should expect a reverse effect, that more industrialized countries are less likely to depend on primary goods for exportation. This argument is skeptical since commodity trade growth may contribute to GDP growth and somehow improve domestic industrial production. We tried to test it and to see whether it intertwines with partisanship on a countries’ export performance. Most Latin American countries are developing countries which tend to be smaller economically than major industrialized countries, and more likely to specialize in the exports of basic commodities. As a result, they can be regarded as price-takers, not just for their import goods, but their export goods as well. That is, the prices of their tradable goods are generally taken as given on world markets. This leads to the second hypothesis.

H2: All other things being equal, left-wing governments export more primary goods when domestic industries have underperformed, and are more protective with more value added as a percentage of GDP in the domestic manufacturing industry.

To model the distributional effects of trade, research in international political economies typically relies on two traditional trade models which predict clear lines of conflict, either along factor-specific class lines between owners of scarce and abundant factors of production,²⁰ or along sectoral lines between workers in tradable and non-tradable industries.²¹ Factor (Stolper–Samuelson) trade model predicts that a left–right party division exists with the left in favor of protection and the right for free trade. In the classic sectoral (Ricardo–Viner) model, the main divide is between exposure to and protection from global competition in general, so industrialization level determines the export, but does not predict any differences based on partisanship. In contrast, our argument predicts that there should be systematic differences not only between the left–right divide but also related to the domestic manufacturing industries’ performance.

3 Analytical Framework

Our central empirical proposition is that contrary to conventional wisdom, Latin American’s left-wing governments take more free trade stances, especially by exporting primary products when domestic industries have underperformed, and are more protective with more value added in the domestic manufacturing industry. We will analyze cross-country data from 15 Latin American²² countries from 1995 to 2019, and focusing both on trade categories and trading partners, to see whether the empirical analyses support our prediction. This argument may be reinforced regarding with trading partners, i.e., developed countries, developing countries excluding China, and China.

Three caveats arise. First, we do not explore whether parties accurately reflect their constituents’ interests. We assume that in democratic settings parties reflect constituents’ preferences. Second, we treat the distributional effects in different industrialization levels as a black box for further investigation. Third, we are not looking at the whole story of causality although we do present a causal mechanism that combines the distributional effects of trade and supply and demand side politics (see Fig. 1). In this paper, we focus on testing the correlation of outcomes, namely partisanship and export performance and the conditional effects brought by domestic industries’ performance.

Fig. 1

Framework. (Source Elaborated by the author)

The dependent variable, the export, captures the volume of merchandise trade based on the SITC commodity classification, Revision 3, at the one and two-digit level, expressed in thousands of US dollars. The data source is UN COMTRADE which provides detailed raw trade data by partner and product. Since the data is also summarized by a group of economies, we are able to get the total all products to developed courtiers, to developing countries excluding China, and to China. Latin American countries seem as mainly primary production exporters, so we specifically look at the export of its primary products, a category composed of SITC rev.3. Within manufactured exports, we look at labor-intensive and resource-intensive manufacturers and manufactured goods.

The primary independent variable for this analysis is the partisan identity of the executive. We test our partisan hypothesis using a dummy indicating the executive partisanship that codes left-wing government leaders as 1, and the others as 0. This variable is taken from the Database of Political Institutions, in the World Bank Data.

We operationalize the domestic industries’ performance using the manufacturing value added as a percentage of GDP of a certain year (World Bank data). The industrialization process in Latin America saw sharp increases in growth after the ISI period, and manufacturing as a share of national income rose.

We include an array of controls for political and economic variables that likely have an impact on trade. Country regime type has been found to have a significant effect since democratic countries may export more. We use polity IV score to control for any effects of the regime type. To capture each economy’s vulnerability to global capital flows and the strength of global market constraints, we include the inflation and exchange rate. We include the natural logarithm of GDP per capita to measure the size of the domestic economy. Finally, we include the FDI as a percentage of GDP to see its influence on trade, and the commodity price indices to examine the commodity boom effect.

Our approach to estimation begins with a panel data spanning 15 Latin American countries over the period 1995–2019. We employ a fixed-effects specification using the Hausman Test, as many of the theoretically important variables reflect historical legacies and are not time-varying. The OLS regression is:

$$\begin{aligned} {\text{Export}}_{{\text{it}}} & = {\rm{\beta }}_0 + {\rm{\beta }}_1 {\text{Partisanship}}_{{\text{it}}} + {\rm{\beta }}_{2} {\text{Manufacturing}}_{{\text{it}}} \\ & \quad + {\rm{\beta }}_3 {\text{Partisanship}}_{{\text{it}}} *{\text{Manufacturing}}_{{\text{it}}} + \gamma \,X_{{\text{it}}} + \varepsilon_{{\text{it}}} \\ \end{aligned}$$

where the growth outcome $\text{Export}$ for a country $\text{i}$ at year t is regressed in the model. ${\upbeta }_{1}$ captures the effects of partisanship of a country i at year t. ${\upbeta }_{2}$ captures the effects of manufacturing of a country i at year t. manufacturing * partisanship is an interaction term to capture conditional effects. A vector (${X}_{\text{it}}$) is country-level controls from various sources. Robust checks and other models are needed in further work to check out problems commonly associated with panel data.

4 Results and Discussion

We begin to present the results of our analysis in Table 4. Models 1–2 examine the effects of partisanship and industrialization on the exportation of primary commodities’ in 15 Latin American countries. Models 3–4 and Model 5–6 change the dependent variable to the export of labor-intensive and resource-intensive manufacturers, and manufactured goods to the world. Looking at Models 1 and 2, we find that the party variable has a statistically significant effect at the p < 0.01 level, with a positive relationship. The manufacturing variable has a positive impact as well. This finding offers support for Hypothesis 1. The interactive relationship between our main independent variables supports our expectations on Hypothesis 2 that partisanship and domestic industries’ performance are conditional on each other, with the effect of each being affected by the value of the other. Although it is difficult to interpret interactive relationships from regression coefficients, we can say that the initial positive effects of partisanship are weakened when domestic industries perform quite well. Models 3–6 show that this heterogeneous effect can also be observed in manufactured goods, but not in labor-intensive and resource-intensive manufacturing exports.

Table 4 Fixed effects models for sectoral exports

Table 5 presents the results for Models 7–12, which categorizes exports by trading partners, i.e., developed countries, developing countries excluding China, and China. Interestingly through Models 7, 8, 9, 10, we find the same effect in developed countries and developing countries excluding China, but they show no significant correlation to exports to China. This indicates that Latin American countries under left-wing governments export more primary goods to developed countries, but do not affect the exportation to China. This coincides with the descriptive data that Latin American countries experience a growth in the percentage of both primary and manufactured goods’ trade with China.

Table 5 Fixed effects models for exporting primary commodities to trading partners

Table 6 shows that partisanship has a positive effect on the labor-intensive and resource-intensive manufacturers’ exportation to developing countries excluding China, the effect is not robust with developed countries and with China. Since correlation is not causality, one interpretation is that regardless of what party, they all export to China and the other way of thinking is that China welcomes Latin American’ exports regardless of the party’s governance. In Table 7, we find the partisanship variable being insignificant on the manufactured goods exportation in the 15 Latin American countries. This agrees with the export-oriented populist literature that left-wing governments are keen to sell primary products.

Table 6 Fixed effects models for exporting labour-intensive and resource-intensive manufactured goods to trading partners

Table 7 Fixed effects models for exporting manufactured goods to trading partners

As for the control variables, GDP per capita has a positive and significant relationship with trade, suggesting the importance of an economies’ size in free trade. Polity IV has no effect, reassuring that the democratic regime seems unimportant in the free trade stance. Exchange rate has a negative effect, meaning that depreciated currency can promote exportation. Especially, FDI has a positive effect on the exportation of labor-intensive and resource-intensive manufacturers, indicating some connections between FDI and trade in a certain kind of value chain production. In most of the occasions, inflation has a negotive effect on export. This is not surprising since this indicator is somewhat related to domestic economic activities. Commodity price index being included, the results are still robust.

5 Case Study

Trade policy has an important role to play in supporting trade diversification. How do Latin American countries respond to the commodity windfall? Using Global Trade Alert data, we record 2664 trade policy interventions from 2008 to 2018 of 10 Latin American countries. The GTA database includes various forms of government action from national legislation to the contract terms of individual state agencies. Each announcement documented by the GTA team includes at least one new and credible promise for change in market conditions at home or abroad. For example, the announcement of a new production subsidy to steel producers would be recorded as one state act with one intervention.

Each GTA database entry provides information about the direction of the change (harmful or liberalizing), the announced policy instrument, its announcement date and, where available, implementation date as well as the sectors and products targeted by the statement. The red column means the intervention almost certainly discriminates against foreign commercial interests. The green column means the intervention liberalizes on a non-discriminatory (i.e., most favored) basis, or improves the transparency of a relevant policy. The amber column refers to the intervention likely involves discrimination against foreign commercial interests. Inward means that the distorted market is the domestic market of the implementing intervention. Outward refers to that the policy change that concerns an outflow restriction. The predominantly affected market is the outflow destination. Outward subsidy is those for export incentives and trade finance interventions (a state aid awarded to an exporting company which generates its sales exclusively abroad may be classified as an outward subsidy). From Fig. 2, we see Brazil and Argentina, based on its market size and population, have the most intervention from the state in trade issues.

Fig. 2

(Source Calculated by the author from Global Trade Alert)

Number of implemented interventions.

A case study of Argentina under two terms of Cristina Kirchner’s administration 2007–2011 and 2011–2015 suggests the dynamic between the left-wing party’s trade incentives and domestic industries’ performance. In Argentina, as Richardson asserts, a new variant of populism combining traditional Latin American populism with export promotion has emerged in Argentina.²³ Since soy-bean cultivation has rapidly expanded, replacing beef and wheat as the country’s leading export commodity, soy-beans and their derivatives have generated three times greater export revenue than beef and wheat products combined in recent years. Unlike beef and wheat, soy-beans are not consumed domestically, hence their export has no direct effect on the effective purchasing power of urban workers. The Kirchner government exploited this fact, undervalued the exchange rate to promote exports, and then taxed the soy exports to subsidize domestic consumption of wage goods, including beef and wheat.

Richardson’s argument explains why the left-wing government prefers to export, but it cannot explain the variance in the left-wing government’s trade incentives. To take a closer look, we find in Cristina Kirchner’s first term of office from 2007 to 2011, there were a large number of trade protection policies aimed at encouraging domestic industry, but such protections declined during her second term of office from 2011 to 2015. That is to say, Kirchner’s government took a more liberalized stand even after the commodity boom from 2002 to 2008. Figure 3 shows the number of trade policy interventions in Argentina since 2007. These reductions in trade protection measures coincide with the worsening of domestic manufacturing value added as a percentage of GDP (Fig. 4).

Fig. 3

(Source Calculated by the author from Global Trade Alert)

Number of trade policy interventions in Argentina.

Fig. 4

(Source World Bank data)

Manufacturing, value added as % of GDP in Argentina.

If we look at the content of each intervention type, we see in her second period of governance, the intervention has been more diversified. Traditional export and import policy instruments, e.g., export ban, tariff and quota, export tax incentive, internal taxation of imports, etc., weighed less in the total state acts. Trade defense instruments such as anti-circumvention, anti-dumping, and anti-subsidy safeguard actions have also been reduced. Instead, there are increases in policies for subsidies and state aid like financial assistance in foreign markets, financial grants, in-kind grants and state loans, etc. (Fig. 5).

Fig. 5

(Source Calculated by the author from Global Trade Alert)

Intervention types under two terms of Kirchner’s administration.

6 Conclusions

Based on the panel data of 15 Latin American and Caribbean countries from 1995 to 2019, partisanship influence has been found, but showing the opposite of what conventional wisdom predicts. The left-wing governments are the ones more favoring exportation, not right-wing parties. Holding many other factors constant, partisanship matters, and is conditional on domestic manufacturing industries’ performance. Left-wing governments take more free trade stances when domestic manufacturing value added as a percentage of GDP is low. By examine the subdivision of trade categories and trading partners, we see the Latin American countries under left-wing governments export more primary goods to developed countries, and do not affect the exportation to China.

This study firstly aims to advance our understanding of trade theory through emphasizing on the role played by political agencies and the conditional effect on the domestic industrialization. We argue and show empirically that the effects of trade performance are more heterogeneous than the classic political economy models predict. Second, it helps to understand South-South cooperation. It is unclear to what extent South-South trade moves beyond the asymmetrical nature of North–South trade. A comparison by trading partner and by categories make progress into the understanding of the dynamics of these emerging trade patterns. Such results call for a further and in-depth study on the supply chains and destination markets among developing countries. Third, it provides a new perspective on Sino-Latin American relations beyond clichés, e.g., China as a threat to the US hegemony in Latin America or the neo-dependent nature of the Sino–Latin America relationship. Some literature interprets the “Beijing Consensus” as an economic and political model that challenges the prevalent “Washington Consensus” in Latin America. Our empirical results somehow show the fact that left-wing governments are keen to sell primary products mostly happened with developed countries and other developing countries excluding China. The economic relations such as trade with China have little influence on Latin American countries’ domestic politics.

As avenues for making further progress into understanding the relationship between domestic political agencies and trade performance, we suggest two areas for further study. First, renewed attention to the state intervention in different stages of development, here referring to industrialization level, is required. How to reflect the mechanism of different trade policy positions of left-wing governments at different stages of development between low and high levels of industrialization is yet unclear. Second, attention needs to be given to the pattern of South-South economic relations and how it differs from or agrees with the nature of North–South relations. This includes the consideration of how supply chains and destination markets within South-South trade reshape the “winners” and “losers” domestically and also unveiling the varied outcomes or uneven geographies of expanding South-South trade.

Appendices

Appendix 1

Summary of explanatory variables

Variables	N	Mean	SD	Min	Max
party left dummy	368	0.522	0.5	0	1
manufacture value added as % of GDP	370	15.187	2.795	9.55	22.487
logGDPpp	370	8.565	0.718	7.006	9.623
exchange rate	368	3.18	2.759	-1.733	9.375
polity IV	360	7.378	5.512	-88	10
inflation	354	1.873	0.916	-2.216	4.749
FDI as % of GDP	369	3.57	2.481	-2.45	12.197
commodity price indices	360	5.781	19.572	-36.232	36.697

Appendix 2

Summary of sources for key variables

Variable	Description	Source
Dependent variables
export	export value of a country in a certain year, can be classified by partners and categories	UNCTAD data
Explanatory variables
exeparty left dummy	dummy, left-wing government leaders as 1, and the others as 0	Database of Political Institutions
manufacture value added	manufacturing value added as a percentage of GDP of a certain year	Word Bank Indicators
Controls
logGDPpp	logarithm of GDP per capita (constant 2010 US$) for certain years	Word Bank Indicators
inflation	inflation	Word Bank Indicators
exchange rate	exchange rate	Word Bank Indicators
polity IV	Measures level of democracy value ranges from -10 to 10	Polity IV
fdi	FDI as % of GDP	Word Bank Indicators
commodity	commodity prices indices, annual (2015=100)	UNCTAD data