Effects of Trade Liberalization on Income
Inequality in Latin America since 1985
Daniela Ospina
daniela.ospinaalonso@student.uva.nl
Student number: 11375647
Supervisor: Naomi Leefmans
Second reader: Dr. Dirk Veestraeten
Number of words: 14,987
2
STATEMENT OF ORIGINALITY
This document is written by Student DANIELA OSPINA ALONSO who
declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business of the University of Amsterdam is
responsible solely for the supervision of completion of the work, not for the
3
Abstract
The notion that trade openness leads to an increase in income inequality gained popularity in Latin-America at the end of the 1990s. This led to the positioning of several left-wing leaders characterized by an anti-trade rhetoric throughout the region in the 2000s. Using a GMM methodology and a panel data for a sample of 17 Latin-American countries, this thesis aimed at estimating the impact of international trade and its different types on income inequality. The most important contribution of this analysis is a simultaneous estimation of the effects of the different types of trade after controlling for the existence of redistributive policies which brings together the works of older and more recent literature on the matter. The main finding of this thesis is the low negative impact of trade openness on income inequality. Other variables such as infrastructure, financial depth and governance seem to provide a better explanation for the fall of income inequality since the 2000 indicating that trade openness has led to a slightly more equal income distribution in Latin America. These results are in line with the findings of recent studies.
4 Table of Contents
1. Introduction ... 5
2. Historical Background of trade liberalization in LA and recent developments ... 8
3. Theoretical Framework ... 14
4. Literature review ... 18
4.1 Early studies on the impact of trade openness on income inequality ... 18
4.2 Recent studies on the impact of trade openness on income inequality ... 19
4.3 Impact of the different types of trade on inequality ... 21
5. Methodology ... 22 6. Data Analysis ... 26 6.1 Dependent variable ... 26 6.2 Explanatory variables ... 27 7. Empirical Results ... 36 8. Concluding Remarks ... 44 9. References ... 46 Tables Table 1 – Relevant FTAs in Latin-America ... 11
Table 2 – Determinants of inequality – Impact of trade openness ... 37
Table 3– Impact of openness to different types of trade on income inequality... 38
Graphs Graph 1 – Gini-index for several Latin-American countries, 1990-2014 ... 6
Graph 2– Average MFN Tariff (%), 1985-2012 ... 9
Graph 3– Non-Tariff Measures Incidence, 1985- 1998 ... 9
Graph 4 – Foreign Direct Investment inflow as % of GDP, 1990-2014 ... 10
Graph 5- Imports and Exports of goods and services (% GDP) in Latin-America, 1985-2015 ... 12
Graph 6 – Extension of the Heckscher-Ohlin model ... 15
Graph 7 – Changes in the Heckscher-Ohlin extension ... 16
Graph 8– Gini coefficient by groups of countries, 1985-2014 ... 27
Graph 9 – Trade openness by groups of countries, 1985-2014 ... 28
Graph 10 – Primary trade by groups of countries, 1985-2014 ... 28
Graph 11 – Components of primary trade: food, fuel, mining and raw agricultural materials trade intensity, 1985-2014 ... 29
Graph 12 – Secondary trade by groups of countries, 1985-2014 ... 30
Graph 13 – Tertiary trade by groups of countries, 1985-2014 ... 30
Graph 14 – Financial depth by group of countries, 1985-2014 ... 31
Graph 15 – Public infrastructure, 1985-2014 ... 31
Graph 16 – Real effective exchange rate (2010=100), 1958-2014 ... 32
Graph 17 – Rule of Law Indicator by group of countries, 1985-2014 ... 33
Graph 18– Inflation CPI by group of countries, 1958-2014 ... 33
Graph 19 – Terms of Trade by group of countries 1985-2014 ... 34
Graph 20 – Secondary Education Enrolment by group of countries, 1985-2014 ... 35
5 1. Introduction
Since the mid-1970s Latin-American countries have opened their economies to the international market through the elimination of barriers to trade and the subscription to numerous Free Trade Agreements (FTAs). In general, trade liberalization in Latin-America was characterized by a rapid reform implementationthat lasted on average two to five years and aimed at dismantling old trade barriers in the form of tariffs and quantitative restrictions on imports. Simultaneously, Latin-American economies also opened their capital accounts through the elimination of restrictions and the subscription to several BITs1 and DTAs2. As a result, cross-border exchange of capital in the region has grown dramatically since the mid-1980s.
Trade liberalization in Latin-America was also achieved through the implementation of regional trade agreements like the reinforcement of the Andean Community (CAN) in 1989. Additionally, Latin-American economies engaged also in a series of bilateral trade agreements with different economies around the world. These two types of FTAs have accelerated the introduction of Latin- American economies to the international market even further.
However, even though these economies have successfully integrated into the global economy, trade liberalization was accompanied by a general increase in income inequality in the region during the nineties decade. Graph 1 presents the Gini coefficient for several Latin-American economies between 1989 and 2014. As shown, in Argentina, Bolivia, Colombia, Dominican Republic, Ecuador and Paraguay income inequality increased significantly in the 1990s just after liberalization reforms were implemented. For this reason, trade openness has been associated with higher income inequality and poverty levels by some. This led to a new wave of an anti-trade rhetoric among politicians throughout the region in the 2000s.
1 Bilateral Investment Treaties
2 Double Taxation Agreements: treaties aimed at preventing double taxation of firms by two different
6 Graph 1 – Gini-index for several Latin-American countries, 1990-2014
Source: World Bank Povcal Database
Due to the unexpected worsening of income inequality in the region, the academic community tried to estimate the impact of trade openness on income inequality in the period that followed trade liberalization. These early studies found a statistically significant positive impact of the former variable on the latter. Additionally, these reports concluded that the new flows of FDI that entered these economies in this period also contributed to the increase in income inequality. These early studies also stressed the need for effective policies to distribute equally the gains from trade across the population, especially among the poor. Nevertheless, this first wave of studies focused on the effect of trade in the period that followed after trade liberalization reforms were implemented. However, since 2000 income inequality has mostly decreased significantlyin these countries as Graph 1 shows. In some economies, e.g. Argentina, income inequality has fallen below its 1991 value. For this reason, and given the few economic studies focusing on international trade and income inequality in this region after 2000, this paper’s main objective is to answer the following research question: What has been the impact of international trade on income inequality in Latin-America for the period 1985-2014? This thesis thus seeks to contribute to the existing economic literature by complementing previous studies by looking at a longer time frame by considering the most recent statistics available. Furthermore, in the estimation process an interaction variable between the period of time and the Gini coefficient is introduced in attempt to capture the individual effect of the decision of liberalizing trade itself on income inequality. Additionally, for this estimation only data regarding Latin-American countries,
35 40 45 50 55 60 65 1990 2000 2010 2014
7 our region of interest, would be considered unlike previous studies which include also information for the Caribbean and other developing economies.
Another contribution of the present study lays in the fact that the methodology used in this estimation is based on the GMM framework presented inPerry & Olarreaga (2006)3 while using the equation estimated by Tsounta & Anayochukwu (2014). The latter study introduces the level of tax revenue as a percentage of GDP as an additional variable to the original group of controls in the former study and is used as a proxy for controlling for the existence of redistributive policies. Such policies lower income inequality and therefore need to be controlled for. Another contribution of the present analysis is to distinguish between liberalization of the three main types of trade (primary, secondary and tertiary) in order to estimate their individual impact on income inequality. The first of these is further divided into trade of food, fuel, mining and agricultural raw materials. This last contribution is innovative since previous studies have estimated the impact of the liberalization of each trade type in separate regressions without considering their individual impact simultaneously in a single regression. This consolidated estimation helps to measure the effect of trade and of each of its components on income inequality in a more precise way. For this reason, the present analysis aims at bringing together the estimation analysis of Perry & Olarreaga (2006), Alderson & Nielsen (2002) and Cassette, Fleury, & Petit (2010) for primary, secondary and tertiary goods trade, in an effort to complement the existing economic literature.
In terms of data, this study employs panel data ranging from 1985 until 2014. The sample of countries4 is the same used by Tsounta & Anayochukwu (2014) in their analysis, excluding non-Latin-American economies since these are not the focus of the present study. A combination of the two methodologies used in Perry & Olarreaga (2006) and in Tsounta & Anayochukwu (2014) is innovative and is expected thus to contribute through the delivering of a more precise estimation.
This paper is organized in the following way: Chapter 2 presents and explains in greater detail the historical background of trade liberalization in Latin-America and recent developments in the region concerning international trade. Chapter 3 explains the theoretical framework considered in the present study, Chapter 4 discusses existing economic literature on the effect of trade openness on income inequality and its main conclusions, while Chapter 5 introduces the methodology that will be used in the present paper’s estimation. Chapter 6 describes the data used and Chapter 7 discusses the estimation results. Finally, Chapter 8 presents some concluding remarks.
3
The methogology of this study is based on López (2003).
4
Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatamala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru and Uruguay.
8 2. Historical Background of trade liberalization in LA and recent developments Trade liberalization of Latin-American economies in the mid-1970s represented a shift towards an export-oriented model in replacement to the Import Substitution Industrialization framework that preceded it. The Import Substitution Industrialization, popular in this region throughout the fifties and sixties, was expected to bring great economic development through the strengthening of domestic production facilitated by high barriers to international trade and capital flows. However, its effectiveness was first questioned in the seventies due to the slow growth and precarious condition of the industrial sector in the region. For this reason, in the mid-1970s Latin-American economies opened themselves to the international market through the elimination of trade barriers and the subscription to new trade agreements. Trade liberalization in Latin-America was a process characterized by a rapid implementation of new reforms for increasing cross-border interaction that lasted on average three to five years per country (Agosin & French-Davis, 1993). The first of these countries to engage in liberalization reforms was Chile in the mid-1970s, quickly followed by Bolivia and Mexico in 1985 and by Venezuela, Brazil and Colombia, among others, in the 1990s. This liberalization entailed the elimination of tariffs and quantitative restrictions.
As Graph 2 suggests, since 1985 Latin-American economies significantly reduced the amount of tariffs applied to imports. For example, Colombia’s average import tariff went from 83 to 6,5 percent between 1980 and 2012. Most of these countries also significantly reduced the number of Non-Tariff Measures (NTM). As can be observed in Graph 3, Mexico, Colombia and Nicaragua present a dramatic decrease in the use of these barriers, whereas in Chile these measures remain unchanged and in Brazil they increased. Likewise, trade liberalization in Latin-America was also accompanied by the opening of the capital account through the elimination of restrictions and the implementation of Bilateral Investment Treaties and Double Taxation Agreements. As a result, cross-border exchange of capital has grown since the mid-1980s. Latin-American economies have received significant inflows of foreign capital, mainly in the form of FDI and Portfolio Investment. As Graph 4 suggests, there was an increase in FDI inflows in several countries such as Chile and Peru, with the exception of Argentina and Brazil. These movements can be explained by the wave of political, social and economic instability in the latter countries when compared to the formerg. The graphs below thus reflect the rapid integration of Latin-American economies into the international market.
9 Graph 2– Average MFN Tariff (%), 1985-2012
Source: Inter-American Development Bank - Lederman, Perry, & Suescún (2002). “Trade Structure, Trade Policy and Economic Policy Options in Latin-America”
Graph 3– Non-Tariff Measures Incidence, 1985- 1998
Source: Lederman, Perry, & Suescún (2002). "Trade Structure, Trade Policy and Economic Policy Options in
Central America" 0 10 20 30 40 50 60 70 80 90
Argentina Brazil Chile Colombia Mexico Perú
Av erag e MN F Tariff (% ) 1985 1992 1999 2012 0 10 20 30 40 50 60
Argentina Brazil Chile Colombia Mexico Nicaragua
NTM i
ncidence
(%
)
10 Graph 4 – Foreign Direct Investment inflow as % of GDP, 1990-2014
Source: UNCTAD
Trade liberalization in the region was also achieved through the subscription to and implementation of regional trade agreements with other economies around the world. The reinforcing of the Andean Community (CAN) in 1989 was the first significant step towards a multilateral approach to trade liberalization. Even though the CAN was established in 1969 with the intention of creating a Custom Union and Common Market, this agreement quickly lost force throughout the seventies and eighties due to the popularity of the ISI economic model in the region. However, its renovation by its then member countries Bolivia, Colombia, Ecuador, Peru and Venezuela consisted of an initial agreement for the elimination of barriers to trade between them, and later on the implementation of an important free trade area in the region in 19935. Nowadays, this FTA has become stronger thanks to the cooperation with important trading partners such as the European Union.
Likewise, in 1991 another regional trade agreement of great economic importance today was implemented. Mercado Común del Sur (MERCOSUR), initially formed by Argentina, Brazil, Paraguay and Uruguay, is a FTA aimed at promoting economic and financial integration between its member countries for boosting trade and investment opportunities. It was later joined by Venezuela in 2006 and Bolivia in 2015. According to the 2014 World Economic Outlook of the IMF, together its member countries represent the fifth largest economy of the world. Later in 1994, Mexico, the United States and Canada implemented one of the most important FTAs today: the North American Free Trade Agreement (NAFTA). This treaty allowed for a gradual removal of barriers to trade, as well as of restrictions on capital flows between these economies. Today NAFTA is considered one of the largest FTAs of the world and since it came into effect in 1994, trade merchandising among these three economies has more than tripled reaching in 2008 a value of US$946.1
5 Organización Comunidad Andina. www.comunidadandina.org
0 1 2 3 4 5 6 7 8 9
Argentina Brazil Chile Colombia Costa Rica Mexico Peru
%
of
GDP
11 billion, according to official reports6. Additionally, the discussion for a possible new free trade area began in 1994 under the name of Free Trade Agreement of the Americas (FTAA). This agreement aimed at expanding NAFTA to all Central and South American countries, as well as the Caribbean countries with the exception of Cuba. However, the negotiations expired in 2005 and the participant members failed to reach a consensus. In addition, this FTA faced the rejection of key economies in the region such as Brazil, as well as the opposition from the general public.
Nevertheless, the creation of the World Trade Organization (WTO) and the subscription of the economies in the region since 1995 to this organization also fuelled the transition towards a liberal economic model. Similarly, in 2011 the Pacific Alliance was implemented by Chile, Colombia, Mexico and Peru as an initiative for establishing a free trade area between these economies. The Pacific Alliance agreement seeks to promote a further economic and commercial integration with the rest of the economies worldwide, especially with the Asian-Pacific region and is expected to deliver promising results. Table 1 lists the most important multilateral FTAs in force today in Latin-America with their member countries and associated economies, as well as their enforcement date.
Table 1 – Relevant FTAs in Latin-America
FTA Year of
implementation
Member countries Associated
members Original Today Andean Community - Comunidad Andina (CAN) 1969 (reinforced in 1989) Bolivia, Colombia, Ecuador, Peru and
Venezuela
Bolivia, Colombia, Ecuador and Peru
Chile, Argentina, Brazil, Paraguay
and Uruguay
Mercado Común del
Sur (Mercosur) 1991 Argentina, Brazil, Paraguay and Uruguay Argentina, Brazil, Paraguay, Uruguay, Bolivia and Venezuela Chile, Colombia, Ecuador, Peru North American Free
Trade Agreement (NAFTA)
1994 Canada, Mexico and USA
Canada, Mexico and
USA None
Pacific Alliance -
Alianza del Pacífico 2011
Chile, Colombia, Mexico and Peru
Chile, Colombia, Mexico and Peru
None, but has 49 observer countries
Source: Official website NAFTA, MERCOSUR, CAN, Pacific Alliance
However, reaching consensus during the negotiation of a multilateral FTA is of great difficulty due to the economic, political and social conflicts of interest between the participating countries. A clear example has been the last round of negotiations of the WTO which is still open today after more than fifteen years and the failure of the FTAA. Because of this, Latin-American economies have also engaged in a series of bilateral trade agreements since the mid-1980s. These bilateral FTAs have not only helped promote a higher trade
12 interaction within the region, but also with developed economies around the world such as the United States, the European Union and China.
In the last forty years, Latin-American economies thus have opened up themselves and have successfully integrated to the international market. The value of exports and imports as a ratio of GDP in the region has grown significantly. As Graph 5shows, this indicator of trade volume for Latin-America has presented an increasing trend since 1985. However, it fell between 2007 and 2009 due to the global economic crisis. Even though it has been upward sloping ever since, these values have not fully recovered to their pre-crisis levels.
Graph 5- Imports and Exports of goods and services (% GDP) in Latin-America, 1985-2015
Source: World Bank
This new openness to international trade was however accompanied by a rise in within-country income inequality in Latin-American economies throughout the nineties decade. The worsening of income distribution in these countries led to the strengthening of an anti-trade position among politicians in the region which blamed exposure to international trade as the main reason why inequality had increased. As a result, during the first decade of the 21st century an anti-trade rhetoric among politicians throughout Latin-American countries quickly gained popularity. Additionally, the ghost of the financial crisis and the economic slowdown of these economies reinforce the attractiveness of this anti-trade rethoric. This led to the positioning of several left-wing leaders as presidents such as Evo Morales in 2006 in Bolivia and Rafael Correa in 2007 in Ecuador, among others.
One of the most recognizable worldwide populist presidents in the region was Hugo Chavez. The former president of Venezuela took power in 1992 and was the first mandatary in Latin-America to publicly condemn free trade and swear to fight against ‘imperialism’. In 2005 at the annual Summit of the Americas, the Venezuelan leader publicly stated“I think we came
5 10 15 20 25 30 GD P rati o (%) Imports (% GDP) Exports (% GDP)
13 here to bury the FTAA and I brought my shovel.”7 Like Chavez, many left-wing leaders in the region shared a common stand regarding international trade which led to the implementation of a series of non-traditional protectionist measures in the 2000s. These new trade barriers surged in the form “of standards, certification procedures and antidumping and antisubsidy measures” against other countries as explained byFischer & Meller (1999). In general, in the 2000s left-wing governments in most of Latin-American economies were characterized by a significant increase in social spending, especially in projects regarding education access and quality. According to World Bank data between 2006 and 2009 government spending on education as a percentage of GDP in Bolivia under Evo Morales mandate increased from 6.3 to 8.1 percent, one of the highest in value and growth in the region. However, at the end of the 2000s, several corruption scandals in Latin-America involving populist political leaders became public which has severely damaged their credibility and has led to a weakening of populism in the region since 2010.
In addition, income inequality in most of Latin-American economies has presented a downward trend since 2000. As Graph 1 suggests, the countries with the most significant decreases in income inequality have been Argentina, Bolivia, Colombia, Ecuador and El Salvador. These countries have engaged in ambitions plans aimed at the reduction of inequality. For example, in 2014 Colombia implemented its National Plan for Development which consists of the improvement of six key aspects for inequality reduction: strategic infrastructure, social mobility, rural areas transformation, security, good governance plus justice for the consolidation of peace, and green growth. Likewise, this project has reinforced the importance of education access and quality to the entire population. Currently, education is its most important goal with more than 37,000 million dollars destined to it. However, this has not been the experience of the other countries in the region such as Brazil and Mexico. In Brazil, political instability, corruption scandals and economic stagnation hava led to a general worsening of the country. Still, amid this downturn, its income inequality has also decreased significantly since 2000, while in Mexico, even though it decreased between 2000 and 2010, its inequality level has remain unchanged ever since.
14 3. Theoretical Framework
This chapter presents and discusses the theoretical basis that explains the mechanisms through which international trade can affect income inequality. First, a brief explanation of the popular Heckscher-Ohlin (H-O) model is made, followed by the introduction of an extension to this framework presented in Leamer (2012).
The Heckscher-Ohlin (H-O) model was the theoretical framework behind the popular conception that trade liberalization in Latin-America would bring great economic growth and social welfare to the countries in the region. The main idea behind this model is that once economies start to engage in international trade, they will gradually specialize in the production of the good whose manufacture makes intensive use of the factor of production their economy enjoys a relative abundance of. Due to the difference in initial endowments of factors of production across economies, trade liberalization will increase the welfare in the countries involved in international trade due the changes in patterns of production which will deliver higher returns to the owners of the input abundant in each economy. On this simple mechanism lays the theoretical basis behind the effect of international trade on income inequality within an economy that the H-O model suggests. Additionally, from this model a theoretical proposition known as the Stolper-Samuelson theorem develops. According to this proposition, a relative increase in the price of a final good, will lead also to a raise in the price of the factor of production in which its production is relatively intense in, while the price of the scarce input will decrease. This movement will happen due to the forces of supply and demand in the market.
As demonstrated by Perry & Olarreaga (2006), Latin-America is a region relatively abundant in capital and skilled workers compared to other developing economies such as China or India. For this reason, due to the intensity of international trade in this territory and in line with the predictions of the H-O model, it could be expected that capital holders and skilled workers will enjoy higher economic benefits than unskilled labour. As a result, the increase in wage inequality between skilled and unskilled workers, as well as the increase in income inequality in Latin-American countries during the nineties decade could have been expected. However, contrary to what this model suggests, the GINI Index for some economies in the region has significantly decreased since 2000, as shown in Graph 1.
Also, the simple version of the H-O model is unable to explain the presence of skill premiums within a same group of individuals, e.g. skilled workers. For this reason, and due to the intention of measuring the impact of international trade on this phenomenon in Latin-America, an extension of the H-O model that explains the existence of skill premiums among skilled workers is introduced. This extension appeared for the first time in Leamer (1995) and was later improved in Leamer (2012).
In this extension, it is assumed that within the skilled workers group, there are heterogeneous educational levels among individuals and that the most educated labour enjoys a higher
15 complementarity with capital that is reflected on its productivity level. This compatibility allows for a higher productivity of skilled workers when in contact with an increase in the amount of capital they have access to. There are four goods in this economy: A, B, C and L. The latter is labour intensive, especially in unskilled workers, while the other three (A, B and C) are capital intensive and employ skilled labour. The difference between A, B, and C is the level of education of the skilled workers involved in the production process, and therefore its productivity level when in contact with capital. A is the good with the most qualified skilled workers, followed by B and C, respectively. Graph 6 shows the initial situation with four different Leontief production functions for A, B, C and L goods. The downward sloping function is the isovalue line which defines the relationship between the production of two or more products for a given market value. Naturally, given that A employs the skilled labour that is the most productive, it is located on an isovalue line that is steeper than the one shared by good B, C and L. This reflects the fact that capital returns, as well as the wage level paid to the inputs involved in the production of good A, are the highest followed by B and C. The wage levels perceived by A, B and C skilled workers corresponds to w(A), w(B) and w(C) in Graph 6. So the initial situation presented in this graph allows for a differentiation in the wage paid to skilled workers in terms of its qualifications and productivity levels.
Graph 6 – Extension of the Heckscher-Ohlin model
Source: Haskel et al. (2012). “Globalization and U.S. Wages: Modifying Classic Theory to
Explain Recent Facts”
Now, it is assumed that as a result of trade liberalization in this economy, there is an increase in the demand for capital intensive goods since it is relatively abundant in capital (skilled workers) in relation to labour force (unskilled workers). Therefore, production in this economy will shift towards the good that makes intensive use of the input its production is relatively intense in. So naturally, production of good L will decrease, as well as the wages
16 paid to its unskilled labour employed. Also, due to the higher return of capital in the international market, the isovalue lines of goods A, B and C will shift toward the origin since production costs have increased as it can be seen in Graph 7. It would be expected for the wage level of the skilled workers employed in its production to rise equally. However, this does not happen due to differences in the productivity levels of these workers.
Graph 7 – Changes in the Heckscher-Ohlin extension
Source: Haskel et al. (2012). “Globalization and U.S. Wages: Modifying Classic Theory to Explain Recent Facts”
As Haskel, Lawrence, Leamer, & Slaughter (2012) explain, even though demand for all capital intensive goods have increased, the wages paid to skilled workers involved in the production of B and C will decrease. This will happen because these workers are insufficiently skilled “to command higher wages in the face of the increased price of capital with which they have to work”. The increase in the cost of capital during the production of good A is completely offset by the increase in productivity of its skilled labour. So under this scenario, the type-A workers are the ones that enjoy a higher wage in comparison to those involved in the production of B and C. This result helps to explain the increase in wage inequality between skilled and skilled labour, but most importantly the existence of skill premiums. As this extension suggest, an increased disparity in the wages paid to skilled workers can be explained through differences in their educational level which is intrinsically related to its productivity.
In sum, the well-known H-O model was the foundation behind the perception that trade liberalization would bring great economic growth and social welfare to an economy. However, due to the relative abundance in capital and natural resources in Latin-America in comparison to other developing economies, income inequality increased. This result is further explained by the theoretical expansion of the H-O model presented by Leamer (2012).
17 In this extension, differentials in productivity among individuals of a same group, e.g. skilled workers, help explain wage inequality within this workers. For this reason, these theoretical frameworks provide an explanation for the increase in wage inequality and for skills premiums among workers of a same group observed in Latin-America in the 1990s after trade liberalization took place.
18 4. Literature review
Estimating the effects of trade openness reforms on income inequality in Latin-America since the mid-1970s has been of great popularity in the economic literature. There was a need among politicians and the academic community to fully understand the mechanisms and the factors that led to the unexpected worsening of the income distribution in the region in the period that followed trade liberalization. In this section an overview of the most relevant economic literature on the impact of international trade openness on income inequality is made. This chapter is divided into three parts: section 3.1 presents the literature that has estimated this impact for the period that followed immediately after trade liberalization (mid 1970s-2000), while section 3.2 presents an overview of the literature that also takes into consideration recent events (1980-2017). The present study divides the relevant literature on the general impact of trade openness on income inequality into two different sections (3.1 and 3.2) due to the strong difference in the estimation results obtained by the two groups of studies. Finally, in section 3.3 a brief overview on the literature analysing the impact of different types of trade on income inequality is made.
4.1 Early studies on the impact of trade openness on income inequality
In a World Bank study, Perry & Olarreaga (2006) estimated the effect of trade liberalization on income and wage inequality, and poverty in Latin-American countries. In this paper, the authors employed a GMM regression analysis using panel-data for Latin-America and the Caribbean (LAC) for a period ranging from 1960 until 2000. From these estimations, the authors were able to conclude that there was a general worsening in income and wage inequality in the region for several reasons related to international trade. First, at the time of the implementation of liberalization reforms in Latin-American economies, these countries presented a relative abundance of capital and of natural resources in comparison to other developing countries which had a much larger pool of unskilled labour, such as China and India. This helps to explain why the benefits of trade were in general absorbed by capital holders and skilled labour, which caused a further increase in income inequality and in skill premiums since capital, but also natural resources extraction, is generally associated with high-skill labour. Likewise, the lack of labour mobility among unskilled workers in these countries reinforced this gap. Finally, the fact that trade liberalization in the region occurred almost 20 years after East Asian countries had opened their economies to international trade also seems to have had an effect on this outcome due to a saturation of the global market by an excessive supply of unskilled workers from East Asia. In general, in their empirical analysis Perry & Olarreaga (2006) found a positive effect of trade openness on income inequality: more intense trade liberalization led to an increase in income inequality. This finding is statistically significant and is the same result obtained by De Ferranti et al. (2003) and other studies that consider the same time frame (Jakobsson, 2006).
Perry & Olarreaga (2006) and De Ferranti et al. (2003) also conclude that, given the worsening in income and wage inequality, there is an urgent need for the implementation of
19 public policies aimed at reducing income inequality that would help reap the benefits from international trade in Latin-America. These policies are divided into two sets. The first of these, is a group of policies aimed at distributing the gains from trade and equalizing the benefits throughout its population, especially to the poor. These policies, as Perry & Olarreaga (2006) explained, are expected to compensate for the skill-biased labour demand in the international market. The second set of reforms is related to a higher government spending on education and social safety nets which will positively affect the development of new skills of its poorest population. As these authors suggest, these policies were expected to reduce skill premiums in the region. This study concludes that trade leads to higher income inequality in Latin-American economies that can be off-set by indirect policies that can help to redistribute more equally trade gains among the population.
4.2 Recent studies on the impact of trade openness on income inequality
The studies described above have focused only on the period that followed immediately after trade liberalization in Latin-American economies took place. However, since 2000 there has been a significant reduction of income inequality in these countries. In a study for the IMF, Beaton, Ceborati, & Komaromi (2017) estimate the impact of trade openness on income inequality in LAC for a period ranging from 1980 till 2013. The authors also employ a GMM analysis with an unbalanced panel database containing observations for 131 countries. During its estimation, Beaton, Ceborati, & Komaromi (2017) introduce previous values of the dependent and explanatory variables for controlling for any possible endogeneity problems. This regression is further complemented with event studies of past episodes of trade liberalization. Using the rate of change of GDP for five years averages as a dependent variable, the authors concluded that “it is hard to find a statistically significant or economically sizable direct negative [effect] of trade on inequality”. As they explain, the effect of international trade on income inequality depends greatly on country specific characteristics such as the level of economic development on the effectiveness of the tax system. Additionally, and in line with the results obtained by Perry & Olarreaga (2006), the existence and the effective implementation of redistributive policies and of safety nets are key for countries to fully benefit from international trade and reduce their income inequality. Due to these findings, during the estimation process in this thesis control variables for the existence of redistributive policies are introduced.
Likewise, a Cluster Report published by the IMF (2017) focusing on the experience of LAC economies, also stressed the need for government interventions for guaranteeing a successful distribution of international trade gains among its population. Additionally, this report found a negative but insignificant impact of trade openness on income inequality. In general, there is limited evidence of the impact of trade on inequality. On one side, greater exports markets and global value chains are expected to decrease inequality while concentration in high-skill manufactured products tends to increase it. As in the study by Beaton, Ceborati, & Komaromi (2017), fixed country specific characteristics matter greatly.
20 Furthermore, in an IMF study Tsounta & Anayochukwu (2014) identify the forces and mechanisms behind the fall in the Gini coefficient since 2000 for these Latin-American countries. They estimated a pooled OLS analysis using unbalanced panel data containing 44 emerging and developing countries for a period ranging from 1990 to 2010. In this regression, the natural logarithm in the Gini index acted as the dependent variable in line with the work of Perry & Olarreaga (2006) and Lopez (2003). Likewise, the authors control for country and time fixed effects. From this estimation, they are able to identify two main forces behind the fall in income inequality in Latin-America since 2000: well-designed redistributive public policies and a strong economic growth. According to their results, the first of these factors is responsible for more than two thirds of the reduction of income inequality in these countries. These results support the concluding remarks of the studies described above regarding the need for redistributive policies. As Tsounta & Anayochukwu (2014) explain, policies based on a higher education enrolment and participation were the key driving force behind this decrease. Additionally, this study also showed that a higher level of tax revenue and of FDI entering the country contributed to the reduction of income inequality. As these authors explain, a higher tax revenue level facilitated government spending and boosted the implementation of public policies designed to improve education attainment in these countries.
In sum, the economic literature that analysed the period that followed immediately after trade liberalization found on average a statistically significant negative effect of trade openness on income inequality. In contrast, most recent studies have found a positive impact that lacks statistical significance. However, the two groups of reports were able to conclude that the impact of trade openness on income inequality depends heavily on the existence of redistributive policies and of country specific characteristics.
For this reason, a combination of the control variables used in the studies described here, are incorporated in the present study to guarantee better results and a more precise estimation of the effect of international trade on income inequality. This also represents a contribution of this thesis to the existing economic literature: an estimation that employs the traditional control variables used by Perry & Olarreaga (2006), with an additional control for the existence of redistributive policies by using as a proxy the level of total tax revenue as a percentage of GDP for an improved analysis as introduced by Tsounta & Anayochukwu (2014). This is innovative since the present analysis simultaneously estimates the impact of trade on inequality while also controlling for the existence of redistributive policies. In addition, another contribution is the introduction of an interactive variable between the time period and trade openness. This new variables seeks to control for any specific effect the time of trade liberalization may have had on the level on income inequality. This interaction is innovative since the studies mentioned above did not estimate this relationship.
21 4.3 Impact of the liberalization of different types of trade on inequality
Perry & Olarreaga (2006) also distinguished between the liberalization of different types of primary trade for estimating its impact on income inequality. They wanted to test the hypothesis that minerals and fuels extraction trade have a stronger negative impact on income inequality than non-extractive activities such as raw agricultural materials production. In their analysis they used interaction variables of trade openness and intensity in trade of food, fuel, mining and raw agricultural materials. In general, these authors found that the coefficient for the main effect of trade openness on income inequality is still positive and statistically significant despite the different types of primary trade. Nevertheless, they also found that the magnitude of this effect was lower when trade intensity in food and raw agricultural materials grew. On the contrary, the magnitude of the main effect of trade openness was higher when trade openness on mining and fuel increased. The impact of fuel trade is the only one that lacks statistical significance though. These results are consistent with the ones reported by UNCTAD (2012): food and agricultural developing exporting countries tend to have on average a lower level of income inequality and poverty, while in mineral and oil exporting economies these indicators were on average higher.
In terms of secondary trade, the latter study found that economies with a higher trade openness in manufacture goods tend to have higher income inequality. On the contrary Alderson & Nielsen (2002) found that higher participation in manufacture trade is correlated with lower income inequality for a sample of OECD countries. As these authors suggest, greater openness to manufacture trade increases the average wage in an economy which can help reduce inequality. However, when it comes to tertiary trade, this variable has a positive correlation with income inequality. Cassette, Fleury, & Petit (2010) also found that for a group of OECD countries trade in services is has a negative impact on income inequality unlike secondary goods.
The division of trade openness between the main three types of trade: primary, secondary and tertiary, as well as a further distinction for primary goods trade into food, fuel, mining and raw agricultural materials represents another contribution of the present analysis. A simultaneous estimation of the effects of the main components of trade and its level of intensity on income inequality behaviour in one single regression is innovative. The present analysis aims at bringing together the estimation analysis of Perry & Olarreaga (2006), Alderson & Nielsen (2002) and Cassette, Fleury, & Petit (2010) for primary, secondary and tertiary trade goods, respectively in an effort for complementing the existing economic literature where these impacts have only been measured separately.
22 5. Methodology
The methodology used in the present study is based on a combination of the GMM methodology in Perry & Olarreaga (2006)8 and the regression equation used in Tsounta &
Anayochukwu (2014) presented below:
𝐺𝑖𝑡 = 𝛼𝐺𝑖𝑡−2+ 𝛽1𝑇𝑖𝑡+ 𝛽2𝑋𝑖𝑡+ 𝜇𝑖+ 𝜀𝑖𝑡 (1.1)
In equation (1.1) 𝐺 is the Gini coefficient of country 𝑖 at time 𝑡, 𝑋 is a set of explanatory variables listed below, 𝜇𝑖 refers to country fixed effects and 𝜀𝑖𝑡 is the error term. In this study, the Gini coefficient is used as an indicator of within country inequality due to the general acceptance and wide use in economic literature, in contrast to other indicators such as the divergence over time of unskilled versus skilled labour income. Additionally, our explanatory variable of interest in this study is the level of trade liberalization between 1985 and 2014 present in equation (1.1) in the form of T.
In this study, trade openness is defined as the sum of the value of exports and imports as a percentage of GDP. This is the same definition used by López (2003). From the basic equation (1.1), an initial distinction is made for trade openness by differentiating between the three main types of trade as shown in equation (1.2). In addition, in an attempt to control for any specific effect the time period of trade liberalization may have had on income inequality, an interaction variable between time and trade openness is introduced in the equation below.
𝐺𝑖𝑡 = 𝛼𝐺𝑖𝑡−2+ 𝛽3𝑃𝑟𝑖𝑚𝑖𝑡+ 𝛽4𝑆𝑒𝑐𝑖𝑡+ 𝛽5𝑇𝑒𝑟𝑡𝑖𝑡+ 𝛽2𝑋𝑖𝑡 + 𝛽6𝑡 ∗ 𝑇𝑖𝑡+ 𝜇𝑖 +𝜀𝑖𝑡 (1.2)
Where𝑃𝑟𝑖𝑚𝑖𝑡 , 𝑆𝑒𝑐𝑖𝑡 and 𝑇𝑒𝑟𝑡𝑖𝑡 refers to primary, secondary and trade respectively. However, following the methodology of Perry & Olarreaga (2006), a further distinction between the different types for primary trade is made in equation (1.3). In the equation below𝐹𝑜𝑜𝑑𝑖𝑡, 𝐹𝑢𝑒𝑙𝑖𝑡, 𝑀𝑖𝑛𝑖𝑛𝑔𝑖𝑡 and 𝐴𝑔𝑟𝑖𝑐𝑢𝑙𝑖𝑡 stand for food, fuel, mining and raw agricultural materials trade respectively. The division of primary trade into these four categories is the same used by the World Bank.
𝐺𝑖𝑡 = 𝛼𝐺𝑖𝑡−2+ 𝛽6𝐹𝑜𝑜𝑑𝑖𝑡+ 𝛽7𝐹𝑢𝑒𝑙𝑖𝑡+ 𝛽8𝑀𝑖𝑛𝑖𝑛𝑔𝑖𝑡+ 𝛽6𝐴𝑔𝑟𝑖𝑐𝑢𝑙𝑖𝑡 + 𝛽4𝑆𝑒𝑐𝑖𝑡+ 𝛽5𝑇𝑒𝑟𝑡𝑖𝑡 +𝛽2𝑋𝑖𝑡 + 𝛽6𝑡 ∗ 𝑇𝑖𝑡+ 𝑔𝜇𝑖+ 𝜀𝑖𝑡 (1.3)
Equations (1.1), (1.2) and (1.3) are estimated separately and the results are presented in Table 2 and Table 3 in Chapter 7. As mentioned before, 𝑋𝑖𝑡 refers to the set of control variables that are used in this estimation. These variables are initial GDP per capita, financial depth,
23 infrastructure, governance, price stability, real exchange rate, terms of trade, time, secondary enrollment and tax revenue. These controls are further discussed below.
-Openness to trade differentiated by primary (agricultural, mining, fuel and raw materials), secondary and tertiary (service) products: in theory, trade is expected to benefit the economy as a whole, but will generate winners and losers as first presented in the Heckscher-Ohlin model. Theoretically, trade is expected to increase inequality in the short-run, but to be off-set and actually decrease in the long-run as stated by the debated Kuznets curve. The mechanisms through which trade openness can impact income inequality are the levels of wages and of capital returns as explained further in the Chapter 3. Also, in the present analysis secondary and tertiary trade refers specifically to manufactures and services trade respectively.
- GDP per capita (logarithm): A higher economic growth is positively correlated with lower inequality levels in the long-run, unlike the short-run. This is related to the much debated Kuznets curve theory which argues that on the first stages of trade openness the income gap between rich and poor will increase since investment opportunities only benefit the former group. Later, as economic growth increases unskilled workers will move to the city and seek for better pay jobs at manufacturing industries which eventually helps to decrease inequality between these two groups in the long-run.
- Price Stability: in theory, higher price volatility is expected to affect in greater proportion the consumption level of the poor since this population spend in comparison more of its income on this goods than the rich. For this reason, an increase in volatility could lead to a worsening in income equality.
- Secondary education enrollment: theoretically, it can be expected that higher human capital leads to higher economic growth due to a rise in productivity and economic benefits. Likewise, an increase in human capital is positively correlated with a lower level of inequality in economic theory. However, this may not be the case if there are significant differences between skilled labour categories as presented in the extension model in Chapter 3.
- Financial development: in theory greater access to financing services are expected to affect positively the income of the poor population. In turn, this could lead to a reduction of income inequality due to a loose in the budgetary constraint of these individuals.
- Infrastructure: In economic theory, access to better infrastructure is expected to generate a higher connectivity between rural and urban areas within a country. Likewise, improvements in infrastructure related to sanitary, telecommunication and electricity are likely to positively affect in a greater proportion the poor population than the more wealthy groups in a country there leading to a reduction income inequality.
24 - Governance: a higher level of governance is related with a most efficient and productive use of public resources, as well a fairer social treatment within a country. With an improvement in governance, income inequality is expected to decrease since the poor segments of society are able to demand better conditions and enjoy better public policies and services provision.
- Total tax revenue: can be positively correlated with lower inequality if the taxation system is progressive. Likewise, higher social transfers are also expected to reduce income inequality.
-Exchange rate: a higher exchange rate can be expected to be correlated with lower income inequality given the shift to an unskilled labour intensive tradable sector following a strong depreciation of the currency.
Using OLS, the coefficients presented in equation (1.1) could be biased due to an endogeneity problem that arises from using panel data models that considers lagged variables and country fixed effects. This problem is visible when considering equation (2). This equation refers to the error term in equation (1.1) before accounting for country-fixed effects and is composed of non-observable country specific effects 𝜇 and an independent shock, 𝜀 . From this equation, it is clear to see that the lagged value of the Gini coefficient 𝐺𝑖𝑡−1 is correlated with the error term through country fixed effects.
𝑢𝑖𝑡 = 𝜇𝑖 + 𝜀𝑖𝑡 (2)
For this reason, a GMM methodology is used since it helps to address this endogeneity problem and delivers coefficients that satisfy the consistency condition of OLS estimators. It does so by acknowledging the existence of country-specific fixed effects and by instrumenting with the second lagged value of the Gini coefficient which is not correlated with the error term in equation (1), 𝑢𝑖𝑡, under the assumption of no serial autocorrelation of the error terms. This correction is already accounted for in equation (1.1), (1.2) and (1.3) when applying the GMM methodology.
In terms of data, the present study will employ panel data ranging from 1985 until 2014. The sample of countries9 is the same used by Tsounta & Anayochukwu (2014), excluding the
non-Latin-American economies since these are out of the focus of the present study10. Likewise, the explanatory variables considered in this thesis are based on the ones analyzed by Perry & Olarreaga (2006) complemented with the level of total tax revenue by the
9
Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatamala, Honduras, Mexico, Nicaragua, Paraguay, Panama, Peru and Uruguay.
10
The countries and variables considered in the present study are further described in the Data Description section
25 government used by Tsounta & Anayochukwu (2014) for controlling for the existence of redistributive policies.
In the methodology employed by López (2003) and Perry & Olarreaga (2006), the sample is reduced by using a non-overlapping five year period averages. However, due to the fact that the present study only considers the period ranging from 1980 till 2014 this will not be replicated and annual data will be used. Additionally, it is of great interest in this study the general trend of income inequality and GDP, and by employing five-year averages these trends would be eliminated which is not desirable.
26 6. Descriptive data Analysis
For the present analysis a dynamic panel dataset is used. This database contains information for seventeen Latin-American countries for a period ranging from 1985 till 2014 for 18 different macroeconomic variables. The countries in the sample are Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru and Uruguay. These 17 countries are divided into two different groups based on the following criteria: Group 1 refers to those whose its Gini coefficient value in 2014 is higher than its 1985 level, while Group 2 refers to those whose its 2014 Gini value is below its 1985 level. In sum, Group 1 is composed by those countries in the sample that have shown an upward-sloping trend in its Gini between 1985 and 2014, whereas Group 2 refers to those with a negative trend for the same period. For both groups, the evolution of all variables will be described in this chapter. However, a common average for the totality of the countries will be presented for those variables for which the two groups do not present a visible difference in trend.
6.1 Dependent variable
Gini coefficient: Following the methodology of Tsounta & Anayochukwu (2014) the dependent variable of the present analysis is the logarithm of the Gini coefficient for guaranteeing a normal distribution. Information regarding this variable was extracted for the World Bank Povcal database. The behaviour of the Gini coefficient for the 17 countries considered in this study for the period 1985-2014 is presented in Graph 8. As shown, the two sets of countries experienced a significant increase in its Gini coefficient in the nineties decade, followed by a fall in this indicator since 2000 till 2014. The decrease in income inequality was more pronounced for countries in Group 2. This behaviour more clearly seen in the general trend line.
27 Graph 8– Gini coefficient by groups of countries, 1985-2014
Source: World Bank Povcal database – Author’s calculations
Note: Sample of countries conforming each group: Group 1 (Colombia, Costa Rica, Dominican Republic, Panama and Paraguay), Group 2 (Argentina, Bolivia, Brazil, Chile, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Peru and Uruguay).
6.2 Explanatory variables
Trade openness: Following the analysis of Perry & Olarreaga (2006) and López (2003), this variable refers to the sum of the total exports and imports as a percentage of GDP. Information for this indicator was extracted from the regular World Bank online database. Graph 9 presents the changes in trade openness between 1985 and 2014 for the two groups of countries considered in the present study. As shown, countries in Group 1 experienced a general decrease in trade openness of approximately 20 points, while Group 2 shows a clear increase in this variable. The trends of these two groups suggest a negative correlation with income inequality since a steep increase in trade openness coincided with a decrease in the Gini in the case of Group 2, while the opposite applies for Group 1. 40 42 44 46 48 50 52 54 56 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
28 Graph 9 – Trade openness by groups of countries, 1985-2014
Source: World Bank – Author’s calculations
Primary trade: refers to the total sum of exports and imports of primary goods as a percentage of total trade. Information for this variable was obtained from the regular World Bank online database. As Graph 10 shows, there was a common movement between countries in the region: a decrease in primary trade openness between 1985 and 2014. This behaviour suggest a positive correlation between primary trade and income inequality since in this same period the latter variable fell too.
Graph 10 – Primary trade by groups of countries, 1985-2014
Source: World Bank – Author calculations
Food, fuel, mining and raw agricultural material trade: Following the analysis of Perry & Olarreaga (2006) primary trade is further divided into its main four components: fuel, food, mining and raw agricultural materials. These variables are measured as the participation of imports and exports of each component in total trade. As Graph 11 shows, there was a general decrease in the region in the level of food and of raw agricultural materials trade.
30 40 50 60 70 80 90 100 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 (X =M )/GD P (% )
Group 1 Group 2 General average
25 30 35 40 45 50 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Pri mary tra de/T ot al tra de (% ) General average
29 This movement could suggest a positive correlation of these two variables and trade openness due to the general down sloping trend in the inequality. On the contrary, while for mining and fuel there is no significant movement from its 1985 value in 2014, both variables experienced a sharp fall between 1985 and 1999, followed by a steep increase since 2000. This movement coincides with the general increase in income inequality followed by its reduction in the 1990s and the 2000s respectively. This opposite behaviour suggests a negative correlation this variable and mining and fuel.
Graph 11 – Components of primary trade: food, fuel, mining and raw agricultural materials trade intensity, 1985-2014
Source: World Bank – Author’s calculations
Secondary trade: This variable refers to the participation of manufacture as a percentage of total trade. Information on this indicator was extracted from the World Bank online data base. As Graph 12 shows, there was general increase in secondary trade throughout 1985 and 2014 in all the countries while there was a downward sloping trend in income inequality This
15 16 17 18 19 20 21 22 23 24 25 Foo d t ra de/To tal trade (% )
Food Trade Intensity
General average 4 6 8 10 12 14 16 Fuel trade/T o tal trade (% )
Fuel Trade Intensity
General average 3 4 4 5 5 6 6 7 M ining t ra de/To tal trade (% )
Mining Trade Intensity
General average 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 Ra w a g riculture trade/T o tal trade (% )
Raw Agricultural Materials Trade Intensity
30 movement suggests the possibility of a negative correlation between secondary trade intensity and income inequality.
Graph 12 – Secondary trade by groups of countries, 1985-2014
Source: World Bank – Author calculations
Tertiary trade: This indicator is measured as the percentage of services trade from total trade and was calculated using information from the World Bank online database. In general, there is a very subtle downward sloping trend as Graph 13 shows. This common behaviour with respect to the Gini suggests the possibility of a positive correlation between tertiary trade and income inequality. Nevertheless, it is worth nothing that the fall in tertiary trade is very low.
Graph 13 – Tertiary trade by groups of countries, 1985-2014
Source: World Bank – Author calculations
Financial depth: Refers to the amount of private credit provided by deposit money banks as a percentage of GDP, as used by López (2003). Information for this variable was extracted
25 30 35 40 45 50 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Second ary tra de/T ot al tra de (% ) General average 14 16 18 20 22 24 26 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Te rtiary tra de/T ot al tra de (% ) General average
31 from the IMF’s Global Financial Development Database. As presented in Graph 14, the countries experienced in general a steep increase in financial deepening between 1990 and 2014. This significant increase could be negatively correlated with the general fall in income inequality in this same period.
Graph 14 – Financial depth by group of countries, 1985-2014
Source: IMF’s Global Financial Development Database
Infrastructure: Following the analysis of López (2003), this variable is measured as the number of fixed telephone subscriptions per 100 people. As Graph 15 shows, there has been a common increase in the amount of public infrastructure in the two groups of countries without any visible fluctuation. This movement suggests a negative correlation between infrastructure and income inequality.
Graph 15 – Public infrastructure, 1985-2014
Source: World Bank 15 20 25 30 35 40 45 50 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Pri va te credi t by domestic ba nks (% GDP) General average 2 4 6 8 10 12 14 16 18 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Fixed telephon e li nes per 10 0 ha bita nt s General average
32 Real exchange rate: This indicator refers to the value of a currency against a weighted average of several foreign currencies divided by a price deflator, as defined by the IMF. In this case 2010 value constitutes the base (2010=100). As Graph 16 shows, there was a general appreciation in the level of the exchange rate in the countries, while there was a common reduction in income inequality throughout 1985 and 2014. This joint movement suggests the possibility of a positive correlation between the level of the exchange rate and income inequality.
Graph 16 – Real effective exchange rate (2010=100), 1958-2014
Source: IMF’s Global Financial Development Database
Rule of Law: This indicator assess how laws are enacted and administered, while also considering how efficient and fair is their enforcement. This variable ranges from -2.5 for the worst and till 2.5 for the best ranked. Information for this indicator was extracted from the World Justice Project official website. As Graph 17 shows, the countries in the sample shared a common positive trend throughout the period considered. This movement suggests the possibility of a negative correlation between the level of governance and income inequality.
80 85 90 95 100 105 110 115 120 125 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Rea l Effectiv e Excha ng e Ra te General average
33 Graph 17 – Rule of Law Indicator by group of countries, 1985-2014
Source: World Justice Project
Price Stability: This variable is defined as the inflation rate plus 100 and is based on the analysis of López (2003)and refers to how intense and volatile inflation is in a given territory at a given time. As Graph 18 shows, countries in Latin-America presented on average a decrease on inflation. This common movement suggests the possibility of a positive correlation between these variables which supports the idea that high volatility leads to more inequality.
Graph 18– Inflation CPI by group of countries, 1958-2014
Source: World Bank
Terms of Trade: This variable refers to the terms of Net Barter terms of trade index and is calculated as the percentage ratio of the export unit value indexes to the import unit value indexes, measured relative to the base year 2000 as explained by the World Bank. As Graph 19 shows, the set of countries do not present a general trend in its terms of trade throughout
-0,9 -0,8 -0,7 -0,6 -0,5 -0,4 -0,3 -0,2 -0,1 0,0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 General average 0 200 400 600 800 1000 1200 1400 1600 1800 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 General average
34 the period considered. However, between 1985 and 2001 there was a deterioration in the terms of trade which coincide with a general rise Gini index. Likewise, after 2000 there has been an improvement in the terms of trade, especially in countries of Group 2 which also experienced a decrease in their inequality levels. This joint movement suggests the possibility of a negative correlation between these two variables.
Graph 19 – Terms of Trade by group of countries 1985-2014
Source: World Bank
Secondary Education Enrolment: is measured as the percentage of gross secondary enrolment of both sexes. The data for this variable was extracted from the World Bank databank. As Graph 20 presents, the general average for this variable presents a common trend throughout the period considered with no significant fluctuations. This behaviour suggests a negative correlation between secondary enrolment and income inequality since the Gini index presented a general decrease in this same period.
80 85 90 95 100 105 110 115 120 125 130 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 General Average
