THE DICHOTOMY OF DEVELOPMENT IN BRAZIL A TEST OF KUZNETS U CURVE HYPOTHESIS

Indian Journal of Economics & Business, Vol. 11, No. 2, (2012) : 485-501 THE DICHOTOMY OF DEVELOPMENT IN BRAZIL A TEST OF KUZNETS U CURVE HYPOTHESIS J. MICHAEL DEINLEIN * Abstract In 1955, Simon Kuznets recognized the inherent dichotomy of development and the parallel process of economic growth and growing inequality at the outset. Kuznets Inverted U-Hypothesis is a depiction of the differential effects that development has on different groups with a country and furthermore how to analyze inequality in the development process. This paper will attempt to elucidate Kuznets principal hypothesis and apply Kuznets findings to the case of Brazil. Particular interests will be paid to structural transformation within a society and how it affects level of inequality, as well as educational and demographic indicators. Lastly, an analysis of the effects liberalization and factor price equalization have on income inequality in developing countries will be put forward and how this affects inequality in Brazil Keywords: Kuznets Curve, Gini Coefficient, Brazil, Factor Price Equalization JEL Code:D63, I24 I. INTRODUCTION Development, at its core, is an uneven process. While poverty can never be alleviated without economic progress, one must keep in mind that development remains only part of the solution for poverty eradication. What development fails to acknowledge is the disparities within the distribution of incomes that accompany economic development and the erosive effect this has on poverty alleviation in the developing world (Kharsru and Jalil, 2002). Simon Kuznets recognized this dichotomy within economic development, and in 1955 developed his Inverted U-Hypothesis of income inequality. He proposed that as a country developed, income disparities would first worsen, even in the presence of economic growth, before theyimproved. His hypothesis provided a reconsideration of development and the differential effects it had on different sectors of society. Kuznets recognized the uneven process of development, and his insights allow us to reconsider the factors that ameliorate income inequality. This dual nature of economic development is none more apparent than in the case of Brazil. Brazil presents an interesting case of study due to its economic * MA Candidate, Josef Korbel School of International Studies, E-mail: mikedeinlein12@gmail.com

486 J. Michael Deinlein preeminence in the world paradoxically paralleled by its habitually high levels of income inequality. Brazil today touts the seventh largest economy, and even still only some very poor African countries like Lesotho, Swaziland, Sierra Leone, and Namibia maintain worse income inequalities. The countries richest one percent less than two million people have 13 per cent of all household income, the same figure that the poorest 50% have (Beghin, 2005). With such drastic income inequality, Brazil still continues to struggle with the pestilence of poverty. 54 million people in Brazil are considered poor, and within this group 20 million are ranked as extremely poor according to the Institute of Applied Economic Research. Therefore, Brazil presents a perfect case for testing Kuznets Inverted U Hypothesis and the effect development has on inequality. Roth (2010)does a similar analysis of the Kuznets Inverted U Hypothesis for Brazil, however, Roth pays particular attention to the historical development of Brazil and fails to give proper attention to the underlying demographic and educational trends that occur over the course of Brazil s economic development, which could have significant distributional effects as argued later in this paper. This paper will analyze the robustness of Kuznets Inverted U Hypothesis when applied to the case of Brazil by measuring longitudinal changes in the country s Gini coefficient over the last thirty years of Brazil s economic development. In order to understand the trends inherent in the data, particular consideration will be paid to the effect of structural change on income inequality, as purported by Kuznets in 1955. According to the theory, income inequality will continue to increase during the early years of development due to fundamental differences in the marginal productivity in the two sectors, the traditional and industrial. After proper consideration of structural change, consideration will be given to the effect educational and demographic transitions have on inequality measures in Brazil. Such analysis stems from the research of M.S. Ahluwalia on Kuznets Inverted U- Hypothesis, who found that both these factors have significant positive effects on diminishing income inequality within a country. Finally this paper will seek to demonstrate the effects liberalization has on income inequality in Brazil through an analysis of Paul Samuelson s Factor Price Equalization Theory. This theory suggests that as an economy opens up to trade, and specializes in the good that intensely uses its abundant factor, there is an equalization of factor prices between the two countries. Due to this effect, the abundant factor will ultimately benefit from trade. If this proves true in the case of Brazil, this could have profound effects on the income distribution between the skilled and unskilled labor forces within the country. II. THEORETICAL FRAMEWORK Measuring Inequality There are numerous measures of income inequality used in the world today. Some measures of income inequality simply look at the discrepancy between the lowest

The Dichotomy of Development in Brazil 487 income quintile and the highest income quintile within a society, however such a measure neglects differences in the middle of the wealth distribution. In his seminal work, Economic Growth and Income Inequality, Kuznets utilized the Gini Coefficient to measure income inequality longitudinally across the three different countries and therefore for our purpose this study will do likewise. Simply put, the Gini Coefficient is the quantification of the Lorenz Curve, to facilitate comparisons across time and countries. The Lorenz Curve is a graphical representation of the income inequality within a country, where cumulative percentage of income is plotted on the vertical axis and the cumulative percentage of the population is plotted on the horizontal axis (Figure 1). The 45-degree angle line that bisects the graph is a representation of the theoretical perfect distribution of wealth. The second curve, or the Lorenz Curve, is the actual distribution of wealth in that country; the greater the separation of the Lorenz Curve and the perfect distribution diagonal, the greater the income inequality in the country. Figure 1: Lorenz Curve and Gini Coefficient However, the Lorenz Curve still remains simply a graphical representation of income inequality. In order to quantify inequality, CorradoGini developed the Gini Coefficient. The coefficient is defined as the ratio of the area that lies between the line of equality and the Lorenz Curve (Area A), over the total area under the line of equality (Area A + Area B). Gini = A / (A+B) x 100 1

488 J. Michael Deinlein A higher Gini Coefficient indicates higher income inequality in that country. Usually the Gini Coefficient is within the range of 0.3-0.7, with anything above 0.6 representing severely unequal distribution of income a value Brazil has hovered around for decades. The following section will examine the longitudinal effects on income inequality (measured by the Gini Coefficient) as it relates to economic development. Special emphasis will be given to Simon Kuznets Inverted U- Hypothesis, and the effects of liberalization on economic development with special attention given to the Heckscher-Ohlin Model and Paul Samuelson s theory of Factor Price Equalization. Kuznets and Economic Inequality In 1955, Simon Kuznets explicated his Inverted U-Hypothesis that supposed that the income distribution would tend to worsen in the preliminary stages of economic growth only to be ameliorated in later stages. For Kuznets, growth in income per capita signaled growing inequality as this wealth was consolidated by the leading sectors of an economy in the principal stages of economic development, but as other sectors in an economy caught up and a multitude of other factors acted on the society, the trend of growing inequality would eventually reverse and begin to decline (See Figure 2). Figure 2: Inverted U-Curve For Kuznets, two factors are integral for increasing inequality in the distribution of income: the concentration of savings in the upper income bracket, and the process of structural change within a country. Kuznets notes, According to all recent studies of apportionment of income between consumption and savings, only the upperincome group saves, and what is perhaps more important is that this inequality

The Dichotomy of Development in Brazil 489 in the distribution of savings is greater than in the distribution of property incomes (Kuznets, 1955). Such disparity in the savings rate of the two groups suggests that as the upper income bracket continues to save the proportion of income yielding assets in the hands of the upper groupings further increases, thus providing a basis for larger income shares for these groups in the future (Kuznets, 1955). Secondly and perhaps more pivotal in the spread of inequality is the structural change that accompanies development. Kuznets analysis of this transition is analogous to the structural transformation purported by Arthur Lewis. For Lewis, differences appear as traditional versus modern sectors (Syrquin, 1988). The traditional, or agricultural, sector is characterized by an excess supply of labor, necessitating a marginal product of labor close, if not equal, to zero. Because marginal product is miniscule, the wage rate is set at the subsistence level. The industrial sector, on the other hand, is designated by a positive marginal product of labor, which precipitates wage rates above the subsistence level. This difference in the wage structure of the two sectors engenders migration to the industrial centers as people attempt to capture the higher price for their labor. Due to rapidly increasing per capita productivity in urban pursuits, inequality in the total income distribution will continue to increase (Kuznet, 1955). These two factors explain the divergence in economic wellbeing of the two sectors but what then accounts for the downward movement of the Kuznets Curve? The first explanation for the second half of the curve utilizes the same logic of structural change. As more and more people migrate to industrial centers, the marginal product of these urban pursuits will slowly decline, and the marginal product of agricultural pursuits will gradually increase, precipitating a convergence in wage rates between the two sectors, thus lowering inequality. Theoretically, migration will stop when the marginal product of labor curves for the two sectors cross, signifying comparable wage rates. However, due to differences in technology, the industrial sector will typically maintain a higher marginal product and there will always be divergence in wage rates between the two sectors, but it is still important to note that convergence does occur in the process of structural transformation. Furthermore, Kuznets maintains the successful great entrepreneurs of today are rarely the sons of the great and successful entrepreneurs of yesterday (Kuznets, 1955). In a dynamic economy technological change is rampant and property assets in an older industry have a decreasing proportional weight as younger and more rapidly growing industries enter the economy. Unless owners of older industries can transfer their assets into these newer pursuits, new entrepreneurs will accrue more and more assets and close the inequality gap (Kuznets, 1955). Building on the findings of Kuznets, Montek S. Ahluwalia conducted a crosssectional data sample of 60 countries to determine whether or not there was evidence supporting Kuznets Inverted U Hypothesis. While Ahluwalia did find evidence in support of Kuznets hypothesis, the additions offered in his study are perhaps more interesting for our analysis of Brazil. Ahluwalia found that there are three aspects

490 J. Michael Deinlein of development that appear to be systematically related to the degree of inequality in a country: intersectoral shifts between the traditional and modern sectors (as expounded by Kuznets), expansion in the educational and skills characteristics of the population, and the demographic transition involving the reduction in the rate of growth of the population (Ahluwalia, 1976). These last two factors are integral in analyzing the trends of inequality in Brazil. Factor Price Equalization and Inequality While not addressed by Kuznets, a third factor could have profound effects on inequality in Brazil, factor price equalization. In 1948, Paul Samuelson explicated his theory of factor price equalization in his piece International Trade and the Equalization of Factor Prices for The Economic Journal. The theory expands on the findings of the Heckscher-Ohlin Model, which maintains that trade will be mutually beneficial if two countries specialize in the production of the good that uses its abundant factor more intensively. What Samuelson s theory of Factor Price Equalization adds, is what happens after two countries engage in international trade. If for example country A is labor abundant, the wage rate to rental rate for capital will be less in country A than in its trading counter part, country B ((w/r) A < (w/r) B ). Under the assumptions of full employment of resources, as country A specializes in the production of the labor intensive good (good x), it will slow down the production of the capital intensive good (good y) to free up more labor and capital for the production of good x. However, lower production of good Y will release capital faster than labor (because Y is capital intensive). Consequently the wage rate in country A will go up (and the rental rate will go down) because the demand for labor now outstrips the capacity of country A to supply it. The opposite is of course true in country B. These changes in the prices of labor and capital in the two countries cause a convergence of the wage to capital ratio of the two. Accordingly the optimal level of trade will be when these factor prices become equal or when (w/ r) A = (w/r) B. Under this logic, trade will ultimately benefit the abundant factor in a country. If country A is labor abundant, the wage rate increases due to factor price equalization, and thus the abundant factor is now relatively better off than it was before the introduction of international trade (Samuelson, 1948). What does this then mean for the case of Brazil? At the end of the 1980s, a consensus was developing that the protectionist policies currently in place were adversely affecting the competiveness of Brazilian exports. The government in Brazil began to commit to a policy of liberalization and protectionist easing. In 1990 a broad trade policy reform was introduced, and in 1991 Brazil signed the Treaty of Asuncion, which created the Southern Common Market with Argentina, Paraguay, and Uruguay. From 1991-1994, intraregional trade was subject to drastic decreases in tariff rates, and a common external tariff (CET) was introduced (Pereira, 2004). Due to their recent commitment to international integration, Brazil should be a rather decent test case for factor price equalization and the effects of economic liberalization.

The Dichotomy of Development in Brazil 491 So how would factor price equalization operate in Brazil?In the country, unskilled labor is the abundant factor, but more importantly for our analysisunskilled labor is a significantly marginalized sector of society where much of the inequality is concentrated. If factor price equalization holds true for the Brazilian case, the wage rate of unskilled labor necessarily has to increase relative to wage rate of skilled labor. Employment would have to shift from skilled to unskilled and factor price equalization would result in an overall positive effect on unskilled labor in Brazil (Gonzales, Filho, & Terra, 2005). This would suggest that a third dynamic operating on inequality in Brazil is the increasing push for liberalization and this factor price equalization. III. BACKGROUND OF BRAZIL Brazil s Economic Development The Brazilian economy has undertaken multiple development strategies since the Post-War Period. From 1949 to 1962 Brazil implemented a growth strategy of policy induced import substitution. Favored industries received exemptions from duties on capital imports and intermediate goods and special lines of credit to bolster their positions in the market (Morley & Smith, 1971). The high protection of domestic industries from import competition led to astounding growth rates in Brazil. Whether or not the counterfactual would have proved otherwise is impossible to determine, but the impressive growth of Brazil from 1900-1973 is indubitably remarkable. Over that time period Brazil s growth performance was only bettered by Japan and Finland (Abreu, Bevilaqua&Pinho, 1997). However, the protectionist policies instituted, also contributed to growing inequalities between the leading and lagging sectors of society. With the onset of the oil shock in November of 1973, Brazil was forced to either implement austerity measures that would ameliorate the deterioration of Brazil s terms of trade, or to continue on its trajectory of growth that would necessitate increased borrowing from abroad and the accumulation of sizeable international debts. Tragically, Brazil chose to sustain its high growth rates under the assumption that future savings of foreign exchange, precipitated by the import substitution policies and development of new export capacity would allow Brazil to produce trade surpluses large enough to service its international debt in the future. However, drastic increases in the current account deficit and substantially higher interest rates, made it almost impossible for Brazil to service its debts, and external debt continued to spiral out of control (Baer, 1995). The past policies of import substitution, subsidies to capital, directed credit, and excessive borrowing to cover fiscal deficits resulted in the debt crisis of the 1980s and 1990s. Productivity in the 1980s and 1990s declined precipitously as macroeconomic instability and high inflation disrupted the production processes throughout the country (Thomas, 2006). The government utilized the printing presses to service their outstanding debt and as a result, inflation became

492 J. Michael Deinlein unsustainable reaching levels over 2000%. The hyperinflation throughout Brazil had substantial negative effects on medium term and long-term growth through the distortion of price signals and consequently the incapacity to efficiently allocate resources throughout the economy (Kahn &Ssnhadji, 2001). The agricultural sector suffered especially due to the price instability of commodities and as a result there was greater income disparities between the rural and urban sectors. The debt crisis of the 1980s and 1990s neutralized and even reversed decades of progress for Brazil (Thomas, 2006). The past 5 decades have been a roller coaster of economic development for Brazil (Table 1). The boom years of the 1960s and 1970s did as much for placing Brazil on a trajectory of economic development as the 1980s and 1990s did for derailing it. Today Brazil is starting to once again show promising signs for sustained development and growth, but under these aggregate economic indicators lays an unnerving truth. Brazil supports the seventh largest economy in the world and the fifth largest population. Gross national income per capita is $9,390, firmly indicating that Brazil is a middle-income country (World Bank, 2010). However, despite the admirable economic performance of Brazil, income inequality continually plagues this South American giant. The causes of inequality in Brazil and the obstacles for its eradication are evident, however policy proscriptions have been slow to materialize in the country. The historically slow response to income inequality in Brazil presents significant difficulties for analyzing Kuznets Inverted U Hypothesis in the country. However, recent policy improvements in combatting inequality may demonstrate the growing pertinence of the hypothesis in the future. The next section will be devoted to explaining the causes of inequality in Brazil and the policies implemented to ameliorate the economic imbalances in the country. Table 1 Growth Rate of Real GDP Decade 1960s 1970s 1980s 1990s 2000s Growth Rates of Real GDP 5.9% 8.5% 3.0% 1.8% 4.5% Inequality in Brazil To begin, the cultural dimensions of inequality in Brazil are rather interesting to investigate. The normative definitions and evaluative perceptions of inequality across Brazilian culture are vast. For many of the elites in society, poverty and inequality are analogous. There exists a conflation of the two, where the elites do not see inequality as an unwillingness to redistribute the wealth Brazil has but rather the incapacity to exploit the available resources and potential wealth of the country (Therborn, 2006). Brazil is a middle-income country with the potential to ameliorate, at least in part, the pestilence of inequality, however the manner in which it is viewed within Brazilian society has precipitated the ever widening gaps between the fortunate and the marginalized.

The Dichotomy of Development in Brazil 493 In a study in 2004, the World Bank outlined four principal agents for inequality in Brazil. 1. The underlying distributions of assets across the population might be more unequal than in other countries. Important assets are educational attainment, land, and capital. 2. Price differentials of these assets notably education might be steeper in Brazil than elsewhere. If the wage differences for each extra year of schooling in Brazil are much higher than in comparable countries, then more income inequality would be generated from the same underlying distribution of education. 3. It could be that Brazil s excess inequality arises neither from unequal distributions of assets nor excessive wage differentials by skill, but from behavioral differences or differential patterns of use of these assets. Labor force participation, occupational choice, and fertility decisions could account for very substantial differences in the distribution of household per capita incomes. 4. The distribution of claims and entitlements to state transfers might be less progressive than in other countries. Particular interest should be paid to retirement pension programs (World Bank, 2004). These causes of income inequality in Brazil are compatible with the analysis of Kuznets, Williamson & Higgins, and Ahluwalia. Educational opportunities are much more concentrated in the urban sectors of the country and this unequal distribution of assets across the population creates a dichotomy between the poor traditional sector and the urbanized industrial sector. In Brazil, a worker who has completed secondary education earns, on average, 3.7 times as much as a worker who has completed one to four years of schooling (Thomas, 2006). Therefore it is clear to see how the price differentials of these assets plays a further role at increasing the gap between the two sectors. Lastly, the distribution of claims and entitlements across the population is far less progressive. Opposed to using public spending as a means for redistribution, Brazil s fiscal policies have perpetuated the unequal economic separation. While Brazil spent 20% of its GDP on social programs, the poorest 20 per cent received only 1.7 per cent of monetary social transfers in 2000 (Thomas, 2006 and Roth, 2010). The causes of economic inequality in Brazil are evident. However an analysis of the Kuznets curve for Brazil and the relevant literature on his findings along with an evaluation of new developments within the country demonstrate that Brazil is certainly working towards alleviating inequality. IV. EMPIRICAL EVIDENCE Brazil and Kuznets Inverted U-Hypothesis As Brazil has progressed on the trajectory of development since the post war period of the fifties and sixties, it has demonstrated a surprising capacity for maintaining

494 J. Michael Deinlein high rates of inequality. By 1960 the Gini coefficient was 50, significantly lower than its 1989 high of 63.0 but still astonishingly high even for a developing country. Growing income disparities throughout the country and an ever-rising Gini coefficient marked each decade up until the 1990s. Figure 3 illustrates the rise of the Gini coefficient from 1981 to its high in 1989, but more importantly it illustrates the downward trend that has occurred ever since. While one cannot definitively say the data from Brazil confirms there is an inverted U relationship between development and inequality the change in the coefficient from 1981 to 2009 is only 3.6 the growing inequality since the 1960s and the downward trend since 1989 certainly suggests that there is at least some support for Kuznets hypothesis, even if the relationship remains weak. From 1960 to 1989 the coefficient increased from 50.0 to 63.0, a considerable increase, and from 1989 to 2009 the coefficient decreased from 63.0 to 53.9. This data suggests that there is at least some relationship between the growing prosperity of the country and the level of inequality based on its growth trajectory. However, it is not clear from the data presented below what factors coincided with the U shaped arc of inequality in Brazil. The following sections will provide some insight. Figure 3: Gini Coefficient Distribution of Household Per Capita Income in Brazil 1981-2009 Structural Change Kuznets maintained that a major reason for growing income inequality was the structural difference between the traditional and modern sectors of society. Due to the divergence in the two sectors marginal products of labor, wage disparities arose. The excess supply of labor in the traditional sector resulted in wage rates at the subsistence level due to a marginal product of labor close, if not equal, to zero. However, a positive marginal product characterized the urban sector and therefore the wage rate was above that of the subsistence level. This incentivized migration to the urban centers as workers sought to capture the increased rents for their

The Dichotomy of Development in Brazil 495 labor. Due to rapidly increasing per capita productivity in urban pursuits, inequality in the total income distribution will continue to increase (Kuznet, 1955). However, this process is mitigated by yet a second force. As more and more people leave the traditional sector and migrate to the modern sector, the marginal product of labor begins to rise in agriculture, and decrease in industry. This could create a convergence in wage rates between the sectors and a decrease in inequality. Table 2 demonstrates the patterns of structural change in Brazil over the period between 1981 and 2010. The effects of structural transformation are clearly present in the data. In 1981 the rural population comprised 32% of the total population. Every subsequent year either showed a decrease in the rural population or no change. In 1990 the rural population had decreased to 25%. By 2000 this figure was down to 19%. And in 2010 the rural population had decreased to just 14% of the population. This data on the migration patterns between the rural and urban sectors certainly demonstrate structural change within the country and certainly lend support to Kuznets Inverted U Hypothesis (World Bank, 2011). Table 2 Structural Change in Brazil from 1981 to 2010 Year Rural Population % Year Rural Population % 1981 32 1996 22 1982 31 1997 21 1983 30 1998 20 1984 29 1999 19 1985 29 2000 19 1986 28 2001 18 1987 27 2002 18 1988 27 2003 17 1989 26 2004 16 1990 25 2005 16 1991 25 2006 15 1992 24 2007 15 1993 23 2008 14 1994 23 2009 14 1995 22 2010 14 However, how does this relate to the patterns of income inequality within the country? As Kuznets suggested the rising income inequality in the 1980s could be attributed to the growing concentration of people in urban centers and the increased wages that accompanied this migration. The structural change resulted in income inequality being concentrated in the traditional sector.while there is no definitive transition between growing inequality and decreasing inequality that can be deduced from the data, patterns do arise that suggest a convergence in wages between the

496 J. Michael Deinlein two sectors and a slowing of migration. Over each decade the decrease in the rural population was less than the decade that preceded it. For 1981-1990 the decrease was 7%. From 1990 to 2000 the decrease was 6%, and from 2000 to 2010 the decrease was only 5% (World Bank, 2010). While these are not substantial disproportions, they still show a decline in the rate of structural change. This could suggest that over this period structural transformation declined due to rising marginal product in the traditional sector as more and more people leave, and decreasing marginal product in the modern sector as more and more people settle in the urban centers. This again causes a convergence in the wage rates of the two sectors and consequently the incentive for migration becomes less and less, as demonstrated by the data in Table 2. This convergence in wage rates manifests itself in the downward trend of inequality since 1990. Education Our second consideration comes from the work of M.S. Ahluwalia. Building off Kuznets Inverted U Hypothesis, Ahluwalia found that there were three distinct aspects of the development process that appeared to be systematically related to the degree of inequality in a country. Among these was the expansion of the educational and skills characteristics of the population (Ahluwalia, 1976). In his analysis, he uses the literacy rate and the secondary enrollment ratio as crude approximations of the level of human development within a country. However, his operationalization of human development is apt for two reasons. The literacy rate acts as a basic measure of the educational level of the stock of the population, while secondary enrollment measures the degree of educational achievement beyond this base level (Ahluwalia, 1976). Ahluwalia finds that there is clear evidence that education is significantly and positively correlated with equality, and furthermore the positive impact of education on relative equality is quantitatively robust. Following from these findings, Table 3 and 4 demonstrate the current trends in adult literacy and secondary school enrollment over the last thirty years. The effects of the literacy rate on economic equality are difficult to measure for Brazil due to the lack of data between 1980 and 2000. However, one should recognize that there has been a significant increase in the literacy rate from 1980 to 2006. The 15% increase over this time is substantial and should be taken into account. However, from this data one cannot definitively say that the increase in the literacy rate led to a decrease in income inequality. The data from the 2000 suggests that there may be some correlation, as the 4% increase in the literacy rate corresponded with a decrease from 58.6 to 55.9 in the Gini coefficient, but the correlation still remains relatively weak. Table 3 Literacy Rate in Brazil for Selected Years Year 1980 2000 2004 2006 Literacy Rate (% of Adult Population) 75 86 89 90

The Dichotomy of Development in Brazil 497 Table 4 Secondary School Enrollment Rates in Brazil - Selected Years from 1999 to 2008 Year 1999 2000 2001 2002 2003 2004 2005 2008 School Enrollment 66 69 71 75 76 78 79 82 Secondary (% Net) The increase in the secondary enrollment rate is rather more substantial. In his analysis, Ahluwalia noted that an increase in the secondary enrollment rate was associated with shifts in income from the top 20 per cent to all other groups (Ahluwalia, 1976). While again, our data is limited to 1999 and beyond, the correlation between secondary enrollment rate and decreased inequality is far more robust than the correlation with the literacy rate. From 1999 to 2008 the secondary enrollment rate increased by 16 per cent. Over this same time period the Gini coefficient decreased from 58.6 to 53.9. This data is compatible with Ahluwalia s findings that secondary enrollment is a significant redistributive agent in developing countries. Demographic Transition The third consideration again comes from the work of Ahluwalia. Along with structural change and the expansion of educational and skills characteristics, Ahluwalia found that a depressed rate of population growth was associated with decreasing inequality within a country. Ahluwalia noted that high growth rates have a significant positive impact on the income share of the top 20% and a negative impact on the income shares of all other groups. A possible explanation for this association between greater inequality and high growth rates is the fact that different income groups grow at different rates. Ahluwalia argues that high growth rates in the total population reflect greater differentials in population growth across groups, with lower income groups growing at a significantly faster rate, which in turn generates greater income inequality (Ahluwalia, 1976). If a country experiences declining growth rates, it typically will be concentrated in the lower income brackets. This in turn decreases the dependency ratio within these groups, which in turn frees up more resources that no longer have to be used on the purchase of necessities (food, clothing, shelter, etc.). Table 5 demonstrates the demographic transition that has occurred in Brazil. In Brazil, the decreases in the population growth rate could have significant effects on income inequality. Over the past thirty years, Brazil demonstrated declining growth rates,and as Ahluwalia noted, this could have significant redistributive impact. In 1989 the growth rate was 1.8% while the Gini index was at an all time high of 63.0. In 2009 the growth rate was half of what it was in 1989 and the Gini index had fallen to a thirty year low of 53.9. These results suggest that the decreased growth rates are in fact capturing the differentials in population growth rates between groups, and the sectors most affected are those with lower

498 J. Michael Deinlein Table 5 Demographic Transition in Brazil from 1981 to 2010 Year Population Growth Year Population Growth (Annual %) (Annual %) 1981 2.3 1996 1.5 1982 2.3 1997 1.5 1983 2.3 1998 1.5 1984 2.2 1999 1.5 1985 2.1 2000 1.4 1986 2.0 2001 1.4 1987 2.0 2002 1.4 1988 1.9 2003 1.3 1989 1.8 2004 1.2 1990 1.7 2005 1.1 1991 1.7 2006 1.1 1992 1.6 2007 1.0 1993 1.5 2008 0.9 1994 1.5 2009 0.9 1995 1.5 2010 0.9 incomes. Due to this demographic effect income inequality in Brazil has shown a noticeable decrease since its all time high in 1989. Factor Price Equalization The last factor that could help explain the inequality trends in Brazil is its recent commitment to liberalization and the resultant factor price equalization as hypothesized by Paul Samuelson in his extension of the Heckscher-Ohlin Theorem. As noted earlier, factor price equalization maintains that as two countries specialize in the good that uses their abundant factor intensively, the relative prices of labor and capital (or labor and land, capital and land, etc.) in the two countries will converge, ultimately benefiting the abundant factor within the country (Samuelson, 1948). For Brazil this would suggest that as the country liberalizes, the abundant factor, unskilled labor, would ultimately benefit from increased wages. In their analysis, Trade Liberalization and the Evolution of Skill Earnings Differentials in Brazil, Gustavo Gonzaga, NaercioMerezesFilho, and Cristina Terra look into this very effect. The three look at Brazil s liberalization period from 1988 to 1995 and employ multiple regression analyses that check the trade transmission mechanism through disaggregated data on tariffs, prices, earnings, employment and skill intensity (Gonzaga, Filho, & Terra, 2005). They find that over this period, employment shifted from skilled to unskilled intensive sectors. The effect on the employment share of skilled labor between industries (that is between skilled and

The Dichotomy of Development in Brazil 499 unskilled labor intensive sectors) fell by 25%, suggesting that trade liberalization was accompanied with a shift away from skilled labor intensive sectors to unskilled labor intensive sectors. This is consistent with Samuelson s theory about factor price equalization. Gonzaga, Filho, & Terra further note that this data is particularly significant because Brazil showed earnings inequality decrease over the liberalization period, which is contrary to evidence for most other developing countries that pursued similar liberalization strategies. When comparing Gonzaga, Filho, & Terra s data to the Gini Index over the same period, it is easy to see how factor price equalization could have played a part in diminishing income inequality in Brazil. From 1989 to 1995 the Gini index fell from 63.0 to 59.24. Furthermore, this only takes into account the intensive adjustment period. Brazil has continued to dismantle its protectionist policies since 1988 and today the Gini index is down to 53.9. This suggests that there is a strong correlation between liberalization, factor price equalization, and declining income inequality in Brazil. V. CONCLUSION The contrasts inherent in economic development precipitate a skewed picture of development s benefits. While a country can never eradicate poverty without first enjoying increased growth, one must be wary in assuming increases in per capita incomes mean that all people or even the majority of people are better off (Kulkarni, 2006). Kuznets Inverted U-Hypothesis recognizes the dichotomy inherent in economic development and a useful tool in analyzing who actually benefits from the process. Kuznets highlights the differential effects growth has on different sectors within society, and while the robustness of his theory is at times disputed, the lessons remain nonetheless pertinent. The additions to Kuznets theory of inequality offered by Ahluwalia and the mechanisms of factor price equalization help to further explain how inequality can be exacerbated and ameliorated within a country. In Brazil, the data is by no means conclusive, however, the patterns that arose through this analysis suggest that the country is well on its way to tackling income inequality. The structural change inherent in the country explains the increases in income inequality from the 1960s to 1989, and those same effects could help explain the downward trend that has occurred ever since 1989. The educational and demographic transitions that have occurred in Brazil, in particular the growing secondary enrollment rate and the diminishing population growth rate within the country, provides strong correlations for why income inequality has diminished over the last 20 years. And lastly, the effects of factor price equalization and the shift to unskilled labor suggests that those subject to immense inequality are in fact starting to reap the benefits of greater openness and liberalization within Brazil, as the period of liberalization that started in 1988 has demonstrated reductions in income inequality in almost every single year since. While income inequality still plagues Brazil, the country certainly has begun the process of eradicating income inequality.

500 J. Michael Deinlein 1. World Bank Definition of GINI coefficient. Note References Ahluwalia, M. S. (1976), Inequality, Poverty and Development. Washington, D.C: World Bank. Baer, Werner (1995), The Brazilian Economy: Growth and Development. Westport, CT: Praeger, Print. Beghin, N. (2008), Notes on Inequality and Poverty in Brazil: Current Situation and Challenges. From Poverty to Power. Oxfam International. Brecher, Richard A., and Ehsan U. Choudhri (1993), Some Empirical Support for the Heckscher- Ohlin Model of Production. The Canadian Journal of Economics 26.2: 272-85. Print. Carneiro, F. (2003), The Impacts of Trade on the Brazilian Labor Market: A CGE Model Approach. World Development 31.9 : 1581-595. Print. Clifton, David S., and William B. Marxsen (1984), An Empirical Investigation of the Heckscher- Ohlin Theorem. Canadian Journal of Economics 17.1 : 32-38. Print. Cárdenas, Enrique, José Antonio Ocampo, and Rosemary Thorp (2000), An Economic History of Twentieth-century Latin America. the Latin American Economies in the Late Nineteenth and Early Twentieth Centuries. New York: Palgrave in Association with St. Antony s College, Oxford, Print. Gonzaga, Gustavo, NaércioMenezesFilho, and Cristina Terra (2006), Trade Liberalization and the Evolution of Skill Earnings Differentials in Brazil. Journal of International Economics 68.2 : 345-67. Print. Higgins, M. & Williamson, J. G. (1999), Explaining Inequality the World Around: Cohort Size, Kuznets Curve, and Openness. New York: Federal Reserve Bank of New York. Inequality and Economic Development in Brazil. Washington, D.C.: World Bank, 2004. Print. Khan, Mohsin S., and Abdelhak S. Ssnhadji (2001), Threshold Effects in the Relationship between Inflation and Growth. IMF Staff Papers 48.1. Print. Kharsu, Syed M., and Mohammad M. Jalil (2010), Revisiting Kuznets Hypothesis: An Analysis with Time Series and Panel Data. Bangladesh Development Studies 30.3-4: 89-112. Print. Kulkarni, K. G. (2006), Readings in International Economics. New Delhi: Serials Publications. Kuznets, Simon (1955), Economic Growth and Income Inequality. The American Economic Review 45.1. Print. Lücke, Matthias (1990), Traditional Labour-intensive Industries in Newly Industrializing Countries: The Case of Brazil. Tübingen: J.C.B. Mohr. Print. Morley, Samuel A., and Gordon Whitford Smith (1971), The Effect of Changes in the Distribution of Income on Labor, Foreign Investment, and Growth in Brazil,. Houston, TX: Program of Development Studies, Rice University. Print. Pereira, L. V. (2005), Brazil Trade Liberalization Program. Washington D.C., World Bank. Roth, Sarah (2010), Economic Development and Income Inequality in Brazil: A Test of Kuznets U-Curve Hypothesis. International Economic Development: Theories, Models and Case Studies of Countries Leading the Change. Edited by Kishore Kulkarni. Matrix Publishers. Samuelson, Paul A. International Trade and the Equalisation of Factor Prices. The Economic Journal 58.230 (1948). Print. Samuelson, Paul A. (1953), Prices of Factors and Good in General Equilibrium. The Review of Economic Studies 21.1. Print.

The Dichotomy of Development in Brazil 501 Syrquin, Moshe (1988), Patterns of Structural Change. Handbook of Development Economics, Volume I. Edited by H. Chenery and T.N. Srinivasan.Elsevier Science Publishers. Therborn, Göran (2006), Inequalities of the World. London: Verso. Print. Thomas, Vinod. From inside Brazil: Development in a Land of Contrasts. Washington, D.C.: Stanford Economics and Finance, 2006. Print.