Peripherality, Income Inequality, and Economic Development in Latin

Download PDF
0 downloads 0 Views 580KB Size Report
Comment
Oct 21, 2018 - Kobe University, 2-1, Rokkodai, Nada-ku, Kobe 657-8501,. Japan. .... Theoreticians of structuralist macroeconomics such as Taylor (1983) formally modelled this ..... activity's association with an accumulation of technological ...
DP2017-08

Peripherality, Income Inequality, and Economic Development in Latin American Countries Yoshimichi MURAKAMI Nobuaki HAMAGUCHI Revised October 21, 2018

Revised on October 21, 2018

Peripherality, Income Inequality, and Economic Development in Latin American Countries Yoshimichi Murakami (Corresponding Author) Assistant Professor, Research Institute for Economics and Business Administration, Kobe University, 2-1, Rokkodai, Nada-ku, Kobe 657-8501, Japan. Email: [email protected] Nobuaki Hamaguchi Professor, Research Institute for Economics and Business Administration, Kobe University, 2-1, Rokkodai, Nada-ku, Kobe 657-8501, Japan. Email: [email protected]

Abstract This study, following a neo-structuralist perspective, presents a development puzzle concept for Latin American countries (LACs) as a triangular relation amongst peripherality (low level of technological progress and low level of global value chain (GVC) integration, with primary commodity dependence), income inequality, and per-capita income. Our empirical analysis based on this triangular relation reveals that a decrease in income inequality and an increase in per-capita income are mutually reinforcing in 16 LACs during 1995–2014. Results show that although primary commodity dependence, technological progress, and GVC integration increase per-capita income, they partly mitigate the increase by increasing income inequality. Additionally, the increasing effects of GVC integration and primary commodity dependence on income inequality are mitigated if the country is integrated into GVCs with higher technological capabilities and if primary commodity exports are accompanied with an increase in social expenditure. Keywords: primary commodity dependence; technological progress; global value chain (GVC); social expenditure; neo-structuralism JEL classifications: F63, O15, O47, O54

1

1. Introduction The first decade of the 21st century must be remembered as a golden age for Latin American countries (LACs): they enjoyed price stability, high economic growth, and decreased poverty and income inequality (Gasparini, Cruces, and Tornarolli, 2011; Lustig, Lopez-Calva, OrtizJuarez, 2013; Amarante, Galván, and Mancero, 2016; Moreno-Brid and Garry, 2016). Their good economic performance depended on the continued strong demand for primary commodities. However, with recent deterioration of external economic conditions, the region’s economy has stagnated and wavered as progress in poverty and inequality reduction has stalled. That economic distress has occurred in tandem with intensification of socio-political tensions, provoking major political changes in some countries: the victory of Mr. Lopez Obrador in the 2018 Mexican presidential election, presidential impeachments in Brazil and Peru, and a shift from populist to conservative administration in Argentina. In Venezuela, the Maduro administration has resorted to human rights oppression to retain power in the midst of social turmoil. The region has undergone stark political transformation since its economies stagnated and destabilized. Why are LACs so vulnerable to global economic changes, reacting similarly to a canary in a coal mine? This big question might demand years of study. It is even puzzling considering the region’s favourable features for economic development: resource-richness, proximity to North American markets, and upper-middle-income status with a fairly developed 2

manufacturing base. Actually, LACs are no strangers to the global economy, with which its development trajectory has remained closely linked since colonial times. To address the question of this puzzling vulnerability of LACs, we can evaluate some peculiar characteristics of the region. In revisiting the literature, we found that Latin American structuralism (henceforth structuralism) provides some clues for revealing the puzzling vulnerability.1 As discussed in the next section, structuralists specifically address two phenomena: income inequality and economic peripherality (i.e. factors that restrict an economy to underdeveloped peripheral position as opposed to highly developed central economies). A salient feature of peripherality is the balance-of-payments constraint (BPC) on economic growth in commodity-dependent LACs, as reported by Cimoli, Porcile, and Rovira (2010), Moreno-Brid and Garry (2016), and Murakami and Hernández (2018).2 Another distinctive attribute of peripherality is technological and productive backwardness. In this regard, structuralist arguments transitioned from a traditional view, which emphasized primary commodity dependence as the fundamental cause (Prebisch, 1950), to a neo-structuralist perspective that also emphasized low levels of technological progress (ECLAC, 1990) and low levels of global value chain (GVC) integration (ECLAC, 2014) as difficulties.3 Regarding income distribution, we adopt a neo-structuralist view to consider that a decrease in income inequality and an increase in per-capita income are mutually dependent, as discussed in greater detail in Section 2. We also devote attention to the influence of 3

peripherality on income inequality to assess whether globalization might increase or decrease the inequality of a peripheral economy. Because peripherality might indirectly affect per-capita income through the mutually interactive relation between inequality and income level, an unequal peripheral economy, as in Latin America, undergoes stronger effects of external shocks on per-capita income. Consequently, a novel contribution of this study is to capture interaction between income inequality and per-capita income, and the effects of peripheral features on both income inequality and per-capita income. Based on the triangular relation, which is discussed in greater depth in subsequent sections, we empirically assess whether a decrease in income inequality and an increase in per-capita income are mutually reinforcing, and assess the manner in which and the extent to which the peripheral features have affected income inequality and per-capita income in LACs since the mid-1990s. Our assessments reveal that a decrease in income inequality and an increase in per-capita income are mutually reinforcing in LACs during the period under analysis. Results show that although primary commodity dependence, technological progress, and GVC integration increase per-capita income, they partly mitigate the increase by increasing income inequality. Additionally, the increasing effects of GVC integration and primary commodity dependence on income inequality are mitigated if the country is integrated into GVCs with higher technological capabilities and if primary commodity exports are accompanied with an increase in social expenditure. 4

This paper is organized as follows. Section 2 explains the triangular relation amongst peripherality, income inequality, and per-capita income, following the neo-structuralist perspective. The section also briefly reviews whether a formal economic model or empirical findings have supported the arguments. Section 3 empirically evaluates the triangular relations amongst LACs in 1995–2014. The final section concludes the discussion and presents some policy implications. 2. Triangular relation amongst peripherality, inequality, and per-capita income in LACs Following the neo-structuralist concept of peripherality, this study assesses a development puzzle concept of LACs as a triangular relation amongst peripherality, income inequality, and per-capita income, as shown in Figure 1. This section explains the triangular relation and briefly reviews the arguments as supported by formal economic models and empirical findings. [Figure 1 near here] 2.1. Relation between income inequality and per-capita income in LACs Traditional structuralists argue that inequality in LACs, largely because of historical factors such as colonialization and land tenure systems, is the driving force behind chronic inflation, thereby harming economic growth and depressing per-capita income (Prebisch, 1950; 1961). Theoreticians of structuralist macroeconomics such as Taylor (1983) formally modelled this traditional structuralist idea to assess inequality as exacerbating inflation through sectoral conflicts. He thereby showed why orthodox macroeconomic adjustments did not work in LACs. 5

Some empirical studies based on cross-national analysis including LACs support the argument that income inequality decreases per-capita income growth (Alesina and Rodrik, 1994; Barro, 1996a; Easterly, 2007). Moreover, Beetsma and Van Der Ploeg (1996) show that the initial level of income inequality is positively associated with subsequent inflation rates. The growth regression literature, including Fischer (1993) and Barro (1996b) also provides evidence that inflation rates are associated significantly with low per-capita income growth. In contrast, neo-structuralists consider that the relation between income inequality and per-capita income is bidirectional rather than unidirectional. They emphasize that decreasing inequality and increasing per-capita income are mutually reinforcing (ECLAC, 1990; Fajnzylber, 1990; Leiva, 2008). On the one hand, they argue that a more equal income distribution can contribute to higher per-capita income directly through promoting human capital and physical capital investment and indirectly through technological progress by human capital accumulation (ECLAC, 1990; Fajnzylber, 1990). Some theoretical studies support this argument. For instance, Galor and Zeira (1993) present a model that predicts that countries with a more equal income distribution will have higher per-capita income because they have a larger share of individuals who can afford to invest in human capital. Galor, Moav, and Vollrath (2009) also present a model that predicts countries with more equal landownership distribution promote emergence of human-capital promoting institutions. 6

On the other hand, neo-structuralists argue that higher per capita labour income can contribute to decreasing income inequality because labour income is more likely to increase than other sources of income if structural change is accompanied by an increase in labour productivity (ECLAC, 1990; Fajnzylber, 1990). Theoretical studies also support this argument. For example, the model proposed by Galor and Zeira (1993) predicts that countries with higher per-capita income will have more equal income distribution because wages of unskilled workers will increase in countries where even unskilled workers can afford to invest in human capital and thereby become skilled workers. Therefore, the model predicts mutual reinforcement of lower income inequality and higher per-capita income; it predicts that countries with a more equal income distribution will have higher per-capita income, whereas countries with higher per-capita income will have more equal income distribution, exactly supporting the neo-structuralist argument. However, empirical studies based on cross-national panel data covering LACs have usually included the assumption that income inequality and per-capita income are determined independently. For example, Székely and Mendoza (2017) empirically evaluate the determinants of Gini coefficient in LACs, but they do not consider the per-capita income effects. Although Morley (2001) and Bucciferro (2010) assess the determinants of Gini coefficient in LACs including their per-capita income level, they do not regard income inequality and percapita income as determined simultaneously. We test whether our empirical analysis supports 7

the neo-structuralist argument in LACs in Section 3. 2.2. Relation between peripheral features and per-capita income in LACs Because traditional structuralists regard commodity-exporting countries as adversely affected by long-term worsening terms of trade, they argue that specialization in exports of primary commodities harms national economic growth and lowers per-capita income (Prebisch, 1950; Prebisch, 1959). However, empirical studies such as those by Cuddington (1992) and Hadass and Williamson (2003) find that long-term deterioration in primary commodity prices has not occurred historically. In contrast, empirical studies based on cross-national growth regression analyses have supported the argument that specialization in primary commodity exports lowers growth of per-capita income in developed and developing countries (Sachs & Warner, 1995; 2001) and in LACs (Agosin, 2009). Moreover, Sachs and Warner (1999) demonstrate that natural resource booms were sometimes accompanied by declining per-capita income in several LACs during 1960–1994. Therefore, some additional channels exist through which commodity dependence lowers national per-capita income. First, natural resource abundance or resource booms particularly shrink high-productivity manufacturing sectors, which have positive externalities for increasing per-capita income. Simultaneously, they expand low-productivity non-tradable sectors because resource booms create excess demand for products of the latter sectors, and move resources away from the former to the latter, which is typically designated as Dutch 8

disease (Corden & Neary, 1982; Corden, 1984). McMillan, Rodrik, and Verduzco-Gallo (2014) demonstrate that such productivity-reducing effects attributable to expansion of lowproductivity service sectors took place in Africa and Latin America following their integration into global trade in the 1990s. Second, natural resource abundance is associated with the deterioration of institutional quality through increasing rent-seeking activities. Some studies based on cross-national analysis find that institutional quality rather than natural resource abundance itself is the proximate factor degrading economic performance (Bulte, Damania, & Deacon, 2005; Mehlum, Moene, & Torvik, 2006). According to the neo-structuralist perspective, low productivity stemming from a low level of technological progress and a low level of global GVC integration, rather than primary commodity dependence per se, is a fundamentally important peripheral feature, as explained in the Introduction. Consequently, whether GVC integration is associated with productivity improvement must be reviewed. Recent studies have directly examined the productivity effects of GVC integration at firm level. For example, Montalbano, Nenci, and Pietrobelli (2014) find that industry-level backward GVC participation, as measured by the share of foreign value added embodied in the industry’s total exports, improves firm level productivity in four LACs: Brazil, Argentina, Chile, and Mexico. Additionally, the literature related to foreign direct investment (FDI) assesses the productivity effects of foreign affiliates on local firms rather than the productivity effects of 9

lead firms on exporting firms that are directly integrated into GVCs in developing countries. However, some studies show that the positive effects of FDI on the productivity of local firms do not occur automatically in developing countries, including LACs. For example, Borensztein, De Gregorio, and Lee (1998) find that the contribution of FDI flows to developing country’s economic growth is positive only when they have more than a threshold level of human capital. As evidence from specific LACs, Kokko (1994) and Kokko, Tansini, and Zejan (1996) report that only local firms with small technological gaps vis-à-vis foreign firms receive positive intraindustry spillovers from FDI, respectively, in Mexico and Uruguay. Moreover, Aitken and Harrison (1999) find that intra-industry FDI has negative effects on the productivity of local firms in Venezuela because negative effects from intensified competition with foreign firms dominate positive effects such as demonstration and imitation. Consequently, these findings indicate that the occurrence of positive spillover effects requires domestic firms to have at least some technological capabilities to adopt new advanced technology used by foreign firms for their own production or to compete with them. By contrast, Jordaan (2008) finds that technological gaps vis-à-vis foreign firms in the same industries have positive effects on the productivity of local firms in Mexico. In summary, although neo-structuralists expect positive productivity effects from integration into GVCs, empirical studies investigating LACs yield inconclusive findings. 2.3. Relations between peripheral features and inequality in LACs 10

We also devote some attention to peripherality effects on income inequality. Neo-structuralists regard peripheral features as associated directly with increasing income inequality (ECLAC, 1990; Fajnzylber, 1990). First, international integration based on natural resource exports is regarded as associated with increasing income inequality because natural resource sectors are often capital-intensive or skilled-labour-intensive. Therefore, integration into global trade is expected to increase wages of skilled workers or capital owners, thereby increasing income inequality. Moreover, because ownership of natural resources is concentrated in a few hands in LACs, an increase in returns to natural resources will further raise the owners’ income (ECLAC, 1990). Because natural resources are sometimes owned by governments in LACs, the governments can distribute the windfall profits from rising natural resource prices, thereby temporarily contributing to a reduction in income inequality. Nevertheless, income inequality will not decrease in the long-term unless such government transfers contribute to improving the labour productivity of low-income groups. Empirical studies show increasing effects of natural resource sectors on income inequality. For example, Leamer, Maul, Rodriguez, and Schott (1999) find that the export shares of raw materials and tropical permanent crops are positively correlated with Gini coefficients in LACs in 1980 and 1990 because the production of such products requires a good deal of physical capital that is complementary to skilled labour. Second, according to the neo-structuralist perspective, international competitiveness based on technological progress can reduce inequality because an increase in value-added 11

exports can improve labour productivity, thereby pushing up workers’ wages (ECLAC, 1990). However, they also consider that technological progress is likely to increase inequality, at least temporarily, because it takes some time until technological progress encompasses all lowproductivity sectors (ECLAC, 1990). Such a temporal increase in inequality derives from skillbiased technological changes (SBTCs), as shown by the model proposed by Acemoglu (1998). Empirical studies have revealed that both effects actually occurred in LACs after their integration into the global economy. On the one hand, many empirical studies show that SBTCs were observed in LACs after the implementation of trade liberalization. Increased income inequality in the 1980s and 1990s is explained to a good degree by SBTCs (Goldberg and Pavcnik, 2007; Murakami, 2018). On the other hand, the SBTC effects faded in the 2000s. The observed decrease in income inequality during this period is largely attributable to a decrease in skill premiums (Gasparini, Galiani, Cruces, and Acosta, 2011; Lustig et al., 2013). Those findings support the argument that technological progress increases income inequality in the short term, although it reduces income inequality in the long term. Third, because integration into GVCs can be a powerful tool for internationalization and productivity improvements of small and medium firms through creation of production linkages with foreign firms, it can be expected to contribute to reduction in inequality in LACs (ECLAC, 2014).4 Moreover, because unskilled labour-intensive tasks are likely to be transferred from developed to developing countries in line with comparative advantage, 12

demand for unskilled workers is expected to increase. Therefore, income inequality is expected to decrease in developing countries. However, empirical studies based on cross-national panel data analysis such as those by Herzer, Hühne, and Nunnenkamp (2008) and Suanes (2016) find that FDI is positively associated with Gini coefficients in LACs. As evidence from a specific country, Feenstra and Hanson (1997) find that the growth in FDI inflows from the US is positively associated with changes in the share of skilled workers’ wages within manufacturing industries in Mexico. Consequently, empirical studies do not support the argument that integration into GVCs is associated with reduced inequality in LACs. 2.4. Summary of the triangular relation amongst peripherality, inequality, and per-capita income in LACs Consequently, according to the triangular relation, a decrease in peripheral features is expected not only to increase per-capita income directly but also to increase it indirectly by reducing income inequality. Conversely, an increase in peripheral features is expected not only to decrease per-capita income directly but also to decrease it indirectly by increasing income inequality (see Figure 1). If negative external shocks are associated with increasing peripheral features (e.g. decrease in GVC integration), then LACs receive a stronger negative effect on per-capita income. This increase of the effect of external shocks on per-capita income within the triangular relation might reveal the puzzling vulnerability of LACs. 3. Empirical analysis of the triangular relation amongst peripherality, inequality, and 13

income level in LACs 3.1. The model In this section, we describe our empirical assessment of the triangular relation amongst peripherality, inequality, and per-capita income, discussed in Section 2, using panel data of 16 LACs from 1995 to 2014.5 We choose this period because data obtained before 1995 are not necessarily reliable because of hyperinflation and currency fluctuations.6 Based on the discussion presented in Section 2, the model includes the assumption that per-capita income and income inequality are determined simultaneously. The three peripheral features affect both per-capita income and income inequality. In the first equation, following the specification for estimating production function, per-capita income is determined by the per-capita capital stock, variables indicating the three peripheral features, and other controls. In the second equation, income inequality is determined using a linear and a quadratic term of per-capita income, variables indicating the three peripheral features, and other controls. Regarding the other controls, we include only the inflation rate in the first equation, whereas we include the inflation rate, share of social expenditure, and survey characteristics (i.e. whether the survey covers only urban areas or the country as a whole) in the second equation because we consider that the share of social expenditure and the survey characteristics affect the country’s income inequality but do not affect the country’s per-capita income. 14

Consequently, they serve as exogenous variables for the identification of structural equations. Regarding the measurements of the peripheral features, we assess primary commodity dependence by the share of primary commodity exports in all exports, technological progress by the number of patent applications per thousand population including residents and nonresidents, and GVC integration by the share of FDI inflow in GDP (FDI),7 the share of intermediate goods (the sum of parts and components) exports in all exports (Intermediate X), and the share of intermediate goods (the sum of parts and components) imports in all imports (intermediate M). The simultaneous equations are

y i ,t = β10 + β11 k i ,t + β12 Ginii ,t + β13 NRi ,t −1 + β14 Patent i ,t −1 + GVC′i ,t −1β15 + β16 Inflationi ,t −1 + α i + e1i ,t Ginii ,t = β 20 + β 21 y i ,t + β 22 y i2,t + β 23 NRi ,t −1 + β 24 Patent i ,t −1 + GVC′i,t −1β 25 + β 26 Inflationi ,t −1 + β 27 Sociali ,t −1 + β 28Urbani ,t + α i + e2i ,t

,

(1)

,

(2)

where i and t indexes respectively denote country and time, y is the log of real GDP per capita,8 k is the log of real capital stock per capita,9 Gini represents the Gini coefficient of household per-capita income,10 NR signifies the share of primary commodity exports in all exports,11 Patent denotes the number of patent applications per thousand population including residents and non-residents,12 vector GVC consists of the three indicators of GVC integration (FDI,13 intermediate X, and intermediate M14), inflation is the inflation rate,15 Social expresses the share of social expenditure to GDP,16 and Urban is a dummy variable that takes the value 1 when the survey for the calculation of the Gini coefficient covers urban areas only. We also 15

include fixed effects of each country α to control for unobservable time-invariant country characteristics that affect both the level of income and inequality. Tables 1 and 2, respectively, present descriptive statistics and the correlation matrix of those variables. Some samples are missing because of data availability issues, especially in cases of the Gini coefficient and the number of patent applications (Table 1). Therefore, we have an unbalanced panel dataset. We estimate structural equations (1) and (2) using threestage least squares (3SLS). Because equation (2) includes a nonlinear endogenous variable (the 2

quadratic term of log of real GDP per capita y ), we use a quadratic term of log of real capital 2

2

stock per capita k and its linear term k as an additional instrument for y , in line with Wooldridge (2002). Based on the neo-structuralist perspective, Table 3 presents the expected signs of the coefficients of the explanatory variables. As discussed in Section 2, we expect that income inequality decreases per-capita GDP. Also, the share of primary commodity exports decreases per-capita GDP. The number of patent applications and the GVC integration indicators increase the per-capita GDP level. Also, the inflation rate decreases according to the per-capita GDP level. We also expect that per-capita GDP decreases the Gini coefficient up to some threshold level. The share of primary commodity exports increases the Gini coefficient. The GVC integration indicators decrease the Gini coefficient. The number of patent applications decreases the Gini coefficient in the long term, but it might increase it in the short term. In 16

addition, the inflation rate increases the Gini coefficient. We also estimate structural equations, which include interaction terms between the number of patent applications and the three GVC integration indicators in equations (1) and (2), and an interaction term between the share of primary commodity exports and the share of social expenditure in equation (1). The former interaction terms in equation (1) are expected to enhance the positive effects of GVC integration on per-capita income because empirical studies such as that by Borensztein et al. (1998) find that the absorptive capacity, such as the technological capabilities of a host nation, enhances positive spillovers from FDI. The former interaction terms in equation (2) are expected to enhance the decreasing effects of GVC integration on income inequality because positive effects from participation in GVCs in terms of productivity and wages are more likely to spill over to other firms or sectors if they have higher absorptive capacity. The latter interaction term in equation (2) is expected to enhance the decreasing effects of social expenditure on income inequality because progressive social programs such as conditional cash transfer are particularly associated with fiscal revenues from commodity exports (Lustig et al., 2013). [Tables 1–3 near here] 3.2. Estimation results Table 4 presents estimation results of the structural equations without interaction terms. With regard to equation (1) (i.e. the determinants of log of real GDP per capita), the Gini coefficient 17

has significant and negative effects, as expected. The number of patent applications, the share of FDI inflows, and the share of intermediate goods exports have significant and positive effects, as expected. Consequently, technological progress and GVC integration increase percapita income, supporting the arguments discussed in this relation in Section 2. However, unexpectedly, the share of primary commodity exports also has significant and positive effects, contradicting the arguments discussed for this relation in Section 2. However, the finding is plausible because LACs experienced a rise in commodity prices in the 2000s, which is associated with both an increase in the income level and an increase in the share of primary commodities. The coefficient for the log of per-capita capital stock is positive and significant, as expected, whereas the coefficient for the inflation ratio is unexpected and positive, but not significant. Related to equation (2) (i.e. the determinants of the Gini coefficient), the coefficient on the linear term of log of GDP per capita is significant and negative, although the coefficient on its quadratic term is significant and positive, as expected. Consequently, the relation between log of GDP per capita and the Gini coefficient is U-shaped, contradicting earlier studies such as those conducted by Morley (2000), which support the traditional Kuznets’ inverted-U hypothesis in LACs. Moreover, the Gini coefficient is a decreasing function of GDP per capita in the range of the data observed in our analysis (i.e. 1101 USD – 14,688 USD; see Figure 2). Therefore, it is apparent that a decrease in inequality and an increase in per-capita 18

income are mutually interactive and mutually reinforcing in LACs, empirically supporting the arguments discussed for this relation in Section 2. The share of primary commodity exports has significantly increasing effects, empirically supporting the arguments related to the relation between primary commodity dependence and inequality discussed in Section 2. The finding suggests that the primary commodity sectors and possibly the service sectors expanded by the commodity boom are likely to be skilled-labour-intensive. The number of patent applications has significantly increasing effects, indicating that the effects of SBTCs still dominate the possible downward pressure on the wages of skilled workers because of an increase in their relative supply in this period. However, unexpectedly, the share of FDI inflows and the share of intermediate goods exports have significantly increasing effects. Therefore, it is apparent that GVC integration increases inequality in LACs, contradicting the arguments discussed for this relation in Section 2, but supporting those of earlier studies such as those of Feenstra and Hanson (1997), Herzer et al. (2008), and Suanes (2016). The finding shows that foreign affiliates and local firms exporting parts and components generate skilled labour demand in this period. Therefore, the findings indicate that although technological progress and GVC integration increase per-capita income directly, they partly mitigate the increase indirectly by increasing income inequality. Calculations for the estimated coefficients in Table 4 demonstrate that the direct positive effects of technological progress and GVC integration dominate the 19

indirect negative effects on per-capita income.17 Consequently, unexpectedly, LACs partly mitigated the negative effects of an increase in the peripheral features associated with external shocks (e.g. a sharp decrease in the share of FDI inflow in 2009 18) on per-capita income during the period under the analysis. The coefficient for the share of social expenditure is significant and negative, as expected. Consequently, progress in social policies during this period contributes to decreasing inequality, supporting the previous studies such as that by Suanes (2016). The coefficient for the inflation rate is positive, as expected, but not significant. The estimation results of the structural equations with the interaction terms are presented in Table 5. Unexpectedly, the interaction terms between the number of patent applications and the three GVC integration indicators are not significant in equation (1). By contrast, it is noteworthy that the interaction terms between the number of patent applications and the GVC integration indicators (the share of FDI inflows and the share of intermediate goods imports) have decreasing effects on the Gini coefficient in equation (2). Therefore, the finding demonstrates that the countries integrated into GVCs with higher technological capabilities mitigate the increasing effects on inequality derived from the GVC integration. This finding concurs with results of earlier studies that have revealed the importance of absorptive capacity in enabling host countries or firms to receive positive spillovers from FDI. The estimated coefficients demonstrate that the threshold number of patent applications 20

necessary to surpass the observed increasing effect of the share of FDI inflows is 0.167 (Table 5).19 However, the average level of the number of patent applications (0.087) is smaller than this threshold level (Table 1). Only the average values of Uruguay (0.177) and Chile (0.170) exceed it. It is also noteworthy that the interaction term between the share of primary commodity exports and the share of social expenditure has a decreasing effects on the Gini coefficient in equation (2). Thus, the increasing effects of the share of primary commodity exports on inequality are mitigated if they are accompanied with an increase in the share of social expenditure. The finding coincides exactly with the experience of South American countries in the 2000s, where progressive social policies were financed by revenues from commodity exports. However, the threshold level of the share of social expenditure that is necessary to surpass the observed increasing effect of primary commodity exports is 0.249 (Table 5); 20 none of the average of the share of social expenditure of 16 LACs exceeds it. [Tables 4–5 near here] [Figure 2 near here] 4. Concluding remarks Although LACs present some favourable features for economic development, they are highly vulnerable to global economic changes. To address the puzzling vulnerability of LACs, we revisited the neo-structuralist idea and presented the development puzzle concept for the region as a triangular relation amongst peripherality (low level of technological progress and low level 21

of GVC integration, with primary commodity dependence), income inequality, and per-capita income. We empirically tested the triangular relation in LACs during 1995–2014. The findings of empirical analyses are explained below. First, decreasing income inequality and increasing per-capita income were found to be mutually reinforcing in LACs. Therefore, the reduction of income inequality promoted an increase in per-capita income, which in turn contributed to decreasing income inequality, exactly supporting the neo-structuralist arguments. The finding provides new evidence related to the growth and inequality in LACs, which earlier studies have not found. Consequently, this study is a novel contribution to the literature. Second, primary commodity dependence unexpectedly increased per-capita income, although it increased income inequality too. Therefore, findings related to the effects on percapita income seemingly contradict results of empirical studies as well as the neo-structuralist arguments. However, because a rise in commodity prices was found to be temporarily associated with both an increase in the per-capita income and an increase in the share of primary commodities in the 2000s, it is plausible that such a relation is observed during the period under the analysis. We also found that the increasing effects of primary commodity exports on income inequality were mitigated if they were accompanied with an increase in social expenditure. However, the threshold level of the social expenditure that is necessary to surpass the observed increasing effects of primary commodity exports on income inequality is unsustainably high. 22

Third, technological progress, as measured by the number of patent applications, and GVC integration, as measured by the share of FDI inflows and the share of intermediate goods exports, increased per-capita income as well as income inequality. Therefore, the findings indicate that technological progress and GVC integration partly mitigated the increase in percapita income indirectly by increasing income inequality. In other words, unexpectedly, because increasing peripheral features such as a sharp decrease in FDI partly mitigated the decrease in per-capita income indirectly by decreasing income inequality, it is apparent that LACs did not receive stronger effects of external shocks through this triangular relation. We also found that the increasing effects of GVC integration on income inequality were mitigated if the country was integrated into GVCs with higher technological capabilities. However, technological progress in most LACs during the analysis period were lower than the threshold level that is necessary to surpass the observed increasing effects of FDI on income inequality. These findings suggest the following implications for LACs. Mexico and some Central American countries do not automatically ensure their higher per-capita income with equitable distribution by integration into North American supply chains. Policies that simultaneously promote GVC integration and technological progress are important requirements for it. South American countries cannot continue to mitigate the increasing effects of primary commodity exports on income inequality by expanding expenditure for progressive social programs after the end of the commodity boom. Primary commodity dependence might 23

serve as a structural factor that indirectly lowers per-capita income by increasing income inequality. The observed positive association between technological progress and income inequality might be explained by SBTCs, but the observed positive association between GVC integration and income inequality, which causes unexpected negative effects on per-capita income by increasing inequality, demands further analysis. We must ascertain why GVC integration increases income inequality in LACs if they have insufficient technological capabilities. One possible explanation is that GVC integration in LACs might not have generated sufficient unskilled labour demand because of the lack of scale attributable to bottlenecks such as insufficient infrastructure and market segmentations. If this is true, then it is crucially important to adopt a policy for the additional promotion of regional integration within LACs to encourage the scale economy to work. Although this issue is beyond the scope of this study, this area represents an interesting subject for future research. Acknowledgements The authors are deeply grateful to João Carlos Ferraz, Atsushi Fukumi, Tatsuji Hayakawa, Katsushi Imai, Mikio Kuwayama, Jorge Mario Martínez Piva, Tomoo Marukawa, Enrique Mu, Sachiyo Murase, Enrique Peruzzotti, Takahiro Sato, Tatsuya Shimizu, Terukazu Suruga, Yuriko Takahashi, and Tsuyoshi Yasuhara for their insightful comments and constructive suggestions. Any remaining errors are the authors’ own. 24

Funding details This work was supported by the [Japan Society for the Promotion of Science (JSPS)] under Grant-in-Aid for Scientific Research (B) Grant [number 16H03313] and Grant-in-Aid for Young Scientists (B) Grant [number 17K17877]. Disclosure statement There are no potential conflicts of interest to declare.

25

Notes 1. Latin American structuralism, which has basically developed in the United Nations Economic Commission for Latin America and the Caribbean (ECLAC), is an important stream of structuralist economics. For further explanation of the origin of the structuralism and the contribution of Latin American structuralism, especially in the field of inflation theory, to structuralist economics, see Arndt (1985). 2. The BPC model predicts that a country’s long-term economic growth rate is determined by the growth rate of real export volumes divided by the income elasticity of demand for imports. Therefore, an increase in the income elasticity of demand for imports and stagnant growth rate of export volumes, which are observed in commodity-exporting countries, serve as constraints on economic growth. 3. Consequently, traditional structuralists have identified industrialization with economic development, often providing justification for the protectionist policies for import substitution industrialization (Bielschowsky, 2009). In contrast, since neo-structuralists consider that an activity’s association with an accumulation of technological capabilities and increased productivity is necessary for economic development, they less emphasize protectionist policies for industrialization (Leiva, 2008; Bielschowsky, 2009; Ocampo, Rada, and Taylor, 2009). 4. Dependence and world-systems theory, on which structuralism has at least some influence, argues that integration of peripheral countries into global economy including reception of FDI 26

from core countries is associated with increased inequality within peripheral countries (Kaulihowa and Adjasi, 2018). 5. The 16 LACs consist of Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. 6. None of the 16 LACs reported the three-digit inflation rate during 1995–2014. Note 15 explains the inflation rate data source. 7. FDI is regarded as one type of relation between lead firms and local firms, designated as ‘GVC governance’ (Gereffi, Humphrey, and Sturgeon, 2005) because shifting of production activities from one country to another can be achieved either by relocating the production base from a parent company to its foreign affiliates or by outsourcing production to local suppliers or third-party providers. Because the former case requires FDI by definition, the share of FDI inflows can be an appropriate indicator of GVC integration. 8. Real GDP per capita data were referred from the International Financial Statistics (IFS) 2016 CD-ROM of International Monetary Fund (IMF). All values were converted from national currencies to US dollars and were deflated by the US GDP deflator (2010 = 1). 9. Real capital stock per capita data were referred from the Penn World Table ver. 9.0. (http://www.rug.nl/ggdc/productivity/pwt/). 10. Gini coefficient of household per-capita income data were referred from the Socio27

Economic

Database

for

Latin

America

and

the

Caribbean

(SEDLAC)

(http://sedlac.econo.unlp.edu.ar/eng/). 11. Primary commodity exports and total exports data were referred from UN-COMTRADE (https://comtrade.un.org/data/). Primary commodities are defined by the sum of Standard International Trade Classification (SITC) Rev. 1 codes 0, 1, 2, 4, 667, and 68, in line with the UNCTAD product groupings and composition. 12. Patent application data were referred from the World Intellectual Property Organization (WIPO) Statistics Database (http://ipstats.wipo.int/ipstatv2/index.htm?tab=patent). 13. FDI inflow data were referred from IFS 2016 CD-ROM of IMF. The FDI inflow is calculated as liabilities minus assets of direct investment. 14. Intermediate goods export and import data were referred from UN-COMTRADE. Intermediate goods include only parts and components, as defined by the sum of Broad Economic Categories (BEC) codes 42 and 53. 15. Inflation rate (annual variation of consumer price index) data were referred from the Statistical Yearbook

for

Latin America

and

the

Caribbean

2015

of

ECLAC

(http://interwp.cepal.org/anuario_estadistico/anuario_2015/en/index.asp). 16. Social expenditure, as defined by the sum of education, health, social security, and housing expenditure,

data

were

referred

from

CEPALSTAT

of

ECLAC

(http://estadisticas.cepal.org/cepalstat/web_cepalstat/estadisticasIndicadores.asp?idioma=i). 28

17. Indirect effects can be calculated by multiplying the coefficient on the Gini coefficient in equation (1) by the coefficient on the variable in question in equation (2). 18. The simple average of the share of FDI inflows in the 16 LACs decreased sharply from 4.0% in 2008 to 2.5% in 2009. Note 13 explains the data source of FDI inflows. 19. The threshold number can be calculated by dividing the coefficient on the share of FDI inflows by the absolute value of the coefficient on the interaction term between the number of patent applications and the share of FDI inflows in equation (2). 20. The threshold level can be calculated by dividing the coefficient on the share of primary commodity exports by the absolute value of the coefficient on the interaction term between the share of primary commodity exports and the share of social expenditure in equation (2).

29

References Acemoglu, D. (1998). Why do new technologies complement skills? Directed technical change and wage inequality, Quarterly Journal of Economics, 113(4) 1055–1089. Agosin, M. R. (2009). Export diversification and growth in emerging economies, CEPAL Review, 97, 115–131. Aitken, B. J., & Harrison A. E. (1999). Do Domestic Firms Benefit From Direct Foreign Investment? Evidence from Venezuela, American Economic Review, 89(3), 605–618. Alesina, A., & Rodrik, D. (1994). Distributive politics and economic growth, The Quarterly Journal of Economics, 109(2), 465–490. Amarante, V., Galván, M., & Mancero, X. (2016). Inequality in Latin America: a global measurement, CEPAL Review, 118, 7–31. Arndt, H. W. (1985). The origins of structuralism, World Development, 13(2), 151–159. Barro, R. J. (1996a). Democracy and Growth. Journal of Economic Growth, 1(1), 1–27. Barro, R. J. (1996b). Determinants of Economic Growth: A Cross-Country Empirical Study, National Bureau of Economic Research, working paper no. 5698. Beetsma, R. M. W. J., & Van Der Ploeg, F. (1996). Does inequality cause inflation?: The political economy of inflation, taxation and government debt, Public Choice, 87(1-2), 143–162. Bielschowsky, R. (2009). Sixty years of ECLAC: Structuralism and neo-structuralism, CEPAL 30

Review, 97, 171–192. Borensztein, E., De Gregorio, J., & Lee, J.-W. (1998). How Does Foreign Direct Investment Affect Economic Growth? Journal of International Economics, 45(1), 115–135. Bucciferro, J. (2010). Economic Restructuring and Poverty/Income Inequality in Latin America. Journal of Sustainable Development, 4(1), 1–16. Bulte, E. H., Damania, R., & Deacon, R. T. 2005. Resource intensity, institutions, and development. World Development, 33(7), 1029–1044. Cimoli, M., Porcile, G., & Rovira, S. (2010). Structural change and the BOP-constraint: Why did Latin America fail to converge? Cambridge Journal of Economics, 34(2), 389–411. Corden, W. M. (1984). Booming sector and Dutch disease economics: Survey and consolidation, Oxford Economic Papers, 36(3), 359–380. Corden, W. M., & Neary, J. P. (1982). Booming sector and de-industrialisation in a small open economy, The Economic Journal, 92(368), 825–848. Cuddington, J. T. (1992). Long-run trends in 26 primary commodity prices. Journal of Development Economics, 39(2), 207–227. Easterly, W. (2007). Inequality Does Cause Underdevelopment: Insights from a New Instrument. Journal of Development Economics, 84(2), 755–776. ECLAC. (1990). Changing Production Patterns with Social Equity: The Prime Task of Latin American and Caribbean Development in the 1990s. Santiago, Chile: ECLAC. 31

ECLAC. (2014). Regional Integration: Towards an Inclusive Value Chain Strategy. Santiago, Chile: ECLAC. Fajnzylber, F. (1990). Industrialization in Latin America: From the ‘Black Box’ to the ‘Empty Box’: A Comparison of Contemporary Industrialization Patterns. Santiago, Chile: ECLAC. Feenstra, R. C., & Hanson, G. H. (1996). Foreign direct investment and relative wages: Evidence from Mexico’s maquiladoras, Journal of International Economics, 42(3), 371–393. Fischer, S. (1993). The role of macroeconomic factors in growth, Journal of Monetary Economics, 32(3), 485–512. Galor, O., Moav, O., & Vollrath, D. (2009). Inequality in Landownership, the Emergence of Human-Capital Promoting Institutions, and the Great Divergence. Review of Economic Studies, 76(1), 143–179. Galor, O., & Zeira, J. (1993). Income distribution and macroeconomics, Review of Economic Studies, 60(1), 35–52. Gasparini, L., Cruces, G., & Tornarolli, L. (2011). Recent Trends in Inequality in Latin America. Economía, 11(2), 147–190. Gasparini, L., Galiani, S., Cruces, G., & Acosta, P. (2011). Educational Upgrading and Returns to Skills in Latin America: Evidence from a Supply Demand Framework, 1990–2010 32

(World Bank Policy Research Working Paper 5921). Washington, DC: World Bank. Gereffi, G., Humphrey, J. & Sturgeon, T. (2005). The Governance of Global Value Chains. Review of International Political Economy 12(1), 78–104. Goldberg, P. K., & Pavcnik, N. (2007). Distributional effects of globalization in developing countries, Journal of Economic Literature, 45(1), 39–82. Hadass, Y. S., & Williamson, J. G. (2003). Terms-of-trade shocks and economic performance, 1870–1940: Prebisch and Singer revisited, Economic Development and Cultural Change, 51(3), 629–656. Herzer, D., Hühne, P., & Nunnenkamp, P. (2014). FDI and Income Inequality—Evidence from Latin American Economies. Review of Development Economics, 18(4), 778–793. Jordaan, J. (2008). Intra- and Inter-Industry Externalities from Foreign Direct Investment in the Mexican Manufacturing Sector: New Evidence from Mexican Regions. World Development 36(12), 2838–2854. Kaulihowa, T. & Adjasi, C. (2018). FDI and Income Inequality in Africa. Oxford Development Studies, 46(2), 250–265. Kokko, A. (1994). Technology Market Characteristics and Spillovers. Journal of Development Economics, 43(2), 279–293. Kokko, A., Tansini, R., & Zejan, M. C. (1996). Local technological capability and productivity spillovers from FDI in the Uruguayan manufacturing sector, Journal of Development 33

Studies, 32(4), 602–611. Leamer, E. E., Maul, H., Rodriguez, S., & Schott, P. K. (1999). Does natural resource abundance increase Latin American income inequality? Journal of Development Economics, 59(1), 3–42. Leiva, F. I. (2008). Latin American Neostructuralism: The Contradictions of Post-Neoliberal Development, Minneapolis, University of Minnesota Press. Lustig, N., Lopez-Calva, L. F., & Ortiz-Juarez, E. (2013). Declining inequality in Latin America in the 2000s: The cases of Argentina, Brazil, and Mexico, World Development, 44, 129–141. McMillan, M., Rodrik, D., & Verduzco-Gallo, I. (2014). Globalization, structural change, and productivity growth, with an update on Africa. World Development, 63(1), 11–32. Mehlum, H., Moene, K., & Torvik, P. (2006). Institutions and the Resource Curse. Economic Journal, 116(508), 1–20. Montalbano, P., Nenci, S., & Pietrobelli, C. (2014). ‘International Linkages, Value Added Trade and LAC Firms’ Productivity’, Departmental Working Papers of Economics, University ‘Roma Tre’, working paper no. 0198. Moreno-Brid, J. C., & Garry, S. (2018). Economic Performance in Latin America in the 2000s: Recession, Recovery, and Resilience? Oxford Development Studies, 44(4), 384– 400. 34

Morley, S. A. 2000. The effects of growth and economic reform on income distribution in Latin America, CEPAL Review, 71, 23–40. Murakami, Y. (2018). Globalization and Income Inequality in Latin America: A Review of Theoretical Developments and Recent Evidence, RIEB Discussion Paper Series, DP2018-16, Research Institute for Economics & Business Administration, Kobe University. Murakami, Y., & Hernández, R. A. (2018). The Impacts of China on Economic Growth: Evidence for Brazil, Chile and Peru. Journal of Post Keynesian Economics, 43(3), 430– 454. Ocampo, J. A., Rada, C. & Taylor, L. (2009). Growth and Policy in Developing Countries: A Structuralist Approach. New York: Columbia University Press. Prebisch, R. (1950). The Economic Development of Latin America, and Its Principal Problems. New York: United Nations, Department of Economic Affairs. Prebisch, R. (1959). Commercial policy in the underdeveloped countries, American Economic Review, 49(2), 251–273. Prebisch, R. (1961). Economic Development or Monetary Stability: The False Dilemma. Economic Bulletin for Latin America, 6(1), 1–25. Sachs, J. D., & Warner, A. (1995). Natural resource abundance and economic growth. NBER Working Paper No. 5398. 35

Sachs, J. D., & Warner, A. (1999). The big push, natural resource booms and growth. Journal of Development Economics 59(1), 43–76. Sachs, J. D., & Warner, A. (2001). The curse of natural resources, European Economic Review, 45(4–6), 827–838. Suanes, M. (2016). Foreign Direct Investment and Income Inequality in Latin America: a Sectoral Analysis. CEPAL Review, 118, 45−61. Székely, M., & Mendoza, P. (2017). Declining Inequality in Latin America: Structural Shift or Temporary Phenomenon? Oxford Development Studies, 45(2), 204–221. Taylor, L. (1983). Structuralist Macroeconomics: Applicable Models for the Third World, New York, Basic Books. Wooldridge, J. M. (2002). Econometric Analysis of Cross Section and Panel Data, Cambridge, Massachusetts, MIT Press.

36

Table 1. Descriptive statistics of the variables Symbol y k Gini NR Patent FDI Intermediate X Intermediate M Inflation Social Urban

Definition log of GDP per capita log of capital stock per capita Gini coefficient ratio of manufacturing exports to total exports number of patent applications per 1000 inhabitants ratio of FDI inflow to GDP ratio of intermediate exports to total exports ratio of intermediate imports to total imports inflation rate (annual change of consumer prices) ratio of social expenditure to GDP urban dummy

Obs. Mean Std.Dev. 344 8.504 0.673 344 10.168 0.589 265 0.509 0.046 324 0.662 0.242 218 0.087 0.062 338 0.033 0.025 315 0.046 0.076 319 0.117 0.064 341 0.093 0.118 300 0.121 0.056 265 0.117 0.322

Min 7.004 8.873 0.413 0.110 0.001 -0.025 0.000 0.014 -0.012 0.025 0.000

Source: See notes 8–16 for details. Note: Variables having fewer than 344 observations include missing samples.

37

Max 9.595 11.229 0.619 0.986 0.268 0.149 0.439 0.314 0.999 0.263 1.000

Table 2. Correlation matrix of the variables y y k Gini NR Patent FDI Intermediate X Intermediate M Inflation Social Urban

k 1.000 0.939 -0.263 -0.354 0.713 0.134 0.362 0.485 -0.079 0.606 0.265

Gini 1.000 -0.274 -0.339 0.722 0.076 0.314 0.468 0.017 0.646 0.337

1.000 0.075 -0.375 -0.104 0.012 0.055 0.110 -0.172 -0.536

NR

1.000 -0.300 0.071 -0.825 -0.710 0.058 -0.353 -0.054

Patent

1.000 0.197 0.352 0.353 -0.104 0.559 0.415

FDI

1.000 0.060 -0.131 -0.186 0.127 -0.109

Intermediate X Intermediate M Inflation Social

1.000 0.785 -0.027 0.274 -0.086

1.000 -0.011 0.196 0.116

1.000 -0.068 0.097

Source: Authors’ calculations based on the sources presented in notes 8–16.

38

1.000 0.066

Urban

1.000

Table 3. Expected signs of explanatory variables

k Gini y y2 NR Patent GVC Inflation Social Urban

Dependent variables (1) (2) y Gini + – – –

+ + + (short term), – (long term)

+ +

–

–

+ –

+ or –

Source: Authors’ elaboration. Note: + and – denote that the explanatory variable has increasing and decreasing effects on the dependent variable.

39

Table 4. Estimation results of equations (1) and (2) without interaction terms

k Gini

Dependent variables (1) y 0.814 *** (0.053) -0.841 *** (0.319)

(2) Gini

y y

-0.573 *** (0.153) 0.025 *** (0.009) 0.068 * (0.036) 0.169 *** (0.051) 0.198 ** (0.092) 0.589 *** (0.150) -0.031 (0.075) 0.010 (0.012) -0.306 *** (0.083) -0.006 * (0.004) 3.503 *** (0.659) 0.912 yes 136 16

2

NR Patent FDI Intermediate X Intermediate M Inflation

0.317 *** (0.112) 0.277 * (0.168) 0.653 ** (0.295) 1.780 *** (0.486) 0.391 (0.240) 0.017 (0.038)

Social Urban Cons R -squared Fixed effects Number of obs. Number of countries

0.327 (0.674) 0.993 yes 136 16

Note: Numbers in parentheses are standard errors. ***, **, and * respectively denote significance at the 1%, 5%, and 10% levels.

40

Table 5. Estimating equation (1) and (2) results of with interaction terms

k Gini

Dependent variables (1) y 0.763 *** (0.062) -1.180 *** (0.374)

y y2 NR Patent FDI Intermediate X Intermediate M Patent * FDI Patent * Intermediate X Patent * Intermediate M Inflation

0.352 *** (0.117) 0.228 (0.566) 0.719 (0.593) 2.123 *** (0.539) 0.242 (0.348) -0.029 (4.593) -2.367 (2.423) 2.072 (3.926) 0.031 (0.038)

Social NR * Social Urban Cons R -squared Fixed effects Number of obs. Number of countries

0.986 (0.781) 0.994 yes 136 16

(2) Gini

-0.768 *** (0.176) 0.036 *** (0.010) 0.174 *** (0.042) 0.694 *** (0.129) 0.619 *** (0.155) 0.704 *** (0.170) 0.137 (0.098) -3.712 *** (1.204) 0.490 (0.670) -3.114 *** (0.974) 0.011 (0.011) 0.341 ** (0.147) -0.697 *** (0.219) 0.000 (0.004) 4.310 *** (0.748) 0.926 yes 136 16

Note: Numbers in parentheses are standard errors. ***, **, and * respectively denote significance at the 1%, 5%, and 10% levels. 41

Figure 1.

42

Figure 2.

43

Figure Captions 1. Figure 1. Triangular relations amongst peripherality, income inequality, and per-capita income. Source: Authors’ elaboration. Note: Arrows indicate that the variable in question affects the other variable in the particular directions seen here. + and – respectively denote that the variable has increasing and decreasing effects. 2. Figure 2. Fitted values of the Gini coefficients. Note: Fitted values are calculated from the estimated coefficients in Table 4.

44