Developing Equity or Structuring Inequality?:

Toward a Political Economy of Income Inequality[1]

 

Matthew C Mahutga

Department of Sociology

UC Riverside

 

Garrett Grainger

Department of Psychology

University of Central Florida

 

Andrew K Jorgenson

Department of Sociology

University of Utah

 

Roy Kwon

Department of Sociology

UC Riverside

Institute for Research on World-Systems

University of California-Riverside

IROWS Working Paper # 55 is available at https://irows.ucr.edu/papers/irows54/irows54.htm

v. 11/17/09

 

 

 

Abstract

Beginning with the seminal work of Kuznets (1955), social scientists have struggled to understand the observed cross-national variation in income inequality.  Conventional wisdom suggests that several factors internal to nation-states explain this variation, including domestic institutions (Esping-Anderson 1990; Lee 2005) and a country’s stage in the process of economic development (Kuznets, 1955). A substantial body of research also demonstrates that other factors, including levels of foreign investment and country’s overall role in an interdependent world-system, also explain variation in income inequality (Mahutga and Bandelj 2008; Bornschier and Chase 1985; Nolan 1983).  This paper reports preliminary results from an analysis designed to determine the most important among the many proposed causes of variation in income inequality between countries and over time.   

 

Introduction

Interest in income inequality is growing across the social sciences and sociology in particular (Morris and Western 1999).  The “big three” sociology journals have all published studies of income inequality, including studies of between country inequality (Firebaugh and Goesling 2004), studies of within country inequality (Nielsen 1995; Alderson and Nielsen 1999; 2002; Lee 2005; Lee et al. 2007) and studies that attempt to assess global inequality as the sum of both (Korzeiniwicz and Moran 1997; Firebaugh 1999; 2000; Goesling 2001).  Of these three types of studies, those purporting to explain cross-national variation in within country inequality dominate the literature.

            There are two primary approaches to the study of within-country income inequality in sociology.  The first approach focuses on the domestic attributes of a given country, including its overall level of development and the institutional structures within it.  The second approach focuses on a country’s relationships to other countries, or its overall position in an interdependent international division of labor.  Moreover, these competing perspectives are rarely treated in juxtaposition (c.f. Alderson and Nielsen 1999; Lee et al. 2007), nor have there been any empirical attempts to understand which among these causal narratives has the most explanatory power.  In what follows, we discuss these approaches to income inequality.  More importantly, we report the preliminary results of a study designed to determine which among these many factors best explains observed differences in income inequality between countries. 

Income Inequality and Economic Development     

Among studies focusing upon internal attributes, the “internal development model,” initially proposed by Francois Nielsen, seems to have become the standard baseline income inequality model.  Taking the classical observations of Kuznets as a point of departure, the internal development approach models cross-national variation in income inequality as a function of a country’s stage of economic development.  Inequality is thought to be low at the lowest levels of economic development when a significant share of the labor force remains in the agriculture sector.  Inequality then rises over the course of development as segments of the labor forces shift out of agriculture and into manufacturing, creating inequality between the agricultural and manufacturing sectors.  However, this upswing in income inequality declines at higher stages of development, when the majority of the labor force is in the manufacturing sector.  This basic narrative of sectoral labor shifts is thought to explain the widely observed curvilinear cross-sectional association between income inequality and development (usually measured as GDP per capita).  Moreover, previous research suggests that the percent of the labor force in agriculture (LABAG) has a negative association with income inequality, while sector dualism (DUAL)—the absolute value of the difference between the percent of the labor force in agriculture and the share of GDP in agriculture—has a positive association. 

            In addition to sectoral labor shifts, Nielsen (1994) emphasized two other mechanisms to explain the relationship between income inequality and development.  The first of these was “generalized socio-cultural dualism” reflected by the uneven spread of the demographic transition.  Generalized socio-cultural dualism refers to the uneven diffusion of personal productivity enhancing cultural and technological aspects of modernization among a country’s population, and is measured as the natural rate of population increase (POPGROW).  The second mechanism was the spread of education.  The spread of education is thought to reduce income inequality by reducing the wage premium to skilled workers, and is typically measured as the percent of the relevant school aged population enrolled in secondary education (2ED).

 

Instituting Income Inequality: Democracy and the State

Another powerful explanation for cross-national variation in income inequality is institutional differences.  Of course one of the largest and well-explicated literatures is the welfare states literature, which focuses on developed, typically OECD countries.  Of particular importance in this literature is the relative extent to which the state is involved in redistributing wealth via social security transfers, decommodification and other fiscal policies (e.g. Alderson and Nielsen 2002; Bradley et al. 2003; Esping-Anderson 1990).  States that have comparatively progressive taxation policies, take on a larger share of the health care and other social reproduction burdens, and have comparatively wider and deeper safety nets also have lower income and wage inequality within them.  Moreover, these institutional differences are often important enough that they can literally reverse the direction of the causal relationship between certain social processes and income inequality across developed countries with different institutional configurations (Rueda and Pontusson 2000). 

            While the theoretical literature implicating institutions in the explanation of cross-national income inequality differences is most rich for developed OECD countries, research increasingly address the institution – inequality link for countries spanning the whole developmental continuum.  Relative levels of democracy and state intervention have been heavily studied, with mixed results.  Some find a straightforward negative association between democracy and income inequality across countries that is premised on the logic that democratic countries give more opportunity for poorer groups of citizens to impact policy construction and thereby redistribute wealth (Cutright 1967; Hewitt 1977; Jackman 1974; Muller 1985; 1988).  One scholar suggests that the relationship to income inequality is not this linear, and that early stages of democracy might give policy access to elites while later stages of democracy grant access to lower classes, thereby creating a curvilinear association between level of democracy and income inequality (Simpson 1990).  Possibly indicative of the lack of a robust relationship between democracy and inequality, however, some studies find no systematic relationship between the two variables (Bollen and Jackman 1985; Weede 1989), or suggest that the causal arrow may run in the opposite direction (Solt 2008).  

            Turning to issues of the state, there is fairly extensive case study literature documenting the role that the developmental state played in comparatively low levels of income inequality in East Asia.  East Asian states (to varying degrees across countries within the region) manufactured lower levels of income inequality by redistributing land ownership in order to increase productivity in the agriculture sector, which was then used as an engine for industrialization (Kay 2002).  Moreover, the East Asian developmental state pursued developmental policies of “midwifery” (Evans 1995) that included the selective use of both export incentives and import restrictions in industries they determined were key to rapid industrialization (e.g. Haggard 1989; Wood 1997).  Despite this rich and growing comparative historical and case study literature, state characteristics have only recently entered into studies of income inequality in large panels of countries, which is most likely due to the paucity of reliable and comparable data on state characteristics for poor countries (eg. Evans and Rauch 1999). Still, two aspects of the state have been shown to be associated with income inequality—the size of the public sector and the percent of GDP accounted for by government spending. 

Lee (2005) argues that the size of the public sector—measured as government tax revenue over GDP—increases income inequality from low to moderate levels, but then decreases it over the transition from moderate to large public sectors.  The mechanism involved is that states with smaller public sectors are primarily “growth-oriented regimes,” while those with larger public sectors are “equity oriented regimes.”  As a consequence, “growth oriented regimes translate larger public sector size into higher inequality, whereas equity-oriented regimes translate bigger public sector size into lower inequality” (Lee 2005: 159).  Moreover, Lee combines this with a theory of the role of democratic institutions to argue that “in regimes wherein democracy is not fully developed, increasing public sector size will be positively associated with income inequality.  On the other hand, in institutionalized democracies, inequality will decrease with larger government size” (159). 

Lee demonstrates the veracity of these claims by showing that public sector size has a curvilinear association with income inequality, and that public sector size decreases income inequality more in fully democratic regimes.  In a forthcoming study on the determinants to income inequality during post-socialist transition in Central and Eastern Europe, Bandelj and Mahutga note that the transition away from socialism has been coupled with the retrenchment of relatively egalitarian government spending in these countries, and find that countries that maintain relatively high levels of government spending also have lower levels of income inequality (Bandelj and Mahutga Forthcoming).  While this study was limited to CEE countries, a similar argument can be made with respect to the diffusion of neo-liberal policy scripts at the global level.  Indeed, as many states pear back welfare state programs, one would expect levels of income inequality to rise within them.  This argument has been shown to explain the historically unique inequality upswing in developed countries—the great “U turn”—since the late 1970s (Alderson and Nielsen 2002).  Thus, we suggest that the inequality reducing effect of public spending should also matter for countries across the developmental spectrum.   

World-System Position and Foreign Direct Investment

Next to the internal development model, the second most widely studied model of income inequality includes variables related to a country’s relationship to the whole world-economy.  While the empirical approaches used to characterize a country’s relationship to the whole world-economy are numerous and varied, we focus on two here because of the ubiquity with which they appear in the literature.  The primary way in which a country’s relationship to the world-economy is conceived is via its overall position in an interdependent international division of labor, or world-system (e.g. Wallerstein 1989).  This interdependent international division of labor is seen as hierarchically organized, with the most dominant states located in the core, the most dominated states located in the periphery, and states that tend to both dominate those below them and be dominated by those above them are located in an intermediate semiperiphery. 

From the world-systems perspective, a “country's world-system position, in a macro-structural sense, is considered the key determinant of the society's capacity for sustained …development,” including rates of economic growth, the structure of a country’s productive forces, state capacity and institutional structure, etc. (Crowly, Rauch, Seagrove and Smith 1998:32; also see Evans 1979; Snyder and Kick 1979).  Just as a country’s level of development is determined by its position in the world-system, so to is its level of income inequality.  As Bornschier and Chase-Dunn note, “it is not a low level of development that creates high inequality, but rather peripherality in the world division of labor…” (1985: 23).  Yet, most of the mechanisms through which income inequality differences across zones of the world-system arise work through developmental dynamics. 

One approach to the explanation for why core countries have lower income inequality than do non-core countries is differences in the structure of the economy across zones of the world-system.  Core countries are seen as the containers of the industries, production processes and innovations that are at the leading edge of the capitalist world-economy.  While the concrete forms of these economic activities change over time—i.e. mechanized textile manufacturing was once at the leading edge but has not become “peripheralized”—the key point is that the leading activities remain within the boundaries of the core of the world-system (e.g. Arrighi and Drangel 1984; Chase-Dunn and Grimes 1995).  The counterpart to this historical dynamic, of course, is that the activities contained within the boundaries of the periphery of the world-system are the least advanced.  As a result, “…organizational forms of production are much more complex [in the core]…whereas in the periphery standardized and routinized production prevails….and this is accompanied by a larger marginalized segment of the population” and higher inequality (Bornschier and Chase-Dunn 1985: 127-8).         

Another key explanation for income inequality differences between zones of the world-system is alliances between ruling classes in core and non-core countries.  While core countries tended to experience rising levels of participatory democracy over the course of economic development, some world-system scholars believe that elites in peripheral countries are able to resist demands for the political integration of lower classes, and thereby “successfully resist demands for redistribution” (Bornschier and Chase-Dunn 1985: 23; also see Rubinson 1976; Rubinson and Quinlan 1977).  Similarly, world-system scholars argue that the resulting class structure in peripheral countries works against the development of the kinds of egalitarian state structures that developed in core countries.  Moreover, successful efforts to redistribute that do emerge in peripheral countries may be shallow or transitory because of the inherent weakness of peripheral states, both internal and vis-à-vis international intuitions and stronger states in the core (McMichael 1996; Rubinson 1976).  This external weakness may be exacerbated by internal weakness, since the fragmented class structure of peripheral states results in a lack of domestic legitimacy, making them politically fragile and prone to instability (Chase-Dunn 1998). 

Empirical results for the relationship between world-system position and income inequality have been fragile.  Part of this has to do with the great number of way sit has been operationlized, but we do not discuss that here (see Lloyd et al. 2009; Mahutga 2006; Snyder and Kick 1979; Bollen and Appold 1983; Babones 2009; Arrighi and Drangel 1984; Rubinson 1977).  Instead, we review a number of findings.  Studies that support the world-system approach to income inequality find that income inequality tends to bear a negative linear association to world-system position—core countries have the lowest inequality, followed by semiperipheral and peripheral countries, as expected.  Sometimes, however, significant differences are found only between core and non-core countries (i.e. the semiperiphery is not significantly lower than the periphery Rubinson 1977; Rubinson and Quinlan, 1977; Nolan 1983).  Yet, an equal number of studies find that the results of measures of world-system position are mixed or become null after controlling for level of economic development or the internal development model (Alderson and Nielsen 1999; Weede 1980; Lee 2005; Lee et al. 2007).  This latter result could be interpreted in a number of ways.  For one, since many of the mechanisms through which world-system dynamics impact inequality work through development, it could be that including variables capturing differential development literally explain observed differences across zones.  Yet, if the causal sequence is world-system position à development à inequality, it could be that including a slew of development variables results in a significant amount of colinearity, and a simpler world-system model might account for a number of developmental processes.  We return to this discussion below and will pursue it further as this project progresses.

Apart from measures of a country’s overall position in the world-system, one of the standard concepts included in world-economy models is foreign investment dependence (PEN).  Initially framed as an indicator that captured inter-country power differentials arising from the structure of the world-system, this literature argues that the penetration of less developed (read peripheral) countries by foreign (read core) capital increased income inequality in the host economies (e.g. Beer and Boswell 2002; Bornscheir, Chase-Dunn and Rubinson 1977; Bornschier and Ballmer-Cao 1979).  Indeed, the impact of PEN is not limited to income inequality, but is also purported to affect development outcomes such as economic growth, urbanization and the composition of the labor market (Bornscheir and Chase-Dunn 1985; Dixon and Boswell 1996; Evans and Timberlake 1980; Kentor 1998, 2001).  Further, a growing body of literature suggests that PEN increases rates of many different type of environment pollution in host economies (Jorgenson 2006a, 2006b; Jorgenson and Kick 2006; Kentor and Grimes 2006).

Early thinking on the mechanisms involved in the generation of inequality provided two different explanations involving income differentials between foreign and domestic sectors.  In one scenario, foreign capital creates capital-intensive production processes that are concentrated in smaller, outward oriented foreign sectors (Frank 1967, Stack 1980).  In other words, foreign sectors are more productive than domestic sectors, such that the income differential between these sectors arises out of the higher productivity of the former vis-à-vis the latter.  On the other hand, there is also a large literature on the growth effects of FDI, which tend to show that countries dependent on FDI have slower growth than do countries with less dependency (e.g. Bornschier and Chase-Dunn 1985; Dixon and Boswell 1996; Kentor 1998; Kentor and Boswell 2003; cf. Firebaugh 1992; de Soysa and Oneal 1999).  These scholars suggest that PEN constrains the host economy’s forces of production to low wage and unskilled manufacturing, and leads to very little beneficial “spill-over” effects such as research and development activities, industrial services or backward and forward linkages with domestic suppliers or distributors (Hirschman 1945; Galtung 1971; Bornschier et al. 1979; Bornschier and Chase-Dunn 1985).  As such, this literature is consistent with the notion that foreign firms could actually pay less than domestic firms so that the income differential between sectors arises out of the lower productivity of the foreign vis-à-vis the domestic one. 

Contemporary research tends to suggest that the first scenario is a more viable explanation for the effects of FDI on income inequality (Aitken, Harrison and Lipsey 1996; ILO 1998; Moran 2002).  Indeed, evidence suggests that foreign firms are more productive because of their greater capital and scale intensity (Aitken, Harrison and Lipsey 1996; c.f. Lipsey and Sjoholm 2001), which increases the demand for less abundant skilled relative to more abundant unskilled labor (Feenstra and Hanson 1997).  Moreover, scholars also believe that PEN constrains the production of human capital within host economies, which further exacerbates income differential between foreign and domestic sectors because of the greater accumulation of human capital in the former (Bornschier and Ballmer-Cao 1979; Bornschier and Chase-Dunn 1985). 

Apart from mechanism that evoke inter-sector productivity differences, scholars also suggest that PEN influences the distributive capacity of nation-states by producing a “race to the bottom” in which governments attract FDI through policies that lower the bargaining power of labor, eliminate provisions that encourage full employment and wage enhancement, such as job training, local purchasing requirements, minimum wages and regulated cost of living increases, and thus remove institutional constraints on rising income inequality (McMichael 1996; DeMartino 1998; Ranney 1998; Beer and Boswell 2002).  In short, some scholars believe that domestic institutions become less egalitarian than they would otherwise be in the absence of FDI.

            While the PEN literature was initially conceptualized as a mechanism responsible for reproducing world-system inequalities—e.g. slowing growth and raising inequality in non-core countries—it has now become a topic of interest in and of itself.  Indeed, it is an unanswered empirical question of whether or not FDI has different effects in different zones of the world-system.  If a straightforward world-systems understanding of FDI were correct, one would expect FDI to either decrease inequality or have no affect on it in the core, but to increase it in the periphery.  On the other hand, if FDI dynamics operate independently from world-system dynamics, one would expect it to have comparable effects across world-system zones.  While we hope to explore this issue further in future renditions of this project, for now we will conceptualize PEN as a part of the “external,” or “world-economy” approach to income inequality and include it along with world-system position as a way to compare and contrast this approach with the development and institutional ones discussed above.   

 

Our Approach

With the exception of one or two excellent studies (Alderson and Nielsen 1999; Lee et al. 2007), there is a paucity of research designed to understand the relative importance of “internal” and “external” factors to cross-national variation in income inequality.  The circumscribed goal of the present study is to engage in just such an empirical evaluation, and we believe two contemporary developments have made this more possible now then ever.  First, there is now much more income inequality data available now than in previous years.  Second, there are much better measures of world-system position that cover a wider number of countries and vary over time.  Third, the econometric literature has advanced to a stage where a happy balance can be forged between taking protections to avoid bias coefficient estimates and maintain as much identifying variation on both sides of the equation, a discussion we turn to below in our methodological section. 

              Before discussing our data and methods, we want to map out the empirical logic that follows.  Again, or goal is to determine which among the three causal narratives outlined above has the most explanatory power for cross-national income inequality differences.  As a consequence, we estimate four models.  The first three estimate the development, institutional and world-economy models independently, while the latter one includes all variables in one final model. 

When taken together, the internal development model is as follows:

(1)                 gini = α + β1LABAG + β2DUALISM + β3POPGROW + β42ED + ε 

where gini is the Gini coefficient of a given country and ε is an error term that will remain undefined for now (see Lee et al. 2007 and Mahutga and Bandelj 2008 for recent applications). 

The institutional model includes the processes discussed above: democracy (DEMOC), size of the public sector (PUBSEC) and government spending (GVTSPND).  This model is equal to

(2) gini =  α + β1DEMOC + β2PUBSEC + β3GVTSPND + ε

where gini and ε are the same as above. 

Finally, the world-economy model will include a measure of world-system position (Babones 2009) along with a model of foreign capital penetration.  A straightforward approach would be to simply include a typical measure of PEN: the stock of FDI / GDP.  However, while many studies in the world systems / dependency literature report the deleterious effect of PEN on economic growth and income equality (Bornschier and Ballmer-Cao 1979; Bornschier and Chase-Dunn 1985; Bornschier, Chase-Dunn, and Rubinson 1978; Evans and Timberlake 1980; Tsai 1995; Dixon and Boswell 1996; Kentor 1998; Alderson and Nielsen 1999), early studies were subject to a major critique by Firebaugh (1992).  In the context of economic growth, PEN researches estimated models that contained both FDI flow (the yearly inflow) and FDI stock (the cumulated inflow) in the same equation, and interpreted a positive coefficient on the former as a short-term beneficial effect, and a negative coefficient on the latter as a long term deleterious one. 

Firebaugh countered that the negative sign on the stock variable could also be interpreted as a beneficial effect of the foreign investment rate: the faster the yearly inflow of foreign investment, the faster the economic growth.  This argument stems from the fact that the foreign investment rate is calculated as the ratio of flow/stock, so that including both flow and stock as separate regressors is equivalent to estimating the effect of the numerator while holding the denominator constant in the case of flow, and vise-a-versa in the case of stock.  Thus, “a negative effect of foreign investment stock in such a model would indicate a positive effect of the foreign investment rate on growth because, keeping flow constant, the larger the stock the smaller the rate” (Alderson and Nielsen, 1999: 612).  Firebaugh concludes by arguing that the correct interpretation for the findings from PEN studies is that the foreign investment rate (the ratio of flow/stock) benefits developing countries, but that it doesn’t benefit them as much as domestic investment.

We should note that Firebaugh’s main critique was directed at interpretations of the effects of foreign investment on economic growth rather than on income inequality. Nevertheless, Firebaugh (1992) argues that the logic holds in the context of “non-economic” outcomes (pp. 123-124), and as Alderson and Nielsen (1999) emphasized, studies measuring the effect of investment dependence on inequality should also avoid misinterpretation due to denominator effects.  Thus, we follow previous analyses (Dixon and Boswell 1996; Alderson and Nielsen 1999) and use a model of FDI that includes both the rate of FDI as well as domestic investment to remove any spurious interpretation of FDI penetration.  Our world-economy model is as follows:

(3) gini = α + β1SEMIP + β2PER + β3FDIPEN + β4FDIRATE + β5DOMESTIC + ε

where gini and ε are the same as above, SEMIP = semiperiphery, PER = periphery, FDIPEN = FDI stock / GDP, FDI rate = FDI flow / FDI stock and DOMESTIC = Domestic Capital Formation / GDP. 

Data and Method

            Dependent and Independent Variables

The data that underlie the following analyses are correlated in Table 1, while their means, standard deviations, computation and sources are reported in Table 2.  While we will discuss these data further in future drafts, we want to point out that the main dependent variable allows us to analyze from 3 to 5 times more cases than the state of the art inequality study in sociology (e.g. Alderson and Nilesen 1999; Lee 2005).            

Table 1: Correlation Coefficients for Variables Included in Analysis (except year). 

 

 

 

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

1

GINI

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2

WS Position

-0.569

 

 

 

 

 

 

 

 

 

 

 

 

 

 

3

FDI Penetration

0.222

0.051

 

 

 

 

 

 

 

 

 

 

 

 

 

4

FDI Rate

-0.102

-0.008

-0.191

 

 

 

 

 

 

 

 

 

 

 

 

5

Domestic Investment

-0.097

0.090

0.022

0.111

 

 

 

 

 

 

 

 

 

 

 

6

GDP pc

-0.601

0.910

0.113

0.033

0.122

 

 

 

 

 

 

 

 

 

 

7

GDP pc sq

-0.613

0.917

0.107

0.027

0.112

0.998

 

 

 

 

 

 

 

 

 

8

Secondary Education

-0.641

0.724

0.146

0.088

0.043

0.822

0.824

 

 

 

 

 

 

 

 

9

Dualism

0.542

-0.707

-0.110

-0.075

-0.048

-0.789

-0.785

-0.726

 

 

 

 

 

 

 

10

% Labor Force in Agriculture

0.526

-0.796

-0.193

-0.071

-0.117

-0.898

-0.889

-0.794

0.937

 

 

 

 

 

 

11

Natural Rate of Pop Increase

0.705

-0.661

-0.002

-0.134

-0.053

-0.758

-0.761

-0.802

0.673

0.719

 

 

 

 

 

12

Democracy

-0.429

0.647

-0.042

0.028

-0.051

0.669

0.672

0.617

-0.580

-0.625

-0.600

 

 

 

 

13

Public Sector Size

-0.418

0.521

0.199

0.019

0.106

0.546

0.550

0.546

-0.474

-0.548

-0.584

0.414

 

 

 

14

Government Spending

-0.325

0.438

0.123

-0.030

0.059

0.416

0.428

0.453

-0.347

-0.424

-0.411

0.306

0.672

 

 

15

Unexplained Unit Effects

0.480

0.015

0.189

-0.080

-0.111

0.016

0.015

0.015

-0.015

-0.018

-0.009

0.019

0.018

0.013

 

16

year

0.061

-0.144

0.421

0.023

-0.120

0.013

0.014

0.245

-0.053

-0.076

-0.190

0.094

0.021

-0.043

0.188


Fixed Effects Vector Decomposition Model

The data we analyze below consist of a pooled cross-section of time series.  Time series cross-section regression models require repeated observations of a given unit over time.  The repeated cross-sections are pooled, such that each unit’s repeated observations of a random variable appears as a unique observation in the dataset.  In the context of ordinary least squares (OLS), the following notation captures the nature of the data:

(4) 

 where i indexes the country, t indexes the time period of observation, α is the classic intercept, and x is a vector of observed covariates that vary across observations and / or over time, and ε is the classic error term.   Thus, 4 yields the pooled OLS estimator of y.  ε may contain information on random variation, in which case 4 is unbiased and efficient because cov(ε, x) = 0, but it may also contain non-random error related to unobserved parameters (θ) that do not enter the model.  This non-random error may vary over time, but not over units (period effects), over units but not over time (time-invariant unit effects) or over both units and time (disturbances).  If we call the first type of unobservable θp, the second type θz and the third type, θo then estimates obtained from 4 are biased if Cov(ε, θ)0 for any θp, θz, or θo.  Bias arising from omitted θo is the classic omitted variable bias for time-varying covariates, and time-series cross-section models contain no special power to account for the omission of these types of variables.  However, pooling time varying observations does allow one to account for θp and θz.

Two common approaches for dealing with these omitted variables are the random effects (REM) and fixed effects (FEM) models.  The REM model is given by

(5) 

In 5, ui is an extra component of the error variance that is specific to each unit in the model, with the assumption that that the unit specific portion of the error term is uncorrelated with the observed covariates.  The consistency of y in 5 hinges on the validity of the assumption that cov(u, x) = 0 for all x.      

The FEM model exploits the variation within cases to account for the unmeasured time invariant unit-specific heterogeneity.  The FEM estimator is obtained by including a set of unit specific intercepts in 4

(6)     

where z is a dummy variable for unit i.

            Choosing between REM and FEM approaches hinges crucially on tests for whether or not the REM assumption of uncorrelated unit effects holds.  If it does, REM estimators are typically more efficient than FEM estimators unless the ratio of units to time periods is significantly less than 1.  If it does not hold, RE estimates of β are biased.  However, a draw back of FEM models is that they cannot identify parameters on time-invariant or nearly invariant variables, which are perfectly or near perfectly correlated with z in 6.  Moreover, FEM give up a significant amount of efficiency in cases where most of the identifying variation resides between rather than within cases, since they effectively remove all between case variation.

            Previous work on inequality has therefore modeled the outcome using REM models, even when the assumption of uncorrelated unit effects was violated (Alderson and Neilsen 1999; Lee 2005; Lee et al. 2007, etc.)  While we concur that it is important to take measures to preserve identifying variation between cases, and to retain the ability to model time invariant or nearly invariant variables, we believe there are now preferable ways of tackling these issues that do not presuppose the violation of the crucial REM assumption.

            Thus, in the models that follow, we employ a fixed effects vector decomposition model (FEDM).  The FEDM is designed to deal with the problem of time invariant or slowly changing variables when the assumption of the REM is not met.  In practice, the FEDM model proceeds in three stages, similar to a three-staged least square (3SLS).  In the first stage, a baseline model is estimated including the fixed effects.  The first stage excludes the time invariant or nearly invariant variables, and ends when the unit effects are estimated and saved for the second stage.  In the second stage, the unit effects are regressed on the time invariant or nearly invariant variables, and the residual values are saved for the third stage.  In the third stage, the dependent variable is regressed on any and all independent variables along with the residual from the second stage.  Substantively, these unit effects are interpreted as the part of the unit effects that is not a function of the time invariant or nearly invariant predictors.  Monte Carlo simulations have shown that the FEDM is preferable to the REM model when the assumption of uncorrelated unit effects is not met, and to FE models when the between case variation is sufficiently large relative to the within variance (Plumper and Troeger 2007)    

Table 2: Descriptive Statistics and Percent Variance Between Cases for Variables Included in Analysis

Variable

Source

Measurement

Mean

Std Dev

% SSE Between Cases

Stage 2

Income Inequality

 

 

 

 

 

 

SWIID Gini Coefficients

Solt (2009)

0 - 1, zero is perfect equality and 1 is perfect inequality

37.488

10.496

88***

N/A

World-System Position

 

 

 

 

 

 

Core, Semiperiphery and Periphery

Babones (2009)

Position in smooth income distribution

1.807

0.825

90.6***

X

Investment

 

 

 

 

 

 

FDI Penetration

World Investment Report (2008)

log 10(FDI Stock / GDP)

0.86

0.785

52.3***

---

FDI Rate

World Investment Report (2008)

(FDI Flow / FDI Stock)

0.192

0.507

5.1**

---

Domestic Investment

World Development Indicators (2009)

(Domestic Capital Formation / GDP)

22.354

9.023

42.9***

---

Internal Development Model

 

 

 

 

 

 

% Labor Force in Agriculture

FAOSTAT (2006) / World Development Indicators (2003)

Labor Force in Agricultre / Total Labor Force

44.506

29.593

92.4***

X

Sector Dualism

World Development Indicators (2003)

 

        p = % labor force in agriculture, L = share of GDP in agriculture

23.111

18.453

84.4***

X

Natural Rate of Pop Increase

World Development Indicators (2003)

Birth Rate - Death Rate

1.935

1.703

77.9***

X

Secondary Education Enrollment

World Bank (2002)

Secondary Education Pupils / Relavent School Aged Population

49.573

33.165

82.2***

X

GDP per capita (PPP)

Heston et al. (2002)

Local currency coverted to international PPP dollars

3.598

0.493

92.6***

X

GDP per capita (FX)

World Development Indicators (2009)

Local currency converted in int exchange rates

3.27

0.672

96.1***

X

Institutions

 

 

 

 

 

 

Democracy

Marshall and Jaggers (2000)

Index of 0 - 10, where 0 is least democratic and 10 is most

3.853

1.208

74.9***

X

Government Spending

World Development Indicators (2003)

Government Spending / GDP

1.164

0.185

65.9***

X

Public Sector Size

World Development Indicators (2003)

Government Tax Revenue / GDP

18.226

11.395

60.9***

X

Table 2 corroborates years of inequality research in the social science in the sense that the majority of variation in the dependent variable and all but three of the predictors resides between rather than within cases (see column 5).  Plumper and Troeger (2007) suggest that the key determinant for whether or not a given covariate should be introduced in the second stage of a FEVD model is the extent to which it does not vary over time, the temporal counter part to large between case variation relative to that within cases.  Thus, all of the independent variables noted with an X in column 6 of Table 2 were included in the second stage described above and are therefore independent of the unit effects that follow, by construction.  However, two of these variables (GOVSPND and PUBSEC) fall below the 2.8 between / within ratio suggested by Plumper and Troeger as a benchmark (2007: 136).  Still, because our subsequent analyses indicated that these were somewhat weak predictors of income inequality and we want to give the institutional argument the largest possible chance in explaining the variation in income inequality we observe, we also included these two variables in the second state.          

            Like any least squares approach, the validity of the FEVD estimator hinges on two additional assumptions—no serial or spatially contemporaneous auto-correlation in the error term. If it can be shown that serial correlation is an issue, then the data need to be adjusted for the appropriate type of serial correlation.  If it can be shown that spatial contemporaneous auto-correlation or groupwise heteroskedasticity is an issue, then this needs to be accounted for by panel corrected standard errors (PCSE).  Preliminary analysis showed that these data had both serial and spatial contemporaneous auto-correlation, and we therefore adjust for a first order (AR(1)) auto regressive process and we conduct inference with PCSE in both the first and third stage of the FEVD model (Plumper and Troeger 2007; Beck and Katz 1995; Prais and Winston 1954).

Results

Table 3: Unstandardized Coefficients from Serial and Spatially Contemporaneous Auto-Correlation Corrected Fixed Effects Vector Decomposition Regression of Income Inequality on Select Independent Variables, 1970-2001.

 

0

1

2

3

4

Internal Development

 

 

 

 

 

GDP pc a

---

40.509**

---

---

38.206*

 

 

(15.125)

 

 

(16.360)

GDP pc squared a

---

-6.501**

---

---

-6.040**

 

 

(1.992)

 

 

(2.150)

% of the labor force in agriculture a

---

-.155**

---

---

-.183***

 

 

(.043)

 

 

(.044)

Sector Dualism a

---

.206***

---

---

.232***

 

 

(.021)

 

 

(.025)

Secondary Education Enrollment a

---

-.036***

---

---

-.040***

 

 

(.006)

 

 

(.006)

Natural Rate of Population Increase a

---

.342***

---

---

.361***

 

 

(.045)

 

 

(.044)

Institutions

 

 

 

 

 

Democracy a

---

---

-.047+

---

.139***

 

 

 

(.027)

 

(.033)

Government Spending a

---

---

-.487

---

-.854

 

 

 

(.756)

 

(.715)

Size of the Public Sector a

---

---

-.017

---

-.043**

 

 

 

(.012)

 

(.014)

World-Economy

 

 

 

 

 

Semiperiphery a

---

---

---

-.339*

-.905***

 

 

 

 

(.174)

(.218)

Periphery a

---

---

---

-1.241***

-2.454***

 

 

 

 

(.314)

(.348)

Foreign Capital Penetration

---

---

---

.269*

.552***

 

 

 

 

(.125)

(.138)

Foreign Investment Rate

---

---

---

-.013

-.002

 

 

 

 

(.168)

(.186)

Domestic Investment

---

---

---

.002

-.005

 

 

 

 

(.009)

(.010)

Unit Effects

---

.415***

-.006

.037

.484***

 

 

(.045)

(.025)

(.026)

(.056)

Constant

41.546

-20.649

42.542***

43.597***

-16.210

 

 

(28.753)

(1.482)

(1.320)

(31.076)

R2

.573

.885

.556

.611

.904

 

 

 

 

 

 

BIC

4502.965

4867.84

4512.49

4384.94

4736.05

 

 

 

 

 

 

ρ

.996

.965

.996

.995

.959

 

 

 

 

 

 

Countries

107

107

107

106

106

Observations

1636

1636

1636

1580

1580

Notes: a variable included in second stage regressions.  Numbers in parentheses are panel corrected standard errors.  First order auto-regressive term denoted by ρ.  + p< .10; * p<.05; ** p<.01; *** p<.001. All coefficients are net of T-1 yearly time dummies that are not displayed. 

 

            Table 3 reports results of the third stage of the FEVD model described above.  Model 0 is a baseline model that includes only t-1 dummy variables for the time periods.  We suppressed the coefficients on those variables for ease of presentation, but it is interesting to note that the time dummies explain 57 percent of the variance in these AR(1) adjusted data.  More importantly, the BIC statistic allows us to compare the development, institutional and world-economy models as stand-alone explanations for the cross-national variation in income inequality we observe.  The BIC prefers parsimonious models to those with “extra” covariates, unless the fit of the model is significantly enhanced by the additional covariates.  Smaller values of BIC indicate a better fit relative to the number of parameters.  Raftery (1995: 139) outlines a general rule of thumb in which BIC decreases from 2-6 indicate “positive” evidence of improved fit; decreases from 6-10 indicate “strong” evidence of improved fit; while decreases greater than 10 indicate “very strong” evidence for improved fit.  We can compare the BIC statistic across the three configurations as a means of comparing the “goodness of fit” of each approach as stand alone models.

            Model 1 includes all of the variables from the internal development model.  All of the models are significant and in the expected direction.  The sign and significance of GDP per capita and its squared term indicate that income inequality rises with income until relatively high levels are reached, after which it falls.  Sector dualism has a positive affect on income inequality, and, controlling for dualism, the percent of the labor force in agriculture has a negative effect.  Likewise, secondary educational enrolment tends to reduce inequality, while the natural rate of population growth tends to increase it.  Despite all of the significant variables, the BIC actually shows a 365 point increase in size, suggesting that the internal development model does not significantly add to the fit of the model given the number of parameters.  This finding particularly surprised us, and it deserves future analysis.  Our tentative explanation is colinearity.  Indeed, the mean variance inflation factor (VIF) is a whopping 47.22 for this model, considerably above the acceptable benchmark of 10. 

            Model 2 estimates the effect of the institutional covariates.  While the variables are all signed in a manner that would be consistent with common sense expectations—active states and democracy lower inequality—only democracy is even marginally significant.  Moreover, the BIC statistics shows a slight increase compared to the baseline model, which probably has more to do with their lack of statistical power than it does with colinearity, since the mean VIF for this model is only 2.67.  The power of these variables is disappointing when compared to that of institutional variation among developed OECD countries.[2]  One potential explanation is that institutional variation is more important for countries at otherwise comparable levels of development. 

            Model 3 introduces the world-economy model.  Inequality’s relationship with world-system position appears to follow the negative linear pattern hypothesized by world-system scholars.  The difference in inequality between the core and semiperiphery is smaller than that between the core and periphery.  Foreign capital penetration has a positive effect on income inequality.  While neither the rate of FDI or domestic investment have significant effects, their inclusion is necessary to assuage concerns over spurious PEN effects.  It is perhaps interesting to note that the BIC for this model is significantly reduced vis-à-vis the baseline model, with a reduction of roughly 216 points, even though neither the rate of FDI nor domestic investment are significantly different from zero.  Contrary to the internal development model, the world-economy model produces the lowest mean VIF of any of those estimated: 1.72.  Tentatively, these results suggest that the world-economy model is the best stand along explanation for cross-national variation in income inequality, and may be a more parsimonious explication of development dynamics than are other development approaches. 

            How do these variables perform in a model that controls for all of them?  Model 4 introduces all three models into the equation, with some surprising results.  All of the variables for the internal development model remain significant in the expected direction, but two of them are slightly less significant.  The sign on democracy switches from negative to positive, and the size of the public sector now has a significant negative association with income inequality.  Clearly, the effect of institutional variation is sensitive to whether or not development and / or world-economy characteristics are controlled for.  On the other hand, the effects of the world-economy variables become more powerful when all else is controlled for.  Both the size of these effects and their ratio to their standard errors increases in the full model.  The BIC statistic is still very much significantly larger than the baseline model, but it is significantly smaller than the internal development model even though the mean VIF for this model is an alarming 91.23.  This suggests two tentative conjectures.  First, the world-economy model is a powerful developmental narrative vis-à-vis the internal development model, as indicated by the improvement of this model over model 1.  Second, including both world-economy and the internal development model results in a significant amount of redundancy, indicated by the high VIF of model 4.  Future work might attempt to decompose the internal development model into redundant and unique sources of variation and proceed accordingly. 

Conclusion

            Thus far we have articulated three very different models of cross-national income inequality.  The internal development model relates cross-national differences in income inequality to a nomothetic developmental path along which each country is bound to progress.  As countries develop from agrarian to industrial societies, inequality is expected to rise and then fall, in turn driven by labor’s shift out of agriculture, a growingly even distribution of educational attainment and the demographic transition.  The institutional model sees cross-national differences in income inequality as the result of conscious political activity on the part of domestic and democratic institutions, resulting in progressive taxation and economically active states that reduce income inequality.  The world-economy model suggests that both development and institutional formation are structured by the positions countries occupy in the larger world-system, and that FDI could have an independent inequality increasing effect.

Figure 1: Association between World-System Position (Babones 2009) and Internal Development / Institutional Covariates. 


Notes: Bars represent average level of development / institutional covariate among countries in each world-system zone. 

 

            The evidence we presented thus far suggests that development does indeed matter—all of the covariates from the internal development model are significant and in the expected direction.  However, there appears to be a significant degree of redundancy in this narrative, and development may be best conceptualized as an intervening dynamic in the larger social-structural determinants of income inequality.  Indeed, as Figure 1 shows, the variation across world-system zones of each of the development and institutional covariates is consistent with the notion that both development and institutional formation is largely structured by a country’s position in the larger world-economy.  The potential primacy of the structural account of development is also reflected both in the pattern of the coefficients presented above and in the associations between the internal development covariates and world-system position displayed in Table 1.  In every case, the association between internal development and world-system position is larger than that between the former and income inequality.  On the other hand, institutions do not seem to matter as much across countries at radically different levels of economic development or positions in the larger world-economy, suggesting again that institutional variation across countries along the continuum from core to periphery may be larger a function of that continuum.  Moreover, controlling for development and world-economy dynamics, democracy may actually increase income inequality. 

            Still, these results and their associated generalizations can only be tentative at best, and they raise a number of additional questions.  For example, which among the internal development covariates best explains income inequality, especially in combination with variables that largely determine development?  One might expect that GDP per capita would have different effects on income inequality conditional on world-system zone.  If the processes outlined by the internal development model are correct, we would expect it to have a positive association with income inequality in the periphery, since these countries are at the lowest levels of development and should therefore line up on the ascending slope of the inverted U.  We would also expect it to have a curvilinear association with inequality among semipeirpheral countries, since this zone contains a relative mix of more “core like” and “peripheral like” countries that straddle the apex of the inverted U.  Further, it would also suggest that it has a negative relationship to inequality among core countries, since these are the most developed countries that span the descending slope of the inverted U. 

Figure 2: Association between GDP per capita and Income Inequality by World-System Zone (Babones 2009)

 

 

Notes: 1 = Periphery, 2 = semiperiphery and 3 = core.  Predictions are fit with locally weighted least square (LOWESS). 

 

However, we do not find this pattern.  Instead, as Figure 2 shows, we find that GDP per capita has a negative association with income inequality in all but the semiperiphery, in which it does have a curvilinear association.  Given that this is entirely inconsistent with a straightforward extension of the internal development model, and world-system position is highly correlated with GDP per capita, this could suggest that “peripheral, semiperipheral and core positions explain the observed pattern between level of economic development and income inequality within countries (Bornschier and Chase-Dunn 1985: 128).  While we feel the panoply of evidence presented thus far is at least suggestive of the explanatory power of the world-economy model vis-à-vis the internal development model, we do believe that more research is necessary. 

 


Bibliography

 

Alderson, Arthur and Francios Nielsen. (2002) “Globalization and the Great U-Turn: Income Inequality Trends in 16 OECD Countries.”  American Journal of Sociology 107: 1244-1299.

 

_______.  (1999).  “Inequality, Development and Dependence: A Reconsideration.” American Sociological Review 64: 606-631.

 

Aitken, Brian, Ann Harrison and Robert E. Lipsey.  1996.  Wages and foreign ownership: A Comparative Study of Mexico, Venezuela, and the United States.  Journal of International Economics 40(3-4): 345-71.

 

Arrighi, Giovanni and Jessica Drangel.  1986.  “The Stratification of the World-Economy: An Exploration of the Semiperipheral Zone.”  Review 10(1): 9-74. 

 

Babones, Salvatore.  2009.  The International Structure of Income: Its Implications for Economic Growth.  Berlin: VDM. 

 

Bandelj, Nina. (2008) From Communists to Foreign Capitalists: The Social Foundations of Foreign Direct Investment in Postsocialist Europe. Princeton, NJ: Princeton University Press.

 

Bandelj, Nina and Matthew C. Mahutga.  2010.  How Socio-Economic Change Shapes Income Inequality in Central and Eastern Europe.”  Social Forces Forthcoming

 

Beck, Nathaniel and Jonathon N. Katz.  1995.  “What to Do (and not to Do) With Time Series Cross Section Data.”  American Political Science Review 89(3): 634-47.

 

Beer, Linda and Terry Boswell. (2002) “The Resilience of Dependency Effects in Explaining Income Inequality in the Global Economy: A Cross-National Analysis, 1975-1995.” Journal of World-Systems Research 8(1):30-59.

 

Bollen, Kenneth and Robert Jackman. (1985) “Political Democracy and the Size Distribution of Income.” American Sociological Review 50: 438-457.

 

Bollen, Kenneth A. and Stephen J. Appold. 1993. "National Industrial Structure and the Global System." American Sociological Review 58:283-301.

 

Bornschier, Volker and Ballmer-Cao, Thanh-Huyen. (1979) “Income Inequality: A Cross National Study of the Relationships Between MNC-Penetration, Dimensions of the Power Structure and Income Distribution.” American Sociological Review 44:438-506.

 

Bornschier Volker and Christopher Chase-Dunn. (1985) Transnational Corporations and Underdevelopment.  New York: Praeger.

 

Bornschier, Volker, Christopher Chase-Dunn and Richard Rubinson.  (1978)  “Cross-national Evidence of the Effects of Foreign Investment and Aid on Economic Growth and Inequality.  A Survey of Findings and a Reanalysis.”  American Journal of Sociology 84: 651-83.

 

Bradley, David, Evelyne Huber, Stephanie Moller, Francois Nielsen, and John D. Stephens. 2003. “Distribution and Redistribution in Postindustrial Democracies.” World Politics 55:193-228.

 

Chase-Dunn, Christopher.  1998.  Global Formation, Updated Edition.  Lanham: Rowman and Littlefield. 

 

Chase-Dunn, Christopher and Peter Grimes. 1995. “World-Systems Analysis.”  Annual Review of Sociology 21: 387-417. 

 

Chase-Dunn, Christopher and Richard Rubinson. 1977. “Toward a Structural Perspective on the World-System.”  Politics and Society 7(4): 453-76. 

 

Cutright, Phillips. 1967. "Inequality: A Cross- National Analysis." American Sociological Review 32:562-78.

 

Crowly, Angela, James Rauch, Susanne Seagrave and David. Smith.  1998. “Quantitative Cross-National Studies of Economic Development: A Comparison of the Economics and Sociology Literatures.” Studies in Comparative International Development 33(2): 30-57.

 

DeMartino, George.  (1998)  "Foreign Direct Investment." In Foreign Policy in Focus, edited by Martha Honey and Tom Barry.  Retrieved April 24, 2005 (http://www.fpif.org/briefs/vol3/v3n14fdi.html)

 

de Soysa, Indra and John R. Oneal.  1999.  “Boon or Bane?  Reassessing the Productivity of Foreign Direct Investment.”  American Sociological Review 64(5): 766-82.

 

Dixon, William and Terry Boswell, (1996)  “Dependency, Disarticulation, and Denominator Effects: Another Look at Foreign Capital Penetration.”  American Journal of Sociology 102: 543-562.

 

Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capitalism. Princeton, NJ: Princeton University Press.

 

Evans, Peter and James E. Rauch.  1999.  “Bureaucracy and Growth: A Cross-National Analysis of the Effects of Weberian State Structures on Economic Growth.”  American Sociological Review 64: 748-765.

 

Evans, Peter and Michael Timberlake, (1980)  “Dependence, Inequality, and the Growth of the Tertiary: A Comparative Analysis of Less Developed Countries.”  American Sociological Review 45: 531-552.

 

Evans, Peter.  1995.  Embedded Autonomy: States and Industrial Transformation.  Princeton: Princeton University Press.

 

_______.  1979.  Beyond Center and Periphery: A Comment on the Contribution of the World-Systems Approach to the Study of Development.”  Sociological Inquiry 49(4): 15-20.

 

FAOUN.  2009.  FAOSTAT Online Statistical Service. Rome: FAOUN.  http://faostat.fao.org.

 

Feenstra, Robert C. and Gordon H. Hanson.  1997.  Foreign Direct Investment and Relative Wages: Evidence from Mexico’s Maquiladoras.  Journal of International Economics 42(3-4): 371-93.

 

Firebaugh, Glenn.  2000.  “The Trends in Between-Nation Income Inequality.”  Annual Review of Sociology 26:323-39.

 

________ 1999.  “Empirics of World Income Inequality.”  American Journal of Sociology 104(6): 1597-1630.

 

_______.  1992.  “Growth Effects of Foreign and Domestic Investment.”  American Journal of Sociology 104(6): 105-30.

 

Firebaugh, Glenn and Brian Goesling.  2004.  “Accounting for the Recent Decline in Global Income Inequality.”  American Journal of Sociology 110(2): 283-312

 

Frank, Andre Gunder. (1967)  Capitalism and Development in Latin America.  New York: Monthly Review Press.

 

Galtung, Johan.  (1971)  “A Structural Theory of Imperialism.”  Journal of Peace Research 2: 81-117.

 

Goesling, Brian.  2001.  “Changing Income Inequalities Within and Between Nations: New Evidence.”  American Sociological Review 66: 745-761.

 

Haggard, Stephan.  1989.  “The East Asian NICs in Comparative Perspective.”  Annals of the American Academy of Political and Social Science 505: 129-141.

 

Heston, Alan, Robert Summers and Bettina Aten.  2002. Penn World Table Version 6.1. Center for International Comparisons at the University of Pennsylvania (CICUP), October.

 

Hewitt, Christopher. 1977. "The Effect of Political Democracy and Social Democracy on Equality in Industrial Societies: A Cross-National Comparison." American Sociological Review 42:450-64.

 

Hirschman, Albert O. [1945] (1980) National Power and the Structure of Foreign Trade. Berkeley: University of California Press.

 

ILO.  1998.  Labor and Social Issues Relating to Export Processing Zones.  Geneva: ILO

 

Jackman, Robert W. 1974. "Political Democracy and Social Equality: A Comparative Analysis." American Sociological Review 39:29-45.

 

Jorgenson, Andrew K.  (2006a) “Global Warming and the Neglected Greenhouse Gas: A Cross-National Study of Methane Emissions Intensity, 1995.”  Social Forces 84: 1777-1796.

 

_______.  (2006b)  “The Transnational Organization of Production and Environmental Degradation: A Cross-National Study of the Effects of Foreign Capital Penetration on Organic Water Pollution Intensity, 1980-1995.”  Social Science Quarterly 87: 711-730.

 

Jorgenson, Andrew K. and Ed Kick (eds.).  (2006)  Globalization and the Environment.  Leiden, The Netherlands: Brill Academic Press. 

 

Kay, Cristobal.  2002.  “Why East Asia Overtook Latin America: Agrarian Reform, Industrialization and Development.”  Third World Quarterly 23(6): 1073-1102. 

 

Kentor, Jeffrey.  (2001)  “The Long Term Effects of Globalization on Income Inequality, Population Growth and Economic Development.”  Social Problems 48(4): 435-455.

 

_______.  (1998) “The Long-Term Effects of Foreign Investment Dependence on Economic Growth 1940-1990.” American Journal of Sociology 103: 1024-1046.

 

Kentor, Jeffrey and Peter Grimes.  (2006)  “Foreign Investment Dependence and the Environment: A Global Perspective.”  Pp. 61-78 In Globalization and the Environment.  Jorgenson, Andrew and Ed Kick (eds.).  Leiden, the Netherlands: Brill Academic Press.   

 

Korzeniewicz, Roberto P and Timothy Patrick Moran  1997.  “World-Economic Trends in the Distribution of Income, 1965-1992.”  American Journal of Sociology 102(4): 1000-1039.

 

Kuznets, Simon. (1955)  “Economic Growth and Income Inequality.” American Economic Review 45:1-28.

 

Lee, Cheol-Sung.  2005.  “Income Inequality, Democracy, and Public Sector Size.”  American Sociological Review 70(1): 158-177.

 

Lee, Cheol-Sung, Francois Nielsen and Arthur S. Alderson.  2007.  “Income Inequality, Global Economy and the State.”  Social Forces 86(1): 77-111.

 

Lipsey, Robert E. and Fredrik Sjoholm.  2001.  “Foreign Direct Investment and Wages in Indonesian Manufacturing.”  NBER Working paper no. 8299.  Cambridge, Mass: National Bureau of Economic Research. 

 

Lloyd, Paulette, Matthew C. Mahutga and Jan de Leeuw. 2009.  “Looking Back and Forging Ahead: Thirty Years of Social Network Research on the World-System.”  Journal of World-System Research 15(1): 48-85.

 

Mahutga, Matthew C.  (2006)  “The Persistence of Structural Inequality? A Network Analysis of International Trade, 1965-2000.”  Social Forces 84(4): 1963-1889.

 

Mahutga, Matthew C. and Nina Bandelj.  2008.  “Foreign Investment and Income Inequality: The Natural Experiment of Central and Eastern Europe.”  International Journal of Comparative Sociology 29(6): 429-454. 

 

Marshall, Monty G. and Keith Jaggers. 2000. "Polity IV: Political Regime Characteristics and Transitions, 1800-1999." Center for International Development and Conflict Management, University of Maryland, College Park

 

McMichael, Phillip.  (1996) Development and Social Change: A Global Perspective.  Thousands Oaks: Pine Forge Press

 

McMillan, Carl. (1987) Multinationals from the Second World: Growth of Foreign Investment by Soviet and East European Enterprises. London: MacMillan Press.

 

Morris, Martina and Bruce Western.  (1999)  “Inequality in Earnings at the Close of the Twentieth Century.”  Annual Review of Sociology 25: 623-57.

 

Moran, Theodore.  2002.  Beyond Sweatshops: Foreign Direct Investment in Developing Countries.  Washington D.C.: Brookings Institution. 

 

Muller, Edward N. 1985. "Income Inequality, Regime Repressiveness, and Political Violence.  American Sociological Review  50:47-61.

 

_______.  1988. "Democracy, Economic Development, and Income Inequality." American Sociological Review 53:50-68.

 

Nielsen, Francois.  (1994)  “Income Inequality and Development: Dualism Revisited.”  American Sociological Review 59: 654-77.

 

Nolan, Patrick.  1983.  “Status in the World-System, Income Inequality and Economic Growth.”  American Journal of Sociology 89(2): 410-419

 

Plumper, Thomas and Vera E. Troeger.  2007.  “Efficient Estimation of Time-Invariant and Rarely Changing Variables in Finite Sample Panel Analyses with Unit Fixed Effects.”  Political Analysis 15: 124-139. 

 

Prais, S and C. Winsten.  1954.  “Trend Estimation and Serial Correlation.”  Cowles Commission Discussion Paper No. 383.  Chicago.

 

Raftery, Adrian E.  1995.  “Bayesian Model Selection in Social Research.”  Sociological Methodology 25: 111-163

 

Ranney, David.  (1998) “Investment Liberalization Agenda.” In Foreign Policy in Focus, edited by Martha Honey and Tom Barry, vol.3, no. 21. Retrieved April 24, 2005 (http://www.fpif.org/briefs/vol3/v3n21trad.html)

 

Rubinson, Richard.  1976. “The world-economy and the distribution of income within states: a cross-national study.”  American Sociological Review 41(4): 638-659.

 

Rubinson, Richard. and David Quinlan.  1977.  “Democracy and social inequality: a reanalysis.”   American Sociological Review 42: 611-623.

 

Rueda, David and Jonas Pontusson.  2000.  “Wage Inequality and Varieties of Capitalism.”  World Politics 52: 350-83.

 

Simpson, Miles. (1990) “Political Rights and Income Inequality: A Cross-National Test.” American Sociological Review 55: 682-693.

 

Solt, Frederick.  2009.  “Standardizing the World-Income Inequality Database.”  Social Science Quarterly 90(2): 231-242.

 

_______.  2008. “Economic Inequality and Democratic Political Engagement.”

American Journal of Political Science 52(1):48-60.

 

Snyder, David and Edward L. Kick. 1979. “Structural Position in the World System and Economic Growth, 1955-1970: A Multiple-Network Analysis of Transnational Interactions.” American Journal of Sociology 84(5): 1096-1126.

 

Stack, Steven (1980)  “The Political Economy of Income Inequality: A Comparative Analysis.” Canadian Journal of Political Science 13:273-286.

 

Tsai, Pan-Long. (1995) “Foreign Direct Investment and Income Inequality: Further Evidence.” World Development 23: 469-483.

 

UNCTAD.  2009.  World Investment Directory.  New York: United Nations.  http://stats.unctad.org/FDI

 

Wallerstein, Immanuel. 1989 The Modern World-System III. Berkeley: Academic Press.

 

Weede, Erich.1989. "Democracy and Inequality Reconsidered." American Sociological Review 54:865-68.

 

Wood, Adrian.  1997.  “Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom.”  The World Bank Economic Review 11(1): 33-57.

 

World Bank. 2003. World Development Indicators [CD ROM]. Washington DC: World Bank

 

World Bank.  2002.  EdStats.  Washington DC: World Bank.

 

World Bank 2009.  World Development Indicators [online].  Washington DC: World Bank. 

 



[1] Draft manuscript.  Please do not cite, reproduce or circulate in any way without express written consent of the lead author.  Direct all correspondence to Matthew C. Mahutga, Department of Sociology, University of California at Riverside.  1226 Watkins Hall, Riverside, CA. 92521.  matthew.mahutga@ucr.edu.  Mistakes, omissions and sloppy prose are entirely the fault of the lead author.  

[2] In unreported analyses, we tested curvilinear effects of public sector size and government spending, as well as interactions between each of these and democracy.  None of these analyses produced significant results.