Income Inequality and Transnational Corporate Penetration

This study examines whether the positive association between national income inequality and transnational corporate penetration found previously by Bornschi cr and Chase-Dunn (1985) and others circa the late 1960s still holds for the mid-l980s. Both methodological and theoretical problems of earlier studies arc discussed and solutions arc offered. Economic development, political-institutional and regional variables arc also included in the analyses. While further research is warranted, the results provid e support for a WorldSystcm/Dcpcndcncy perspective in understanding income inequality cross -nat ionally.


INTRODUCTION
Bornschicr and Chase-Dunn's seminal 1985 work, Transnational Corporations and Underdevelopment, provided a revitalized foundation for quantitative crossnational research on income inequali ty from a World-System /Dependency perspective.The authors focused on the rela tionship between transnational corporate penetration and underdevelopment as a primary mech anism of capitalist exploitation and maint enanc e of the core/periphery hierarchy in the world economy.Foreign economic involv ement b y transnational corporations (TN Cs) in the economics of less -developed countri es (LDCs) was argued to be one means by which powerful world actors maintain an inher ently unequal global division of labor.Using linear regression techniques on an impro ved cross-na tional data set, Bornschicr and Chase-Dunn (1985) presented empirical tests of their hypotheses, which confirmed a positive relationship between levels of foreign corporate penetration and income inequali ty.
With one important exception (Tsai 1995), most recent work on cross-national income inequality has moved away from the examination of the impact of foreign dir ect investment (FDI) dependence on income distribution, focusing instead, for example, on the effects of economic and socio-cultural dualism (Williamson 1991;Nielsen and Alderson 1995) or tcchnoccological heritage (Lenski and Nolan 1985;Crenshaw and Ameen 1994).This is surprising because foreign direct investment has dramaticall y increased in importance over the past two decades and is currently the primary source of resource flows to developing nations (Froot 1993;Tsai 1995).Indeed, LDCs arc encouraged to attract foreign investment as one route to economic growth and well-being in the contemporary world-economy.Foreign investment is promoted by development professionals and lending agencies such as the World Bank as an efficient way to add to existing domestic pool.., of capital, technology and entrepreneurial talent (Rothgeb 1996).
Specification of the relationship between FDI and income distribution in LDCs is far from complete.Further examination of the processes specified by World-Systcm/Dcpcndcncy researchers is necessary, and it is especially crucial to examine the effect of FDI over the past two decades.This study takes the important step of quantitatively examining the relationship between TNC penetration and income inequality during the mid-1980's.The majority of previous studies have used data from the latc-1960's and carly-1970's and calls for more contemporary analyses arc frequent (Krahn and Gartrell 1985;Ragin and Bradshaw 1992 ).In addition, while most previous studies indicate that high levels of foreign control over the economi cs of LDCs arc associated with income inequality during the 1960s and 1970s, Tsai (1996) found that, with the specification of geographical region, the effect.., of foreign penetration were marginal.This is a significant contribution to the body of literatur e on the subject, and the following study will address this critique by incorporating regional variables into the analyses.

Journa I of World-Systems Research
Many researchers have examined recent changes in the world economy, but quantitati ve studies have largely ignored the possible impact of these changes on the relation ship between transnational corporate penetration and income inequality.As Gcrcffi (1989) and others have argued, the declining significance of industriali zation in national economics has implications for development theory.In addition, the rise of certain newl y industrializ ed countri es has affected traditiona l global spaciocconomic hierarch ies (Haggard 1990).
Recent improvements in data accessibility and quality allow us to mor e reliably test World-System /Dependency arguments concerning the deleterious effect ofTNC penetratio n on income distribution .The analyses presented here also incorporat e the insights of various researchers into the study of national incom e inequality.Th e data arc an improv ement over most previous studies; indicato rs arc available for a greate r numb er of nations and comparability issues have been addressed.Additionally, th e correctl y specified economic development variables and regional indicators arc included in the equations, in order to ensure that the empirical results arc not spurious.Moreover , I us e the percentage of income accruing to the top 10% a.., one of the two inequalit y indicators in the models, which is unprecedented by previous literature on the subject.Theoretically, this is a better test of World-System /Dependenc y theories, a.., thes e arguments focus on the concentration of income for the elite segment.., of national populations.
The research presented in the following sections is a first step in the process of incorporating temporal change into our understanding of the structural procc sscs of crossnational inequality; its limited aim is an updated empirical test of the World-System/Depcndcncy model.After an overview of the theoretical model, I briefly discuss the variables used and related methodological issues.Then, the updat ed Born schier and Cha..,c-Dunn regression model is analyzed .Following this, additional politicalinstitutional variables, which some have argued that Dependenc y research ha.., ignored (i.e.Muller 1988), will be added to the equations for a more rigorous test of th e mod el.This will also serve to further specify the redistributive effects of national governm ent structure on income inequality, a.., well as provide a more complete picture of the processes affecting the relationship between TNC penetration and national we lfarc.In th e final section, I present a discussion of the results of the study, a.., well a.., my conclu sions regarding directions for future research on this topic.

Modernization/Developmental Theories of Income Inequality
Modernization or developmental theories of incom e distributi on predict that developing nations will exhibit high er level.., of income inequality rel ative to both non-indu strial and industrialized countries.Howev er, these scholars argue that, a.., economic growth continues, incom e distribution within these nations becomes mor e equitable.This approach is most frequently a ... sociated with the work of Simon Kuznets (1955Kuznets ( , 1963Kuznets ( , 1976) ) who found a curvlincar a ... sociation between income inequality and economi c growth, and wa.., among the first to develop a theoretical argum ent to explain this findin g.

Journa I of World-Systems Research
Modernization theorists argue that wealth concentrate s in the hands of a few entrepreneu rs in the early stages of indus trialization, a.., this is the most efficient use of scarce capita l (Crenshaw and Ame en 1994).Incrca..,ing the rate of capital investm ent, both foreign and domes tic, depends on the development of mod ern economic segments of the economy.This process entails the expropriat ion of surplus cap ital from other segments , and is harde st on the extracti ve and agricultural economic sectors.At this early stage of industrialization, the state plays a critical role in social capital accumulation by implementing policies which ca~c the burden on these sectors a~ their surplus is transferred to modern industries in urban area~ (Rostow 1960).
The link between economic growth and income inequality from a Modernization/Developmental perspective is the notion of spacial disparities in growth.These theorist~ argue that industrialization creates uneven development among sectors of the national economy.A~ urban modern industries grow, individuals involved in these sectors become relatively advantaged compared to other portions of the population, especially those in rural traditional industries such a~ agriculture.This creates disparities not only in income, but in social welfare, political power and integration into markets.Therefore, economic growth leads to economic dualism and high levels of inc omc inequality among national populations; the mechanism which links industrialization and income distribution is sectoral and spacial disparities in the distribution of wealth.
A~ national development and urbanization continue over time, Modernization theorists hypothesize that income inequality dccrca~cs.Capital concentration becomes less urgent in the latter stages of industrialization, a~ urbanization economics arc realized and modern economic activities and institutions diffuse from urban to rural area~.The process of spacial disarticulation found in early stages of development gives way to spacial integration in the later stages.
Theorists in this tradition argue that continued economic growth expand~ the middle cla~s and incrca~cs employment and saving rates among the poor, leading to decreasing income disparity ( Paukert 1973;Cheney and Syrquin 1975;Ahluwalia 1976;Kuznets 1976).This approach a~scrts that industrialization eventually expands employment opportunities for the entire population, generates occupational specialization which incrca~cs the bargaining power of labor and improves quality of life by dccrca~ing the cost of living (Deane 1979).In addition, economic growth provides incentives for selfintcrcstcd elites to voluntarily divest portions of their wealth to the population in order to incrca~c consumption by creating ma~s markets for their good~, dccrca~c conflict and legitimate the power structure (Lenski 1966).
New research in this tradition seeks to explain the Kuznct's "invcrtcd-U" by reference to a comprehensive dualistic explanation which secs a~pccts of development a~ transitional.Nielsen (1994) and Nielsen and Alderson (1995;1997) argue that the curvlincar relationship between income inequality and development is due to "transitional development processes related to the dualism (both economic and generalized) of traditional and modern sectors of developing societies."(Nielsen 1994).Sector dualism, and hence the high levels of income inequality which arc brought about by sue h processes, arc a temporary consequence of the labor force shifts and wage differentials between, and within, traditional agrarian and modern industrial sectors of developing societies.Inequality is also an unintended consequence of the demographic transition and resultant generalized dualism, which includes the uneven diffusion of sociocultural traits a~sociatcd with earning power, such a~ education.It is not development per sc, but dualism and diffusion processes which arc argued to be the keys to cxpl aining income inequality (Nielsen and Alderson 1995).

Journal of World-Systems Research
The empirical evidence from the literature which test<; for the relationship between economic development and income inequality cross-nationall y is mi xed.While man y have found a significant Kuznets-type inverted U-shapcd association (Crenshaw and Ameen 1994 ;Wccdc 1993Wccdc , 1989;;Crenshaw 1992;Muller 1988;Johnson 1986;Wccdc and Ticfenbach 1981 ), others have found development to be insignificant when other variables arc included in their models (Nielsen 1994;Simpson 1990;Chan 1989;Muller 1989;Prcchel 1985;Stack 1980).More recent studies have failed to confirm the Kuzn ct hypothesis using improved longitudinal data (Dcningcr andSquire 1997, 1996) , timeseries analyses (Sasaki ct al 1997), or examination of case studi es (Bowman 1997).Clearly, further empirical work is needed in this area.While the present study is primarily focused on examining the effect of transnational corporate penetration on national income distribution, the statistical model<; will include a curvlincar specification of economic development.In addition to subjecting the Kuznct's "invcrtc d-U" hypothesis to furth er examination, this will also ensur e that the findings concerning the effec ts of FD I dependence on inequality arc not spurious.

World-System/Dependency Theories of Income Inequality
A" opposed to Modernization theory's emphasis on the internal dynamics of economic growth, World-System and Dependency theories arc nco -Marxist persp ectiv es that focus on the global structure of the capitalist world econom y.Most com pletel y elaborat ed in the work ofWallcrstcin (1974,1979,1989) and Cardoso and Falctto (1979), among others, this approach argue s that nation al economic growth, inequality and sociopolitical change can only be und erstood through the analysis of a nation's relati ve position in the spaciocco nomic hierarc hy of the world system.That is, the relation ship between economic growth and incom e inequality within any single nati on is d cpcnd cnt on that soc iety's relational position in the world division of labor and global power structure.It is asserted that the dynamics of capitalist accumulation in developin g countries arc different than the processes observable in core nations (Prcchel 1985).The issue that World-Systcm /Dcpcndcncy analyses po int our attention to is not the lack of economic growth in developing nations, but the type of grow th their dependent status affords them and it's consequences.
In the World-System /Dependency perspective, capitalist deve lopm ent is depende nt on social and material inequality and this inequali ty is in turn a result of incorp orat ion into the wo rld system.National econom ic growth and income distribution arc in large part determined by growth potentials of productive activities in the larger global structur e. Ther efore, this approach hypot hesizes that strat ification of incom e will correspo nd with the world division of labor and position in the world economy.Thi s is a difficu lt concept to operationalize and, while some researchers have found that position within the spac iocconomic hierarchy affects inequality independent of other factors (Nolan 1983;Bornschicr and Chase-Dunn 1985), most recent studies have found insignificant effects when this concept is operationalized as core /periphery dummy variables includ ed in cross-national regression models (Muller 1988;Simpson 1990;Crenshaw 1992).
Although not fully settled in the literature, the relationship between spacio-cconomic position and inequality is in all probability more complex than can be captured by dichotomous variables, and other researchers have argued that core /periphery stat us is likely to work indirectly toward increasing inequality through the mechanisms discussed below (Simpson 1990).
There arc variants to the World-System /Dependency approach regarding the creation of income inequality, some of which emphasize concentration ofland ownership (Furtado 1970;Muller and Sdigscn 1987;Boswell and Dixon 1990) or national export-structure (Baran 1957;Frank 1967;Galtung 1971;Prcchel 1985).The following st udy focuses on the effect of cross-national capital transfers, which some have argued arc mor e indicativ e of dependency status during the past twenty to thirty years than arc trade -bas ed meas urcs (Prcchel 1985;Chan 1989).This strand of the literature emphasizes foreign dir ect investment as the primary means through which the modern capitalist world-system creates, and maintains, intra-and international socioeconomic inequiti es.Many empiri cal studies of this relationship have confirmed a significant association between foreign corporate penetration (an indicator of the amount of foreign own ership and control over a host economy) and inequality (Evans and Timberlake 1980;Kohli ct al 1984 ;Bornschicr and Chase-Dunn 1985;Chan 1989;London and Robinson 1989;Crenshaw and.Ameen 1994;Dixon and Boswcll 1996).Other studies find this association only in certain geographical regions (Rothgeb 1993;Tsai 1995).Even those scholars which fail to confirm this relationship generally report their conclusions with reservation and do not dismiss TNC penetration as a potentially important determinant of income inequality (Wc cdc and Ticfcnbach 1981;Crenshaw 1992).

Journa I of World-Systems Research
In the World-System/Dependency perspect ive there arc three mechanisms that arc hypothesized to link foreign investment and social inequality (Crenshaw and Ameen 1994).First, foreign investment in developing countries generates large secto ral dispariti es in the national economy , crea tes labor aristocracies and results in the underutilization of indi genous labor.Second, transnati onal corporations operating in developing na tions accrue a disproportio nate share of local sources of credit and repatriate profits rather than reinvesting them in the local economy.Finally, the governments of these nations, moti vated by the necessity (generated by their incorporation into the capitalist world economy) of attracting and maintainin g foreign inves tment , implemen t polici es and strateg ics that decrease the pow er of labor and inhibit vertical mobility.These include tax concessions, guarantees of profit repatriation, and labor laws unfavorable to workers (London and Robinson 1989).
Scholars in the World-System/Dependency tradition argue that the relationship between foreign investment and internal income inequality ha~ different effects on various sectors of the economy, but in all segment~ it creates and sustains income inequality in the national population (Crenshaw and Ameen 1994).Foreign capital investment in the agricultural sector destroys traditional production processes and leads to unemployment and ovcrurbanization through its capital intensive means of organization (i.e.labor shedding, land enclosure).In the extractive sector of the economy, foreign investment benefits only a small portion of the national population and thereby incrca~cs income inequality.This is because TNC penetration in this sector creates only a small well-paid labor force and because ownership of natural resources is typically concentrated.
World-System theorists argue that foreign investment in the manufacturing sector ha~ the most harmful effect on national income distribution.National economics in non-core nations with large manufacturing sectors have high level~ of income inequality because profits in this sector arc incrca~cd by the maintenance of a large, surplus low-wage labor force.Therefore, high rates of income inequality arc in the interest of transnational corporations and national elites who benefit from foreign investment; they have little incentive to take action to distribute income more equitably.Contrary to the hypotheses of Modernization theorists, the World-System perspective argues that the uneven development of highly penetrated developing economics benefits transnational corporations in that the only segment of the population which can afford to buy these manufactured goods is the wealthy elite (Evans and Timberlake 1980;Prcchel 1985;London and Robinson 1989).Domestic demand for these goods depends on the concentration of wealth and high level~ of income inequality.Although redistribution of wealth and the resultant expansion of markets may be in the long term interest of foreign corporations, they arc driven primarily by the short-term profit logic of capitalism (Bornschicr and Cha~c-Dunn 1985).
Furthermore, there is a convergence of interests between transnational corporations and the wealthy elite segments of the national population in maintaining income inequality which creates barriers to the "trickle-down" effect of industrialization predicted by Modernization theories.In addition to the incentives for inequity for foreign investors discussed above, the national elite strive to maintain their power and higher income so a~ to maintain privileged consumption patterns and access to status symbols.A common international cla~s interest in the persistence of high level~ of inequality thus link foreign investors and indigenous elites, leading these powerful groups to support ( and in some ca~cs attempt to incrca~c) the existing unequal income distribution and to coo pt and repress opposition from other segments of the population (Bornschicr, Cha~c-Dunn and Rubinson 1977;Bornschicr and Ballmer-Cao 1979;Nolan 1983;Bornschicr and Cha~c-Dunn 1985;Stokes and Anderson 1990)..Touma I of' World-Systems Research World-System/Dependency theories take issue with the Modernization argument that all capital investment is beneficial, distinguishing between the conscqucn ccs of different forms of investment (Dixon and Boswell 1996).Moreover, th ey reject the hypoth esis that the benefits of development to national elites inevitably "trickle-do wn" to the larger population, cmpha-;izing how intranational ela-;s interests and structural barriers impact the possibilities for improvements in social welfare.This approach directs our att ention to the connections between countries, and how these link-; structure the potentials for growth and well-being in discrete nations.In addition, the approach addres ses microfoundations by showing, for example, how income inequality benefits the shorttcrm interests of powerful actors within developing nations.

DATA AND MEASURES The Measurement of Income Inequality
Many researchers have recognized problems with cross-national mca-; urcs of income inequality (e.g.Ahluwalia 1993Ahluwalia (1974]]; Muller 1993Muller (1984]]).A-; Hoover (1989Hoover ( : 1008) ) notes in his work assessing the comparability of various types of income inequali ty data, most cross-national studies of this phenomena have been "relative ly insensiti ve to issues of data quality and comparability."Fortunately, data collection procedur es have improved in recent years, and much work ha-; been done in a-;scssing the str engths and weaknesses of various mca-;urcs of income inequality, a-; well a-; their collection and interpretat ion.Ahluwalia (1993Ahluwalia ( (1974]]) notes that income inequality is a difficult concept to m ca-;urc quantitatively and is frequently incorrectly specified.The two most commonly used mca-;urcs of incom e inequal ity arc the Gini coefficient, which look-; at the disp arity between equal and actual distribution of income among quintile shares, and the proportion of incom e received by the top 20% of the population 1. Use of the Gini coefficient ha-; been extensi vely critiqued on both methodolo gical a-; well as theoretical fronts (Braun 1991;Hoov er 1989;Chan 1989;Muller 1993Muller (1984]]) .
It is, how ever, generally for theor etical rca-;ons that man y researchers opt to use shares of income instead of the Gini coefficient.Muller (1993Muller ( (1984]]) argues that percentile shares of incom e arc the most appropria te mca-;urcs because World-S ystem/Depend ency arguments point to the concentration of income in the upper end of the distribution a-; the crucial indica tor of income inequality.Th e Gini score is a mca-;urc of the variation of quintile scores from an equal distribution.A-; there is little variation in the bottom 20%, the differenc e between concentration in the top percentiles and the Gini scor e is due almost entirely to the distribu tion in the middle.In addition, the use of upp er proportional shares of income ha-; been argued to indirec tly mca-;ure a-;sct inequalit y, another significan t dimensi on of economic stratification (Boswell and Dixon 1993).Conside rin g these theoretical concerns, I will use both the top 10% and top 20% percentage shares of total incom e to mca-;urc income inequality.Moreover, a-; stated in the introdu ction, th e use of top decile shares of income is unprecedented in previous literature and is a better test of World-System /Dependency argument.:;.The measurement of fractile shares of income as an indicator of income inequality also presents methodological problems, most importantly regarding the reliability and comparability of the data.One of these issues concerns the temporal ordering of variables.Cross-national data on percentage share of income is usually not available for a substantial number of countries for a short time span, for example over two or thr ee years.Frequently, data on income inequality in cross-national studi es covers an extend ed period of time, often temporally preceding the independent variables.As income distribution is a relatively stable structural characteristic most researchers agree that th e variable can be validly included even where the measures temporally precede the independent variables to some degree (Chan 1989;Nielsen 1994).While incom e inequality docs change, this change is usually relatively slow.That is, income distribution measures typically exhibit only small changes from year to year.

Journa I of World-Systems Research
Scholars have debated the appropriate length of the time span, however, with some arguing for five year (Muller 1993(Muller [1984]]), nine year (Hoover 1989), or eleven year (Chan 1989) time periods.This issue is important because, a-; Hoover (1989) indicat es, while incom e distribution is relatively stable, we cannot uncriticall y a.:;sumc that factors influencing its variability operate in the same way at all points in time.Other scholars have argued that the need for a larger sample justifi es the use of a wider time period, and have found the same empirical patterns using different time restricti ons (Chan 1989).For this rca.:;on, I limit the mca.:;urcmcnt of the income inequality variable to a 14 year period (1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993).This decision is ba.:;cd on the need for a larger sample, a-; well a-; the fact that this work examines the Dependency model in the mid-80s 2 .With this time restriction there is no overlap with Bornschicr and Cha.:;c-Dunn (1985), and little with pa.:;t Dependency studies.
Another issue in the measurement of incom e inequality concerns the comparability o fthc data, which researchers often collect from various sources in order to incrca.:;csampl e size.This can lead to problems if, for example , the data is not national in coverag e or is ba.:;ed on different income-receiving units or income concept.:;(Hoover 1989).Fortunately, sinc e income data collection has improved in recent years, I obta ined the data used in this study from a single source, the 1995 World Bank Developm ent Report (WBDR).The WBDR ha-; been cited a-; the best available source of income data and ha-; less comparability problem s than other data sets, although standardized data collection procedures have not yet been established in all countries (Hoover 1989;Mull er (1993Mull er ( [1984]]) .
Hence, this data set improves upon those used in previous cross -national research on income inequality in three wal s.First, the data is updated and derived from one of th e most recent sources available .Use of this dataset in the present study is an important first step in updating the modeling of national income distribution beyond the late-l960' s and early-l 970's data used in the majority of previous studies (important exceptions arc Johnson 1986 andNielsen andAlderson 1995).As stated prcvio usly, this updat e will allow us to determine if the relationships specified in the late 1960s were still operating in the mid-l980s.In addition, I take theoretical concerns into account by using percentile shares of income rather than the Gini coefficient, since the testing of World-Systcm/Dcpendency arguments directs our focus on the distribution of income in the upper portions of the population.Furthermore, the use of top 10% shares of income is more relevant to these concerns as it indicates greater concentration, and the use of this measure is rare in previous studies of this topic.Finally, I have addressed methodological issues by using data from a limited time frame and from a single source to insure that validity and comparability problems arc minimized.

Measurement of the Independent Variables
Following the full Dependency model specified by Bornschicr and Chase-Dunn (1985), I use five independent variables in the linear regression equations predicting incom e inequality in models # 1-#4 (Table l) and #8-# 11 (Table 2).Logg ed real GDP per capita in 1985 (RGDPPCLN) (from Summers and Heston's 1995 PENN World Table Mark 5.6 data set) is included to control for level of economic developm ent.This measure is an improvement over those studies, including Bornschicr and Chase-Dunn's, which use logged GNP per capita without purchasing power parities.In a correct specification of th e development/inequality relationship, logged real GDP per capita squared (RGDPPC2) is also included in the model, as level of economic development and income inequality are hypothesized by Modernization/De velopmental theorists to exhibit a cross -sectional curvlinear association (Kuznets 1955(Kuznets , 1963(Kuznets , 1976)).

[Page 8] Journal of World-Sys tems Research
The mcmmrcment of transnational corporate penetration (PEN ) is the ratio betw een foreign direct investment inward stock ( from the World Investment Report 1996) and market GDP (from the World Tables 1995) in 1985.As a measur e of foreign control over the host economy , this variable is both a methodological and theoretical improv ement over those used previously (Dixon and Bosw ell 1996).One memmre of th e effect of state socioeconomic and sociopolitical processes, a dummy-coded variable of communist government in 1985 (COM) is included in the model.The coding of this variable is ba.., ed on The World in Figures : Editorial Information Compiled by The Economist (1987).This operationalization is not the same a.., that used by Bornschicr and Cha..,c-Dunn (1985).However, it is primarily includ ed in their study a.., a control variable .Both variables mca..,urc the same und erlying concept and address the argument that Muller (1993Muller ( [1984]]) and others have made, which states that the redistributive effects of government mitigat e the relationship between income inequality and other variables.In addition, I created a core country dummy-coded variable (CORE) a<; a mea<;ure of position in the spacioeconomic hierarchy in 1985 1 .However, this variable wa<; not included in the equations separately because it in itself ha<; no substantive meaning net of the economic development and capital dependency mea<;ures.Rather, an interaction term betwe en CORE and PEN( CORE *PEN) wa<; included in the model to control for the possible negative a<;sociation between core position and income inequality.It is this mea<;ure that is the theoretically relevant test of World-System /Dependency arguments because it captures the different processes at work in the core relative to developing nations.
Political-institutional variables are added to the regression equations in model<; #5-7 and #12-#14, to more rigorously test the Dependency model specified by Bornschier and Cha<;e-Dunn.A'l World-System/Dependency theorists argue that both position within the world economy -a<; well a<; sociopolitical processes -have an impact on level<; of state economic inequality, further specification of these relationships is warranted.In addition, the effects of national state structures on income inequality is a subject of contentious debate in the literature.The results of studies examining the relationship between democracy and inequality are inconclusive and contradictory (Hughes 1997).Many studies find no causal relation between the two (Bollen and Jackman 1985;W eede 1989, 1993), while others find a negative a<;sociation (Stack 1980;Muller 1988Muller , 1995;;Nielsen 1994).Still other researchers conclude that democracy ha<; a net positive effect on income inequality (Simpson 1990;Crenshaw 1992).Clearly, this proc ess needs furth er specification.
For this rea<;on, I include two additional dummy variables in the regression models.Th e first is an indicator of democratic regime, derived from Gurr 's Polity III data set.This variable wa<; constructed by creating an averag e democracy score for each nation for th e period 1980 to 1985.The distribution of the data revealed a clear threshold point which distinguished democratic regimes from non-democratic ones, an average democracy score of 6.66 or higher.Therefore, for ea<;e of interpretation, a dummy variable wa<; constructed (DEM) from these average democracy index scor es.The second is a variable indicating social democratic representation in parliament (SOCDEM).Some researchers have found positive effects of social democratic parties on incom e distribution (Hewitt 1977;Huber, Ragin and Stephens 1993), but most cross-national studies have not included regime type systematically.Separating the effects of social democratic versus non-social democratic government may explain the contradictory findings surrounding the effects of democracy discussed above .Many studies have found the cooperative public-private structures that characterize social democratic regimes promote overall social welfare, increasing national economic security and equalit y (Hicks 1988;Kenworthy 1995).The social democra tic government indicator is dummy coded 1 for majority participation in parliament betwe en 1980 and 1985, ba<;ed on information obtained from The Statesman's Yearbook (1987).

Journa I of World-Systems Research
Table 1 presents the OLS regression models of the relationship betwe en the top decile income share mca<;urc of income inequality and the independent variabl es for countri es with populations over one million in 1985 .I will discuss these equations first, and then contra<;t the results with those of the models predicting top quintile income shar es (Table 2).No outliers were found for any of the regression analyses and a full list of all nations included in the models is included in Appendix A. Following the Dependency model presented by Bornschicr and Cha<;c-Dunn (1985), the mca<;urcs wer e ent ered into the equation one at a time in models # 1-4 in order to examine the theoreticall y sp ecified effects of the independent variables.State government variables arc introduced into the equation in models # 5-7.
[Page 10] Journal of World-Systems Research  (Kuznets 1955(Kuznets , 1963(Kuznets , 1976)), but is also consistent with the World-Systcm prediction of higher income inequality in the semi-periphery.
The World-System/Dependency paradigm is tested by bringing the transnational corporate penetration (PEN), communist government (COM), and interaction term between core status and PEN (CORE*PEN) variables into the regr ession equations (equations 2, 3 and 4).The relationship between PEN and TOPlO is, as expected, positive and significant, even when controlling for other factors.The communist governm ent variable has a significant and negative relationship with TOPlO.That is, nations with communist governments have relatively less income inequality than non-communist states.The positive association between TOP 10 and PEN is stable with the inclusion of other variables and provides strong support for the World-System/Dependency contention that nation s whose economics arc highly penetrated by foreign corporations exhibit greater degrees of income inequality.The inclusion of the TNC penetration measure increases the variance explained by the equations from .24 to .34.However, contrary to the findings ofBornschicr and Chase-Dunn for the late l 960's, the interaction term between the core and penetration measures is not significant, nor docs it offer any additional explanatory power to the equation.I discuss the potential explanations for, and implications of, the disparate empirical findings of this study and that ofBornschicr and Chase-Dunn in the conclusion.
Moving to the equations which include the additional political-institutional variables (#5-7), we sec that economic development continues to exhibit a curvlincar association with inequality.Importantl y, PEN remains significant with the addition of increased specification of governmen t type; the positive association between high transn ational corporate penetration and top decile income percentage share seems to be robust.Democratic status appears to have an insignific ant effect on income distribution.Both COM and SOCDEM are significantly negativel y a-;sociated with TOPlO ; these form s of government have less income inequality relative to non-communist non-d emocra cies and liberal democracies.Together with the development and penetration mea-;ures, th ese variables explain 38% of the variance in top decile income shares (#7).
[Page 12] Journal of World-Systems Researc h Turning to the models predicting top quintile income share (Table 2), we sec that th e results arc similar to those predicting decile share.The economic development variables continue to exhibit a significant curvelincar association with income inequality .As in th e models predicting TOPl 0, the communist government variable has a positive relati onship with TOP20 and the CORE*PEN interaction term is insignificant.The robust and positive PEN effect found in the models predicting decile income share arc replicated in these models as well, as arc the effects of government type .I present a discussion of these results and explore their implications for the study of cross-national income inequality in the next section.

Journa I of World-Systems Research
A<; alluded to previously, Tsai (1995) argues that previous findin gs of a positive relationship between foreign corporate penetration and incom e inequality may well be the result of geographical differences in inequality.The empirical model<; of that study indicated that the effects of FDI were insignificant when regional variables were included in the regression equations.In order to explore this possibility for the present analysis, I introduce regional dummy variables into the core regression equations in Table 3.Following Tsai (1995), I include a Latin American and an Ea<;t /Southca<;t A<;ian indicator in the models predicting TOPlO (#15) and TOP20 (#16) 2 .Th e addition ofregional specification substantially increa<;cs the amount of variance explained, from .38 ( #7) to .48 (#15) for the equation predicting top decile share and from .36 (#14) to .45 (#16) for the equation predicting top quintile share.The development indicators become insignificant, no doubt reflecting systematic differences in regional economic development which arc captured by the geographic mca<;urcs.The leftist politicalinstitutional variables continue to be negatively related to top decile and quintile income share.The result<; also indicate that Latin American nations have substantially greater income inequality than their counterparts in the reference group.Importantly, however, the FDI penetration measure remains significant and continue s to exhibit a positive a<;sociation with inequality.This indicates that, while the inclusion ofrcgional indicators may be crucial for a full specification of any model predictin g income inequality , foreign corporate penetration ha<; an independent deleterious effect on national income distribution, at lea<;t for the mid~80s.fi .

DISCUSSION AND CONCLUSIONS
The empirical findings of this study concerning the relationship between transnational corporate penetration and income inequality provide continued support for the World-System/Depcndcncy model, although they arc somewhat different than research focusing on an earlier time period.In accordance with the theoretical argument, foreign economi c penetrat ion is a significant and robust predictor of the concentration of incom e in the upper portion of the population.The empirical analyses reveal the "negative externaliti es" of capital dependency for national income distribution (Dixon and Boswell 1996).Th ese results hold even where other relevant variables, such a<; level of development or regional specification , arc includ ed in the models.Perhaps the greatest contribution of this study is that it takes advantage ofreccnt imp rovements in data collection and availability, allowing for an examination of the effects of FD I dependence on inequality durin g a more recent time period than the majority of previous research.However, the result<; arc not unequivocal.The insignificanc e of the core/penetration interact ion term raises interesting questions for future research.Central to the World-System/Dependcncy approach is the notion that development processes arc different in developing nations (the periphery) relative to industrialized ones (the core).Various explanations can be offere d; empirica lly testing the issues raised by this study would be a significan t contribution to the literature on inequality and dependency.
TNC penetration wa.., at a low point in the mid-80s, which might explain the instab ility of the penetration effect (Miner 1997).Also, global inequality ha.., increa..,ed since the 1965, especially during the l980's (Ram 1992;Korzeniewicz and Moran 1997).An examination of changes in penetration and inequality over time should be undertaken in order to explore this possibilit y-.Alternately, it may be that core nations are becoming increa..,ingly subject to the same negative externalities of foreign penetration that have previously been found for less developed countries.That this might have changed between 1967 and 1985 would not be surprising.De industriali zat ion and globalization ar 6 ruments a ... sert that changes in the structure of the world economy have resulted in the increa..,ing convergence between the core and the periphery in terms of capital mobility (for example, Gereffi 1989).Researchers have noted an increa..,e in income inequality in many core nations during recent decades, especially in the United States (Bluestone and Harrison 1988;Braun 1991;Nielsen 1994).Increa..,ingly, many researchers have focused their attention on the economic and social consequences ofrelati ve ly recent developments in industrial production and the structure of global capital, focusing on their implications for the core (i.e.Bowles et al 1984;Piore and Sabel 1984;Ross and Trachte 1990).The rise of some of the newly industrialized economies, especially in Southea..,t Asia where income inequality is relatively low, is further indication of the changes in the relations of global capitalism that have taken place since 1967 (fo r example, Haggard 1990).

Journa I of World-Systems Research
In conclusion, the positive effect of transnational corporate penetration on income inequalit y is still supported by the empirical evidence.Even accounting for n ational regime type, foreign penetration is a ... sociated with a greater concentration of inc ome in the top fractiles of the population.The quantitative examination of cross-national income inequalit y during the pa..,t two decades is long overdue.The pres ent research represents an necessary first step towa rd this end.However, this study raises more issu es than it settles.The results leave many question unanswered, especially with rega rd to changes in the struc ture of exploitation and dependence in the world economy.Bornschier and Cha..,e-Dunn (1985) provided evidence of the change in dependenc e fro m trade to capital investment relati ons .The research presented her e indicat es another change in this process, although the model s do not empirically test this hypothesis.I su ggest that the explanation mi ght be found in an examination ofrecent hist orical develop mentsfluctuations in foreign capital penetrat ion , mediation in core /periphery relations, var iation in the mechani sms of capitalist exploitati on, the changing nature of effective industrial production, changes in migration patterns -deve lopme nts which have influ enced the relatio nship between income inequality and transnational corporate penetrat ion, perhaps altering but not eradicating the posit ive a ... sociation between the two.Furthermore, the inclusion of additiona l political-institutional va riables into the equat ions is an important contribution toward fully specifying the relationship betwe en form o f government and income inequality.While liberal democratic status had no significant effect on income distribution, social democracies and communist governments exhibited lower shares accruing to both the top 10% and top 20% of the population.These result-; are somewhat intuitive; leftist governments with manifest redistributive ideologies and policies decrca-;c income inequality within their national boundaries.This issue needs exploration, a-; in many nations where communist govcrnmen ts failed and/or their economics opened, income inequality increa-;ed a-; nations became more highly penetrated by foreign capital.
Future studies should more fully address and explain the issues raised by the empirical results of this research.Specifically, changes in the levels of income inequalit y and transnational corporate penetration should be added to the models to determine the impact of time on these processes.Different mca-;ures of income inequality should be compared.Other factors which have been found to affect income inequality should also be included to explore the interactions between variables, a-; well a-; to increase the explanatory power of the model.For example, ecological evolutionary theorists point our attention to agrarian population density a-; an important determinant of incom e inequalit y (Crenshaw and Ameen 1994).Additionally, further exploration of the impact of government regime type on income inequality is warranted.Given the incrca-;cd accessibility and reliability of data, important questions concerning the effect of globalization on dependency processes may now be empirically tested.The research presented here represents an important first step in exploring the effect of temporal change on the relationship between global structures and underdevelopment, providing evidence for the continued utility of World-System/Dependency explanations of these phenomena.
[Page 17] Journal of World-Systems Research NOTES l .Income level-; of the bottom 20% arc rarely used due to lack of variation.Most nation s have aro und 4-5% .
2_ .Fifteen observations arc for years prior to 1985 (ranging from 1979 to 1984, sec Appendix A), but were retained in the equations because of the increa-;c in sample size.Removing these nations from the analysis results in a loss of variation which leads to increa-;cd standard errors , thereby dccrca-;ing t -scorcs.While the a-;sociations between the variables remain the same a-; in the models examining the full data set, the significance of these relationshi ps is reduced in the equations examining the reduced data set.It is possible that this is a result of model underspecification, but is unlikely to be an indication that the reported relationships arc solely due to influence of those inequality mca-;urcs tem porally preceding the dependent variables .Moreover, a variabl e mca-;uring the year for which the income distribution data wa-; memmred was not signific ant nor found to alter the relationships between the remainder of variables in the model.

J.
Recently the World Bank ha-.made available an expanded income inequalit y data set which includes both Gini and quintile data over time for a large numb er of nation s, paying close attention to mea-.urement and comparability issues (Deinin ger and Squire 1996).The use of this data in future research seems quit e promi sing, a-. it allows for an examination of income inequalit y over time with a minimi zation of the methodolo gical problems which have plagued previous quantitati ve inequalit y research .
.1_ .A list of the nations designated a-.core countries in the creation of thi s dummy vari able is provided in Appendix A, and is taken from Crenshaw (1992 ) .
.5. .An indication of regional designation is provided in Appendi x A, and follows Tsai (1995).
Q.This is not to argue that the effects of FD I might not vary accordin g to geographical region.Tsai (1995) found evidence of a positive penetration effect on inequali ty for Ea-.t/Southea-.tA-.ia during the 70s.While a full examination of thi s issue is beyond the scope of th e present study, future res earch should explor e this issue.
1.I am currently workin g on a project concernin g this issue.

Journa I of World-Systems Research
.Touma I of ' World-Systems Research

Table 1
Regression of Income Percentage Share (1985)

Table l ,
equation l indicates that income share percentages do indeed exhibit a curvlincar relationship with logged real GDP per capita.That is, income inequality is greater where the national economy is at middle levels of development.This indicates support for the "level of development" paradigm, most commonly associated with Simon Kuznets and the "inverted U" curve