Commerce & Oppositions

Transcription

1 Commerce & Oppositions The Political Responses of Globalization s Losers Yotam Margalit Department of Political Science Stanford Univesity ymargalit@stanford.edu Abstract The integration of markets has greatly benefited some while leaving others behind and discontent. As the disparities between the gainers and losers sharpen, the political pursuit of the votes of the discontented is growing increasingly evident. In some countries the losers rally in support of parties offering economic compensation, but in other countries they support parties centered on a non-economic appeal, emphasizing religion, traditionalism or nationalism. This paper seeks to explain the variation in the political responses of the losers. I argue that many individuals view the economic changes that come with international market integration as inseparable elements of a broader "openness package" that includes also social and cultural changes. The linking of the two dimensions of change leads the losers to support programs that offer protection from both economic and cultural exposure. However, the translation of these preferences into votes is constrained by the electoral rules in place. Analyzing cross-national individual-level data as well as party positions and platforms, the paper offers evidence in support of the argument.

2 The representatives of the World Bank had better listen to us The World Bank and IMF and the transnationals must stop looting our natural resources, and stop privatizing basic services They must respect human life - Evo Morales, Bolivia s presidential candidate (November 2003) We need a president who will change the trade policy of this country. That s what I will do. And the folks who are 55 and 60 won t be losing their jobs to China and other countries around the world. - Richard Gephardt, Democratic Presidential Candidate Debate (October 2004). "[Rich countries ] approach is very simple. Should there be any new ideas in the social, economic or political field, they quickly come up with their own interpretations designed to benefit them. Thus we see them foisting on the world their moral values, political systems and economic globalization. - Mahathir Mohammad, Malaysia s Prime Minister (April 2000) Agricultural production in Poland is now lower than it's been in any time in the last 50 years.the reason is that the European Union and America and other countries have turned Poland into a dumping ground for their over-production. Andrzej Lepper, Leader of Poland s Self Defense, (January 2002) Introduction The effects of the liberalization of national economies, and of their opening up to international trade, have greatly benefited many but left many others behind and discontented. As the disparities between the gainers and losers from global integration sharpened, the political pursuit of the votes of the losers has grown increasingly evident. Electoral campaigns in different countries are increasingly dominated by anti-openness appeals, such as attacks on the loss of jobs to outsourcing, opposition to the privatization of public services sold to foreign multinationals, or calls for curtailing immigration flows. Political denouncements of the effects of global integration have become a common feature in many electoral campaigns, so much so that in some countries the successful mobilization 2

3 of the discontented is arguably one of the most notable political developments in recent decades. In Bolivia, for example, the Movement for Socialism won the support of the masses, its leader Evo Morales famously declaring himself the representative of the victims of neoliberalism and denouncing the influence of the US and of multi-nationals on the country. In France, Le Pen s National Front rallied voters by blaming immigration as the root cause of the country s problems and portraying it as a threat to the national way of life. In Poland, Andrzej Lepper s Self Defense successfully attracted many of the rural and smalltown voters by fiercely condemning the consequences of integration with Europe on Polish society and warning of its devastating effect on the agricultural sector. And in Israel, the religious party Shas successfully appealed to many of the country s left behinds by offering expanded welfare services as well as by advocating a return to traditional Sephardic values to counter secular foreign influences. As these examples suggest, the type of appeals by which political entrepreneurs succeed in shoring up the support of globalization s discontented appear to vary across countries. In some countries those that feel harmed by economic openness rally behind candidates promising extended social safety nets and protection from the vicissitudes of the international market, while in other countries they support parties centered on nationalist, traditionalist or even religious appeals. How can one make sense of these seemingly very different manifestations of discontent with globalization? To what extent can the globalization s self-perceived losers be considered a political constituency with coherent policy preferences, and if so, how can their voting behavior be explained? 3

4 The extant literature does not say much about the effects of global economic integration on actual voting behavior 1. The political response of the losers is even more of an understudied phenomenon. When the issue of the effect of economic integration on domestic politics is addressed, the standard international trade models are almost invariably used as the theoretical basis. Several such studies examine whether the patterns of domestic political alignments are in line with the theoretical predictions about the likely gainers and losers from trade liberalization 2. The evidence they provide suggests that trade theory can help explain some broad historical shifts in political coalitions, such as transitions from classbased cleavages to geographic or sector-based divides. Importantly though, in these analyses the political alignments are studied as a response to an economic phenomenon; the direct material effect of market openness on individuals is assumed to determine their policy preference. This paper suggests, in contrast, that by focusing almost exclusively on changes in relative prices and the resulting distributive aspects of economic openness, these analyses leave out an important part of what losing out from economic integration is about. In many countries, I contend, the constituency of the self-perceived losers contains many more individuals than merely the owners of the disadvantaged production factors or the workers in the sectors adversely affected by economic integration. Rather, the selfperceived losers also include individuals who feel that their social status is diminishing or that they are becoming marginal members in the increasingly open, fast transforming society of their countries. This sense has to do with a variety of causes such as feeling ill-equipped to cope because of the lack of basic proficiencies (e.g. proficiency in English, computer skills or even literacy), liberalization and shifting moral codes, or changing societal structures. 1 Throughout the paper I shall use the terms economic integration, economic globalization and market openness interchangeably. More specifically, these terms will refer to the integration of the markets for good, capital and labor. 2 See, for example, Hiscox 2002, Magee 1978, Rogowski

5 Crucially, these individuals, some of whom are not, in objective material terms, losing from global economic integration, still strongly feel as such. Equally importantly, these individuals are a voting constituency which is susceptible to party messages that appeal to such a sentiment, and that offer a sense of protection from the economic and social changes that their country is undergoing. Politically, this constituency is larger and more potent than the narrower group of workers directly disadvantaged by economic integration, those toward whom international trade models point. The prevailing approach in the literature to the study of gain and loss from economic integration appears limited in accounting for some of the political responses we observe, where the self-perceived losers vote choice seems to be poorly explained on strict economic grounds. This may be a result of the literature s focus on the distributive implications of trade liberalization, largely ignoring non-economic factors that could generate opposition to economic openness. This critique is the point of departure for this paper, which seeks to develop a new argument regarding the voting behavior of the constituency I am here referring to as the losers of globalization. From a definitional standpoint, it is important to stress that this paper focuses on explaining the political responses of the self-perceived losers from economic globalization, a group that does not necessarily consist only of those that are losers in a strict material sense. It is probably reasonable to expect that most of the objective losers also perceive themselves as such, but as I argue below, material loss is not a necessary condition for such perceptions of loss to come about; other (non-economic) aspects of global integration are also consequential in shaping people s beliefs regarding the effects of economic openness. Throughout the analysis then, when discussing the mobilization of the losers, I am 5

6 referring to the broader group of individuals that perceive themselves to be adversely affected by the process of global market integration. In its simplest form, the argument of the paper can be formulated as follows. Global market integration generates pressures for two types of changes that produce anxiety and discontentment: economic change and social-cultural change. For many individuals these two types of changes appear inherently tied; economic liberalization and the entry of foreign businesses are viewed as inseparable elements of a wider openness package, which also includes a variety of processes such as immigration and a changing ethnic composition, increased Americanization or secularization. Opposition to economic openness thus often reflects deep-level apprehension about these other social and cultural changes that people associate with market integration 3. Politically, these linked concerns mean that the constituency of self-perceived losers is characterized by unique support for policy platforms that, in broad terms, offer protection from both economic and cultural exposure. In terms of economic policies, the losers are more likely to support protectionist programs (higher tariffs on certain imports, and subsidies to shield local industries) and more expansive employment insurance schemes. But crucially, many of the losers also want protection from social change, and thus support a conservative or more exclusionist social agenda such that will maintain (or revive) the traditional social order. That is to say, the preferred policy bundle of the losers can be crudely described as combining the economic protection that the classical left provides, on the one hand, with the cultural protection that is typically offered by the social conservative right, on the other. The political translation of this preference bundle is constrained, I argue, by the electoral institutions in place. Electoral rules have implications for the likelihood that the 3 The mutual reinforcing relation between economic driven and culture-based concerns is the focus of Margalit (2006), which utilizes cross-national survey data and a controlled experiment to demonstrate this point. 6

7 losers become a distinct political constituency, as well as for the type of program along which they mobilize. In countries with a system of proportional representation, political entrepreneurs or parties espousing the preferred policy bundle of the losers are likely not only to be available but also to receive disproportionate support among the constituency of the losers. In countries with a majoritarian system, where the institutional incentives are such that only two parties effectively compete, the electoral competition is typically reduced to a single left-right dimension of competition 4. In those cases, the losers are likely to be constrained to vote on a single dimension of protection the economic or the cultural. Given that both dimensions could be of importance for the voters, the losers would not necessarily vote for the left party (i.e. for economic protection), but could instead vote for a party on the right that offer greater cultural protection, even if that party is more free-trade oriented. The result of being constrained by the electoral system to vote on a single dimension of compensation, I hypothesize, is likely to yield a lower articulation of the losers as a political constituency. In other words, the losers vote would often not differ substantially from that of the rest of the population. In sum, the paper makes the case that a framework that comprises the social-cultural dimension as well as the economic one is crucial for understanding the political repercussions of economic integration. I argue that the combination of preferences on the two dimensions, and the incentives generated by the institutional constraints, provide the key to explaining the voting behavior of the losers. The rest of the paper is organized as follows. The first section describes the existing literature on the domestic political implications of global integration and the main insights it offers for understanding the political preferences of the losers. Section II develops the 4 For example, Rae 1971, Cox

8 argument of the paper. Section III describes the data used in the analysis and discusses measurement issues. The next two sections present empirical tests of the argument: Section IV examines the differences between the policy preferences of the gainers and of the losers; Section V analyzes how these preferences translate into vote choice. Section VI discusses alternative explanations. The final section concludes and assesses the broader implications of the argument. I. Economic Integration and Domestic Politics The extant literature studies the political effects of global economic integration mostly at the national level. Much attention has been given to questions such as whether market openness leads to the decline of the welfare state, to an ideological convergence between left and the right, or to a decline in the economic accountability of incumbent governments 5. Noticeably, little has been written about the political repercussions of economic integration on subgroups in the population, whether in terms of policy preferences or in terms of actual voting behavior. There are thus no explicit hypotheses offered in the literature regarding the political preferences that economic openness generates, beyond the general expectation that the losers will support parties opposed to trade and financial liberalization and the gainers will support parties in its favor. In this section I summarize some of the main arguments the literature does offer regarding the political repercussions of global integration and derive a number of possible hypotheses on the losers vote choice. In doing so, I draw on three main bodies of literature: the research on the domestic sources of trade policy; comparative studies of the relation between globalization and political cleavages; and lastly, the institutional literature on the effects of electoral formulae on party competition. 5 Such studies include Garrett 1998, Hellwig and Samuels 2007, Milner and Judkins 2004, Rodrik

9 Models of international trade have been shown to offer some important insights on the political alignments that form in response to economic integration. The logic underlying most of these insights is the Heckscher-Ohlin model (and the Stolper-Samuelson theorem which followed it). Put simply, the model implies that the (objective) losers from trade liberalization are the owners of the relatively scarce factors under conditions of a closed economy; the gainers are those who own the relatively abundant factors. In face of pressures to liberalize the local economy, the owners of the scarce factors are thus likely to form political coalitions that oppose economic openness. This logic has been effectively developed by Rogowski to account for a number of historical political coalitions, ranging from Germany s marriage of iron and rye to those built in the U.S. around the New Deal 6. As these historical case studies demonstrate, trade integration has led potential gainers and losers to form competing political coalitions to advance their respective interests. Other studies have shown that the same H-O framework can be extended to account, at least partially, for individuals trade policy preferences 7. These studies give rise to some hypotheses regarding the losers vote-choice. The H-O model suggests that the characteristics of the losers from economic integration are likely to differ across countries as a function of their economies source of relative advantage in production 8. According to the model, the losers are the unskilled laborers in developed (skill-abundant) countries and the skilled labor in the developing (skill-scarce) countries 9. Since individual level of skill is highly correlated with income, the H-O logic suggests that 6 Rogowski See, for example, Mayda & Rodrik 2005; Scheve and Slaughter Importantly, the model defines the gainers and losers from trade strictly in objective material terms. I later examine the extent to which the implications of the model successfully predict the economic characteristics and the political preferences of the self-perceived losers. 9 The original version of the H-O model focuses on capital and labor as the factors of production. Later modifications to the model include land and a division of the labor category to skilled and unskilled labor. In recent specifications of the H-O model, countries are categorized as either skill-abundant or skill-scarce, based on their supply of skilled workers relative to other countries. 9

10 the losers in advanced industrialized economies are the unskilled poor, but in less developed economies the losers are mostly skilled laborers. The economic interests of the losers might therefore differ accordingly: In countries where the losers are predominantly from amongst the poor, political parties espousing leftist economic policies (e.g. income redistribution, broad welfare safety nets) would gain the disproportionate support of the losers. In less developed economies, the losers, mostly from the middle class, are less likely to rally in support of such a leftist economic agenda. Other hypotheses for explaining the losers vote stem from the research analyzing the political cleavage structure in modern democracies. A number of time-series studies examining the political competition in advanced industrialized countries have argued that recent decades of globalization have seen the waning of traditional political divides most prominently class-based voting - and the rise of new cleavages instead 10. Indeed, this is the expectation underlying Inglehart s argument that new cleavages that reflect post-materialist concerns are arising as societies grow wealthier 11. Instead of traditional class divisions, new wedge issues revolving personal-lifestyle issues or environmental policies become central. Kitschelt also holds that the political landscape is undergoing a fundamental shift, but offers a different account, by which changes in advanced industrialized societies have generated a new realignment that has shifted the main axis of voter distribution from a simple alternative between socialist (left) and capitalist (right) politics to a more complex configuration opposing left-libertarian and right-authoritarian alternatives 12. In a more 10 See Brooks, Nieuwbeerta, and Manze 2006; Franklin, Mackie and Valen 1992; Knutsen Inglehart Kitschelt 1994, p

11 recent cross-national study, Flanagan and Lee offer further evidence of an emerging values cleavage that contrasts holders of authoritarian and libertarian values 13. These studies do not offer clear hypotheses regarding the expected vote choice of the losers. However, by linking concerns about economic integration with an emerging values cleavage, these arguments open up the possibility that the constituency of the losers might be voting on a non-economic dimension. It could be that in some countries the losers vote for leftist parties that offer economic compensation but that in other countries they vote for what Kitschelt describes as the winning formula for attracting the holders of authoritarian values, namely a conservative social agenda combined with a free market economic platform. Case-studies of recent political competitions further confirm that economic integration is generating demonstrable repercussions on political behavior at the domestic level. For example, market openness is argued to have contributed to the emerging salience of ethnic cleavages in Latin America 14 as well as to have been prominent in the rhetoric of politicians competing for office in the region 15. In the Western European context, it is argued to have instilled new content into old contested political issues, and in that sense to have redefined the salient political cleavages 16. The fact that such analyses concentrate on a specific geographic region with little institutional variation, perhaps explains why less attention has been given to the role electoral institutions play in shaping or mediating the effects of economic integration on domestic politics. 13 Flanagan & Lee See Brysk Stokes Ignoring the campaign promises, Stokes argues, is possible because support for economic liberalization is often achieved ex post, i.e. after the reforms bear fruit. By this view, attracting the support of the losers is perhaps consequential in terms of electoral competition, but less so in terms of its effect on actual policy outcomes. 16 Kriesi et al

12 However, the growing literature on the relationship between institutions and political outcomes gives reason to expect that electoral formulae will have a significant effect on the political behavior of the losers 17. In majoritarian or single member district systems requiring plurality, the incentives of the voters and of the elites are such that only few parties compete 18. The political articulation of the preferences of smaller constituencies thus tends to be low. In contrast, in systems with high proportionality of vote-to-seats, parties representing smaller sub-sections of the population face a lower effective electoral threshold and are consequently more prevalent. This insight may have important implications for the type of party and program the losers vote for in different country contexts. I return to discuss the hypothesized implications of the electoral rules in the next section. In sum, the literature gives rise to several different expectations about the likely voting behavior of the losers. The hypothesis derived from the H-O model suggests that losers in advanced economies are likely to vote disproportionately for parties on the left, but that losers in developing economies will instead support parties that represent middle class concerns, some of which are parties located on the ideological right. The literature that links modernization to a shift towards a new dominant values cleavage, suggests that losers are not necessarily likely to vote for parties offering economic protection. Instead, a clash of values could be the central theme separating the vote choice of the losers from the gainers. These different arguments are examined in the empirical section of the paper. First, I present an argument of my own to explain the losers voting behavior. 17 See, for example, Austen-Smith 2000; Persson and Tabellini 1999 and For example, Cox 1997; Duverger 1954; Rae

13 II. Theory For many individuals, economic openness is viewed as having not only material consequences but also social and cultural consequences. Those individuals do not assess the effects of trade liberalization on their well-being, as separate from the social and cultural processes they associate with openness, such as secularization, a change in society s ethnic composition, or Americanization. The key to explaining the vote choice of globalization s losers lies therefore in recasting the analysis from that of a single dimensional phenomenon, focused solely on the economic effects of openness, to that of a two dimensional phenomenon, which comprises also a social-cultural dimension. The political implication of this argument is consequential, since it suggests that the self-perceived losers from economic integration often share preferences not only on economic policy but also on social-cultural issues. Seeking protection from economic exposure, the losers will tend to support policies that offer more substantial insulation from the open market, including increased government funding of retraining programs and more generous employment insurance schemes. They are also more likely to support economic protection in the form of higher tariffs on certain imports and subsidies aimed at shielding local industries from foreign competition. But seeking to curtail social changes that come with integration, the losers will also tend to support conservative agendas that oppose new forms of cultural heterogeneity, curtail foreign influences, and ultimately support the traditional social order. In some countries, such as Turkey and Israel, policies of cultural protection take the form of opposition to trends of secularization; in other countries, such as France, Austria or Switzerland, they take the form of rejection of immigration and its symptoms (e.g. opposing the construction of minarets in Mosques). The preference bundle of the losers can thus be somewhat crudely described as wanting the economic 13

14 protection that the left provides, along with the cultural protection that the conservative right typically offers. This preference bundle translates politically in varied ways across countries. The political response of the losers, I hypothesize, is constrained by the electoral rules in place. In countries with a proportional representation system, one can expect (a) that parties espousing a policy bundle of economically protectionist and socially conservative policies will be available, and (b) that such parties will receive a disproportionate support amongst the constituency of the losers. In countries with majoritarian systems, fewer parties typically compete 19. If the losers are not a constituency sizable enough to gain plurality, which is the case in most countries 20, one would not expect a party or candidate offering their preferred policy bundle to be common 21. The losers are thus likely to be constrained to vote for a single dimension of compensation, the economic or the cultural. As a result, the hypothesis is that in majoritarian systems the losers will either vote for a party offering a single dimension of compensation, or that there would not be an articulation of the losers as a unique voting bloc, i.e. their vote would not substantially differ from that of the rest of the population. Before proceeding, an important clarification is in order. The argument advanced here does not assume that global integration (economic or cultural) is the sole, or even the principal, concern for all the losers when choosing for whom to vote. Clearly, other issues such as national security or crime rates may be of greater concern for some. My qualified 19 This refers not only to purely majoritarian systems but also to mixed-member systems with a low degree of vote share-to-seats proportionality (see Shugart and Wattenberg, 2001). 20 The losers constituency is typically in the range between 10%-25% of the population (see Author, 2006). Although variation across countries exists, the losers group is rarely of a size sufficient to obtain victory in a plurality setting. 21 This expectation is borne from an empirical regularity, not from a theoretical proposition. Theoretically, the preferences for economic and cultural protection can be aligned together in a majoritarian setting too, but empirically, as appendix Table A2.1 shows, in majoritarian systems the bulk of the electorate is typically concentrated either on the (economic and social) left, or the (economic and cultural) right. See Stimson (2005) for a theoretical account for the process by which political competition in a majoritarian system converges into a single left-right divide that encompasses both economic and social policy dimensions. 14

15 hypothesis therefore is that the losers' support for parties that offer the policy bundle described above will be disproportionately large, as compared to the rest of the population, but not that all losers will, in absolute terms, necessarily support parties that offer this platform. The measure of interest is therefore the relative share of losers in the support base of a given party, as compared to the share of the losers in the total voting population. In sum, my argument rests on three related claims. First, that global integration generates pressures for change on two dimensions, economic and the social-cultural, and that many people view these changes as inseparably tied; Second, that the anxieties produced by these changes translate into a preference bundle for protectionist economic policies along with conservative social policies. And thirdly, that the actual manifestation of these preferences in terms of voting behavior is closely linked to institutional constraints, i.e. to the electoral system in place. The first claim was the center of investigation of a previous study 22. The rest of this paper focuses on the empirical testing of the second and third claims. III. The Data & Measurement Issues The analysis developed here requires data on a number of related issues. We first need to be able to classify who the self-perceived losers from economic integration are and to compare their policy preferences to those of the non-losers. The next task is to identify the parties disproportionately supported by the losers and to measure the positions these parties take on the policy dimensions of interest. And lastly we need to be able to rule out alternative explanations that attribute the vote of the losers to other policy dimensions. Unsurprisingly, there is no single source of data that enables all these different classifications and analyses. 22 Margalit

16 This section therefore briefly explains the different data sources I use to address these empirical challenges. As noted, the first empirical challenge is to examine the degree to which the losers and gainers differ in regards to their policy preferences. The key point of interest in this comparison is whether the losers share preferences not only on economic issues but also on questions dealing with cultural and social policy. To conduct this analysis, I use the Pew Global Attitudes Survey, which was carried out in forty four countries in The dataset is suitable for the purpose of this study because it includes data on respondents views on aspects of economic integration as well as on a host of policy-related issues. It therefore allows us to identify the self-perceived losers as well as examine their positions on a range of policy questions. The self-perceived losers from globalization are classified using respondents answer to the following question: Now thinking about you and your family do you think the growing trade and business ties between our country and other countries are very good, somewhat good, somewhat bad or very bad for you and your family? I code as losers those that answered the question as either very bad or somewhat bad. All other respondents are non-losers by this definition. Next, I examine the voting preferences of the losers. This analysis requires a dataset that enables the classification of respondents as either losers or not-losers but also provides information about the party for which they voted. The International Social Survey Program (ISSP) study from 2003 is the only cross-national dataset that meets these requirements in a satisfying manner 23. The ISSP data includes responses of 44, The PEW Global Attitudes survey used for analyzing the policy preferences of the losers does not include information about the political party for which the respondents voted. 16

17 individuals in thirty four countries 24. The question I use to categorize gainers and losers from economic integration reads as follows: How much do you agree or disagree with the following statement: [Respondent s Country] should limit the import of foreign products in order to protect its national economy. Responses where categorized on a five point scale, ranging from agree strongly to disagree strongly. The use of this question for classifying gain and loss from economic integration is not without problems. First, the question asks about preferences on trade policy, rather than a more direct question (as the one in the Pew study) about the respondent s perceived effect of economic openness on herself and on her family. Second, as has been shown by Hiscox (2006), responses to questions on trade are susceptible to framing effects. By posing the question as a tradeoff between free trade and the strength of the national economy, survey respondents are likely to be more supportive of at least some form of restrictions on trade. To counter this problem, I only categorize as losers those who strongly agree with the statement in question, i.e. those that hold unambiguously negative views about the effects of trade openness. This classification also generates results that are more comparable with the Pew data in terms of the size of the group of self-perceived losers 25. An important advantage of using the ISSP survey is that the exact same survey was run close to a decade earlier (in 1995), which enables some degree of inter-temporal analysis in addition to a cross-sectional one. 24 The figure of 34 countries is obtained after consolidating the data for East and West Germany into a singlecountry sample, and doing the same with the Jewish and Arab populations in the sample of Israeli citizens. In two countries (South Africa, Bulgaria) respondents were not asked for the political party of their choice. In two additional countries (Taiwan and the Philippines) only about fifteen percent of the respondents answered the question regarding their party preferences. These four countries are therefore excluded from the analysis that pertains to the political preferences of the losers. 25 The ratio of losers in the PEW data is approximately one in seven. In the ISSP survey the ratio is closer to one in five. 17

18 The key question of interest in this analysis is what parties the losers support disproportionately, compared to the rest of the population. To ensure that we classify only parties that patently attract disproportionate support among the losers, and to avoid analyzing esoteric parties, I code a losers party only if it fulfills two conditions: (i) The losers support the party at a rate that is at least 33% higher than their share in the voting population; and (ii) The party received at least 5% of the actual national vote 26. With these rules, the loser s disproportionate support is calculated as: Over-representation For example, if 30% of a party s supporters are losers, but the losers represent only 20% of the national voting population, then the over-representation of the losers support is calculated as: (30/20-1)*100= 50%. The analysis then focuses on whether the losers parties take policy positions that are consistent with the paper s theoretical prediction. The argument advanced here suggests that these parties are likely to represent the losers preference bundle of a protectionist economic policy and a conservative social policy. To examine whether that is indeed the case, I use the Party Positions in Modern Democracies (PPMD) dataset 27, which places parties along a 20-point scale on a number of policy dimensions. The coding of the parties policy positions is based on a survey of country experts. The n-size for each country ranges from 3 (Bulgaria) to 167 (US), and the country median for coding positions on each dimension is twenty experts. 26 Any choice of a threshold is obviously somewhat arbitrary. I therefore report in Section V and in the Appendix results of sensitivity analyses that test the robustness of the findings to different threshold specifications. As is shown, all substantive findings are not sensitive to the changes of the threshold. 27 Benoit & Laver,

19 The economic policy dimension that the PPMD codes ranges from promotes raising taxes to increase public services to promotes cutting public services to cut taxes. This continuum does not capture the entire dimension of economic protection, which includes additional issues such as subsidies for local industries or support for imposition of tariffs, but it gets at a key aspect of economic protection, namely the extent of governmentfunded programs 28. The social policy dimension ranges from favors to opposes liberal policies on matters such as abortion, homosexuality and euthanasia. I use parties positions on this standard social conservative-liberal continuum as a proxy for their cultural protection appeal 29. Lastly, to examine not only party positions on the issues but also the degree of salience each policy issue has had in a party s platform, I use the Comparative Manifestos Project (CMP) dataset. The CMP classifies each quasi-sentence in a party manifesto into one of 56 categories 30. Such categories include social services limitation, nationalization, or environmental protection". The relative frequency that each category appears in the manifesto is then calculated as a percentage of the total number of quasi-sentences in the manifesto. And so, for example, if a certain party's manifesto has fifty such quasi-sentences, ten of which deal with the need to advance legislation against gas emissions and in 28 Unfortunately, parties positions on the globalization dimension were coded only for parties in France. No other dimension of economic protection can thus be used. 29 Ideally the social policy dimension should have covered issues such as traditionalism or national values. Unfortunately the only dimension which could have offered a useful alternative to test (the nationalism dimension ) was coded for only half the countries covered in the ISSP survey. The nationalism dimension coded parties on a 20-point scale ranging from Strongly promotes a cosmopolitan rather than a national consciousness, history and culture to the extreme opposite : Strongly promotes a national rather than a cosmopolitan consciousness, history and culture. The correlation between the social dimension and nationalism dimension in the countries included in the ISSP survey is This seems to support the notion that the social dimension coded in the PPMD dataset is a reasonable proxy for party s stance on the cultural policy dimension. 30 The Comparative Manifestos Project defines quasi-sentence as an argument. An argument is the verbal expression of one political idea or issue. In its simplest form, a sentence is the basic unit of meaning. For a detailed description of the coding see: 19

20 advocating government investments for fighting climate change, then that would be coded as 20% under the Environment category 31. This data offers some indication of the degree to which different policy issues are central in the offerings of the different parties. As noted earlier, the same ISSP survey was carried out in 1995 and again in I code the party positions in each period using the manifestos from the closest election campaign preceding the time in which the survey was fielded. The CMP dataset codes manifestos for twenty-two of the countries covered in the 2003 ISSP survey and of twenty of the countries covered in the 1995 survey. In total, sixteen countries have complete ISSP data coded for both survey periods. 32 [See Table A5 in the Appendix for complete list of countries]. The range of data sources described above is necessary for substantiating the claim that the combination of concerns on the two dimensions, the economic and cultural, is the key to explaining the losers policy preferences and vote choice. The empirical sections advance this claim in the following manner: First, I show that the losers share a distinct preference bundle that is protectionist on both economic and social-cultural policy issues. Second, I show that this preference bundle is reflected in the platforms of the parties for which the losers vote, under the constraints of the electoral system. Lastly, I address alternative explanations and show that other policy dimensions hypothesized in the literature to explain voting behavior do not account for the vote choice of the losers. 31 The calculation is: 10/50= Russia and Latvia had a parliamentary election coded in one of the periods and a presidential in the other. A comparison across time in these two cases is thus not possible. 20

21 IV. The Policy Preferences of the Losers The argument advanced in the paper is based on the contention that people s apprehension about the economic aspects of market openness are strongly linked to their concerns about social and cultural changes they associate with global integration. The relationship between the two, I have argued elsewhere, is mutually reinforcing. The first hypothesis is therefore that the losers are likely to be more opposed to aspects of openness not only on economic issues but also on social and cultural issues. To test this hypothesis, I present data from the Pew Global Attitudes Survey carried out in To examine the differences in preferences on economic policy, I compare the support of losers and the non-losers to a free market economy as well as their respective answers to question about whether the government is responsible for the needy 33. (See appendix for exact wording of the questions). Figures 1a - 1b graphically present the differences in views. The X-axis measures the total percent of respondents that hold a certain view on any given question. The blue circles denote the percent of losers who answered the question in a given way out of the total population of losers; the hollow diamonds denote the corresponding figure for the non-losers, again as a percentage of the total population of non-losers. The countries for which the data corresponds are ordered on the Y-axis, and sorted in ascending order by the responses of the non-losers. [Figure 1 about here] Figure 1.a presents the percent of respondents who disagreed with the statement: Most people are better off in a free market economy, even though some people are rich and 33 As described in the data section, the self-perceived losers are classified using respondents answer to the question: Now thinking about you and your family do you think the growing trade and business ties between our country and other countries are very good, somewhat good, somewhat bad or very bad for you and your family? I code as losers those that answered the question as either very bad or somewhat bad. 21

22 some are poor. As one might expect, in most countries the gaps between the preferences of the non-losers and the losers are evident and statistically significant (p<0.001). In a majority of the countries the losers were at least fifteen percentage points more opposed to a free market. In France (40%) and the Slovak Republic (31%) the gaps are most substantial. Among US respondents, who are overall most supportive of a capitalist system, the losers report opposition to a free market economy at a rate that is still almost twice as high as that of the gainers (35% vs. 18%). Figure 1b presents the percent of respondents that completely agreed with the following statement: It is the responsibility of the (state/government) to take care of very poor people who can't take care of themselves". In most countries there is a clear majority of losers that seem to hold the view that the government is responsible for taking care of very poor people 34. The differences between the two groups in their views on the question are in many countries quite small and statistically insignificant, though consistently in all but two cases (Italy and Turkey) the losers are more supportive of an active government role in supporting the poor. Turning to positions on social issues, let us compare the views on cultural aspects of globalization as well as on a traditional cultural wedge issue. Figure 1c compares respondents answers to the question What about the way movies, TV and music from different parts of the world are now available in (survey country) do you think this is a very good thing, somewhat good, somewhat bad or a very bad thing for our country? The results are sorted by the percent of gainers that described the effect as bad. The differences are very clear; in all countries the losers are much more negative in their views of foreign cultural influences (p<0.001). The differences are most notable in Eastern European countries, such as Bulgaria and Russia, where the percentage gap is over forty percentage 34 Respondents had the option of selecting one of four answers: completely agree, somewhat agree, somewhat disagree and completely disagree with the statement in question. 22

23 points. Proponents and opponents of economic integration clearly are in disagreement about much more than just economic policies. Figure 1d shows differences in views on homosexuals. The chart presents the percentage of respondents agreeing with the statement Homosexuality is a way of life that should not be accepted. The chart shows that in all countries, the losers population holds a more conservative view (significant at p<0.001). In some countries (e.g. Mexico, Argentina) the differences are small, while in other countries (e.g. Bulgaria, Japan) the gaps are substantial. Importantly though, the gaps on this question, a classic litmus test of conservatism, are much narrower than those in the previous question that dealt directly with views on cultural exposure to the world. This indicates that the difference in views between gainers and losers is not merely a reflection of a conservative-liberal cleavage, but one that is most acute on issues revolving the effects of global integration. Importantly, the differences in views between losers and non-losers remain statistically significant in most cases also when controlling for respondents demographics such as income, educational attainment, gender and age (see appendix Table A1 for significance levels in a multivariate analysis). In sum, the data suggests that the self-perceived losers from globalization tend to hold distinct views not only on economic policy questions but also on social and cultural issues. In fact, the distinctness of these views is even more apparent on social issues than on economic ones. Given that the stance of the losers is generally more leaning towards the left on economic policy questions but more to the right on social policy questions, it is not obvious what parties they would actually vote for. Addressing this question is the focus of the next section. 23

24 V. The Vote Choice of the Losers The policy preferences of the losers do not fall along the classic left-right cleavage but instead appear to cross-cut it. The losers partisan affiliation is therefore somewhat of a puzzle. In this section I use data from the International Social Survey Program (ISSP) to address this question and examine the extent to which my argument helps account for how the losers vote. Let us start with the null hypothesis derived from the Heckscher-Ohlin model. Assuming that people s skill level is positively correlated with their earnings and that wealthier nations are more skill-abundant, the H-O model suggests that the losers from economic openness are the poor in advanced economies but members of the middle class or higher in developing countries. If the poor tend to favor policies of redistribution and government funded welfare safety nets, we would expect the losers to disproportionately support the left in advanced economies, but less so in developing countries 35. Figure 2.1 presents the relationship between the countries economic development (measured by GDP per capita) and the mean income quintile of the losers. Very much in line with the expectation of the H-O model, there clearly seems to be the expected negative relationship: in less developed economies the mean income of the losers is around the third (middle) quintile while in advanced economies it is closer to the bottom of the second quintile. The light grey lines, denoting the 95% confidence interval clear indicate that the negative relationship is statistically significant. Alternatively, if we examine the percent of losers located in the bottom two income quintiles, the same negative relationship between the economic standing of the losers and the economic development of the country remains. 35 The logic of the H-O model generates clearer expectations regarding the vote of the losers in advanced economies, but less so for the losers in developing economies. One expectation is clear: the relationship between skill-abundance (proxied by economic development) and the support of the losers for the left should be positive. 24

25 [Figure 2 about here] However, whereas the H-O prediction regarding the income level of the losers generally holds, the voting behavior hypothesized to flow from it does not. Figure 2.2 presents the relationship between per capita GDP and the losers disproportionate support of the left (compared to the non-losers) in each of the countries included in the ISSP survey 36. Strikingly, the correlation is negative, the exact opposite of the hypothesized relationship: In advanced economies, where the losers are located lower down the income ladder, they vote less for the left than the non-losers do. Contra to the expectation, in less developed economies the losers vote disproportionately in support of the left. This same trend holds also when examining the absolute percent vote of the losers to the left compared to that of the non-losers 37. The H-O logic, in other words, does not extend to account for the observed voting behavior of the losers. An alternative approach to understanding the vote choice of the losers is to analyze the specific parties that the losers support. Table 1 lists all the losers parties, defined as parties that have an over-representation of losers of 33% or higher amongst its voters, and that obtained at least five percent of the national vote 38. In a majority of the countries (62%) included in the ISSP survey, at least one party received such support. Reassuringly, an examination of the list of parties in Table 1 seems to successfully pass a face validity test. In almost cases, the parties identified in the list are those which most people familiar with the 36 The losers disproportionate support for the left is calculated as the ratio of the losers percent of vote for parties on the left divided by the non-losers vote for the left. For example, if the 60% of the losers vote for parties on the left but only 40% of the non-losers do, then the disproportionate support of the losers would be 60/40= Disproportionate support for the left amongst the losers is calculated as the ratio of the losers vote for the left divided by the left vote among the non-losers. A ratio of 1 in a given country means that the losers and the non-losers supported the left at the same rate. The classification of left-right parties is based on the ISSP coding. 38 The method for calculating the losers over-representation was explained in Section III. 25