Why do the less educated oppose trade openness? A test of three explanations in the Netherlands Open Access

In both the United States and North-Western Europe the introduction of free trade treaties is highly debated. Some favour the deregulation of flows of goods and services across national borders, while others oppose it. This is, for instance, reflected in current debates surrounding the introduction and make-up of the Transatlantic Trade and Investment Partnership (Charlemagne, 2015), and in those of more than two decades ago on the introduction of the North American Free Trade Association (NAFTA). In the latter case opponents could be especially often found among American workers. One of their arguments was voiced by Ross Perot during a 1992 US Presidential debate. He predicted that the introduction of NAFTA would lead to ‘a giant sucking sound going South’, a metaphor referring to the relocation of industrial production from the United States to Mexico.

In both political science and sociology, scholars have paid ample attention to public disapproval of such forms of economic openness, which can be referred to as opposition to trade openness. Time and again, this line of research has demonstrated that the less educated are less in favour of economic openness than their more educated counterparts. This has been established for various advanced economies: in numerous studies focusing on trade openness in the United States (Fordham & Kleinberg, 2012; Hoffman, 2010; Mansfield & Mutz, 2009; Scheve & Slaughter, 2001), in seventeen countries including Great Britain, France, and Italy (Edwards, 2006), in ‘twenty-three countries, including the United States, Canada, Japan, many Western and Eastern European countries’ (Hainmueller & Hiscox, 2006, p. 481; cf. Mayda & Rodrik, 2005), and in Australia, Norway, West Germany, Spain, Switzerland, and the United States (Kaltenthaler, Gelleny, & Ceccoli, 2004), as well as in research on self-perceived loss as a result of international economic integration using ‘40 national surveys carried out in five continents in 2002’ (Margalit, 2012, p. 488).

From the earliest studies onwards, opposition to free trade among the less educated has been interpreted in economic terms (with reference to the ‘factor endowment model’ or the ‘Stolper-Samuelson theorem’), reminiscent of Perot's argumentation addressed above. According to this line of reasoning, the highly skilled (read: the more educated) in advanced economies are winners from free trade, while the low skilled (read: the less educated) are losers. The latter would, therefore, oppose open borders from motives of economic self-interest (see e.g. Hainmueller & Hiscox, 2006; Mayda & Rodrik, 2005; Scheve & Slaughter, 2001). However, various empirical studies cast doubt on this explanation (Hainmueller & Hiscox, 2006; Mansfield & Mutz, 2009; Margalit, 2012). Drawing on the literature on support for trade openness and the wider literature in cultural and political sociology and political science, we therefore formulate and test two alternatives: dereification theory, which focuses on the role of cultural capital and resistance to cultural diversity, and a knowledge-based theory that centres on an understanding of the macro-economic gains of free trade. The first one moves beyond studies demonstrating an empirical association between opposition to trade openness and a general resistance to cultural diversity (see e.g. Edwards, 2006; Kaltenthaler et al., 2004; Mansfield & Mutz, 2009; Mayda & Rodrik, 2005; O’Rourke & Sinnott, 2001; Wolfe & Mendelsohn, 2005). Inspired by recent pleas for taking seriously the ramifications of differences in status and cultural repertoires for social inequalities (Lamont, 2016; Lareau, 2015; Ridgeway, 2014), we provide a systematic cultural explanation for why the less educated oppose cultural diversity. The second alternative explanation further develops the suggestion that the education gap in support for trade openness is in fact a knowledge gap (Hainmueller & Hiscox, 2006), by positioning it in the broader literature on political knowledge.

In what follows, we discuss and test the prevailing explanation, as well as the two alternatives.

The dominant explanation of the low support for trade openness among the less educated in developed economies is the so-called ‘factor endowment model’. This argues that the highly skilled (that is, the more educated) in advanced economies have more to gain from free trade than their lower skilled (that is, the less educated) counterparts. The reason advanced is that the (relative) demand for less skilled workers declines when trade is conducted with countries with an abundance of low-skilled workers, while demand for more highly skilled workers increases. As a result, the real wages of the more educated rise, while those of the less educated fall.

The positive relationship between the level of education and support for trade openness has, in accordance with this line of argument, often been ‘seen as straightforward confirmation that high-skilled individuals in skill-abundant economies perceive that trade openness raises their real wages while low-skilled individuals calculate that their real wages decrease with trade liberalization’ (Hainmueller & Hiscox, 2006, p. 491). According to the dominant explanation, education thus functions as an indicator of one's labour-market position – that is, as an ‘economic variable’ (Mayda & Rodrik, 2005, p. 1418) – in relation to support for trade openness. Opposition to trade openness among the less educated is consequently interpreted as a matter of the self-interest of those whose economic standing falls as trade openness increases.

We will refer to this line of argument as the ‘economic-insecurity explanation’. If this explanation is empirically valid, opposition to trade openness would be a redistribution issue, and the higher levels of opposition among the less educated would be attributable to their low income, their weak labour-market position, and their job insecurity.

It should, however, be noted that previous research casts doubt on these hypotheses. There are various indications that (a) a weak economic position does not affect support for trade openness, and (b) support is not consistently lower among those who are active in the labour market than among those who are not, such as students and retirees, as demonstrated by studies including dozens of advanced economies (Hainmueller & Hiscox, 2006; Mansfield & Mutz, 2009; Margalit, 2012). Alternative explanations for the opposition to trade openness among the less educated are, therefore, required.

Dereification theory, the first alternative explanation, starts with the observation that the economic-insecurity explanation is based on two interrelated assumptions that can be questioned. First, support for trade openness is interpreted as a redistribution issue. Secondly, it is assumed that a low level of education simply functions as an indicator of a weak labour-market position and, therefore, indicates economic insecurity. Rethinking these assumptions informs an alternative explanation for the greater opposition to trade openness among the less educated that does not revolve around their economic self-interest, but instead focuses on their well-documented higher average level of aversion to cultural diversity.¹

First, trade openness entails processes that transcend national borders and cultural divides. The public might therefore consider it to be a ‘cultural-order’ issue rather than a redistribution issue. This has been hinted at in studies which demonstrate that opposition to economic openness is related to nationalism, ethnocentrism, and the like, while also pointing out that these sentiments can most notably be found among the less educated (e.g. Edwards, 2006; Kaltenthaler et al., 2004; Mansfield & Mutz, 2009; Mayda & Rodrik, 2005; O’Rourke & Sinnott, 2001; Wolfe & Mendelsohn, 2005). Accordingly, support for trade openness might be related to the well-known cultural-value dimension that pits those in favour of a rigid socio-cultural order against those who appreciate individual freedom and cultural diversity. This is instead of support being linked to the economic egalitarianism versus economic laissez-faireism value dimension, which pits the economically weak against the economically strong.

It is crucial to distinguish between these two value dimensions, which can be labelled as ‘economic egalitarianism/laissez-faireism’ (or ‘economic egalitarianism/anti-egalitarianism’) and ‘cultural progressiveness/conservatism’ (a descriptive label that could also be read as ‘cultural libertarianism/authoritarianism’ or ‘social liberalism/conservatism’). This distinction was already established half a century ago in Lipset's (1959) seminal work on the social bases of politics, and ever since then many survey studies on liberal democracies have consistently shown that these two value dimensions are empirically unrelated among the public at large (for a recent corroboration of this finding and an overview of previous studies reporting it, see Achterberg & Houtman, 2009). One's level of economic egalitarianism does not, therefore, predict one's level of cultural conservatism: they represent distinct ideological domains. If the public considers trade openness to be a cultural-order issue instead of a redistribution one, an explanation for opposition towards trade openness might, as a consequence, need to be found in the sources of resistance to cultural diversity instead of in those underlying egalitarianism.

Secondly, it is important to note that sociologists have demonstrated that one's level of education not only indicates one's economic position, but also one's status position referred to as ‘cultural capital’ (Bourdieu, 1984; Houtman, 2003; Manevska, Van der Waal, Achterberg, Houtman, & De Koster, 2010; Van der Waal & Houtman, 2011). It indicates the extent to which one appreciates, demonstrates, and masters the lifestyle of the cultural elite. This lifestyle is encompassing, ranging from specific cultural tastes and manners to child-rearing practices and food preferences. The most notable example, and therefore the most frequently used indicator for cultural capital, involves the extent to which one appreciates and understands intellectually demanding cultural expressions such as highbrow art forms (Bourdieu, 1984, 1986).²

According to most class analyses, cultural capital is not a class marker but a status or prestige indicator (Wright, 1997, 2005), in accordance with the Weberian distinction between class and status (Weber, 2006 [1922]). However, according to Bourdieu, cultural capital is an intrinsic part of class, as his class analysis entails an ‘endeavor to rethink Max Weber's opposition between class and Stand [i.e., status]’ (Bourdieu, 1984, p. xii). Bourdieu's point is that the higher classes employ their high levels of cultural capital to exclude from privileged positions those with a low amount of cultural capital. That is undeniably so, and we recognise that cultural capital does indeed serve exclusionary practices and the reproduction of class and status differences (Jæger & Breen, 2016). Yet, for the issue at hand (distinguishing between the sources of cultural conservatism and those of economic egalitarianism), merging class markers – such as labour-market positions and the income differences these result in – with status markers – such as cultural capital – obscures the fact that these can have fundamentally different antecedents, functions, and meanings, and consequently underlie different attitudes for altogether different reasons.

Class markers indicate one's economic standing as determined by one's labour market position or position in the production process (Wright, 1997, 2005). Together with the associated indicators income and job security, these can be used for empirically testing the economic-insecurity explanation. As outlined in the previous section, this theory states that opposition to free international trade reflects the economic self-interest of those with weak economic positions (indicated by weak labour-market position, low income, and high job insecurity). In contrast, dereification theory holds that the status marker of cultural capital affects support for free trade, but for very different reasons than those assumed in the economic-insecurity explanation.

The first and crucial analytical step is that the status marker of cultural capital – the extent to which one appreciates, demonstrates, and masters the cultural codes of the elite – also indicates ‘the ability to recognize cultural expressions and to comprehend their meaning’ (Achterberg & Houtman, 2006, p. 78; see also: Houtman, 2003; Houtman & Achterberg, 2010; Houtman, Achterberg, & Derks, 2008; Manevska & Achterberg, 2013; Manevska et al., 2010; Van der Waal & Houtman, 2011). This pertains to both institutionalised and embodied cultural capital (Bourdieu, 1984, 1986). The former indicates the extent to which one has developed cultural capital by means of the education system, as the lifestyle of the cultural elites is particularly salient in institutions of higher education. Embodied cultural capital, on the other hand, refers to cultural capital in the form of long-lasting dispositions of the mind and body that result from lifelong socialisation in certain milieus, most markedly in one's family from early childhood onwards. Although institutionalised and embodied cultural capital are substantially related, they partly operate independently from one another: some individuals who have been socialised in cultural elite milieus do not attain the highest levels of education, while some of those who are well educated do not acquire elite cultural codes from their academic environment and/or may have been born and raised in more modest milieus. As noted, both institutionalised and embodied cultural capital are not only relevant for the reproduction of social inequalities, but also indicate the ability to recognise cultural expressions and to comprehend their meaning.³ This is crucial for the issue at hand, because it results in a so-called ‘dereified worldview’ (Gabennesch, 1972). In what follows, we will discuss what this means, how it comes about, and how it relates to opposition to cultural diversity in general and opposition to trade openness in particular.

Those with more cultural capital have a more dereified worldview, because the ability to recognise cultural expressions and to comprehend their meaning is conducive to what can be termed the ‘denaturalization of culture’ (Manevska, 2014; Manevska et al., 2010; Van der Waal & Houtman, 2011). It stimulates the idea that cultural categories do not exist as givens of nature, but are social constructions (Manevska, 2014; Manevska et al., 2010, p. 7). Accordingly, those with high levels of cultural capital do not tend to conceive of ‘other’ cultures as ‘deviant’ or ‘problematic’: they are more inclined to observe different, changing and, ultimately, arbitrary value patterns and lifestyles (cf. Gabennesch, 1972). On the other hand, those with less cultural capital are generally less inclined to conceive of their own cultural expressions as ultimately arbitrary ways of dealing with the complexities of everyday life and as equally valid as ‘other’ or ‘foreign’ cultural expressions. Instead, they are more predisposed to sanctify their own cultural expressions as absolute, pre-given, and pertaining to an infallible order to which one ought to adapt. They have, in other words, on average a more reified worldview than those with higher levels of cultural capital. It is, therefore, not a surprise that various studies in the Dutch context demonstrate that low levels of cultural capital – measured in terms of either a low level of education or a low appreciation and consumption of highbrow cultural expressions – on average go hand-in-hand with conservative stances on various cultural-order issues, ranging from authoritarianism (Houtman, 2003; Houtman et al., 2008) to ethnocentrism (Manevska, 2014; Van der Waal & Houtman, 2011), welfare chauvinism (Manevska et al., 2010), and gender conservatism (Houtman, 2003).

The observation that cultural capital generates a dereified worldview that underlies culturally progressive attitudes is further underlined by longitudinal and cross-national research, which demonstrates that this is not a universal pattern, but one that is characteristic of modern liberal democracies. In less modern and less liberal-democratic countries, those with ample cultural capital are barely more progressive on cultural issues than those with low amounts of it (cf. Farnen & Meloen, 2000). This is not a surprise if one takes into account the fact that in those countries the arts and (higher) education are both directed at legitimising and reinforcing the established cultural order, instead of challenging it, as discussed by Manevska and colleagues (2010, p. 8). Mediaeval art, for instance, was religiously inspired and directed at legitimising and re-confirming the Christian worldview (Wilson, 1982). Furthermore, in authoritarian regimes (such as Nazi Germany), modern, avant-garde art is often considered to be a threat to the established order, as it aims to ‘denaturalise’ established cultural patterns and meanings by deconstructing, critically challenging, and disturbing them (Bell, 1976; Jensen, 1995).

A similar line of reasoning can be applied to the differences between educational institutions in modern liberal democracies and those in authoritarian regimes. Whereas education in the latter is directed at maintaining the status quo, in contemporary liberal democracies it stresses the importance of freedom in the cultural domain (Gabennesch, 1972). In the curricula in liberal democracies, especially at the highest educational levels, questioning the established social and cultural order may even be encouraged (e.g. via ‘critical’ or ‘reflexive’ thinking). This is reflected in internationally-comparative studies that demonstrate that the effect of education on cultural progressiveness is stronger in more established democracies (e.g. Coenders & Scheepers, 2003; Hello, Scheepers, & Gijsberts, 2002).

In summary, according to the dereification theory discussed here, the well-documented finding that the less educated report higher levels of cultural conservatism (see e.g. Emler & Frazer, 1999; Hainmueller & Hopkins, 2014; Stubager, 2008, 2009) cannot be understood from their weak economic position, but rather from their low levels of cultural capital. This is in line with previous studies which show that cultural capital indicators and indicators of one's economic position can be disentangled both analytically and empirically (Houtman, 2003; Van der Waal & Houtman, 2011), and that low levels of cultural capital underlie ethnocentrism and other measures of resistance to cultural diversity, such as ethnic threat, welfare chauvinism, and authoritarianism, while a weak economic position barely explains the cultural conservatism among the less educated, or does not do so at all (Houtman, 2003; Manevska, 2014; Manevska & Achterberg, 2013; Manevska et al., 2010; Van der Waal & Houtman, 2011).

An additional indication that the higher level of resistance to cultural diversity among the less educated needs to be understood according to dereification theory instead of theories centring on economic self-interest, is the fact that the latter cannot apply to aspects of cultural diversity where self-interest in an economic sense is not at stake. Opposition to trade openness and immigration might be driven by economic self-interest: due to their weak labour-market position, the less educated bear the economic burden of these processes – be it perceived or real – and it is therefore in their self-interest to oppose these on the basis of their weak economic position (cf. Hainmueller & Hopkins, 2014). However, for equally salient aspects of opposition to cultural diversity among the less educated, for example opposition to non-hegemonic sexual identities, favouring the death penalty, authoritarian child-rearing practices, and a preference for law and order policies, one can only wonder what the relationship with a weak economic position might be. These issues simply have no clear distributional consequences, and a theoretical rationale for why the conservative preferences of the less educated on these issues would be driven by a weak economic position is therefore lacking (Houtman, 2003, p. 27). Their cultural conservatism can, on the other hand, be understood by a desire for a rigid socio-cultural order brought about by a low amount of cultural capital, as implied by dereification theory.

Applied to the issue at hand, dereification theory implies that the greater opposition to trade openness among the less educated is not driven by the self-interest of the economically weak, but entails the opposition to cultural diversity of those with low levels of cultural capital. This theory, therefore, predicts that opposition to trade openness pertains to the cultural-value domain, and that the higher levels of opposition among the less educated can be attributed to their, on average, lower levels of cultural capital.

By providing a systematic theoretical rationale, this line of argument could crucially add insight to prior work suggesting that opposition to trade openness among the less educated might somehow be value-driven instead of self-interest-driven. Most of this work is based on the finding that opposition to economic openness and various indicators of cultural conservatism go together (e.g. Edwards, 2006; Kaltenthaler et al., 2004; Mansfield & Mutz, 2009; Mayda & Rodrik, 2005; O’Rourke & Sinnott, 2001; Wolfe & Mendelsohn, 2005). Some studies go a step further and show that the effect of education on pro-trade preferences is (partially) explained by national pride (Mayda & Rodrik, 2005) or isolationism, ethnocentrism, and nationalism (Mansfield & Mutz, 2009). However, as proponents of economic self-interest explanations have argued in reply to the latter study, ‘the meaning of a strong relationship between one attitude and another is far from obvious’ (Fordham & Kleinberg, 2012, p. 316). In other words, it could well be that the less educated are simultaneously ethnocentric and averse to trade openness, while their aversion cannot be causally explained by their higher levels of ethnocentrism and other culturally conservative attitudes (p. 323). However, a recent study focusing on self-perceived loss from economic openness (Margalit, 2012) has advanced the cultural argument in an important way by demonstrating (by means of experimental research in the United States) that cultural threat rather than economic threat partially explains the education effect. Even though this study focuses on a dependent variable that, strictly speaking, is not an exact measure of support for trade openness, it strongly suggests that the negative relationship between such support and education should be interpreted culturally instead of economically.

If dereification theory is empirically corroborated, this would significantly advance this cultural argument: it would provide a systematic cultural explanation for why the less educated oppose trade openness, instead of merely indicating that their opposition is substantially related to their resistance to other aspects of cultural diversity or cultural threat. Moreover, it would indicate that their opposition to free trade, just like their cultural conservatism in general for that matter, is rooted in their low status position, which would resonate with recent arguments that status stratification is crucial for understanding contemporary inequalities (Lamont, 2016; Lareau, 2015; Ridgeway, 2014).

A second alternative explanation for the opposition to free trade among the less educated revolves around their lower levels of what might be considered ‘political knowledge’. The political-knowledge explanation posits that those with less political knowledge are less able to formulate well thought through opinions because they are less acquainted with political and policy issues and processes. This has been empirically corroborated several times in studies on the United States: unlike those with ample political knowledge, their counterparts with less are, for example, less able to translate their interests, attitudes, and ideology into support for the ‘corresponding’ policy, political candidate, or political party (Delli Carpini & Keeter, 1996; Knight, 1985; Lavine & Gschwend, 2007; Stimson, 1975). Such studies have, for instance, demonstrated that people with higher levels of political knowledge more strongly translate their economic egalitarianism (laissez-faireism) into support for a party that favours (opposes) egalitarian state intervention in the economy (Delli Carpini & Keeter, 1996). Similarly, those with higher levels of political knowledge more strongly translate their cultural conservatism (progressiveness) into support for a political party striving for more order (freedom) in the cultural domain (Delli Carpini & Keeter, 1996).

Given our focus on the education gap in support for trade openness, it is important to note that research has consistently reported lower scores on various measures of political knowledge among the less educated, in both the United States (e.g. Delli Carpini & Keeter, 1996; Highton, 2009) and the Netherlands (Achterberg & Houtman, 2009; Manevska et al., 2010), and that experimental research in the United States indicates that this relationship between education and political knowledge is causal (Sondheimer & Green, 2010, p. 186). Moreover, the main argument underlying the literature on political knowledge is echoed in research focusing on the relationship between education and attitudes toward international trade. As previous findings have suggested that opposition to economic globalisation among the less educated could not be interpreted according to the dominant economic-insecurity explanation outlined above, Hainmueller and Hiscox (2006) formulated an alternative viewpoint: ‘the positive relationship between education and support for trade-liberalization might [ … ] reflect the [fact] that more educated respondents tend to be more exposed to economic ideas about the overall efficiency gains for the national economy’ (p. 470). Although they do not refer to the broader literature on political knowledge, their suggestion can be interpreted as a specific application of the more general theoretical rationale of the political-knowledge explanation.

A dominant line of reasoning in the field of macro-economics is that free international trade stimulates economic growth in the countries involved due to comparative advantages and efficiency gains. These ideas, which date back to the seminal writings of Adam Smith and David Ricardo, are part and parcel of the curricula at higher education institutions. As a consequence, free-trade policies are often installed and legitimised with reference to these ideas. If people prefer economic growth, knowledge of this ‘free trade leads to economic growth’ argument would lead them to be more supportive of economic globalisation.

In this light, it is important to note that studies in the United States show that the economic beliefs of the more educated are more similar to economists’ beliefs than is the case for the less educated. This holds in general, as well as specifically, for beliefs about the macro-economic consequences of trade agreements (Caplan, 2002, 2006). Moreover, research in the United States demonstrates that such macro-economic beliefs influence support for liberal trade policies (Fordham & Kleinberg, 2012; Mansfield & Mutz, 2009). Furthermore, it reports positive correlations between economic knowledge and support for trade openness (Walstad, 1997; Walstad & Rebeck, 2002). All this resonates well with the title of Hainmueller and Hiscox's (2006) article, Learning to love globalization, which suggests that if one becomes acquainted with text-book economic ideas on the benefits of free trade, one will support free trade.

Nevertheless, this title also reflects another theorised link between education and support for economic openness, which draws on research demonstrating that ‘increased education [ … ] tends to socialize students to have more tolerant, cosmopolitan views of the world’ (Hainmueller & Hiscox, 2006, p. 470). However, their empirical research focuses on the direct effect of education and so does not allow for the disentangling of this strand of cultural reasoning from the explanation related to differences in knowledge. The present study aims to improve on this by empirically assessing a knowledge-based explanation simultaneously with the economic insecurity and dereification theories discussed above. We will do so by testing whether the greater opposition to trade openness among the less educated can be explained by their lower level of political knowledge, a measure that is closely related to economic knowledge.⁴

In order to test our hypotheses, we collected survey data in the Netherlands in 2012 (Achterberg et al., 2012). To do so, we hired the services of CentERdata (http://www.centerdata.nl/en), which maintains a panel representative of the Dutch population aged 16 and older. Questionnaires are completed online, and respondents who do not have access to a computer with an internet connection are provided with the necessary equipment. In total, 1302 people completed the questionnaire (a response rate of 76.3%). We excluded from the analyses the six respondents who completed the survey in 10 minutes or less, as they had not spent the time that was reasonably needed to provide valid responses.

Support for trade openness is measured by means of an item inspired by the one that is used most often in the relevant literature (e.g. Hainmueller & Hiscox, 2006; Scheve & Slaughter, 2001):

In order to uncover whether support for trade openness is considered to be a redistribution issue or a cultural-order issue, we constructed multiple scales. One multi-item scale represents the economic egalitarianism/laissez-faireism dimension: economic egalitarianism. Two multi-item scales measure the cultural progressiveness/conservatism dimension: authoritarianism and ethnocentrism. Each scale is reliable and has been validated in previous research (Appendix A1 presents the item phrasing, factor loadings, and reliability coefficients). Economic egalitarianism is measured by means of five commonly-used items (Achterberg & Houtman, 2009; De Koster, Achterberg, & Van der Waal, 2013; Houtman, 2003; Manevska et al., 2010). Authoritarianism is measured by means of a seven-item F-scale (Adorno, Frenkel-Brunswik, Levinson, & Nevitt Sanford, 1950; for similar scales used in the Dutch context, see Achterberg & Houtman, 2009; De Koster, Achterberg, Houtman, & Van der Waal, 2010; De Koster et al., 2013; De Koster & Van der Waal, 2007; Houtman, 2003). Ethnocentrism is measured via six items indicating negative prejudice toward out-groups, borrowed from Eisinga and Scheepers (1989) and recently used in Dutch research (De Koster et al., 2010; De Koster, Achterberg, Van der Waal, Van Bohemen, & Kemmers, 2014; Van Bohemen, Kemmers, & De Koster, 2012).

The independent variable education is measured as the number of years needed to attain the highest level of education of the respondent, ranging from 8 (primary education only) to 18 (university degree). Higher scores thus equate to higher levels of education.

The three aspects of economic insecurity reflected in hypotheses 2a to 2c are measured as follows. First, we measure income by asking respondents their net household income in euros, which we divided by thousand in order to aid the interpretation of the regression analyses.⁶ A lower score thus indicates a weaker economic position. Second, we measure labour-market position with four dummy variables, which are ordered by assessing whether one is dependent on the labour market and, if so, to what extent one's labour-market position is weak: (1) not in labour market (retired, student, housekeeping, or disabled); (2) employed; (3) partially employed; (4) unemployed. Third, job insecurity is measured by asking respondents to estimate the risk of becoming involuntarily unemployed during the next 12 months. This is coded as: (1) very small risk; (2) small risk; (3) neither small nor large risk; (4) large risk; (5) very large risk. Since this item logically only has valid scores for respondents who are partially or fully employed, it had a large number of missing values. In order to maximise the number of respondents in our analyses, we coded those who are not in the labour market (retired, student, housekeeping, or disabled) as 0 (since these individuals, by definition, cannot experience job insecurity), and those who are unemployed as 6.⁷ A higher score on job insecurity thus stands for a weaker economic position.

The crucial variable needed to test dereification theory is the standard operationalisation of embodied cultural capital: highbrow cultural consumption (for similar measurements, see Achterberg & Houtman, 2006; DiMaggio & Mohr, 1985; Houtman, 2003; Houtman et al., 2008; Manevska et al., 2010; Van der Waal & Houtman, 2011). It is measured with a reliable multi-item scale, asking respondents (1) how many books they own, and how often they (2) attend art exhibitions, (3) discuss art and culture, and whether they (4) consider themselves to be lovers of art and culture (Appendix A1 provides details). Along with the indicator of institutionalised cultural capital education, this form of cultural capital indicates one's ‘ability to recognize cultural expressions and to comprehend their meaning’ (Achterberg & Houtman, 2006, p. 78). The key to uncovering whether education serves as a vessel for one's cultural capital in its relationship with support for trade openness – as predicted by dereification theory – is to assess whether this other form of cultural capital has a similar effect in a multivariate analysis controlling for indicators of one's economic position.

To measure political knowledge, we used 21 questions about the European Union (details in Appendix A2). Each respondent answered a randomly selected subset of seven questions, and each question had two answer categories: true and false. We recoded incorrect answers as 0 and correct ones as 1. By summing up the resulting scores for participants with valid answers on all seven questions, each respondent was assigned a score between 0 (no correct answer) and 7 (all answers are correct). In order to increase the comparability of our study with others that analyse the impact of political knowledge, we measure knowledge of information or ‘minor facts’, and not understanding of complex political issues. Nevertheless, as Converse (2000) argues after reviewing the bulk of the research on political knowledge: ‘It does not take much imagination to realize that differences in knowledge of several such “minor” facts are diagnostic of more profound differences in the amount and accuracy of contextual information voters bring to their judgments’ (p. 333).

In our regression analyses, we control for age (in years), female (males 0, females 1), and non-native, which codes respondents whose parents were born in the Netherlands as 0, and others as 1 (the standard classification in the Netherlands).

First, we test whether opposition to trade openness pertains to the economic value domain (hypothesis 1, derived from the economic-insecurity explanation) or the cultural-value domain (hypothesis 3, the first one deduced from dereification theory). Table 1 presents the correlations between support for trade openness on the one hand and economic egalitarianism, authoritarianism, and ethnocentrism on the other. Support for trade openness proves substantially negatively related to both authoritarianism and ethnocentrism. As such, it can be clearly identified as pertaining to the domain of cultural-order issues. Those who oppose cultural diversity most strongly also reject trade openness most strongly, which corroborates hypothesis 3. Note, however, that support for trade openness is also negatively related to economic egalitarianism, albeit less strongly than to cultural-order issues. Hypothesis 1 is therefore also corroborated.

These findings suggest that education-related differences in support for trade openness might be explained economically and/or culturally. Whether this is indeed the case, and what the role of political knowledge is, will be analysed later. First, we will demonstrate that education-related indicators of one's economic position can be empirically disentangled from education-related indicators of one's cultural position, which provides a basis for the analyses that test our remaining hypotheses.

Table 2 reports the bivariate correlations of the indicators for economic and cultural position above the diagonal, and their partial correlations controlled for level of education below the diagonal. Level of education is indeed strongly positively related to both the economic indicator par excellence (income) and the indicator for embodied cultural capital (highbrow cultural consumption). This is not very surprising, and practically undisputed. Level of education indicates both economic position and cultural position. It is thus an ambiguous indicator. Therefore, without additional analyses it cannot be decided whether an association between level of education and a specific attitude should be interpreted in economic or cultural terms.

It needs to be noted that the only reason that embodied cultural capital (measured as highbrow cultural consumption) is related to income (above the diagonal) is because both are related to level of education: the partial correlation below the diagonal demonstrates that these unambiguous indicators for, respectively, cultural capital and economic position are not empirically related once level of education is controlled for (cf. Houtman, 2003; Van der Waal & Houtman, 2011). This is theoretically salient, as it means that a positive effect of these variables on support for trade openness can be unequivocally interpreted as, respectively, support for dereification theory and the economic-insecurity explanation. Hence, the share of the effect of education on support for trade openness that is mediated by income can unequivocally be interpreted as rooted in people's economic positions, as the factor endowment model predicts (depicted in the central part of Figure 1). A similar line of reasoning dictates that the share of the education effect that is explained by its relationship with embodied cultural capital must be rooted in people's cultural positions, as dereification theory predicts (depicted in the upper part of Figure 1).

Prior to empirically testing which theory can account for greater opposition to trade openness among the less educated, we take an intermediate step to demonstrate that redistribution and cultural-order issues are very differently affected by the economic and cultural aspects of education, in line with the economic-insecurity explanation and dereification theory.

Table 3 reports regression analyses on economic egalitarianism (first two columns), authoritarianism (columns 3 and 4), and ethnocentrism (columns 5 and 6). Each dependent variable is affected by education: the higher the level of one's education, the less egalitarian (which resonates with economic-insecurity theory), and the less authoritarian and ethnocentric one is (in line with dereification theory).

The first two columns indicate that high (low) egalitarianism of less (more) educated individuals can be attributed to weak economic position, and not to low cultural capital. Due to the inclusion of the economic indicators, which show significant effects, the effect of education declines, while highbrow cultural consumption does not have a significant effect and does not explain part of the education effect. Note, however, that the difference in egalitarianism between labour market categories is not in accordance to economic-insecurity theory, which predicts that the unemployed should be more egalitarian than the employed, and that both categories should be more egalitarian than those who do not depend on the labour market, which is not the case.

Things are different where stances on cultural-order issues are concerned. The higher levels of authoritarianism and ethnocentrism among the less educated prove to be rooted in their lower cultural capital (supporting dereification theory), and not in their weaker economic position (for similar results, see Houtman, 2003; Manevska, 2014; Manevska & Achterberg, 2013; Van der Waal & Houtman, 2011). The only economic variable that is significantly related to authoritarianism and ethnocentrism indicates that the employed are less, rather than more, authoritarian and ethnocentric than those who do not depend on the labour market. This is contrary to what one would expect on the basis of the economic-insecurity explanation.

The analyses performed thus far provide several insights that are relevant for the final step we need to take in testing the three explanations for the greater opposition to free trade among the less educated. First, opposition to trade openness proved most notably a cultural-order issue, but also, to a minor extent, a redistribution issue. Second, the economic and cultural aspects of education cannot only be disentangled theoretically, but also empirically. Third, education's economic aspect merely affects redistribution issues (in accordance with the economic-insecurity explanation), while education's cultural aspect merely affects cultural-order issues (in line with dereification theory).

Taken together, these insights suggest that opposition to trade openness among the less educated is more likely to be explained by dereification theory than by the economic-insecurity explanation. Whether this is indeed the case, and what the role of political knowledge is, is scrutinised by means of the analyses reported in Tables 4 and 5. Table 4 displays the results of ordered-logit regression analyses of support for trade openness. Table 5 presents a decomposition analysis that shows in more detail to what extent the effect of education on support for trade openness can be attributed to economic position, embodied cultural capital, and political knowledge.

Table 5.

Decomposition of total effect of education on support for trade openness into direct and indirect effects via indicators for economic-insecurity explanation, dereification theory, and political–knowledge explanation.^a

Total effect education	0.17*** (0.02)
Direct effect education	0.10*** (0.02)
Indirect effect education	0.07*** (0.01)
Indirect effects of education via …		Explained percentage of initial effect education
		Per separate indicator		Total per explanation
… indicators economic-insecurity explanation
Income	0.02** (0.01)	9.72%
Employed	−0.00 (0.01)	−1.27%
Partially employed	0.00 (0.00)	1.16%		9.36%
Unemployed	−0.00 (0.00)	−0.50%
Job insecurity	0.00 (0.00)	0.25%
… indicator dereification theory
Highbrow cultural consumption	0.05*** (0.01)	27.42%		27.42%
… indicator political-knowledge explanation
Political knowledge	0.01* (0.00)	3.91%		3.91%
N				1079

Total effect education	0.17*** (0.02)
Direct effect education	0.10*** (0.02)
Indirect effect education	0.07*** (0.01)
Indirect effects of education via …		Explained percentage of initial effect education
		Per separate indicator		Total per explanation
… indicators economic-insecurity explanation
Income	0.02** (0.01)	9.72%
Employed	−0.00 (0.01)	−1.27%
Partially employed	0.00 (0.00)	1.16%		9.36%
Unemployed	−0.00 (0.00)	−0.50%
Job insecurity	0.00 (0.00)	0.25%
… indicator dereification theory
Highbrow cultural consumption	0.05*** (0.01)	27.42%		27.42%
… indicator political-knowledge explanation
Political knowledge	0.01* (0.00)	3.91%		3.91%
N				1079

Source: calculations on Achterberg et al. (2012).

* = p < 0.05.

** = p < 0.01.

*** = p < 0.001 (two-sided tests).

^aControlling for concomitants female, age, and non-native.

View Large

Model 1 in Table 4 simply reproduces previous research findings: the more educated prove more supportive of free trade than the less educated. With Model 2, we assess the empirical validity of the economic-insecurity explanation. Entering our three indicators of economic position – income, labour-market position, and job insecurity – allows testing hypotheses 2a, 2b and 2c respectively. Only income significantly affects support for trade openness, as expected. Moreover, its inclusion leads to a significant decline in the strength of the coefficient of the level of education (see Table 5, below). Although the economic-insecurity explanation also predicts that those with a weak labour-market position and high job insecurity will also be more opposed to trade openness, this proves not to be the case. Hence, only one of the three hypotheses deduced from the economic-insecurity explanation (2a) is corroborated.

We now turn to testing the hypothesis derived from dereification theory, which states that opposition to trade openness among the less educated can be attributed to their low level of cultural capital (hypothesis 4). Model 3 includes highbrow cultural consumption, an unambiguous indicator of cultural capital. It proves, as expected, to be positively and significantly related to support for trade openness. What is more, the effect of the level of education significantly declines in strength (see Table 5, below). The fact that the education effect remains substantial accords with previous research on the antecedents of opposition to cultural diversity: as discussed, level of education and highbrow cultural consumption are two indicators of cultural capital that partly function independently of each other (cf. Manevska et al., 2010; Van der Waal & Houtman, 2011). Notwithstanding the fact that they partially overlap as indicators of cultural capital, as is indicated by their substantial correlation, this indicates that each covers a dimension of cultural capital that is not covered by the other. An alternative interpretation of the remaining effect of level of education as representing economic position seems less plausible: the economic aspects of level of education are already teased out by simultaneously modelling income, labour-market position, and job insecurity.⁸ Overall, our analyses clearly indicate that the opposition to trade openness among the less educated can be interpreted in terms of opposition to cultural diversity of those with little cultural capital, which corroborates hypothesis 4.

Having tested the hypotheses deduced from both the economic-insecurity explanation and dereification theory, we shall now examine the empirical validity of the political-knowledge explanation. Model 4 indicates that political knowledge has a positive and significant effect, just as hypothesis 5 predicts. Those with more political knowledge are more supportive of trade openness, and including this effect also modestly but significantly contributes to the decline in the impact of level of education (see Table 5). This corroborates hypothesis 5: the lower levels of political knowledge among the less educated partly account for their greater opposition to trade openness.

To assess the relative explanatory value of the three explanations, we have applied the ‘KHB method’ (Karlson, Holm, & Breen, 2012). This decomposition method enables us to distinguish between direct and indirect effects of education, and indicates for each mediating variable used for testing our hypotheses to what extent it explains the education effect. As such, we can determine to what extent each of the three explanations accounts for the education gradient in support for trade openness.⁹ The first row in Table 5 reports the initial strength of the effect of education, while controlling for relevant background characteristics (compare Model 1 in Table 4). The second row reports the strength of this effect if all indicators of the three explanations under study are modelled (compare Model 4 in Table 4), while the third row indicates the indirect effect of education (that is, the part of the education effect that is mediated by the indicators included for testing the three theories). The rows below indicate the contributions of both the separate explanatory variables and the three explanations under scrutiny.

The final column of Table 5 shows that the economic-insecurity explanation accounts for 9.36% of the education gradient in support for trade openness, while differences in political knowledge account for 3.91% for differences in support for trade openness between the less and the more educated. Embodied cultural capital, the indicator derived from dereification theory, proves by far the most important, explaining 27.24%. Because the remaining direct impact of education probably needs to be understood as an effect of institutionalised cultural capital (as discussed above), the explanatory value of dereification theory is likely even to be substantially higher.

In both academic and public debates, opposition to trade openness is predominantly interpreted as opposition by so-called ‘losers of globalization’ (cf. Kriesi et al., 2008). Therefore it does not seem surprising that the high levels of opposition found among the less educated have often been interpreted in economic terms. After all, they are the ones who bear the burden of open borders: they see their wages decline (Okkerse, 2008), their jobs disappear (Alderson, 1999), and their employment opportunities diminish (Van der Waal, 2015). They also experience more job insecurity (Geishecker, Riedl, & Frijters, 2012). Against this background, the greater opposition to trade openness among the less educated has often been interpreted as driven by the self-interest of the economically weak. Yet our findings indicate that this is only a small part of the explanation, at least in the Netherlands.

We have developed two alternative explanations, which do not centre on economic self-interest, by combining suggestions from the literature on education and support for trade openness with those from the broader scholarly literature. The first is dereification theory, which focuses on cultural capital as an indicator of the ability to recognise cultural expressions and comprehend their meaning. This theory resonates with recent pleas for scrutinising the ramifications of differences in status and cultural repertoires for social inequalities (Lamont, 2016; Lareau, 2015; Ridgeway, 2014). The second is a political-knowledge explanation, which concentrates on individuals’ understanding of political and policy issues and processes. Both alternative theories are empirically corroborated, although their salience differs: dereification theory explains the education gradient in opposition to trade openness to a substantially greater extent than either the widespread economic-insecurity explanation or the alternative political-knowledge explanation.

These findings inform the debate on the education gap in support for trade openness. First, our systematic simultaneous test of three explanations adds strength to the suggestion that the explanatory value of the economic-insecurity explanation is modest at best (e.g. Hainmueller & Hiscox, 2006; Mansfield & Mutz, 2009; Margalit, 2012). Second, the development of the political-knowledge explanation entails an elaboration of the suggestion made by Hainmueller and Hiscox (2006) that being unacquainted with text-book economic ideas might underlie smaller support for trade openness among the less educated. The empirical corroboration of this explanation suggests that being unaware of the macro-economic gains of free trade does indeed hamper support for trade openness and partly accounts for the greater opposition among the less educated. Third, our findings on the relevance of dereification theory add up to an empirically-validated and theoretically-developed interpretation of the suggestion that the greater opposition to trade openness among the less educated is somehow related to their often-reported opposition to cultural diversity (Edwards, 2006; Kaltenthaler et al., 2004; Mansfield & Mutz, 2009; Mayda & Rodrik, 2005; O’Rourke & Sinnott, 2001; Wolfe & Mendelsohn, 2005), and the finding that their self-perceived loss due to economic openness is triggered by cultural threat (Margalit, 2012). Two interrelated observations are relevant in this respect.

First, the empirical corroboration of dereification theory indicates that the often-reported opposition to trade openness among the less educated is an aspect of their opposition to cultural diversity (‘cultural conservatism’) in general. Second, by demonstrating that opposition to cultural diversity in general and trade openness in particular is rooted in low levels of cultural capital instead of in a weak economic position (see also: Houtman, 2003; Manevska, 2014; Manevska & Achterberg, 2013; Manevska et al., 2010; Van der Waal & Houtman, 2011), dereification theory illuminates why the less educated oppose these forms of cultural diversity.

More generally, dereification theory provides an empirically corroborated cultural explanation of the well-documented opposition to immigration among the less educated. Many scholars agree that in research scrutinising attitudes on immigration, ‘the labor market competition hypothesis [ … ] [is] something of a zombie theory’ (Hainmueller & Hopkins, 2014, p. 241). Meanwhile, in alternative explanations, ‘[t]oo frequently, culture operates as a residual category’ (Hainmueller & Hopkins, 2014, p. 242). In contrast, dereification theory offers a systematically developed explanation for why and how culture is relevant. It is highly plausible that opposition to immigration among the less educated – just like their opposition to trade openness, and their authoritarianism and ethnocentrism – is rooted in their low level of cultural capital, which inspires a reified worldview (Manevska, 2014). As cultural capital functions as a status marker, and proves crucial for understanding the intergenerational reproduction of class and status stratification (Jæger & Breen, 2016), this is anything but ‘culture operating as a residual category’. On the contrary, cultural capital is one of the mainstays in contemporary stratification and its corollaries (cf. Bourdieu, 1984; Jæger & Breen, 2016; Lamont, 2016; Lareau, 2015; Ridgeway, 2014).

An additional way in which culture might be relevant for the issue at hand calls for further scrutiny, since it goes far beyond the scope of this study. Shared meanings and frames of interpretation often differ across groups, and this might also be the case for issues relevant for opposition to trade openness. First, the political-knowledge explanation implicitly assumes that macro-economic growth is in the general interest and that all people support this interpretation of what the general interest is. Yet some people may oppose, rather than support, free trade because they perceive or value its consequences differently – for instance because they have environmental concerns relating to economic growth, or fear economic crises, rising poverty, increased inequality, and the violation of human rights resulting from unregulated international trade (cf. Sen, 2000; Stiglitz, 2002). Second, in accordance with dereification theory, cultural capital proves positively related to support for trade openness, but high levels of cultural capital might also inspire opposition to trade openness. Take, for instance, academics supporting alternative globalisation or global justice movements. They oppose free trade for reasons such as environmental concerns or international solidarity, rather than resistance to cultural diversity as in the cases of those with low levels of cultural capital. Future research could focus on uncovering various trade-related belief systems and their social bases, for instance by applying promising methods such as Correlational Class Analysis (Boutyline, 2015). For now it is important to note that the above-mentioned possibilities imply that the explanatory value of the political-knowledge explanation and dereification theory is likely to be underestimated by our analyses.

Overall, the results reported in the present study are intended to add key insights to the discussion of how relatively greater opposition to trade openness among the less educated can be understood, and the same applies to discussions on the explanation of other aspects of their resistance to cultural diversity. It is, however, for future research to decide how far our findings extend beyond the Dutch case, and whether and among which social groups different configurations of political knowledge, cultural capital, and opposition to trade openness exist.

Previous versions of this paper were presented at the annual meeting of Dutch and Flemish sociologists (Dag van de Sociologie) in 2013, the 72nd Annual Conference of the Midwest Political Science Association in 2014, and the European Consortium for Political Research's General Conference in 2015. We thank Éric Bélanger, Josje ten Kate, Daniëlle van de Pavert, Miriam Youngerman Miller, Sally-Ann Ross, Niels Schenk, Lyle Scruggs, and the members of Erasmus University's LOBOCOP and Brown Bag seminars for their constructive comments on earlier versions of this paper.

No potential conflict of interest was reported by the authors.