Concern for enviroment and individual attitudes forward international trade in developing countries

Such industry pressure has induced governments of countries with high environmental protection standards to seek environmental clauses in PTAs with countries that have lower environmenta

Concern for the Environment and Individual

Attitudes towards International Trade in

of trade liberalization Furthermore, and in contrast to prevailing government rhetoric, the majority of respondents support environmental clauses in trade agreements The main policy implication is that – irrespective of green protectionism rhetoric – there exists room for more ambitious efforts to include environmental standards in international trade agreements

1 Introduction

A closer look at the academic literature on international trade and the environment (e.g., Anderson 1997; Anderson and Blackhurst 1992; Bechtel et al 2012; Bhagwati 2002; Ferrantino 1997; Franzen and Meyer 2012; Spilker 2013; Whalley 1991) and

on the positions of governments in international trade liberalization efforts raises interesting questions about preferences of developing countries in this regard Political leaders from advanced industrialized countries have, over the past decades, increasingly insisted on linking environmental protection issues with trade liberalization (Bechtel and Tosun 2009; Drezner 2005; Hultberg and Barbier 2004; Schott 2004) This insistence has led to an institutionalized debate in the World Trade Organization (Trade and Environment Committee), but has thus far not resulted in any amendments of WTO agreements in this direction1 (Dagne 2010; WTO Secretariat 2014)

Outside the WTO, however, trade liberalization efforts have increasingly focused on bilateral or minilateral preferential trade agreements (PTAs) In contrast to the WTO, where negotiations and agreements involve the entire WTO membership (currently

159 member states), countries are free to select any (willing) other country worldwide for an agreement And they are de facto free to design the contents of PTAs according

to their preferences, as long as the new PTA does not clearly violate WTO commitments (Mansfield and Milner 1999: 612-15; WTO 2011) The possibility to discriminate among potential PTA partner countries (which in the WTO is precluded

by the most favored nation (MFN) principle) has opened the door for countries that seek to condition trade liberalization on environmental protection standards (Author) Based on economic and ideological reasoning, many PTAs that have been concluded

in the past decades include environmental protection standards – according to a recent WTO count for the years 1958 to 2011, 46 out of 100 PTAs include environmental provisions (WTO 2011).2 The economic reasoning for such standards derives from a level playing field argument (e.g., Bernauer and Caduff 2004; Prakash and Potoski

1

While no amendments to the GATT/WTO agreements have been adopted, there are legal exceptions that allow members to take regulatory measures, including trade restrictions, to achieve legitimate policy objectives (e.g., the protection of human, animal or plant life and health, and natural resources) (WTO Secretariat 2014)

2

These include “the development of environmental standards; enforcement of national environmental laws; establishment of sanctions for violation of environmental laws; publications of laws” (WTO 2011)

2006; Vogel 1995; 1997) On the assumption that higher environmental standards increase production costs in industries producing tradable goods, producers in country

A that are subject to higher environmental standards will insist that their competitors

in country B should be subject to similar standards Whether higher environmental standards do in fact lead to comparative disadvantages, and in what industries and under what conditions this may be the case, remains contested (Jaffe et al 1995; Mulatu et al 2001; Xu 1999) Nevertheless, it has become common practice for industries in countries with higher environmental standards to make such claims in the context of trade negotiations and to insist on addressing this issue in PTAs Such industry pressure has induced governments of countries with high environmental protection standards to seek environmental clauses in PTAs with countries that have lower environmental standards (e.g., the US and Canada vis-à-vis Mexico in NAFTA)

The ideological motivation derives from widespread sentiments in many advanced industrialized countries that many poor countries suffer from major environmental degradation problems Assuming that trade with these countries may worsen these problems, they believe that trade with poor countries should be avoided, or PTAs should be used to push the governments of such countries into accepting improved environmental protection standards, and to prevent relocation of dirty industries from the North to the South This is where an interesting question about potential differences between elite preferences and ordinary citizen preferences arises Recent research on the relationship between environmental and trade preferences in advanced industrialized countries finds that citizens with greener preferences are more skeptical about trade liberalization (Bechtel et al 2012) With this in mind, it appears that governments of industrialized countries are – besides being influenced by industry pressure for a level playing field – catering to widespread voter demand when insisting on environmental clauses in PTAs and placing trade-environment issues on the WTO agenda

What about developing countries? Not surprisingly, a statement by Laurence Summers, then at the World Bank, in a leaked email (December 12, 1991) was strongly condemned by many policymakers and NGO leaders from industrialized countries Summers noted:

“… Just between you and me, shouldn't the World Bank be encouraging MORE migration of the dirty industries to the LDCs [Least Developed Countries]? … The measurements of the costs of health impairing pollution depends on the foregone earnings from increased morbidity and mortality… the economic logic behind dumping a load of toxic waste in the lowest wage country

is impeccable and we should face up to that…I've always thought that under-populated countries

in Africa are vastly UNDER-polluted, their air quality is probably vastly inefficiently low compared to Los Angeles or Mexico City…The demand for a clean environment for aesthetic and health reasons is likely to have very high income elasticity….Clearly trade in goods that embody aesthetic pollution concerns could be welfare enhancing…The problem with the arguments against all of these proposals for more pollution in LDCs (intrinsic rights to certain goods, moral reasons, social concerns, lack of adequate markets, etc.) could be turned around and used more or less effectively against every Bank proposal for liberalization.”

More interestingly, however, it appears that the statement by Summers lines up quite well with the preferences of many political and economic leaders from the Global South These leaders have, on countless occasions, condemned what they consider green protectionism by industrialized countries At the 1994 GATT Marrakesh Ministerial Meeting the then-acting Minister of International Trade and Industry of Malaysia, Dato’ Seri Rafidah Aziz, stated that environmental issues are being used to promote protectionist motives, particularly to keep out imports from countries that have a competitive edge and comparative advantage (Trade Negotiations Committee 1994) More recently, at the preparatory meeting of the UN Conference on Sustainable Development in May 2010 in New York, the Chinese delegation stressed that “(…) the international community should create an enabling environment for green development We should resolutely oppose the practice of erecting “green barriers” and engaging in trade protectionism under the pretext of environmental protection.” (Third World Network, May 22nd 2010) In yet another example, India’s Special Envoy on Climate Change, Shyam Saran, warned that doors “should not open for protectionism under [the] green label That is something which would be a very negative development.” (The Economic Times, March 25th 2009)

Environmental and development activists tend to interpret such statements as evidence for transnational capitalist collusion at the expense of ordinary citizens in developing countries For instance, they claim that “[e]lites in the [S]outh may benefit from GATT but experience suggests that the majority of people in the [S]outh will lose” (InterPress Third World News Agency, December 27th 1993) They believe that developing country leaders tend to profit personally from attracting dirty foreign

investment, but pass on negative environmental consequences to (usually poorer parts of) their population Accordingly, while advocates of this view demand that citizens

of poor countries need to be protected from such abuse, by avoiding trade liberalization, via environmental clauses in trade agreements, or by other means, proponents of the Environmental Kuznets Curve argue (similar to Summers’ argument above) that poor people may actually not mind more pollution if polluting foreign investment and trade generates additional income for them (Dasgupta et al 2002; Frankel and Romer 1999; Grossman and Kruger 1995; Johnson et al 2007) Interestingly, we know only very little about how citizens in developing countries evaluate the trade-environment relationship Results from the World Public Opinion Survey (2006/’07), which includes samples from both developed and some developing countries3, suggest that citizens from developing countries are, like their counterparts in rich countries, also concerned about the effects of trade on the environment and in fact support environmental standards within trade agreements (The Chicago Council on Global Affairs 2007) However, and despite the high salience of this topic (especially with regards to the international trade governance debate), research on the trade-environment linkage has thus far focused almost exclusively on the macro-level and has bypassed this issue This is the gap this paper seeks to address

We start by developing a theoretical argument on how citizens in developing countries are likely to evaluate the trade-environment relationship The argument is constructed from two perspectives that are motivated by existing political debates on the issue First, we shed light on whether individuals with greener attitudes are more (or less) likely to hold pro-trade preferences While recent research points to a negative relationship in advanced industrialized countries (Bechtel et al 2012) we hypothesize that this relationship is likely to be positive in poor countries Second, because the trade-environment issue has become particularly salient in the context of PTAs, we develop an argument concerning partner countries for preferential trade agreements We hypothesize that developing countries are more likely to favor PTA partner countries with higher environmental standards, and to favor PTAs that include environmental protection clauses

3

The surveys were conducted in 19 countries: Argentina, Armenia, Australia, China, France, India, Indonesia, Iran, Israel, Mexico, Palestinian Territories, Peru, Philippines, Poland, Russia, South Korea,

To test these hypotheses we fielded a combination of standard surveys and conjoint experiments in three developing countries that differ in their economic and political system characteristics as well as their environmental protection record: Costa Rica (mature democracy, upper-income developing country, high environmental standards), Nicaragua (partly democratic, low-income developing country, low environmental standards), and Vietnam (autocracy, low-income developing country, low environmental standards)

The results show that, especially in the two poorer countries in our sample, which also have lower environmental protection standards (Nicaragua and Vietnam), individuals with greener preferences are more supportive of international trade They also show that individuals prefer PTAs with countries that have higher environmental standards, and that they prefer environmental clauses in PTAs

The main policy implication that derives from these findings is that – irrespective of green protectionism rhetoric on the part of political and economic elites – there exists room for more ambitious efforts to incorporate environmental standards in international trade agreements

The remainder of the paper is structured as follows After a brief review of the existing literature (section 2) we develop two hypotheses, one focusing more generally on how environmental and trade attitudes are related, the other focusing on environmental standards and PTAs (section 3) We then present the empirical design (section 4) and the results (sections 5 and 6) and end with suggestions for further research (section 7)

2 Economic and non-economic determinants of trade preferences

Most research on individual trade preferences focuses on the micro-level ramifications of standard trade theories Its arguments are derived from the factor endowments (Stolper-Samuelson) and the specific-factors (Ricardo-Viner) model These theories posit that the ways in which trade policies affect wages and employment opportunities are the main conduits through which individuals experience globalization That is, individual trade policy preferences are shaped by how trade affects income and overall economic welfare Individuals who stand to gain materially from economic openness tend to support it, while those who are expected

to lose tend to favor protectionism

To the extent that individuals are motivated by material self-interest, the two models make clear-cut and distinct predictions about who should support trade and who should oppose trade Early scholarship on individual trade preference formation has found at least partial empirical support for the arguments derived from economic trade theory (e.g., Beaulieu 2002; Hiscox 2002; Magee et al 1989; Mayda and Rodrik 2005; O’Rourke and Sinnott 2001; Rogowski 1987; Scheve and Slaughter 2001) Recent research has challenged the view that trade preferences are mainly a function

of expected distributional implications of economic openness One frequently voiced criticism is that explanations based on material self-interest neglect effects of ideological or cultural factors (Hainmueller and Hiscox 2006; Mansfield and Mutz 2009; O’Rourke and Sinnott 2001) For example, Hainmueller and Hiscox (2006) question whether observed positive effects of skill levels (usually measured by education level) on trade preferences reflect the factor endowments logic They argue that the positive effect of education on (pro-) trade preferences may rather reflect a situation where individuals who have benefited from more education are more aware

of the economic benefits of trade, hold a more cosmopolitan worldview, and exhibit greater tolerance of different cultures and out-groups (Hainmueller and Hiscox 2006) Other studies point in the same direction They have examined the influence of non-economic factors on trade preference formation, for instance fairness considerations, nationalism, and xenophobia (Hainmueller and Hiscox 2006; Herrmann et al 2001; Kaltenthaler et al 2004; Lu et al 2012; Mansfield and Mutz 2009; Mayda and Rodrik 2005; O’Rourke and Sinnott 2001)

More specific to the question pursued in this paper, Bechtel, Bernauer and Meyer (2012) have examined how individual attitudes towards the environmental affect trade preferences They argue that as a result of intense public debates over the environmental consequences of global economic exchange political leaders from advanced industrialized countries have, over the past two decades, increasingly insisted on the inclusion of environmental protection standards in international trade agreements (Bechtel and Tosun 2009; Drezner 2005; Hultberg and Barbier 2004) To test this presumption, they rely on survey data from Switzerland The analysis shows

that individuals with stronger pro-environment attitudes are significantly more likely

to hold more negative views on international trade

The latter line of research has two limitations that we seek to address in this paper First, it focuses on advanced industrialized countries While we know very little about the determinants of trade preferences in developing countries more generally, there is virtually no systematic evidence on whether and how individuals in developing countries view the trade-environment nexus The World Public Opinion Survey (2006/’07) includes two questions on the trade-environment relationship It is the only survey we know of that gauges, among other issues, public opinion on the trade-environment nexus Its results indicate that citizens from various countries around the world are concerned about the environmental effects of trade In none of the countries polled do large majorities believe that trade is good for the environment However, China (57%) and India (51%) rank among the most optimistic with respect trade impacts on the environment Furthermore, the results reveal that public support for the inclusion of “minimum standards for the protection of the environment” in trade agreements is very high in developing countries In China and India endorsement of environmental standards amounts to 85% and 60%, respectively

However, the construction of these items and the survey as a whole does not allow for

an in-depth micro-level analysis of the trade-environment nexus Other surveys, including the standard multinational surveys that are widely used in the political science literature (e.g., the World Values Survey) do not include the required information either Second, the existing evidence is correlational We therefore add to research in this area by focusing on developing countries and by using a combination

of correlational and experimental approaches

3 Environmental attitudes and trade preferences

In this section we develop arguments that shed light on citizens’ views on the environment relationship We do so from two distinct perspectives First, we focus on whether individuals with greener attitudes are more (or less) likely to hold pro-trade preferences While recent research points to a negative relationship in advanced industrialized countries (Bechtel et al 2012), we hypothesize that this relationship is likely to be positive in poorer countries This argument will be empirically evaluated

trade-within a correlational framework.4 Second, we examine the environment-trade linkage through the perspective of trading-partner choices, focusing on whether individuals prefer their country to trade with countries characterized by higher (or lower) environmental standards, and whether they prefer environmental standards in PTAs, even if imposed from the outside This argument is empirically evaluated by means of a conjoint experiment

3.1 Effect of environmental attitudes on trade preferences

A considerable body of scientific evidence suggests that trade and environmental degradation are causally related, though the relationship is quite complex Increased trade expands the scale of economic activity, which tends to cause more environmental degradation However, it also tends to increase welfare and, as suggested by the environmental Kuznets curve (EKC) literature, is therefore conducive to environmental protection (Anderson 1997; Bechtel et al 2012; Ferrantino 1997; Franzen and Meyer 2010) The EKC literature in fact argues, for these reasons, that pollution first increases when poor economies grow and at some point starts to decrease The turning point of the EKC can result from several processes, the main determinants being changes in economic structure (from manufacturing to services) and changes in public demand and government supply of environmental protection (Bernauer and Koubi 2009)

What does this argument imply for linkages between environmental attitudes and trade preferences in developing countries, as opposed to advanced industrialized countries? Let’s start with the latter type of countries Bechtel et al (2012), the only study on the subject to date, focuses on advanced industrialized countries They argue that environmental concerns help us understand how individuals evaluate the costs and benefits of economic integration Specifically, they hypothesize that individuals holding stronger pro-environment attitudes are more likely to be skeptical about international trade and more likely to favor protectionist policies Using survey data from a high-income country, Switzerland, they find that respondents’ reported concerns for the environment indeed correlate positively with protectionist attitudes

4

We explored various experimental designs with convenience samples However, we found it nearly impossible to construct treatments (e.g information priming for environmental problems) that could

This argument rests on a specific causal structure between environmental attitudes and trade preferences, which is not discussed in that study It implies that individuals form their environmental attitudes prior to evaluating the pros and cons of international trade and thus forming trade preferences We do not know of any study that has explicitly tested this assumption Nonetheless, it appears plausible because exposure to the natural environment, and thus also to environmental degradation problems, tends to start very early in childhood, and many environmental problems can be experienced visually and physically (e.g., land degradation, deforestation, water and air pollution) In contrast, international trade is a more complex and remote phenomenon, and preferences with respect to international trade are thus likely to take shape later in life and might also be less stable – again an assumption that has not yet been assessed empirically

If we accept that environmental attitudes have an effect on trade preferences, rather than the other way around, we can think about the implications of different income levels for the environment-trade nexus To start with, existing studies using multi-level analysis of individual level data for several countries show that individual level income matters less for environmental attitudes than national level income (Franzen and Meyer 2010) This finding makes sense intuitively, because when individuals are asked about their preferences concerning environmental protection, the frame of reference is usually the country And this means that they are likely to think more in terms of whether the country could or should afford more environmental protection, rather than whether they personally could or should afford it

How are different national income levels then likely to impact on the perceived environment-trade relationship? Bechtel et al (2012) draw mainly on research on postmodern values and value change (e.g., Inglehart 1997; Inglehart and Flanagan 1987) to argue that individuals with greener attitudes are more likely to prefer trade protection However, the postmaterialism argument appears to focus mainly on explaining environmental attitudes and leaves it somewhat unclear how postmaterialist attitudes could affect trade preferences We thus outline a more complete argument on how (national) income levels could affect the trade-environment relationship as perceived by individuals

Following the EKC argument, individuals in rich countries are likely to exhibit stronger preferences for environmental protection At the same time, they are facing

lower (relative to poorer countries) marginal opportunity costs from some additional restrictions on trade, or from refraining from further trade liberalization To the extent they consider trade to have a negative impact on the environment, they are likely to be more willing to impose restrictions on trade to protect the environment This implies that individuals with stronger green attitudes are likely to hold more negative views

on trade and are more likely to favor trade protection

How will this argument play out in poor countries? Assume, for a moment, that contrary to the EKC argument individuals in poor countries hold equally strong green attitudes and are equally aware that trade has a negative effect on the environment Even in that case, individuals in poor countries are likely to face higher (relative to rich countries) marginal opportunity costs from additional restrictions on trade, or from refraining from further trade liberalization Existing survey research in fact shows that while enthusiasm for economic globalization has waned considerably over the past two decades in many wealthy nations, notably as a result of anxieties over social and environmental consequences of increased trade, there is strong confidence

in many developing countries that trade liberalization improves economic growth and economic welfare (Pew Global Attitudes Project 2008)

If this holds true, we should at least not observe a negative correlation between environmental attitudes and trade preferences If, however, environmental conditions

in poor countries are (because of the EKC logic) rather problematic and marginal opportunity costs of restricting trade are high, we are likely to observe a positive association between environmental attitudes and trade preferences That is, it may well be that citizens of poorer countries prefer more environmental protection, are aware that trade causes environmental problems, but still favor trade liberalization Indeed, if one subscribes to the EKC argument, citizens of poorer countries are likely

to be aware that environmental quality tends to improve as income in their respective country reaches higher levels, both due to the availability of better technology and more funding for environmental protection And because trade tends to have welfare-increasing effects (at the aggregate national level) citizens of poor countries may view trade liberalization as a conduit towards environmental improvements, at least in the middle- to long-term These arguments are summarized in Hypothesis 1

H1: Individuals who are more concerned about the environment are more likely

to support trade liberalization

3.2 Preferences concerning environmental standards of trading partners

While the above argument views the trade-environment relationship from a rather general perspective, recent political debates have centered on whether or not to include environmental protection clauses in international trade agreements Surprisingly, we could not find any observational or experimental public opinion research on this issue

Again, we start with a discussion on what the preferences of citizens from advanced industrialized countries are likely to be and then explore whether we should expect citizen preferences in developing countries to differ from that, and in what respect To begin with, the stringency or ambition level of environmental policies differs quite strongly across countries (Bernauer and Böhmelt 2013) These differences result from differences in economic development levels, factor endowments, natural resource endowments, political regime type, and other factors How does trade openness impact on these environmental regulatory differences? Some studies regard international trade as a mechanism that may, through so called trading-up processes, lead to higher environmental standards in countries that start off with lower standards (e.g., Bernauer and Caduff 2004; Prakash and Potoski 2006; Vogel 1997) Other studies expect competition in environmental regulatory laxity and point to negative consequences for domestic environmental policy (e.g., Damania et al 2000; Dean et

al 2009; Javorcik and Wei 2003)

While academic research has not yet resolved the issue of whether trade promotes or undermines (or is of minor relevance to) environmental protection efforts, environmental standards are increasingly being added to international trade agreements, and in particular preferential trade agreements (PTAs) PTA negotiations have become particularly prominent fora for linking trade and the environment, arguably because PTAs are more flexible legal instruments, compared to global trade policy-making in the WTO (Cottier 2002; Hafner-Burton 2005; Spilker and Böhmelt 2013)

Various scholars note that this phenomenon appears to reflect voter and corporate demand Voters may fear that trade could have negative effects on the natural environment and undermine national environmental policy Import competing firms may fear that foreign competitors could benefit from lower environmental regulatory

costs Both could then combine into “baptist and bootlegger” coalitions (e.g., Gallagher 2008) that drive up environmental standards in trade partner countries For advanced industrialized countries that have been at the forefront of this trend, these arguments are likely to hold true Recent research finds that US citizens are more likely to prefer PTAs with countries that have similar or higher environmental and labor standards (Author) This result is likely to reflect a preference for avoiding harmful competition by means of environmental deregulation as a result of economic openness However, it is also consistent with an argument that citizens use environmental standards as cues for more general social value systems, and that this reflects a preference for minimization of psychic distance in trade relations

As noted further above, for developing countries, casual observation suggests that political and economic elites from these countries tend to oppose environment-trade linkages and environmental clauses in PTAs But we know of no study that gauges citizen preferences in this regard We submit that, similar to advanced industrialized countries, citizens of developing countries prefer PTAs with countries that have similar or higher environmental and labor standards Economically, to the extent individuals view environmental standards as production cost increasing, they will regard higher standards of their trade partner, relative to their own country’s standard,

as an economic advantage Conversely, individuals should be less supportive of trade with partner countries that have weaker environmental standards The reason is that countries with similar or lower environmental protection standards are regarded as direct competitors in attracting foreign investments in sectors that produce pollution-intensive goods These arguments apply only if the respective trade partner does not use the PTA to impose higher environmental protection standards on the developing country

Non-economically, to the extent the environmental preferences of the median citizen (voter) are greener than the environmental preferences of the median political and economic elite member, citizens may view PTAs as an instrument to “import” higher environmental standards and also related technology and funding from their trade partners to obtain a public good their government is unwilling to deliver This argument is reminiscent of the trading-up logic (e.g., Bernauer and Caduff 2004; Saikawa and Urpelainen 2014; Vogel 1996) The economic and the political leverage argument differ in the underlying causal mechanism But they push in the same

direction: We should expect individuals to favor PTAs with countries that have at least similar or have higher environmental standards, all else equal

H2: Individuals are likely to prefer PTAs with countries that have equal or higher environmental standards They are also likely to welcome demands by richer countries for inclusion of environmental protection clauses in PTAs

4 Empirical Design

We start by describing the general survey design and then identify the key variables in the empirical models, as well as the methods used for data analysis

4.1 Survey design and data

Existing surveys do not provide the data that is needed to test the two hypotheses set forth above Hence we implemented population-based surveys in three developing countries: Costa Rica, Nicaragua and Vietnam

We focused on Costa Rica, Nicaragua, and Vietnam because these three countries differ strongly in their level of development, political system characteristics, and environmental protection standards We are interested in whether the hypothesized effects hold across very different types of developing countries, or whether they are moderated by country characteristics Costa Rica is clearly the richest of the three countries Nicaragua and Vietnam’s per capita income level is about one quarter of the income in Costa Rica In contrast, the trade to GDP ratio is much higher in Vietnam than in Costa Rica, with Nicaragua located in the middle between the two Environmental standards, however, are far lower in Vietnam than in Costa Rica, with Nicaragua again located in the middle (see Table A1 in the Appendix)

The surveys were carried out between August 2013 and February 2014 The data was collected in face-to-face interviews We used a stratified random sampling approach that is based on a multi-stage probability sampling design (see Table A2 in the Appendix) Respondents were 18 to 64 years old Sample sizes were 820 in Costa Rica, 800 in Nicaragua, and 1400 in Vietnam The samples for Costa Rica and Nicaragua are from the entire country In Vietnam we restricted the sampling to five key areas: Hai Phong, Hanoi, Da Nang, Ho Chi Minh City, and Can Tho The surveys

in Costa Rica and Nicaragua included all items we use to test Hypothesis 1, as well as the conjoint experiment we use to test Hypothesis 2 (see further below) For logistical reasons, the conjoint experiment in Vietnam had to be implemented separately We started with an English version of the questionnaire and, after translation into Spanish and Vietnamese, used backwards translation to make sure that questions were clearly understood and the information obtained is comparable across countries

4.2 Dependent and explanatory variables

Trade is a multi-facetted phenomenon, and so are preferences concerning trade, which need to be measured for testing our first hypothesis We go beyond the standard approach in the existing literature, which measures trade preferences with a single survey item, and use a composite measure based on twelve survey items The first set

of items taps into respondents’ spontaneous associations with international trade We confronted respondents with six word pairs and asked them to indicate which of these words in a given pair they associated international trade more strongly with In each pair, one word has a positive and the other word has a negative connotation This approach is somewhat reminiscent of (but much simpler than) an Implicit Association Test The second set of items asks respondents to evaluate the consequences of international trade from an egotropic, sociotropic, and a consumer perspective Table 1a provides summary statistics of the composite measure for trade preferences Its values are standardized on a 0…1 scale, with higher values indicating more support for free trade Question wordings and the results of confirmatory factor analysis, which supports aggregation of the 12 items into one measure, are shown in Tables A3 and A4 in the Appendix

Table 1 Overview of key variables

Table 1a Dependent variable: Trade preferences

12 survey items For question wording and results of confirmatory factor analysis, see Tables A3 and A4 in the Appendix

Table 1b Independent variables: Environmental attitudes

natural environment should be implemented, even if they cause some job losses.”

environment, we all should be willing to restrict our current standard of living.”

Govt

Spending

for protecting the natural environment in VN/CR/NIC be increased, decreased, or kept about the same?”

As shown in Figure 1, respondents in all three samples express rather strong support for free trade

Figure 1 Trade preferences

Our measure for environmental concern, the main explanatory variable in Hypothesis

1, is based on three survey items The first two items differentiate between individual

“willingness to pay” for environmental protection in both private life and at the collective level (Table 1b) Both items are worded so as to induce a tradeoff Respondents are asked to indicate whether they would be willing to make a sacrifice

to protect the environment We chose to focus on the widely recognized tradeoff between job protection and environmental protection and the tradeoff between living standards and environmental protection The third item captures policy-oriented environmental preferences We asked whether respondents prefer government spending for environmental protection to be increased, stay the same, or to be decreased Higher values on the three variables indicate stronger pro-environment

attitudes We tried to aggregate the three variables (as well as additional items on environmental attitudes we had included in the surveys) However, low correlation coefficients (see Table A5 in the Appendix) and unsatisfactory values for Cronbach’s alpha indicate that they should not be aggregated

As shown in Figure 2 below and Table 1b above, at first glance this measurement exercise produces somewhat surprising results Environmental attitudes are rather strong in all three countries However, given the much higher average income level in Costa Rica, relative to the other two countries, we would have expected stronger pro-environment attitudes there This is not the case We observe the lowest willingness among Costa Ricans to increase government spending for environmental protection Similarly, Costa Rican respondents seem relatively less willing to implement measures to protect the environment if this leads to job losses Yet, both in Nicaragua and Costa Rica, people are more willing to restrict their living standards in order to protect the environment, compared to the Vietnamese sample

Figure 2 Environmental attitudes

4.3 Other variables

Previous studies have examined a wide range of factors that could explain variation in public support for trade liberalization For example, they have used educational attainment as a measure of skill level, arguing more skills implicate better labor market prospects in a more open economy, which in turn leads to more positive attitudes vis-à-vis trade (e.g., Kaltenthaler et al 2004; Mansfield and Mutz 2009;

Mayda and Rodrik 2005; O’Rourke and Sinnott 2001) Others (e.g., Hainmueller and Hiscox 2006) have argued that the educational attainment effect may reflect an ideological impact (more education leads to more awareness that trade can be beneficial, and to more cosmopolitan attitudes), rather than the standard factor endowments effect derived from trade theory We are agnostic about the underlying theoretical logic, but simply control for education levels of respondents The lowest level on this variable (1) indicates that the respondent has no formal education, while the highest level (7) indicates a postgraduate degree

In addition, we include employment status and household income as these could affect both environmental attitudes and trade preferences We asked respondents to indicate whether they currently have a paid employment (1), are self-employed (2) or are currently not working (0) To capture household income levels, respondents were asked to indicate their income range on a 10-point scale However, descriptive statistics revealed that the distribution of responses to this question differs considerably from the population distribution (based on national census data5) Many people did not respond and many appear to have understated their income We therefore use income satisfaction in our main analysis, relying on a 4-point scale, with higher values indicating greater satisfaction We also include a variable that indicates whether the respondent lives in an urban or rural area

Furthermore, we control for age, as several studies have found that older people are more supportive of protectionism (e.g., Kaltenthaler et al 2004; Mansfield and Mutz 2009; Mayda and Rodrik 2005; O’Rourke and Sinnott 2001) We also control for gender The existing literature suggests that women hold less favorable views on free trade (e.g., Beaulieu and Napier 2008; Burgoon and Hiscox 2008; Kaltenthaler et al 2004; Mayda and Rodrik 2005; O’Rourke and Sinnott 2001) Table 2 summarizes the control variables

5 Vietnam Census: 2009, Costa Rica: 2011, Nicaragua: 2005

Table 2 Control variables

Coun-try

Obs Mean Std

dev

Min Max Description

education that you have attained?

Employ-ment

are currently working or not

financial situation of your household?

In a typical conjoint experiment, respondents are asked to rank or rate two hypothetical profiles (options) These profiles are generated through combination of a set of randomized attributes For our experiment we generated the profiles of potential trading partner countries that vary with respect to several characteristics (see Author for the complete list of attributes) The conjoint experiment started with a short

6

The survey and sampling procedures were exactly the same as the ones employed for the survey that generated the data for testing Hypothesis 1 Also, the same group of enumerators administered the

introductory text 7 and instructions Attribute values were generated using a completely independent randomization approach We focused on the effect of the attribute that describes the potential trade partner’s environmental protection

standards The description of this attribute reads: “The environmental protection

standards in partner countries may be stronger, similar, or weaker, compared to the standards in [respondent’s country].” Thus, the randomized attribute values were

stronger, similar, or weaker We are interested in whether respondents favor trade partner countries that have lower, similar, or higher environmental standards, relative

to their own country’s standards Answers are captured in both binary terms (i.e., whether the respondent prefers country A over country B) and on a seven-point scale indicating how much the respondent prefers (or opposes) each of the two countries

We rescaled these two rating measures to range from 0 (“never support”) and 1 (“always support”) Each respondent was asked to complete five choice tasks In each choice task the respondent was confronted with two country profiles Thus, the unit of analysis is the country profile per choice task, which generates ten observations per respondent.8

The analysis of the data thus generated allows us to identify the causal effect – the Average Marginal Component Effect (ACME) – of country attributes on the probability that a particular country is preferred, or on how positively (or negatively) the potential trade partner country was rated (Hainmueller et al 2014) Because the attribute values are randomly assigned, this analysis can tell us to what extent trade preferences with respect to a particular country are driven by the (potential) partner country’s efforts to protect the environment

7 “[COUNTRY] is currently negotiating international trade agreements with other countries The purpose of such trade agreements is to make it easier for producers from other countries to sell their goods and services in [COUNTRY] (imports), and to make it easier for producers based in [country] to sell their goods and services in other countries (exports) [COUNTRY] is considering different partner countries for such trade agreements These partner countries may differ with respect to their characteristics For a start, please look at the following table very carefully It describes some basic characteristics partner countries for international trade agreements with [COUNTRY] may have.”

8 Number of observations = 2 (country profiles)*5 (choice tasks)*number of respondents This results

in a maximum of 8200 observations for Costa Rica, 8000 for Nicaragua, and 7000 for Vietnam