Nationalism in the produce aisle: Using country of origin labels to stir patriotism and animosity


 This study aims to determine how feelings of nationalism or patriotism for one's own country and feelings of animosity for countries that are in political conflict affect consumers’ willingness to refrain from buying products from various origins. Four hundred forty-two shoppers participated in a survey that included questions about ethnocentric and patriotic feelings and a choice task between local and foreign produce. We find that a significant proportion of Israeli consumers are not willing to buy imported produce from countries that are in conflict with Israel. The proportion of consumers who would not purchase imported produce increases substantially with the intensity of the conflict. Consumers who avoid buying produce from friendly countries are those whose ethnocentric and economic-centric relationship are strong.


Introduction
Consumers' preferences over local products are affected by perceptions of superior local quality, concerns over environmental or economic impacts, and emotional drivers like the feeling of attachment one has to their community. Such emotional drivers may include feelings such as ethnocentrism, patriotism, and nationalism. Local producers may be able to build a competitive advantage based on a reputation for functional attributes (e.g. quality, freshness, etc.) or based upon non-material considerations (e.g. pride of place, ethnocentrism, and animosity) (Gao, Zhang, and Mittal 2017;Steenkamp 2014). The latter emotional factors can jointly affect both choice and the perceptions of functional attributes (Shiv and Fedorikhin 1999). For example, emotional attachments to a brand or product could alter the relative weight given to functional and affective attributes of the product in considering a purchase (Keller and Lehmann 2006).
The value granted to a product by originating in a specific country via such emotional attachments has been referred to as 'country equity' (Shimp, Samiee, and Madden 1993), and the literature contains a few attempts to measure this equity. Positive or negative emotions toward one's country of residence or other countries may add or subtract from the products from issues related to functional quality attributes. We use a choice between produce items of identical quality rating as the context for our study. Israel is an ideal choice for such study due to the ongoing conflicts between Israel and several of their neighboring nations. The Israeli and Palestinian conflict is perhaps the most visible example of how nationalist feelings can lead to extreme and dangerous actions in the name of promoting a nation's social identity (Kelman 1997). By using choices involving countries in varying levels of conflict with Israel, we can identify the impact of such political conflicts independent of considerations of environmental impact. We find evidence that many Israeli consumers are unwilling to purchase produce from countries in conflict with Israel no matter how large the discount in price. Moreover, we find some support for the notion that the nature of price sensitivity can be altered by political conflict. For example, price sensitivity is weakest among consumers in our survey when choosing between Israeli and Palestinian goods (the two have been in active and intense conflict for more than 70 years), with many consumers unwilling to buy Palestinian goods at any price. The relationship between price sensitivity and intensity of political conflict appears to be related to the consumer's political views and intensity of religious feelings-which are themselves correlated. While these results may not be directly generalizable to all political conflicts, they demonstrate the plausibility of using such conflicts to the economic advantage of local goods, and may in some instances provide incentives for some producer groups or policymakers to stoke the flames of political conflict. Political leaders seem recently to have changed the traditional code of behavior and instead of bridging between social groups are using and fueling existing conflicts to gain support from those on the extremes (or their base supporters). The Pew Research Center's analysis of the 2016 US Presidential election displays how division motivated voters (https://www.pewresearch.org/politics/2016/06/22/ partisanship-and-political-animosity-in-2016/). This practice of stoking division can draw attention away from more nuanced and detailed discussions of the real problems, and often provide cover for poor handling of the conflict. In a similar manner, if a certain politician, such as the newly appointed Israeli Agriculture minister, wanted to strengthen demand for local production, he may advocate for stricter COO labeling-an Israeli may prefer to pay more for produce that comes from the Gaza Strip than from Turkey.

Literature review and conceptual model construction
A country's image can generate associations with the quality of functional attributes, reputation of products (Mariutti and de Moura Engracia Giraldi 2019), and emotional and social benefits. General perceptions of quality, reputation, unique characteristics, history, and authenticity help to construct a coherent nation brand (Kotler and Gertner 2002;Steenkamp 2019b). These associations allow COOL to serve the three key functions of a brand: psychological-supporting the consumer's internal identity (Silberhorn, Boztug, and Hildebrandt 2017), sociological-allowing the consumer to signal their identity to others (Escalas and Bettman 2005), and reducing uncertainty by signaling quality to the consumer (Erdem and Swait 1998). Consumers' expectations regarding production quality, safety, and environmental impact are tightly connected with consumer perceptions of specific country attributes and image (Roth and Diamantopoulos 2009). While COO perception, image, and associations affect perceptions, they do not always affect consumers' choices (Roth, Diamantopoulos, and Montesinos 2008;Samiee 2019). A country's image and perceived quality may not be grounded in the actual quality and are likely to be affected by impressions or stereotypes of the COO and its residents. COO image may be related to the political situation (Steenkamp 2019a) and context of the country. The relationship between the political situation and a country's image was highlighted in a recent survey tracking changes in the quality image of different countries between 2016 and 2017 (Strauss 2017). The survey found that countries that were engaged in a current conflict, such as Iran, Iraq, Russia, Ukraine, and Syria, declined in consumer ranking of product quality relative to previous years (Strauss 2017).
In the context of produce, a unique set of inputs associated with a local production region may contribute to a perception of superior quality. In addition to superior quality, local products are associated with lower transportation costs, a lower environmental cost (Grunert, Hieke, and Wills 2014), and a more ecologically sustainable consumption pattern (Tobler, Visschers, and Siegrist 2011).
Even without having quality differences between the local and non-local products, consumers still may prefer local production. These preferences may originate from a variety of sentiments. Ethnocentricity, nationalism, patriotism, and within-group support may lead one to more highly value local production on the one hand, while nationalism or animosity may lead one to value foreign production less on the other hand. For example, Japanese consumers were found to be much more favorable toward a Japanese-made producteven if the product appeared to be below market standards for quality (Gürhan-Canli and Maheswaran 2000). Similarly, mainland Chinese ranked the quality of local products highest, while China was ranked last out of forty-nine in a quality of product image index by non-Chinese respondents (Strauss 2017). The later example hints that perceptions of quality may be affected by the emotions felt toward the COO. Thus, emotions toward a specific COO affect choice indirectly by affecting perceptions and directly by causing utility or disutility that is associated with the congruity of the COO image and consumers' self-and social identities. When the direct effect of the image of COO on social identity and self-identity is very strong, it can completely moderate quality concerns if any exist. Thus, even if local produce is not easily differentiated in terms of product quality, building a local brand is still feasible if the local identity provides some psychological and social benefits.
We take elements from Akerlof's model of caste (Akerlof 1976(Akerlof , 1980(Akerlof , 1997 to form a conceptual model of utility. This model forms the basis of the research hypotheses we test empirically. Akerlof suggested a framework in which consumers' actions affect selfperception and social feedback. Both congruity of self-perceptions with actions and positive feedback from social groups increase utility. Consumers' utility is a function of a vector of the quantity and quality of a product (which can be local or imported), the quantity of other goods, and self-perceptions and social feedback (resulting from adherence to social norms).
Consumption of local produce may help consumers in building a stock of self-perception as being a virtuous individual, for example, by supporting local community, businesses, or labor (Maronick 1995). Utility of self-perception is a function of the degree of congruity between choices made and the self-mindset or conative part of decision making (Johansson 1989). In the context of group identity, concepts of free trade, globalization, and political views affect the level of patriotism, nationalism, and feelings of animosity. The idea that possession of branded products defines who you are (Belk 1988) and that purchase of a particular brand can shape a consumer's identity (Park et al. 2010) can be applied in the context of COO. COOs in themselves can act to some extent as a brand, and can contribute to a consumer's self-perception. Consumers are likely to choose brands that match their identity (Reed et al. 2012). A country's image affects consumer attitudes toward the country, and that attitude then moderates the purchasing choice (Zeugner- Roth, Žabkar, and Diamantopoulos 2015). In addition to the attitude toward specific countries, the choice of imported products may be affected by self-perception of globalization. Consumers who perceive themselves as global citizens of the world support their identity by purchasing global products and brands (Alden, Steenkamp, and Batra 1999). Alternatively, consumers who perceive themselves as locals will prefer local production.
Consumer's benefit is also affected by the social feedback from obeying (or disobeying) social norms. Consumption of produce from a specific origin helps to reinforce or signal a consumer's belonging (Cordes et al. 2003). Figure 1 illustrates the structural relationships between economic variables (quality, price) and self-and social elements of COO.
When the quality of the local produce is not superior to that of the non-local products or when the price ratio between local and non-local products is higher than the ratio of marginal utility from quality, preferences toward local products may persist due to either positive or negative feelings regarding own country and competing countries. This study tries to quantify the power of sentiments toward homeland and specific COOs as forces that shape consumer preference, in addition to quality and price differences.
Previous studies have devoted almost equal attention to research focused on positive feelings toward local products versus negative feelings toward global products. A Google Scholar search of keywords 'ethnocentrism and COO' and 'animosity and COO' reveals 4,440 sources related to the former pair and 4,120 sources related to the latter pair, suggesting a similar importance and research attention given to ethnocentrism and animosity.Positive sentiments for homeland strengthen the preferences for local products, while consumers who have a global identity strengthen preferences for purchasing global brands. However, our model allows for much greater diversity of sentiments. For example, one may support locally grown foods out of love of country (positive), or out of intolerance for outgroups and hate for a foreign country (negative). Shimp and Sharma's (1987) index of ethnocentric feelings, which was quantified in a seventeen-item scale (CETSCALE), is negatively correlated with both intention to purchase imported products and actual purchasing behavior. The correlation with actual purchasing behavior is weakened, however, by price and income effects. Thus, a higher level of ethnocentrism is likely to reduce the sensitivity to price.

Research hypotheses
Nationalism and animosity may serve functionally similar roles in the marketing of domestic versus imported products. Though nationalism and patriotism are recognized as powerful forces that have shaped geopolitics over the last two centuries (Gellner 2008), the psychological foundations and implications of both are still heavily debated. Originally, patriotism and nationalism were used as synonyms, only taking on distinct meanings at the beginning of the twentieth century (Viroli 1995). Patriotism is generally defined as a love for one's country (Merriam-Webster 2021), while nationalism is usually characterized as pairing this love of country with a feeling of superiority and desire for power over other nations. As such, while both are tied to a national identity, these feelings lead to distinct goals (Blank and Schmidt 2003). For example, Blank and Schmidt (2003) find that while the newly reunified Germany was identified by Germans as being associated with inclusiveness, individual feelings of nationalism were associated with intolerance for minorities. Feelings of patriotism, on the other hand, were associated with greater tolerance for minorities. In much of the literature, the ideas of nationalism or patriotism are closely associated with the individual's conception of their own identity (Blank and Schmidt 2003;Kelman 1997). These feelings can be strong and can motivate individuals to take specific actions (McCauley 2001). In the context of branded products, patriotic feelings would suggest highly valuing the products made locally for their contribution to the nation. Alternatively, nationalist feelings would be associated more closely with both valuing the domestically produced products and a devaluing of products made in other countries. These may be seen as being associated with a social dominance orientation, a group-based feeling of dominance and an opposition to social equality (Kemmelmeier and Winter 2008;Kosterman and Feshbach 1989).
Nationalism becomes a particular driver when countries come into conflict. In this case, nationalism is associated with animosity toward specific countries that have come into conflict with the home country (Jung et al. 2002). The negative relationship between ethnocentricity and nationalism and the positive relationship between nationalism and animosity were previously examined using a survey of mainland Chinese consumers who demonstrated a decided bias against Japanese products (Klein et al. 1998). In this work, we push the boundaries of this relationship both by examining a range of levels of conflict between countries and by measuring the likelihood to avoid purchases from a country that is in active conflict. This allows us to quantify the potential revenue effects of animosity resulting from national conflicts. In addition to ethnocentrism, patriotism, and nationalism, the magnitude of the willingness to accept imported produce is expected to be affected by animosity-a different psychological construct from ethnocentrism (Klein et al. 1998;Nijssen and Douglas 2004;Riefler and Diamantopoulos 2007). That is, the relative preference for local produce is motivated by hate of the other rather than preference for the local. Because hate is the strongest emotional motivation (Cameron 2009), negative emotions are a potentially stronger motivator than positive emotions (Baumeister et al. 2001). Such negative asymmetry (Rozin and Royzman 2001) leads to a lower willingness to pay (higher discount demanded) for imports from countries that consumers dislike. COOL provides buyers with information about origin, with negative information likely to have a stronger impact on preferences than positive information. In general, it has been found that negative advertising is more effective at reducing sales of competitor brands than increasing sales of the promoted product (Anderson et al. 2016). Empirical estimation integrating considerations of both ethnocentrism and animosity creates an econometric challenge since negative dispositions toward a COO are positively related to the strength of individual's patriotic feelings (Meas et al. 2014;Orth and Firbasová 2003). Animosity is a function of the intensity of unresolved disputes, perceived economic threats, and cultural factors. Animosity is expected to be reduced for countries that are perceived to be culturally similar to the importing country, increasing the willingness to purchase from those countries (Sharma et al. 1994;Watson and Wright 2000).
Following this discussion, we hypothesize that animosity is likely to have a stronger effect on sales relative to ethnocentric and patriotic feelings. Hypothesis 1 is therefore suggested: H1: Individuals who have strong feelings of ethnocentrism, patriotism, nationalism, or animosity toward global produce are likely to be less sensitive to price when choosing between local goods and goods from countries for which they harbor animosity. In the extreme case when feelings of ethnocentrism, nationalism, and animosity are very strong, consumers may refrain from buying imports altogether regardless of price differences. religious beliefs (Gorsuch and Aleshire 1974;Lantz and Loeb 1996). In a similar manner, animosity may be related to religiosity and differences between the religion of the origin and the target country (Schmidt 2019). Israeli consumers who hold more traditional and conservative views about the world tend to be more religious and display more pride in their country. A larger proportion of ultraorthodox and conservatives tend to identify themselves as supporters of the middle spectrum of the political range, 52 and 49 per cent versus 47 per cent (in both segments) who support the right wing. Orthodox Jews are more right leaning in their political views, with 56 per cent of the orthodox defining themselves as supporting the political right and 41 per cent supporting the middle, while only 2 per cent report support for the political left. Most of those identifying as secular support the political middle parties, 62 per cent, while 24 per cent support the right wing and 14 per cent support the left wing (for more details see see https://www.pewresearch.org/wp-content/ uploads/sites/7/2016/03/israel_survey_overview.hebrew_final.pdf (in Hebrew)). Thus, it is hypothesized that H2: Consumers who are more religious and those who support politically right leaning parties are more likely to refrain from buying imported products from countries that are in active political, cultural, or economic conflict with their home country.
Scholars find that higher levels of education and income, which are naturally likely to be positively related, are negatively correlated with the level of ethnocentrism. Educated individuals are also less conservative (Ray 1990) and report less pride in their country (Rose 1985). Age was found to be positively related with ethnocentrism, suggesting older consumers tend to value domestic products more highly (Chryssochoidis, Krystallis, and Perreas 2007;Han 1989), while younger consumers tend to be more cosmopolitan (Bannister and Saunders 1978) and therefore are expected to be less ethnocentric. The findings regarding the effect of gender on preferences for local products are mixed. On the one hand, females are less individualistic (Singelis et al. 1995) and differ in their views of group attachment (Baumeister and Sommer 1997) leading to a potential higher acceptance of global imports. On the other hand, females are more ethnocentric (Sharma et al. 1994), which could lead to a preference for local products. Thus, it is hypothesized that H2: Younger and more educated consumers are more likely to be more open to imports, while the effect of gender is unclear. Consumers who hold right-wing views are expected to hold a stronger preference for local produce.
In order to explore the role of non-price and income variables in shaping demand for local products and test our research hypotheses, we conducted a survey of 422 Israeli consumers. In the next section, we describe the methodology of the survey.

Data
The data underlying this article and code used are available in the online supplementary material. Data for this study were collected using face-to-face surveys held in shopping centers in Israel. The places in which the shoppers were interviewed (i.e. large cities, suburban, peripheral, etc.) were selected in order to increase the variance of respondents' socioeconomic variables, with a particular focus on two major variables that are the focal point of our research interest. Our goal is to estimate the effect of religiosity and political support (which are closely related) on the choice between local and imported produce. The identity of the COO is assumed to moderate the choice. Accordingly, the sample's relative size in each city was selected as a function of the heterogeneity of the population. The places that are characterized by the largest variance in these two variables were preferred over places with low variance. By design, the sample is not a representative sample of the Israeli population.
However, in the essential variables that we hypothesized affect the choice between local and imported produce, the distribution of our sample is close to that of the Israeli population, and reflects the general heterogeneity present in the country. Specifically, the self-definitions of religiosity, gender, and education are in close proximity to that of the general population. We believe that a high-powered face-to-face interview is sufficient to answer the questions of interest.
Respondents were randomly selected and approached as they exited grocery stores, and were asked to participate in a survey that would take approximately 10 min to complete. Those who agreed to participate were given survey questionnaires. Conducting the survey in close proximity to supermarkets and shopping is believed to increase the relevance of the answers and increase their validity (Babin and Darden 1995;Häubl and Trifts 2000). This added validity and relevance is somewhat of a counterbalance to the hypothetical bias inherent in survey data. Alternative approaches using economic experiments can be used to address hypothetical bias, but may suffer from poor external validity due to non-random selection of participants, or due to failures to place participants in realistic choice contexts (Roe and Just 2009). It is often difficult to find an experimental subject pool that is similar to the policy-relevant population, and even more difficult to simulate the relevant decision environment. We thus opt for interviewing real buyers over potentially obtaining more accurate responses from a less representative sample of participants. We recruited 442 participants, 255 of which were females and 197 males. The ages of participants ranged from 18 to 78. We present the sociodemographic characteristics of our sample in Table 1 and compare them to the Israeli population.
Notably, this sample skews toward singles, more educated, more secular, and left party supporter respondents.

Questionnaire
The conceptual model suggests that the choices are a function of quality, price, and sentiments. In order to crystalize the effect of sociodemographic and political support on choices, the responders were instructed that they needed to select between local and non-local products that are identical in quality. Price, per se, is not the primary object of our study, but rather the willingness to buy at all. Consumers' preferences in this context are a function of differences in quality, ethnocentric and patriotic feelings, and feelings of animosity toward specific COOs. Consumers were instructed that quality was identical; however, consumers' prior beliefs about quality and price differences may be stronger than our instructions, and the perceptions of these two variables were thus included in our estimation equations.
Consumers' ethnocentric sentiments were elicited by collecting information about their perception of the importance of buying local, the economic damage of buying imported produce, and the negative effect of buying imported produce on the agricultural sector. Responders were asked to state their level of agreement with four statements each measured using a seven-point Likert-type closed-form question, for example 'Buying imported fresh agricultural produce is unpatriotic (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)'. Additional Likert rated questions include 'Local agricultural sector is related to national identity', 'Purchasing imported vegetables hurts the agricultural sector', and 'Country of origin is important for me'.

Contingent valuation method
We used the contingent valuation method (CVM) (Boyle 2003;Hanemann 1984) to determine consumer valuation of all the produce covered in the survey. CVM is preferred over other methods in the case of hypothetical choices (Boyle et al. 2001), and leads to less bias than closed-form quantity categories where the way the category is divided a We aimed to compare different levels of political support and levels of religiosity variables. The representativeness of the Israeli population is less concerning given the focus in the comparison between the social groups.
b The per cent of singles varies significantly according to city, gender, level of religiosity, and age. The average number is retrieved from https://www.cbs.gov.il/he/mediarelease/DocLib/2019/181/19_19_181b.pdf. c Education in the age group 25-64 found in the OECD report: https://data.oecd.org/eduatt/ adult-education-level.htm#indicator-chart. d The average reported is in the Jewish population. The average in the general population is 29.8. conveys information that may bias respondents' answers (Krosnick and Presser 2010;Schwarz et al. 1985). CVM is less cognitively demanding than alternative methods (Louviere et al. 2000). Because consumers are not likely to allocate significant cognitive resources to a survey (Thaler and Sunstein 2008), the CVM method is commonly believed to dominate the hypothetical choice method. Consumers were given a choice between local and imported apples, tomatoes, cucumbers, grapes, and garlic. The foreign countries of origin were the USA (apples), Italy (grapes), China (garlic), Jordan (tomato), and the Palestinian Authority (cucumber). The five largest produce imports in terms of volume were selected with the COO chosen so that the produce item is the largest share of imports from the country selected. These countries can be ordered by their relative conflict with Israel (from most friendly to least) as the USA, Italy, China, Jordan, and the Palestinian Authority. The ranking in terms of level of conflict was derived from a pre-test held in a class of undergraduate students in Israel (around seventy participants; see Appendix A in the Supplementary Data). This research design increases the relevance of the products and countries but with some loss in identification. There is a one-to-one relationship between the importing country and the imported product, which eliminates our ability to separate country and produce item effects.
In a similar manner to Davvetas, Sichtmann, and Diamantopoulos (2015), the open-ended CVM questions asked consumers to report the maximum price of the imported vegetable (fruit) that would cause them to prefer the import, for example 'The price of Israeli grapes is 20 NIS (5 Euro) per kilo, what is the highest price of Italian grapes that would cause you to prefer Italian grapes over local grapes? _____ (write zero if you are not willing to purchase imported grapes from Italy for any price)'. Similar questions were asked about the other four products (apples, cucumbers, tomatoes, and garlic) that are associated with different COOs (the USA, Palestine, and Jordan, respectively).
We created a new dummy variable that indicated whether a consumer indicated they would always refrain from buying the produce from a specific COO. The variable assumed the value of zero for consumers who responded to the CVM that they are not willing to buy the produce from a specific COO at any price, and a value of 1 otherwise. Given the open nature of the question, we coded three answers as indicating that a consumer would always refrain from buying the imported good: an acceptable price of zero or any negative value, and explicit text answer in which responders specified that they would not purchase from the specific COO.

Measuring ethnocentrism
In addition to economic and sociodemographic questions, we included commonly accepted scales to measure ethnocentricity that were adapted from Darby et al. (2008) and Shimp and Sharma (1987), and rephrased to fit the Israeli culture, for example 'Buying imported fresh agricultural produce is an unpatriotic act (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)' and 'I prefer buying local vegetables and fruits as it supports the agricultural sector and employment (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)'. We also included a battery of questions about perceptions of price and quality. Questions about quality were adapted from Darby et al. (2008) and Jekanowski, Williams, and Schiek (2000), and phrased as 'Israeli vegetables are of superior quality (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)' and 'I prefer local vegetables and fruits as they are fresher (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)'. The questions about price were phrased as 'The price of vegetables in Israel is too high (1 strongly disagree, 2 do not agree, …, 6 agree, 7 strongly agree)'.

Measuring religious beliefs and political views
The intensity of religious beliefs and political support were elicited using closed-form questions. Religiosity was measured by 'I consider myself (secular 1, conservative 2, …, ultraorthodox 5)' and 'I support the ______ party (Right 1, Center 2, Left 3)'.
As the subject of this study is to identify the choice process for different COOs, we employed a two-stage type of analysis for that purpose. The first stage consisted of dimensionality reduction of the data using a principal component-based factor analysis, identification of underlying factors. The principal component-based analysis is used to avoid potential collinearity between similar perceptions. The second step uses the new dimensions as an input for the choice to refrain analysis (for an application of factor scores in a choice model, see Gensch and Ghose 1992).

Factor analysis
Using factor analysis on the perceptions of eight variables, four represent ethnocentrism, three represent perceptions of quality, and one represents perception of price; three factors were identified that affected consumers' choice whether to refrain from buying imported produce from specific COO. The four variables that capture the level of ethnocentricity are 'Purchasing imported vegetables is unpatriotic', 'Purchasing imported vegetables hurts the agricultural sector', '(The) Country of origin is important for me', and 'Local agricultural sector is related to country identity'. The three perceptual attributes that are expected to capture the quality of the Israeli produce relative to imports from unspecified COO are 'The quality of Israeli vegetables is superior (the best)', 'Israeli vegetables are tastier', and 'Imported fruits are better'. The first component reflects ethnocentric feelings. The second factor represents quality perceptions, while the third represents perception of price. The variance explained by the three factors was 66.1 per cent; analysis yielded a three-factor solution, which is presented in Table 2.
Following the identification of the underlying dimensions involved in the choice whether to purchase from a specific COO, we applied the binary logistic regression to estimate the choice model with the three factor scores from the previous stage, political support, and sociodemographic set of variables.

Econometric approach
The utility of consumer i from buying product j either local (L) or non-local (N) is denoted by U i=L,N . Recall that there is one-to-one identity between a product and COO. The utility function can be separated into a deterministic component, V j , and unobserved random component, ε j , that is assumed to be the distributed extreme value and has zero mean. Specifically, (1) Following Lancaster (1966) and Kim, Allenby, and Rossi (2007), it is assumed that the deterministic element, V j , can be approximated by a linear weighted sum of perceptions that are captured by the three factors. Specifically, where F kij represent respondent i's perceptions of factor (product dimension) k, through factor scores, of product (COO) j, for k = 1, 2, 3 (ethnocentrism, quality, and price). α kij are the COO (product) dimension coefficients to be estimated. The distribution of ε i j is assumed to be exponential (Gumbel type II extreme value or logistic) and thus the odds ratio that COO j will be chosen by consumer i is represented by (see Aldrich and Nelson 1984). We construct a binary logistic regression model for individual i = 1, . . . , N, which identifies her choice of whether to purchase from COO j at any price (the choice to refrain from buying from j): where θ i j is the odds ratio of buying from country j; 3 k=1 α kij F kij are the three factors that represent individual i's perceptions for ethnocentrism, quality, and price; SocioEcon ii is a vector of characteristics for individual i that are uncorrelated with country j, which include age, gender, and education; and Polit i is a vector of political views.
In order to identify the effect of sociodemographic characteristics, strength of ethnocentric feelings, and animosity, we run five binary logistic regressions for each COO separately. The dependent variable is coded as 1 if a consumer is willing to purchase from the specified COO and the independent variables are a set of socioeconomic variables and three factors that represent the level of ethnocentrism, and perceptions about quality and price. The three factors were extracted from a set of seven questions using the principal component analysis extraction method and explain 66.1 per cent of the variance. Since respondents made their choice separately for each COO, it is legitimate to assume that the choice whether or not to purchase from each COO for every individual is independent. This procedure does not allow us to compare the sizes of the marginal effects between the regressions and therefore results speak mainly to significance of the explanatory variables.

Empirical results
Analyzing consumers' choices reveals that about 61 per cent of the respondents are not willing to purchase Palestinian cucumbers regardless of their price. As expected, this is high relative to the per cent of consumers who are not willing to consider purchases from Italy (8.6 per cent), China (15.6 per cent), and the USA (16.1 per cent). The proportion of consumers who refrain from buying imported produce reveals that although prices are an important consideration in consumers' choices between local and imported fresh products, under certain conditions animosity may have a stronger effect than price. This finding supports the second part of Hypothesis 1, and signals a substantive hatred premium among Israeli consumers. Table 3 presents the results of the marginal effects from the five binary logistic regressions.
The trade-off between ethnocentric, patriotic, and nationalistic feelings and income does not appear to have much influence over consumers' expressed preference to refrain from buying from a certain COO. Our results suggest that political views and only political views are significant in affecting whether or not to purchase from the Palestinian Authority and Jordan. These two nations have been in open conflict with Israel for many years and can be considered the ones that stir the greatest animosity among Israeli consumers. Consumers who lean to the left of the political spectrum and liberals tend to have a higher likelihood of buying from Palestine and Jordan relative to right-wing supporters. Ethnocentric feelings affected only the choice of buying from the USA with those who are more ethnocentric being more willing to buy from the USA. These findings support the first part of Hypothesis 1. Age affected the choice of whether to refrain from buying from the USA and China in opposite directions; older consumers tended to refrain more buying from the USA, while younger adults revealed a higher tendency to refrain from buying from China. Finally, gender affected the choice from the USA, China, and Italy. Females tended to refrain more from the selection of imports from the USA and China while being more positive about imports from Italy. These results partially support Hypotheses 2 and 3. Our results reveal that ethnocentrism and sociodemographic profiles are relevant only when animosity is not activated (among more friendly countries). In choices between local production and imports from COO that are in conflict with home country, political views dominate the choice. In contrast, the choice whether or not to refrain from imports from friendly countries is affected by ethnocentric feelings and socioeconomic variables.

Conclusions
Globalization and the diffusion of global brands is a threat to local brands. The threat to local industry, occupation, and community may stimulate emotions, leading to an increase in the willingness to pay premium for local products (Batte et al. 2007). This increase can also be viewed as a higher discount required for the non-local produce. The extent of the discount for the non-local products is likely to be moderated by active and historical political conflicts. Beyond marginal changes in willingness to pay, this paper highlights the potential for political conflicts and political leanings to lead some consumers to avoid products from specific COOs altogether. This underscores the potential for such conflicts to be used to erect effective non-tariff trade barriers for political or trade advantage. In a chaotic and contentious world, examples of political conflicts are not in short supply and they affect both perceptions of quality and choices. The effect of animosity on the choices of consumers and the effect of consumers' political support on animosity have not received much attention while it seems particularly relevant in the current political climate. This paper addresses this research void by analyzing the role of animosity and ethnocentrism in shaping preferences toward local produce. We found that consumers are more likely to avoid products from COOs that are in a higher level of conflict with Israel, such as the Palestinian Authority. At the extreme, consumers' negative sentiments are so strong that they set an infinite price discount for countries that are in active conflict with their homeland. About 60 per cent of the Israeli consumers will avoid purchases from the Palestinian Authority at any cost. The size of the segment who refrains from buying from other countries that are in lower intensity of conflict (i.e. ex-enemy, have not been in active war, etc.) is smaller relative to the proportion of people who would refrain from buying Palestinian produce.
Our results suggest that among variables we examined, political views alone are significant in affecting whether or not to purchase from countries that have been in a conflict with Israel for many years, while ethnocentric and quality consideration affect choice only when the COO is in a friendly relationship with the homeland. Thus, examining overall variation may often hide the hatred premium effect that appears to transcend most demographic and economic characteristics. These findings address a common empirical problem of separating between ethnocentrism and animosity that are likely to be correlated (Orth and Firbasová 2003). Consumers who support right-wing parties display stronger support for local production. In addition, there is a high correlation between political views and the intensity of religious beliefs.
An obvious limitation of the current study is the representativeness of the sample. Our sampling design aimed to capture heterogeneity in the two major variables we hypothesized would affect ethnocentrism and animosity. Following this principle, the cities and rural places in which we held our survey and the relative sample sizes were chosen to increase the variance of self-defined religiosity and political support. The sample is not a representative sample of the Israeli population; however, in a few essential variables, our sample is proportional to the distribution of the Israeli population. Specifically, self-defined religiosity, gender, and education are in close proximity to those of the general population. The sample succeeded in representing the proportion of center (liberal) party supporters, but the proportion of left-wing supporters is much higher than in reality. Our sample underrepresented non-Jews.
Two other limitations of our survey are the one-to-one identity of country and product and that we did not track consumer's animosity level to each of the COOs in the study independently. We tried to represent the actual imports of produce in Israel as much as possible. Thus, we chose products from each COO that are actually exported to Israel (e.g. apples from the USA). This does not allow us to include products as fixed effects. Since we did not ask implicitly about animosity, we could not test Hypothesis 1 directly. However, we were able to identify which consumers assign high importance weights to the identity of the COO. Future research should address more precisely the level of animosity and ethnocentrism as a way to distinguish between these two sentiments. Our findings point to the potential for raising non-tariff trade barriers with nations who are in conflict through the simple promotion of COOL, drawing attention to offensive origins at the point of purchase. While this could have price effects among countries that are in mild conflict-or not in conflict-with the home country, much larger effects would prevail for countries in deep conflict. Moreover, it would appear that the political climate would significantly impact the effectiveness of such a policy.

Supplementary data
Supplementary data are available at QOPEN online.