The Impact of Economic Inequality on Preferences for Minority- vs. Majority-Endorsed Products: The Case for Bulgaria

(1)

The Impact of Economic Inequality on Preferences for

Minority- vs. Majority-Endorsed Products: The Case for

(2)

2

University of Groningen

Faculty of Economics and Business

MSc Marketing Management

Master Thesis

Date of submission: 14.06.2019

First supervisor: Dr. Sumaya Albalooshi

Second supervisor: MSc A. Schumacher

(3)

3

Abstract

This research examined the effects of perceived economic inequality on consumer behavior. More precisely, whether a condition characterized by high levels of economic inequality can lead consumers to prefer majority-endorsed product over a minority-endorsed product. The main hypothesis of this study is that people exposed to high inequality will express lower levels of general trust and subsequently will favor majority-endorsed products over minority-endorsed products. Previous research suggests that lonely consumers worry about being negatively assessed by others and conform to the majority when in public and that happy non-lonely consumers prefer majority-endorsed products. The research design is a 2(economic inequality: high vs. low) x 2(endorsement type: majority vs. minority) between-subjects factorial design. The results display no significant difference in attitude responses. A moderated mediation analysis was performed in order to identify whether general trust can have an interaction effect with endorsement type, but there was no significant interaction effect. The only significant effect of the moderated mediation analysis was that high inequality condition is a significant predictor of general trust and leads to low levels of general trust.

(4)

4

Acknowledgements

(5)

5

I. Introduction

Economic inequality is composed of two different distributions. It can relate to income inequality which is the uneven distribution of income among a population. It can also refer to wealth inequality which is the degree to which wealth in terms of assets is concentrated in the hands of a small part of the population. This paper focuses on the latter distribution.

Furthermore, inequality is measured by the Gini coefficient. A Gini coefficient of 0 suggests perfect equality, where everyone has the same wealth or income. In contrast, a Gini coefficient of 1 suggests perfect inequality, where only one person has all the income and wealth in a country. Inequality and its effects on our global society is a topic of wide discussion. Rising inequality leads to many unfavorable consequences and is a big issue around the world. The gap between the rich and the poor in many developed countries such as the USA is at its highest level in decades. However, in other developed countries such as Japan inequality is at relatively way lower levels. Hence, global inequality trends are mixed because some countries experience a decline in inequality while others suffer from rising inequality. This paper investigates the effects of perceived economic inequality in Bulgaria.

According to Peshev (2015), there is a rising trend in income inequality in Bulgaria and its position is among the highest in EU in terms of income inequality. However, in terms of wealth inequality Bulgaria is exhibiting a relatively low wealth Gini compared to other EU members. Moreover, the wealth Gini of Bulgaria is approximately 0.68 according to Allianz Global Wealth Report (2018). Wealth Ginis above 0.68 are common for most of the other EU countries.

Researchers and policymakers are investigating the drivers of inequality and seek ways in which this global problem can be solved. Growing inequality is a big disadvantage for the poorer segments of the society and results in lack of opportunity for these segments to climb up the social ladder. Furthermore, this harmful phenomenon also has notable implications for

macroeconomic stability and growth. More importantly, inequality can grant the top one percent of the wealth distribution with political and decision making power which can lead to suboptimal use of resources and lead to economic instability.

(7)

7

Organisation for Economic Co-operation and Development also published a report concerning inequality in the member countries. This report presents a rising trend in income inequality for most of the members.

The main goal of this research is to identify what are the effects for consumer decision making when a consumer experiences economic inequality. This study’s objective is to

understand the relationship between perceived macro level inequality and consumer preferences between majority- and minority-endorsed alternatives. What is the effect of perceived economic inequality (high vs. low) on preferences for endorsement type (majority vs. minority)? Could this relationship be mediated by general trust?

II. Literature review

Economic inequality has been a topic of wide discussion for the last couple of decades. It is a common phenomenon both in developed and developing countries. According to Smeeding (2005) and Payne et. al. (2017), in the past decades there has been a trend of rising economic inequality in most of the developed countries. Previous literature looks into different dimensions of economic inequality such as income, wealth and pay inequality. These dimensions have all been examined relative to issues of risk taking, poverty, generosity, personal relative deprivation, and more. In this paper, economic inequality is defined as the degree to which wealth is

concentrated in the hands of a small part of the population. Wealth can be characterized as the total amount of assets owned by a household or single individual, including bonds, stocks, pension and real-estate.

Economic inequality has been associated with many adversities. Pickett et. al (2005) explains that there is a positive relationship between inequality and homicide rates, and births to

adolescents within and outside of the United States. Wilkinson & Pickett (2009) argues that higher levels of inequality result in higher levels of drug use, obesity, violence, poorer

(8)

8

individuals according to Côte_{́ et. al. (2015), to excessive risk taking according to Payne et. al.} (2017), and to reduced happiness according to Oishi et. al. (2011).

Without a doubt, economic inequality results in many different forms of human threats, can cause negative feelings and emotions, and can have an adverse implication for all of the social classes in a community. It is interesting to see how inequality can influence people’s mind and decision making when purchasing a product or service. There is little research on the effects of inequality on consumer behavior in the consumption domain. Even though, the instruments linking economic inequality to the behaviors remain poorly examined, there are some studies that try to investigate inequality on the individual level. Barford (2017) states that sadness; anger and hope for change are feelings that are linked to inequality. Inequality contributes for many

injustices and the negative feelings that it evokes form from the need to challenge these

injustices. Furthermore, Côte_{́ et. al. (2015) suggests such negative feelings are not only common} for the lower social classes but also for the higher-income individuals. Côte_{́ et. al (2015) states} that higher-income individuals in societies characterized by inequality are concerned about losing their privileged position. Jetten et. al. (2017) also supports this claim that economic

inequality leads wealthier groups to experience negative emotions such as fear because they want to protect and maintain their status and wealth. Hence, inequality leads individuals from all social classes to feel anger and sadness even though that these feelings have different roots for the different social classes.

This research investigates the effect of economic inequality on product attitude in terms of preference between majority- and minority-endorsed products. Attitudes and beliefs have one underlying characteristic and that it is that they are subjective. Furthermore, there is no objective method of examining the correctness of an attitude. However, people can seek or draw

information with regards to the validity of their attitudes from the social environment. According to Festinger (1954), people access the correctness of their beliefs and attitudes by considering other people’s opinions. According to Asch (1956), people want to identify with the right position and the degree of popular support is considered to be an important indicator of

correctness. Therefore, people should be more easily influenced by a majority-endorsed message compared to a minority-endorsed message. According to Wang et. al. (2012), lonely consumers prefer minority-endorsed products when their preferences are kept private and shift their

(9)

9

hand, non-lonely consumers prefer majority-endorsed products in both cases. In contrast, minority-endorsed products fit better with feelings of loneliness or other negative feelings. On the other hand, lonely consumers worry about being negatively assessed by others and conform to the majority when in public. Fritsche et. al. (2017) suggests that economic turmoil can often enhance collective action intentions. Therefore, the first hypothesis of this paper is that in societies characterized by high economic inequality consumers exhibit a stronger preference for majority-endorsed products compared to minority-endorsed products.

Previous research accommodates several suggestions of causal instruments through which economic inequality could influence trust. Moreover, most of the studies point to a negative relationship. Hardin (2006) presents mechanisms through which economic inequality could influence trust and calls these mechanisms the encapsulated interest conception of trust. In other words, Hardin (2006) suggests that trust can be seen as the anticipation of cooperative behavior developed from the truster’s knowledge of the incentive framework facing the trustee. Wilkinson & Pickett (2009) also presume that the impact of economic inequality is mediated by a decline in social trust. The main argument why inequality leads to a decline in trust is that as differences between people are greater, uncertainty rises and consequently social trust in people and

(10)

10

and is the belief that endorsement by several people is more trustworthy than the idiosyncratic or false view of one individual. If trust is low then consumers should value more options that convey more trust or majority-endorsed alternatives. This research incorporates one more hypothesis. This hypothesis suggests that participants in the high inequality condition will display preference for majority-endorsed products, which is sequentially mediated by experiencing low levels of generalized trust.

III. Hypotheses and Conceptual model

This research integrates one independent variable, one dependent variable, and one mediator variable. This study hypothesizes that participants in the high economic inequality condition will favor majority-endorsed products. In other words, the higher the economic inequality, the higher the preference for majority-endorsed alternatives. This study also hypothesizes that the expected positive relationship between the high inequality condition and the preference for majority-endorsed products can be mediated by participants experiencing low levels of generalized trust in the high inequality condition. More formally, the hypotheses and the conceptual model (see

Figure 1) are as follows:

𝐻₁ : Participants in the high economic inequality condition will display a higher preference for majority-endorsed products relative to participants in the low economic inequality condition.

𝐻₂ : Participants in the high economic inequality condition will display a higher preference for majority-endorsed product relative to participants in the low economic inequality condition, which is mediated by experiencing low levels of generalized trust.

(11)

11

IV. Methodology

a. Participants and design

The experiment was designed and conducted using the online platform Qualtrics. The distribution method of the survey used was snowball sampling. The total number of people that took part in the experiment is 200 (119 Male, 81 Female). The design of the study was a 2(economic inequality: high vs. low) x 2(endorsement type: majority-endorsement vs. minority-endorsement), between-subjects factorial. The participants were all residing in Bulgaria. There was no monetary compensation for participation. After initial analysis of the data, a total of 118 participants were excluded from further analyses due to the fact that their responses had violated the criteria for a proper response. Those participants either reported that they do not understand the information provided on the comprehension check questions, or did not report the correct percentage of the wealth distribution pie charts, or failed to answer the attention check questions, or used a smartphone which decreases the effectiveness of the manipulation, or they left the computer during the study. The subsequent analysis was performed with the remaining 82 participants.

b. General procedure

(12)

12

participants’ comprehension of the information provided in the pie chart. The comprehension questions were followed by questions that were created in order to measure the participants’ perception of the wealth distribution in their province of residence (see Appendix B). The next section was composed of five statements and participants had to select how much they agree or disagree with these statements (see Appendix C). These statements were employed in order to estimate the degree of general trust of participants. After this general trust scale, participants were once again randomly assigned to an endorsement type: one including a majority-endorsed movie and the other a minority-endorsed movie. After observing the movie description (see

Appendix D), respondents were asked to indicate their attitude towards the movie (see Appendix E). Afterwards, participants were also asked to report their mood in terms of how happy, sad,

relaxed, calm, angry, and stressed they are (see Appendix F). The final section was composed of questions concerning how attentive the participants were (see Appendix G). The survey finished with a debriefing in order to provide participants with information about the real purpose of the survey and the actual wealth distribution statistics of Bulgaria. Respondents were then kindly thanked for participating and allowed to exit the page.

V. Manipulation and measurement

a. Inequality manipulation

(13)

13

The two conditions (high vs. low) were coded into one variable. The high inequality

condition was coded as “1” and the low inequality condition was coded as “0”. This aggregated variable serves as the first independent variable in the two-way ANOVA analysis.

b. Generalized trust scale

This study incorporates a mediator – general trust. The 5-item generalized trust scale

employed in the online survey was adopted from Yamagishi (1986). The scale is constructed out of five statements and participants were asked to provide information about how much they agree or disagree with these five statements (see Appendix C).

The measurement is done on a scale from 1 (Strongly disagree) to 5 (Strongly agree) and is kept continuous. Items 3 and 5 were selected from the “fear” scale presented by Yamagishi (1986). Furthermore, this 5-item trust scale is based on the two most important factors

recognized. The first factor considers the belief that people are mostly honest (items 1 and 4), and the second factor relates to the belief that trusting others is risky (items 2, 3 and 5). Item 4 is reverse scored and was recoded after the data was imported in SPSS. Therefore, a high score on this scale indicates that the participant expressed a low level of general trust. In addition, a reliability analysis was performed and the Cronbach’s alpha (α = .731) proved the scale to be reliable. Moreover, the items were aggregated into one variable that was the mean score of all the items and were kept for the moderated mediation analysis.

c. Endorsement type

(14)

4-14

item 7-point scale (1 p unappealing, uninteresting, unsatisfying, and unlikely to watch; 7 p appealing, interesting, satisfying, likely to watch). Next, a reliability analysis was performed in order to verify that the scale is reliable. The Cronbach’s alpha (α = .966) proved that the scale is a reliable representation of attitude. All of the items were kept and aggregated into one that represents the mean of all the items. In the main analysis the aggregated variable was selected as the dependent variable.

In addition, the majority and minority conditions were coded into one variable. The majority condition was given a “1” and the minority condition was given a “0”. This aggregated variable represents the second independent variable in the two-way ANOVA analysis.

d. Mood

In this section, participants were asked to report their current mood (see Appendix F). After importing the data from Qualtrics into SPSS, the items for “Sad”, “Stressed”, and “Angry” were recoded into different variables which were reversed. Therefore, high score on this scale

represents a participant that is in an extremely good mood. Then a reliability analysis was conducted and the scale was proved to be reliable due to a high Cronbach’s alpha (α = 0.922). After that the items were aggregated into one variable representing the mean of all scores.

VI. Results

a. Initial analysis and Manipulation check

(15)

15

some as unemployed and not looking for a job (8.5%). The majority of respondents (41.5%) identified as having achieved a bachelor degree, those that identified as having achieved a master degree were (29.3%), those that identified as high school graduates were (25.6%), one

participant claimed to have achieved a doctoral or professional degree (1.2%), and the rest identified as “Other” (2.4%). The majority of participants were single (64.6%), some were married (19.5%), there was one participant that identified as divorced (1.2%), and the rest

identified as “Other” (14.6%). The majority of respondents described their political orientation as liberal (67.1%), the second biggest group as conservative (14.6%), and the third biggest group as socialist (13.4%). The level of religiosity could be described as diverse, where the biggest group (36.6%) claimed to be religious, and the second biggest group (25.6%) identified as not being at all religious. Most of the participants claimed to belong to the lower-middle social class (52.4%), then there were some that identified as upper-middle social class (26.8%), some identified as working class (19.5%), and one participant identified as poor (1.2%).

With the intention to ensure that the manipulation of economic inequality was performed properly and accomplished the desired effect, it was tested if the participants in the high and low inequality conditions perceived the distribution respectively more unequal and more equal. There were three questions employed in order to determine this. More specifically, “How equally distributed do you perceive the wealth of your province of residence”, “To what extent do you feel that your province is suffering economically”, and “How satisfied are you with the economic status of your province of residence”. These three questions were measured on a 7-point slider scale and the answers for the three questions respectively were “Extremely Unequal” –

“Extremely Equal”, “Far Too Little” – “Far Too Much”, and “Extremely Dissatisfied” –

“Extremely Satisfied”. The results of an univariate analysis of variance supported the claim that the manipulation was successful. Furthermore, respondents randomly assigned to the high inequality condition perceived the wealth distribution to be less equal (M = 1.81, SD = 1.43), relatively to the respondents randomly assigned in the low inequality condition (M = 4.05, SD = 1.46, t(1,80) = 6.954, p < .001). Participants in the high inequality condition reported that their state was economically suffering more (M = 5.95, SD = 1.32), as compared to those randomly assigned to the low inequality condition (M = 4.65, SD = 1.41, t(1,80) = 4.269, p < .001).

(16)

16

participants randomly allocated in the low inequality condition (M = 3.63, SD = 1.49, t(1,80) = 4.306, p < .001).

In addition, it was tested whether the participants in the high and low inequality condition reported more negative and more positive mood respectively. The aggregate variable of mood was used as a dependent variable and the inequality condition was used as an independent variable in an univariate analysis. The results supported the claim that experiencing the high inequality condition leads to a worse mood. Moreover, respondents randomly assigned to the high inequality condition reported a worse mood (M = 4.67, SD = 1.64), as compared to those randomly assigned to the low inequality condition (M = 5.74, SD = 1.06, t(1,80) = 3.417, p = .001).

b. Two-way ANOVA

The first part of the main analysis is composed of a 2(economic inequality: low vs. high) x 2(endorsement type: minority endorsement vs. majority endorsement) between-subjects factorial analysis of variance. This analysis aims to establish the effect of inequality and endorsement type on product attitude. The results of the two-way ANOVA show no significant main effect of inequality on product attitude (F(1,78) = 0.065, p = .799, η2p = .001). A significant main effect

for endorsement type on product attitude with (F(1,78) = 23.118, p = .000, η2p = .229). More

importantly, there was no significant effect found for the interaction between inequality and endorsement type (F(1,78) = .109, p = .743, η2p = .001). The results of the two-way ANOVA

(17)

17

Figure 2. Estimated marginal means of product attitude per condition

c.

Moderated serial mediation analysis

Even though, the interaction effect from the two-way ANOVA was not significant, this study expected to find a significant interaction effect between generalized trust and endorsement type. A moderated serial mediation analysis was conducted using Andrew Hayes’ PROCESS Macro v. 3.1, Model 15. Inequality condition was employed as the independent variable and product attitude as the dependent variable. Furthermore, general trust was included as the mediator and endorsement type as the moderating variable. The results display that inequality is a significant predictor of the mediator – general trust (b = .4907, t(1,80) = 3.3550, p = .0012, 95% CI [.1996; .7818]), such that high inequality leads to lower general trust levels.

However, the results from the regression table of product attitude do not show that inequality is a significant predictor of product attitude (b = -.1284, t(5,76) = .3084, p = .7586, 95% CI [-.9577; .7009]). General trust is also an insignificant predictor of product attitude (b = .4445,

t(5,76) = 1.4663, p = .1467, 95% CI [-.1593; 1.0483]). The interaction term between inequality

condition and endorsement type is also not significant (b = -.2072, t(5,76) = .2488, p = .8042, 95% CI [-1.8657; 1.4514]). Likewise, the interaction term between general trust and

endorsement type is not significant (b = .9084, t(5,76) = 1.4981, p = .1382, 95% CI [-.2992; 2.1160]). The only significant predictor of product attitude is endorsement type (b = 2.0082,

(18)

18 t(5,76) = 5.0943, p = .0000, 95% CI [1.2230; 2.7933]). Additionally, the index of moderated

mediation (.4458) and its confidence interval that included zero [-0.0803; 1.1764], suggested that there is no moderated mediation.

VII. Discussion

The main purpose of this research was to determine that individuals exposed to a condition of high inequality show a more favorable attitude towards majority-endorsed products. The model also featured a mediator of this relationship – general trust. The results from a 2x2 between-subjects design experiment rejected the first hypothesis. Therefore, no difference between the conditions was found. Likewise, the results from the moderated mediation analysis did not support the second hypothesis. In this section, possible reasons for these insignificant results are discussed. Furthermore, the potential biases in this study could also contribute for a more

plausible future research.

Lack of awareness and knowledge regarding the Inequality Phenomenon

One potential explanation for the non-significant results is that participants were able to observe only one of the inequality conditions and probably most of the respondents had limited idea of what inequality actually is and its consequences for the society. Respondents were unable to compare the condition to which they were assigned with anything else. Therefore, participants had limited awareness and knowledge of what inequality in other countries and provinces might be and its features and consequences. This might have affected participants’ understanding of the inequality phenomenon, which then leads them to have a false interpretation of the inequality condition to which they were assigned. A clear description of what economic inequality is in the beginning of the online survey might have contributed for a better understanding of the concept.

Endorsed product being too specific

(19)

19

preferences for movies. Therefore, the endorsement type could have had lower relevance for participants due to their predetermined disapproval to these specific movie genres.

Sample size, language, and distribution method

Firstly, the sample size of the experiment is quite small due to a high number of respondents being excluded from it and short time constraint for conducting the research. The lack of

participants might be one of the reasons for the non-significant results. A 2x2 between-subjects should include at least 200 responses while this research had only 82 after filtering the data. Therefore, increasing the sample size might have resulted in different findings.

Secondly, the survey was conducted in English which is not the native language in Bulgaria and thus could have had a big impact on the participants’ comprehension leading to their

exclusion from the sample. Indeed, a big part of the respondents had inserted wrong percentages in the pie chart comprehension questions or had reported that they did not understand the

information on the comprehension check questions.

Thirdly, snowball sampling has some disadvantages. Since the researchers can only refer to those whom they know, this sampling method can have a possible sampling bias and margin of error. In other words, the researcher might only be able to reach to a small group of people and this could lead to inconclusive results. Another disadvantage of this sampling method is that without compensation for the time spent by participants, they might not be cooperative and decline participation in the research.

Self-reported data

(20)

20

perceive as emotional and personal. For instance, such topic might be the inequality situation in one’s hometown area. These biases could affect the results from the analysis.

VIII. Conclusion

This study examined the impact of economic inequality on product attitude for majority- and minority-endorsed products. The research hypothesized that when consumers experience the high inequality condition they will express a more favorable attitude towards majority-endorsed products compared to minority-endorsed products. In addition, the study hypothesized that this relationship can be mediated by general trust. In order to examine these relationships, an online survey was developed and distributed to individuals residing in Bulgaria. The distribution method used was snowball sampling and there was no monetary compensation for the

participants. There were 200 participants initially, but after filtering the data from responses that did not meet the predetermined criteria there were 112 excluded participants. Subsequently, the valid data (N=82) was analyzed on SPSS program in order to conclude whether the hypotheses are supported or not. The inequality manipulation was found to be successful by initial

manipulation check tests. However, the findings of this paper do not support the two hypotheses. Therefore, there is no difference between the inequality condition and the endorsement type for the product attitudes. In addition, general trust was not found to mediate the relationship. The only significant effect found in the two-way ANOVA was that of endorsement type on product attitude but nothing can be concluded from that. In addition, from the moderated mediation analysis it was found that inequality is a significant predictor of general trust and that the high inequality condition leads to lower levels of general trust, but still nothing can be concluded from this relationship as well. Therefore, both hypotheses must be rejected.

However, this research still contributes to the existing literature. There were no previous studies that examined the effect of inequality on consumer preferences for majority- and

(21)

21

References

Alesina, A., & Perotti, R. (1993). Income Distribution, Political Instability, and Investment.

European Economic Review, 40(6), 1203-1228. doi:10.3386/w4486

Allianz Global Wealth Report 2018

Asch, S. (1956). Studies of independence and conformity: I. A minority of one against a unanimous majority. Psychology Monographs, 70 (9)

Barford, A. (2017). Emotional responses to world inequality. Emotion, Space and Society, 22, 25-35. doi:10.1016/j.emospa.2016.10.006

Bauhoff, S. (2011). Systematic self-report bias in health data: impact on estimating

crosssectional and treatment effects. Health Services and Outcomes Research Methodology,

11(1- 2), 44-53.

Boyd, R., & Richerson, P. (1985). Culture and the evolutionary process. Chicago Ill., etc.:

University of Chicago Press.

Bricker, J., Ramcharan, R., & Krimmel, J. (2014). Signaling Status: The Impact of Relative Income on Household Consumption and Financial Decisions. SSRN Electronic Journal.

doi:10.2139/ssrn.2503557

Christen, M., & Morgan, R. (2005). Keeping up with the joneses: Analyzing the effect of income inequality on consumer borrowing. Quantitative Marketing and Economics, 3(2), 145-173. doi:10.1007/s11129-005-0351-1

Claidie_{̀re, N., & Whiten, A. (2012). Integrating the study of conformity and culture in humans} and nonhuman animals. Psychological Bulletin, 138(1), 126-45. doi:10.1037/a0025868 Coffe_{́, H., & Geys, B. (2006). Community heterogeneity: A burden for the creation of social} capital? Social Science Quarterly, 87(S1), 1053-1072. doi:10.1111/j.1540-6237.2006.00415.x Côte_{́, S., House, J., & Willer, R. (2015). High economic inequality leads higher-income}

individuals to be less generous. Proceedings of the National Academy of Sciences of the United

(22)

22

Durand, M., & Murtin, F. (2015). The Relationship Between Income and Wealth Inequality : Evidence from the New OECD Wealth Distribution Database.

Festinger, L. (1954). A theory of social comparison processes. Human Relations, 7, 117-140. Fischer, J.A.V. & Benno, T. (2006). The effect of relative income position on social capital.

Economics Bulletin, 26(4), 1−20.

Fritsche, I., Moya, M., Bukowski, M., Jugert, P., Lemus, S., Decker, O., Navarro-Carrillo, G. (2017). The great recession and group-based control: Converting personal helplessness into social class in-group trust and collective action. Journal of Social Issues, 73(1), 117-137. doi:10.1111/josi.12207

Hardin, R. (2006). Trust (Key concepts). Cambridge: Polity Press.

Jetten J, Wang Z, Steffens NK, Mols F, Peters K, Verkuyten M. A social identity analysis of responses to economic inequality. Current Opinion in Psychology, 18, 1-5.

Oishi, S., Kesebir, S., & Diener, E. (2011). Income inequality and happiness. Psychological

science, 22(9), 1095-1100.

Payne, B., Brown-Iannuzzi, J., & Hannay, J. (2017). Economic inequality increases risk taking.

Proceedings of the National Academy of Sciences of the United States of America, 114(18),

4643-4648. doi:10.1073/pnas.1616453114

Peshev, P. (2015). Analysis of the Wealth Inequality Dynamics in Bulgaria: Different Approach.

Economic Alternatives Journal, 4, 29-33.

Pickett, K., Mookherjee, J., & Wilkinson, R. (2005). Adolescent birth rates, total homicides, and income inequality in rich countries. American Journal of Public Health, 95(7), 1181-3.

Piketty, T. (2014). Capital in the Twenty-first Century. Cambridge, Massachusetts: Harvard

University Press.

Rothstein, B., & Uslaner, E. (2005). All for all: Equality, corruption, and social trust. World

(23)

23

Saez, E. (2014). Income Concentration and Top Income Tax Rates. Presentation at the Tax Policy Center & USC Conference: Growing Income Inequality: Is Tax Policy the Cause, the Cure or or Irrelevant? USC Gould School of Law, Los Angeles, February 7.

Smeeding, T. (2005). Public policy, economic inequality, and poverty: The united states in comparative perspective*. Social Science Quarterly, 86(S1), 955-983. doi:10.1111/j.0038-4941.2005.00331.x

Wang, J., Zhu, R., & Shiv, B. (2012). The lonely consumer: Loner or conformer? Journal of

Consumer Research, 38(6), 1116-1128. doi:10.1086/661552

Wilkinson, R., & Pickett, K. (2009). The spirit level : Why more equal societies almost always do better

Wisman, J. D. (2009). Household Saving, Class Identity, and Conspicuous Consumption.

Journal of Economic Issues, 43(1), 89-114. doi:10.2753/jei0021-3624430105

Yamagishi, T. (1986). The provision of a sanctioning system as a public good. Journal of

Personality and Social Psychology, 51(1), 110-116. doi:10.1037//0022-3514.51.1.110

Zhao, J. & Tomm, B. (2018). Psychological responses to scarcity. Oxford Research

(24)

24

Appendix

(25)

25

(26)

26

(27)

27

(28)

28

(29)

29

(30)

30

(31)

31

(32)

32

(33)

The Impact of Economic Inequality on Preferences for Minority- vs. Majority-Endorsed Products: The Case for Bulgaria